DISTRIBUTECH 2024: Transformer Monitoring in the Context of Grid Modernisation

DISTRIBUTECH is one of the largest annual conferences in the electric utility sector, drawing thousands of engineers, executives, technology vendors, and policymakers to engage with the operational and technical dimensions of grid modernisation. The 2024 edition was shaped by a common thread: the grid is changing faster than its physical infrastructure is being renewed, and the assets most under pressure are the ones that have been in service the longest.

Transformers sit at the centre of that pressure. Delta-X Research attended DISTRIBUTECH 2024 and participated in conversations about what transformer condition monitoring needs to look like in a grid that is being asked to do fundamentally new things.

The Grid Modernisation Pressure on Transformer Fleets

The energy transition creates specific physical stresses for transformers that conventional asset management frameworks were not designed to handle. Understanding these stresses is necessary for calibrating how aggressive a DGA monitoring programme needs to be.

Distributed energy resources and reverse power flow. Large-scale integration of rooftop solar, utility-scale wind, and storage systems creates bidirectional power flows through transformers that were designed for unidirectional operation. Distribution transformers, in particular, are seeing load profiles that bear little resemblance to the demand curves their thermal models were based on. Thermal cycling accelerates cellulose insulation degradation [1], and its chemical signature, CO and CO₂ accumulation in the oil, is detectable by DGA. A programme that is not tracking CO and CO₂ trends is missing a significant signal.

EV charging and electrification load peaks. Electric vehicle charging at scale creates sharp, synchronised load peaks on distribution circuits. Even if average loading remains within nameplate rating, peak loading events drive disproportionate hotspot temperatures due to the non-linear relationship between temperature and thermal ageing described in transformer loading guides [1]. Transformers exposed to repeated thermal stress events may show accelerating fault gas generation that a threshold-based DGA programme, which evaluates concentration against a fixed limit at each sample without integrating the trajectory, can fail to detect until deterioration is advanced.

Ageing infrastructure and accumulated risk. A large fraction of the North American transmission and distribution transformer population was installed in the 1960s through 1980s. Many units have now exceeded their original design life of 30–40 years. For ageing transformers, the relationship between gas concentration and remaining life is non-linear and population-dependent. Research by Lapworth and McGrail [2] established that DGA condition assessment requires population-context calibration. The same gas profile that is acceptable in a ten-year-old unit may be a precursor to failure in a unit with 50 years of service and accumulated insulation degradation.

Asset Intelligence: What That Actually Means for DGA

"Asset intelligence" was a recurring framing at DISTRIBUTECH, the idea that utilities should derive more decision-relevant insight from the data their equipment already generates. The concept is sound, but the value of asset intelligence depends entirely on what analytical method processes the raw data.

Dissolved gas analysis has been standard practice for transformer condition assessment for more than four decades [3]. The analytical methods applied to DGA data, however, have not evolved uniformly. Many utilities continue to rely on the concentration thresholds specified in IEEE C57.104-2019 [3] as the primary interpretation framework, a method that evaluates each sample in isolation against limits that do not account for individual transformer history, fleet population, or the statistical distribution of gas data in real populations.

Reliability-based DGA addresses this gap by grounding interpretation in statistical analysis of a large, validated transformer population [4]. The method computes two metrics from each transformer's full DGA history:

CSEV (Cumulative Severity) is the accumulated fault energy indicated by the transformer's dissolved gas record, normalised against the reference population. A CSEV value expresses how unusual this transformer's cumulative gas profile is relative to the entire population database. Critically, CSEV integrates the full sample history rather than treating each reading independently; this makes it sensitive to gradual deterioration that never crosses any single-point threshold but represents a consistently abnormal pattern [4].

HF (Hazard Factor) is a reliability-engineering metric derived from the empirical relationship between CSEV level and observed transformer failure probability in the population data [4]. HF provides the link between a condition measurement and the question asset managers actually need to answer: what is the failure risk, and how urgent is intervention?

Together, CSEV and HF produce a continuously updated fleet priority list. As loading patterns change and sampling continues, each transformer's rank updates automatically. This is the analytical layer that turns DGA data into genuine asset intelligence: not just flagged concentration exceedances, but a ranked, defensible view of relative fleet risk.

Online Monitoring and Data Integration

A significant portion of DISTRIBUTECH discussion focused on online monitoring deployment: permanently installed sensors that provide continuous or near-continuous DGA readings from critical assets. For transmission transformers on highly loaded circuits, online monitoring enables earlier detection of rapid fault development. For distribution transformers exposed to new EV charging load profiles, it provides a continuous thermal and chemical record that periodic laboratory sampling cannot match.

Monitor Watch extends TOA's R-DGA analytical capabilities to online monitor data streams. The challenge of online monitoring is not data acquisition, since modern sensors are reliable, but interpretation. Online sensors measure dissolved gas concentrations continuously, and the resulting data carries noise from temperature cycling, dissolved air variation, and sensor drift that is not present in carefully controlled laboratory extractions. Applied directly to R-DGA calculations, raw online monitor data produces spurious CSEV and HF excursions that undermine the programme [4].

Monitor Watch applies signal processing calibrated for common online DGA sensor architectures before computing R-DGA metrics. The cleaned data stream is then processed using the same method as laboratory samples, producing a single consistent risk picture for each transformer regardless of whether its DGA record is built from lab samples, online monitor readings, or both.

For utilities managing mixed fleets, with a subset of critical units under online monitoring and the rest sampled periodically, the ability to apply a single analytical framework across all data types is critical for maintaining a coherent fleet view.

The Maintenance Strategy Implication

CIGRE Technical Brochure 445 [5], produced by Working Group A2.34, frames transformer maintenance strategy as a function of asset criticality, condition, and remaining life. In the grid modernisation context, criticality is changing: transformers that were once lightly loaded distribution units now sit on heavily used circuits serving EV charging infrastructure or connecting distributed solar. Maintenance intervals and monitoring intensity calibrated for the previous load regime may be inappropriate for current operating conditions.

R-DGA methodology adapts to this reality because it is population-based and history-integrated rather than static. A transformer whose CSEV trajectory accelerates following a change in loading pattern will be identified by the methodology before its gas concentrations reach published thresholds, providing the advance warning needed to manage risk before failure becomes imminent.

Further Reading

For the technical basis of R-DGA methodology and its relationship to the broader DGA standards framework, visit the Science page. For product details on how TOA implements R-DGA for fleet management, see the TOA product page. For Monitor Watch capability specifics, visit the Monitor Watch page. To discuss how R-DGA applies to your organisation's specific fleet management challenges, contact us.