Using Online DGA to Safeguard a Transformer's End-of-Life Operation

The first gassing patterns appeared in 2010. In 2013, an unidentified event drove ethylene from 35 ppm to 100 ppm within weeks; 2017 brought another significant event. In 2018 the transformer was relocated and its oil degassed — and in May 2019 a problematic DGA sample appeared again. A visual scope inspection in October 2019 found carbon-like debris on the preventive autotransformer's surfaces, and that same month the gassing began escalating continuously.

The underlying problem was temporal resolution: with laboratory sampling every two weeks, the utility could not tell when each gassing event occurred, nor distinguish whether increases were gradual or driven by discrete incidents — exactly the information that defines a fault's severity.

In October 2019 a Vaisala Optimus™ OPT100 online DGA monitor was installed, replacing the biweekly sample with hourly data and operating alongside frequent laboratory sampling for validation. The gas signature was consistent: dominating ethylene (58.7 to 77.4%), methane (16.1 to 32.6%) and ethane (6.5 to 10.0%), with carbon monoxide and dioxide stable. Duval triangle and pentagon analysis indicated a T3 thermal fault — above 700 °C, in oil only, with no paper involvement.

The decisive information was not the level but the pattern: continuous monitoring showed the gassing was not growing incrementally but in clear steps — separate incidents, not continuous degradation — and with no correlation to load, which stayed at just 40% of nameplate rating. As a utility engineer summed it up: replacement was already in their minds, but they were able to demonstrate that the issue was evolving — not steady state.

Between October and December 2019 the T3 fault indicators remained stable. In January 2020 the monitor revealed accelerated, continuous gas formation: the fault had stopped evolving in incidents and was progressing steadily. With that picture — a very high risk of imminent failure — the transformer was permanently removed from service in early January, as a planned outage rather than an emergency.

The inspection after opening the tank confirmed the diagnosis: no obvious overheating in the main parts and no severe paper degradation, but severe carbonization on the lower and middle wood spacers adjacent to the preventive autotransformer core, in multiple zones — consistent with separate incidents and with fault spot temperatures above 700 °C. The case's conclusion: laboratory DGA can identify that a fault exists, but only continuous monitoring reveals the actual gassing pattern — and that pattern is what makes it possible to run an end-of-life asset safely, right up to the correct moment to retire it.