Online DGA monitoring is already recognized by tranformer operators – and increasingly by insurers – as essential formaintaining the reliability of high-voltage electrical systems.

Early fault detection and clear assessment of severity are essential for prioritizing maintenance and preventing transformer failures. While dissolved gas analysis (DGA) is the preferred fault detection and diagnostic method, online monitoring solutions are often either too expensive or too limited in diagnostic value, thus limiting the Smart Grid approach, where all critical transformers in a grid can be monitored in real time with central software.

This article introduces a cost-effective online monitoring strategy using hydrogen (H₂), acetylene (C₂H₂), and dissolved moisture. This has been enabled by the adoption of Tunable Diode Laser Spectroscopy for precise acetylene measurement in transformer monitoring. While hydrogen provides broad early detection, acetylene is a marker of high-energy faults, and moisture reflects insulation and dielectric risk. Supported by thermodynamic gas-formation principles and field data, the approach achieves strong diagnostic coverage that is not possible with fault-detection methods based on hydrogen-only, hydrogen & carbon monoxide, or composite gas monitoring.

Introduction

Unexpected transformer failures can lead to blackouts, equipment damage, and costly repairs. While offline laboratory DGA remains the definitive method for assessing the in-oil active part health of in-service transformers and performing fault diagnostics, the online DGA monitoring is already recognized by transformer operators – and increasingly by insurers – as essential for maintaining the reliability of high-voltage electrical systems. This is because, in a digital power system environment, online dissolved gas analysis (DGA) offers advantages that complement traditional laboratory testing, particularly for detecting fast developing faults that occur between periodic offline oil sampling.

In fact, early online systems from the 1990s have demonstrated the value of continuous monitoring of transformers, however, their design generally centered on using hydrogen as a “smoking gun” for detecting anomalies.

To improve diagnostic capability including the ability to detect and measure the trend of acetylene as highest-risk gas, multi-gas online monitors were later on developed, offering detailed insight into fault types and severities. Although considered the gold standard, their cost and complexity make widespread deployment difficult, especially across large fleets or remote installations.

To achieve wider coverage and risk reduction at fleet level, many transformer operators have adopted simpler one-gas, two-gas, or composite-gas monitors. However, this approach presents several challenges:

• False alarms caused by non fault hydrogen generation (stray gassing, oil aging, or sampling errors) [3].
• Ambiguous carbon monoxide (CO) interpretation, as CO may result from both benign oil oxidation and critical cellulose degradation [4].
• Delayed or missed detection of severe electrical faults when arcing gases such as acetylene are not monitored.

In addition, composite-gas monitoring masks individual gas behavior, preventing clear fault diagnosis.

Therefore, a growing need exists among transformer operators for:

• Early warning of developing conditions that could lead to catastrophic failure, so operators can determine whether investigation and maintenance crews should be dispatched in emergency mode or not; and
• Affordability and simplicity, to enable monitoring of a larger number of transformers.

In other words, transformer operators need: Effectiveness, Reliability, and Affordability in a DGA monitoring solution.

DGA – Background and Diagnostic Principles

Transformer oil DGA is based on the principle that electrical and thermal stresses within transformers decompose insulating materials – both oil and solid insulation – generating gases characteristic of the type and severity of the fault.

Origins and Significance of Dissolved Gases

Each gas carries diagnostic value as described in Table 1.

Gas	Gas Source	Diagnostic Value
Hydrogen (H₂)	Low-energy heating, partial discharges, arcing, Stray gas, etc.	Early warning, but not highly specific
Acetylene (C₂H₂)	Arcing, high-energy discharge	A definite indicator of severe electrical faults/high temperature
Methane (CH₄)	Low-energy thermal faults	Context for thermal faults
Ethane (C₂H₆)	Moderate overheating	Context for thermal faults
Ethylene (C₂H₄)	High-temperature thermal faults	Context for severe overheating
Carbon Monoxide (CO)	Paper degradation, oil oxidation	Insulation ageing indicator, but prone to false positives
Carbon Dioxide (CO₂)	Paper degradation, oxidation	Insulation ageing indicator
Oxygen (O₂) / Nitrogen (N₂)	Air ingress	Leak detection and atmospheric contamination

Table 1 – Gas sources and their principal diagnosis value

Thermodynamic Basis of fault Gas Generation

Gas generation in transformers follows the thermodynamic decomposition of insulating materials under different stress levels. Figure 1 explains the fundamental mechanisms through which varying amounts of energy lead to the formation of specific gases.

Figure 1 – Simplistic thermodynamic model for gas formation from an alkane as mineral oil

The Simplistic thermodynamic model for gas formation described in Figure 1 confirms that:

• H₂ and CH₄ are generated at relatively low thermal activation energies.
• C₂H₆ and C₂H₄ require higher energy, typical of moderate overheating.
• C₂H₂ formation requires the highest energy input, correlating directly with arcing and high-energy faults.

While all gases and ratios between some of them provide diagnostic information, H₂ and C₂H₂ are most critical for the early detection of serious faults. From Figure 1 and Table 2, it can be observed that hydrogen and acetylene are the two main gases associated with electrical faults and high-temperature conditions.

Moisture, though not a gas, plays a significant role in assessing insulation health and predicting dielectric failure risk.

Energy Level	Gas Source	Diagnostic Value
Low (Corona, Partial Discharge)	H₂, CH₄	Partial discharge, stray gassing
Medium (Overheating, Hot Spots)	C₂H₆, C₂H₄, CH₄	Thermal faults (T1, T2)
High (Arcing, Severe Overheating)	C₂H₂	Arcing, high-energy discharge and thermal faults (T3)

Table 2 – Gas generation at each energy level and typical fault type

The Case for H₂, C₂H₂ and Moisture Monitoring

Focusing on H₂, C₂H₂, and dissolved moisture addresses operators’ need for a monitoring strategy that enables fleet-level risk reduction and investigation/maintenance prioritization.

This can be achieved by detecting anomalies and clarifying fault severity, as follows:

• Hydrogen (H₂) – a universal early indicator of many faults, including partial discharge, low-energy heating and the more benign stray gassing.
• Acetylene (C₂H₂) – confirms the presence of arcing or high-energy discharges with overheating above ~700°C. C₂H₂ is rarely generated in benign conditions, making it a decisive diagnostic parameter, unlike other gases.
• Moisture – provides insight into insulation aging and the risk of dielectric breakdown or bubble formation under thermal stress.

By combining early fault detection, straightforward diagnostics and affordability, this methodology offers a practical and scalable solution for safeguarding transformer fleets in today’s rapidly evolving power systems.

Combining Theory and Practical Experience

In addition to developing the well established geometric models for DGA diagnosis, Dr. Michel Duval formulated a thermodynamic model characterizing gas evolution across different temperatures and the associated stresses linked to these gases. The model is described in [5] and illustrated in Figure 2.

Figure 2 – Correlation of gas formation versus temperature and actual stress [5]

In a separate investigation, a Korean research group [6] established a correlation between various stress conditions and the probability of failure.

Table 3 integrates the Korean study, the Table C.3 “Occurrence of Types of Faults or Stresses Identified by DGA” [8] and the key-gas method.

Figure 3 – The result of fault cause analysis by fault parts [6]

Gases vs Fault/Stress	H₂ (%)	C₂H₆ (%)	CH₄ (%)	C₂H₄ (%)	C₂H₂ (%)	Failure Probability (%)
PD	95	2	2	1	0	1
S	85	10	5	0	0	0
T1	46.7	23.3	23.3	6.7	0	4
O	40	20	24	16	0	0
C	33.3	16.7	20.8	25	4.2	0
T2	29.2	12.5	16.7	33.3	8.3	6
T3	25	8.3	12.5	41.7	12.5	30
D2	40	4	5	16	32	40
D1	50.7	2.2	3.6	7.2	36.2	13

Table 3 – Gas Signatures and Fliure Risk by Fault Type

The primary conclusion derived from the table is that two gases- acetylene and hydrogen – are consistently associated with all documented failure cases. In most instances where failures occurred, acetylene served as the principal precursor, reliably indicating the presence of high-risk, potentially catastrophic
fault conditions. Hydrogen provided complementary diagnostic value by capturing additional failure modes not exclusively identified by acetylene.

This observation supports the development of the diagnostic strategy outlined in section Diagnostic Flow Logic.

Diagnostic Flow Logic

The proposed online monitoring decision logic integrates H₂, C₂H₂, and moisture measurements to provide clear, actionable guidance. The suggested thresholds and actions are described in Table 4.

Condition	Hydrogen (ppm)	Acetylene (ppm)	Recommendation
No alarm	< 50 AND	< 0.5	Continue monitoring
Non-critical alarm	> 50 AND	< 0.5	Schedule laboratory DGA
Critical alarm – possibly incipient electrical discharge fault	< 50 AND	> 0.5	Schedule laboratory DGA within maximum 24 hours
Critical alarm – steady fault*	> 50 AND	> 15	Schedule urgent inspection and laboratory DGA; prepare for load reduction or outage
Critical alarm – rapid rise	> 10 AND	> 1	Immediate response and investigation; consider emergency outage

Elevated moisture in insulation system represents risk of fault formation (or escalation if fault already exists)

*During the first years of transformer’s life, ig hydrogen >25ppm AND Acetylene >5 ppm then recommendaiton is to schedule urgent inspecion and laboratory DGA; prepare for load reduction or outage

Table 4 – A Decision Framework for Transformer Monitoring Based on Hydrogen and Acetylene Alarm Thresholds

Routine conditions require no action, while elevated hydrogen prompts offline DGA to investigate potential low energy faults. Concurrent increases in hydrogen and acetylene, or rapid rises in both, trigger immediate maintenance actions to prevent severe failures. This approach enables timely intervention while minimizing unnecessary outages.

When combined with hydrogen and acetylene monitoring, moisture measurement provides critical insight into dielectric margin and insulation stress, completing the picture needed to assess both the likelihood and the potential severity of transformer failures.

Economic Analysis

Based on the CIGRE Technical Brochure 783 [7], Table 5 provides an indication regarding the cost-revenue for different types of monitoring approaches.

Table 5 compares the diagnostic coverage and relative cost of various online DGA configurations.

Monitor Type	Estimated cost (relative prices)	Coverage of Severe Faults	Comment
9-gas	$$$$	~95-98%	Best diagnostics, high cost
Composite 4-gas	$$	~80-90%	Good compromise, but poor specificity
H₂ + CO	$-$$	~60-75%*	Commonly used, but poor specificity
H₂	$	~60-75%*,**	Request an offline DGA before a diagnosis
H₂ + C₂H₂ (proposed)	$$	~80-90%	High value and specificitiy; detects arcing with same performance as fault diagnostic monitoring technologies

*Coverage after offline tests; **Some utilities report as low as 50-60% faults coverage

Table 5 – Comparative Diagnostic coverage and Cost of Online DGA Monitoring Configurations

Coverage percentages represent estimated detection capability for severe fault conditions, based on published data and industry experience. Cost levels are relative and reflect typical market pricing for each monitoring type.

Routine conditons require no action, while elevated hydrogen prompts offline DGA to investigate potential low-energy faults

While hydrogen-only and H₂ + CO solutions offer basic fault detection with limited specificity, the addition of acetylene significantly enhances the ability to identify high-risk discharge faults without the complexity and cost of complete multi-gas systems.

Table 5 shows that expanding the number and type of monitored gases improves fault detection capability, with H₂ + C₂H₂ providing a balanced approach between fault detection, diagnostic performance and system affordability.

Use cases of hydrogen, acetylene and moisture monitoring approach

Since online DGA equipment is used both to monitor individual unhealthy transformers and to reduce fleet level risk by deploying on all critical transformers regardless of current health, the hydrogen, acetylene, and moisture monitoring approach can be considered in the following cases:

• Critical, healthy substation distribution transformers – risk mitigation at fleet level
• Healthy generation and transmission transformers – risk mitigation at fleet level
• Healthy industrial and other mission-critical transformers – risk mitigation at fleet level
• Critical, healthy renewables – risk mitigation at fleet level
• Gassing transformers where real time complete DGA diagnostics is not required – monitoring where budget is constrained

Conclusion

As leading utilities begin implementing fleet-level risk-reduction strategies for medium-power transformers, the need for a well-balanced cost–benefit monitoring approach is becoming increasingly clear.

The H₂ + C₂H₂ + moisture monitoring strategy presented in this article provides the fault detection and diagnostic coverage that matters for preventing critical transformer failures, at a cost level that finally enables fleet-wide deployment and true risk reduction. Used in conjunction with the industry’s benchmark for final decision-making – laboratory oil testing – the approach can support what is likely the most effective transformer monitoring strategy available today.

Bibliography
[1] Wagner HH. Pennsylvania TCG Transformer Fault-Gas Continuous Monitor. Doble Conference Index of Minutes. Sec. 6-701; 1967.
[2] Duval M. Dissolved Gas Analysis, It Can Save Your Transformer. IEEE Electric Insulation Magazine 1989; 5:22-27.
[3] Grisaru M. Transformer maintenance: Hydrogen–the most measured and monitored transformer parameter. Transformers Magazine. 2018; 5(4):42-49.
[4] Höhlein-Atanasova I, Frotscher R. Carbon oxides in the interpretation of dissolved gas analysis in transformers and tap changers. IEEE Electr Insul Mag. 2010; 26(6):22-26.
[5] Ongoing Activities at IEEE, IEC and CIGRE on DGA. EPRI-TSUG Conference, St. Louis; 2013.
[6] Jung JR, et al. Advanced Dissolved Gas Analysis (DGA) diagnostic methods with estimation of fault location for power transformer based on field database; 2016
[7] CIGRE Technical Brochure 783. DGA Monitoring Systems; 2019
[8] CIGRE Technical Brochure 771. Advances in DGA interpretation. 2019