Delta-X Research: Over Three Decades of DGA Innovation

Delta-X Research was founded in Victoria, British Columbia in 1992 with a single purpose: to bring more rigorous, quantitative methods to dissolved gas analysis for power transformer condition monitoring. More than three decades later, that purpose remains at the centre of everything the company does, but the analytical tools, the industry context, and the scale of the challenge have all changed substantially.

The Problem That R-DGA Was Designed to Solve

Dissolved gas analysis for transformer condition monitoring dates to the 1960s. By 1992, the field had established frameworks for gas measurement and threshold-based fault detection. IEEE C57.104 [1] and IEC 60599 both provided guidance on gas concentration limits and ratio-based fault classification methods. What they did not provide, and what the industry lacked, was a method for quantifying how serious a transformer's condition was relative to the population of transformers, or for tracking whether that condition was progressing at a rate that indicated genuine risk accumulation.

The limitation of threshold comparison is structural: it assesses whether a value exceeds a boundary, but it does not assess how typical or atypical that value is relative to the population of transformers. A transformer at 200 ppm hydrogen might be a concern in one context and entirely normal for its design, age, and service conditions in another. Threshold methods treat both identically. Similarly, a transformer whose hydrogen is rising at 5 ppm per month from a zero baseline is more concerning than one that has been stable at 200 ppm for ten years, but threshold comparison provides no mechanism to reflect this.

Jim Dukarm's approach, documented in an early 1993 IEEE paper [2] and developed over subsequent years of research and validation, was to ask a different question: rather than comparing a gas concentration to a fixed limit, how unusual is this transformer's gas record relative to the population of transformer histories? This population-normalised framing is the conceptual foundation of R-DGA.

The CSEV and HF Metrics

The formal mathematical specification of R-DGA methodology, including the definitions and validation of CSEV and HF, is documented in Dukarm, Draper, and Arakelian [3], published in IEEE Electrical Insulation Magazine in 2012.

CSEV (Cumulative Severity) integrates the total fault energy implied by a transformer's complete dissolved gas history, normalised against a validated reference population. Each DGA result contributes to the CSEV calculation based on the gas generation rate indicated, weighted by the thermodynamic energy content corresponding to each gas species and fault mechanism. Over successive sampling intervals, CSEV accumulates, reflecting the actual history of fault energy deposition in the transformer's insulation system rather than only the most recent sample.

This accumulation property has important practical consequences. A transformer that has been generating gas slowly and consistently for 20 years may have a high CSEV even if no individual sample ever triggered a threshold alert. Conversely, a transformer with a single anomalous sample caused by a transient event (oil sampling contamination, a through-fault that has since resolved) may show a high single-sample value but low CSEV because the historical record does not corroborate sustained fault activity. Dukarm et al. [3] demonstrated both effects in utility data: threshold methods produced false positives for stable high-concentration profiles and false negatives for progressive low-concentration accumulations.

HF (Hazard Factor) maps the CSEV value onto the empirical relationship between condition severity and observed failure probability in the population database [3]. This link, from a dimensionless condition metric to an operationally meaningful risk indicator, is what makes R-DGA directly useful for fleet prioritisation rather than requiring expert judgment to translate a CSEV number into a maintenance action.

The population database against which these metrics are normalised represents decades of accumulated transformer histories from utility fleets across North America. Its size and diversity is a core part of why the methodology is reliable: a reference population drawn from a narrow geographic or operational context would not generalise; a large, diverse population supports robust statistical normalisation.

From Research to Production: TOA Software

The commercial expression of R-DGA methodology is Transformer Oil Analyst (TOA), software designed specifically for utility engineers and asset managers who need to manage DGA programmes across large transformer fleets. TOA automates the calculation of CSEV and HF for every transformer in the database, generates ranked fleet risk views, and applies fault classification methods, including the Duval Triangle and IEC ratio analysis, alongside R-DGA metrics within a single analytical environment.

The software design reflects the operational reality of how utility DGA programmes work. Laboratories submit results; engineers review exceptions and trends; asset managers use ranked outputs for capital planning and maintenance scheduling. TOA's workflow supports each stage of this process, from result import and data quality validation through fleet ranking outputs that can be communicated to operations and finance leadership.

TOA is currently used by 15 of the top 20 US utilities, a market position built over decades of demonstrated reliability. The user base includes utilities managing fleets of hundreds of transmission transformers, cooperative utilities with smaller distribution fleets, and industrial operators with critical process transformers. The same analytical methodology applies across this range: R-DGA is not calibrated to transformer size or voltage class, because the underlying physics of gas generation in mineral oil insulation is consistent across transformer types.

Monitor Watch and the Extension to Online Monitoring

The development of reliable gas-in-oil sensor technology created an opportunity to extend R-DGA methodology beyond periodic laboratory sampling. Monitor Watch integrates online sensor data from a range of commercially available dissolved gas monitors with the same CSEV and HF analytical framework applied to laboratory samples in TOA. The result is a unified view of transformer health in which laboratory samples and continuous sensor readings are analysed consistently, using the same population-normalised methodology.

CIGRE TB 630 [4] frames the deployment case for online monitoring: units combining high operational consequence with elevated condition risk indicators justify continuous monitoring. Monitor Watch's value in that context is that it does not require engineers to develop and apply different interpretation criteria for online data versus laboratory data. The R-DGA framework does the work, generating updated fleet risk rankings as sensor data arrives and flagging rate-of-change patterns that indicate progressing fault conditions between scheduled sampling intervals.

Scientific Grounding and Standards Contributions

The methodology underlying Delta-X Research's products is grounded in peer-reviewed science. The 2012 paper by Dukarm, Draper, and Arakelian [3] provides the formal specification and validation of CSEV and HF. Jim Dukarm's participation in CIGRE Study Committee A2 working group activities has contributed to the development of CIGRE TB 771 [5], which provides updated DGA interpretation frameworks for alternative insulating fluids and improved fault diagnosis criteria, and to the international technical discussion around DGA methodology more broadly.

CIGRE TB 812 [6], the most comprehensive international survey of power transformer reliability data, provides the statistical context in which R-DGA fleet risk ranking operates. Its documentation of failure rate distributions across age cohorts, failure mode frequencies, and the relationship between condition and reliability underpins the design of any rigorous condition-based maintenance programme and provides the empirical grounding for the population-normalised approach that R-DGA implements.

2022 Frost & Sullivan Recognition

In 2022, Frost & Sullivan recognised Delta-X Research with the North American Enabling Technology Leadership Award for TOA with R-DGA, acknowledging the methodology's impact on industry practice and the quality of the product's implementation. This external recognition affirmed what Delta-X Research's utility customers had been demonstrating through long-term adoption: that population-normalised, history-integrating DGA analysis produces better maintenance decisions than threshold-only methods.

The Path Forward

The challenges facing transformer asset managers in 2026, including fleet ageing accelerating across North America [6], constrained manufacturing capacity for replacement units, energy transition demands increasing loading stress on existing equipment, and growing industry interest in alternative insulating fluids requiring different analytical frameworks [5], represent exactly the environment in which rigorous condition monitoring adds the most value. The combination of extended service life requirements and elevated replacement risk makes the quality of the analytical evidence base more consequential, not less.

Delta-X Research's commitment for the next decade is the same as it was for the last three: to develop and deliver the most reliable, scientifically grounded analytical tools for transformer DGA interpretation: tools that allow utility engineers to make better decisions with the data they already have.

For the technical foundation of R-DGA, visit the Science page. For product information on TOA and Monitor Watch, visit their respective product pages. To discuss what better DGA analysis could mean for your fleet in 2026, contact us.