Understanding emergency levels in transformer faults: A guide for maintenance teams

Transformer fault detection has evolved significantly over the past century, from the rudimentary Buchholz relay of the 1920s to today's sophisticated online dissolved gas analysis (DGA) systems. While these technological advances have greatly improved our ability to monitor transformer health, they've also introduced a challenge: determining the true emergency level when an alarm is triggered.

For maintenance teams, this challenge is far from academic. Misclassifying transformer faults has serious implications for both operational costs and safety. When a minor issue is mistakenly classified as critical, resources are wasted on unnecessary emergency responses. Conversely, when a severe fault is underestimated, catastrophic failure can lead to complete transformer loss, collateral damage and safety risks to personnel.

The financial impact of these misclassifications is substantial. Unnecessary emergency responses can cost utilities tens of thousands of dollars per incident, while a catastrophic transformer failure can cost millions.

The maintenance classification challenge

Maintenance teams responsible for transformer fleets face several common scenarios that highlight the difficulty of accurately classifying faults. A typical situation involves receiving an alarm from a hydrogen monitoring system with no clear indication of the fault severity. Without additional data, maintenance teams often default to conservative responses, treating many alarms as potentially serious.

Traditional fault detectors contribute to this challenge because of their inherent limitations. While cost-effective, hydrogen-only monitors detect a wide range of faults without distinguishing their nature or severity. Composite gas detectors are similarly limited in their ability to differentiate between fault types. Even some monitors capable of detecting acetylene - the critical indicator of high-energy arcing faults - have poor sensitivity (>3 ppm), which is above the 2 ppm threshold recommended by CIGRE Technical Bulletin 783.

These limitations lead to "false alarm fatigue." As noted in CIGRE TB 783, "Such false alarms due to stray gassing will be much more frequent than real alarms of arcing. So after repeated such false alarms, the operator may come to disregard them." This normalisation of deviance creates a dangerous situation where maintenance teams become desensitised to alarms.

High-energy arcing faults that generate acetylene often develop rapidly and can cause catastrophic transformer failure if not addressed promptly. As CIGRE TB 783 warns, "Arcing may be picked up only in its late values, when larger spikes of acetylene and hydrogen are formed over short periods of time, sometimes too late to avoid a catastrophic failure if arcing is in windings..." A real-world example of this occurred when an InsuLogix G2 at a critical data centre recorded acetylene increasing from 0 ppm to approximately 30 ppm. Unfortunately, the InsuLogix wasn’t connected to SCADA, and the transformer failed after approximately six months - a situation that might have been prevented.

The impact of misclassification on resources

Misclassifying transformer faults creates a cascade of resource inefficiencies. When non-critical issues trigger emergency responses, maintenance teams face unnecessary dispatches that pull resources away from scheduled activities. These unplanned interventions typically come at premium costs, requiring overtime labour, emergency equipment mobilisation and possible production interruptions.

Delayed response to truly critical issues creates an even greater financial risk. A major transformer failure can cost millions in replacement equipment, emergency repairs and energy redispatch costs. For industrial operations, the production losses from an unplanned outage often dwarf the equipment replacement costs.

Beyond financial considerations, misclassification also introduces significant personnel safety risks. High-energy arcing faults can lead to catastrophic transformer failures, potentially resulting in fires, explosions or oil spills. Maintenance crews may inadvertently place themselves in danger if the fault has been underestimated.

Oil sampling expenditures represent another hidden cost of misclassification. Without clear fault indicators, maintenance teams often resort to increased frequency of manual oil sampling and laboratory analysis. Each sample costs between $300-$500 when accounting for collection, shipping, analysis and reporting. A utility with hundreds of transformers might spend tens of thousands annually on oil sampling that could be avoided with more precise online monitoring.

A revealing case study comes from a utility that recently upgraded its online DGA monitoring program by replacing several dozen composite gas monitors with InsuLogix G2 systems. Previously, the composite gas monitors would detect general gassing trends but couldn't distinguish between fault types, leading to a standard response protocol for all alarms. After implementing monitors capable of specifically detecting acetylene with high precision, they reported a significant improvement in their maintenance efficiency. Acetylene-specific alarms enabled clear emergency level classification, allowing for calibrated responses based on actual risk.

Modern approaches to emergency classification

The evolution of transformer fault classification has centred on developing more sophisticated gas monitoring approaches, with acetylene detection emerging as the critical differentiator. Acetylene is uniquely valuable because it forms only at temperatures exceeding 500°C, making it a definitive indicator of high-energy faults, including arcing.

Modern monitoring systems with laboratory-grade precision for acetylene detection (0.5 ppm sensitivity) have transformed emergency classification. This level of sensitivity allows detection of high-energy faults at their earliest stages, often weeks before they would trigger conventional monitors.

The relationship between hydrogen and acetylene concentrations provides valuable diagnostic information. When hydrogen levels rise without accompanying acetylene, this typically indicates lower-energy faults such as partial discharge or localised overheating.. However, when acetylene appears alongside hydrogen, especially if acetylene levels are increasing rapidly, this signals a high-energy fault requiring urgent attention.

CIGRE Technical Bulletin 783 specifically notes that "in case of an arcing fault D1 or D2 in windings, which is potentially the most dangerous fault in transformers, the IEC/CIGRE typical value of acetylene that should be detected is ~2 ppm." This establishes a clear threshold for early detection of critical arcing faults, emphasising the importance of monitoring systems capable of detecting acetylene below this level.

Implementing effective response procedures

Effective emergency classification is only valuable when paired with appropriate response procedures. The key to successful implementation lies in integrating advanced monitoring capabilities with existing maintenance programs. This integration begins by mapping the emergency classification levels to specific maintenance workflows already established within the organisation.

Creating tiered response protocols based on gas levels allows maintenance teams to allocate resources proportionate to the actual risk. A practical approach might include:

• For detected hydrogen without acetylene: Increase monitoring frequency and schedule the next routine oil sampling
• For low acetylene levels (0.5-2 ppm): Perform additional diagnostic tests during the next scheduled outage window
• For moderate acetylene levels (2-5 ppm): Plan an inspection within 1-4 weeks based on the rate of increase
• For high acetylene levels (>5 ppm) or rapid increase: Implement emergency procedures for potential transformer failure

A persuasive case study comes from a large industrial oil refinery where routine laboratory testing detected 1.5 ppm of acetylene in a critical transformer. Rather than immediately taking the transformer offline, they deployed an InsuLogix G2 high-sensitivity acetylene monitoring system to continuously track acetylene levels between laboratory tests.

The refinery implemented a response protocol specifically calibrated to acetylene levels, with maintenance actions triggered at predefined thresholds. When acetylene fluctuations occurred, laboratory analysis confirmed them, validating the monitoring system's accuracy. What is particularly notable is that in several cases, hydrogen concentrations remained below 40 ppm - a level that would not have triggered concern in a hydrogen-only monitoring setup, despite the presence of acetylene indicating a developing high-energy fault.

Conclusion

Clearly classifying transformer fault emergencies delivers three interconnected benefits: cost reduction through fewer unnecessary emergency responses and the prevention of catastrophic failures; safety improvements when maintenance teams have accurate information about fault severity; and resource optimisation when teams can confidently prioritise their workload based on actual transformer conditions.

For organisations looking to improve their transformer fault classification, several practical next steps deserve consideration. Begin by assessing your current monitoring capabilities, particularly regarding acetylene detection sensitivity. Next, develop a clear emergency classification framework with defined thresholds and corresponding response protocols. Finally, integrate these classifications into existing maintenance management systems to ensure consistent application across the organisation.

By implementing precise monitoring coupled with clear emergency classification, maintenance teams can convert transformer care from a reactive exercise fraught with uncertainty into a confident, proactive discipline that optimises both reliability and resources.