What does online condition monitoring add to transformer maintenance?
An online monitor measures continuously what a laboratory measures a few times a year, so it shows how fast a gas, the moisture or a temperature is changing between samples. It does not replace the laboratory: a monitor reading that would change a maintenance decision is normally confirmed with a laboratory sample first. Condition-based maintenance works when each trigger names the measurement, the rate, the confirmation step and the action it starts.
Most transformers are sampled once or twice a year. Between two laboratory reports a developing fault can grow for months without anyone seeing it. Online monitoring closes part of that gap. It also produces a stream of numbers that someone has to read, and alarms that someone has to answer. This guide covers what the instruments measure, what continuous data changes in the diagnosis, why the laboratory keeps the last word, and how to turn readings into maintenance triggers.
What an online monitor measures
IEEE C57.143-2012 is the guide for applying monitoring equipment to liquid-immersed transformers and their components. The quantities fall into a few families, each with its own sensor and its own limits:
- Dissolved gases. A single-gas monitor usually reports hydrogen, sometimes combined with carbon monoxide: it is an early-warning channel. A multi-gas monitor reports the individual hydrocarbons, carbon oxides and often oxygen and nitrogen, which is what fault-type methods such as the IEC 60599:2022 ratios and the Duval triangles need.
- Moisture in oil, usually as relative saturation together with oil temperature.
- Temperatures: top oil, and a winding hot-spot that is either calculated with a thermal model such as IEC 60076-7:2018 or measured with fibre-optic probes.
- Components: bushing capacitance and dissipation factor through the test tap, tap-changer parameters, partial discharge with electrical or acoustic sensors.
A hydrogen-only monitor tells you that something is generating gas. It cannot tell you whether the source is partial discharge, overheating or stray gassing of the oil. That question needs the full gas profile.
What continuous data changes: the rate
IEEE C57.104-2019 made the rate of change a diagnostic in its own right. Table 3 compares consecutive samples; Table 4 evaluates the rate over several samples taken between 4 and 24 months. A gas below its 90th percentile can still be rising fast enough to matter. With yearly sampling, the rate is computed on two or three points. With a monitor it is computed on hundreds, so a change in slope becomes visible weeks or months earlier.
Continuous readings need cleaning before they are read as a rate. Gas solubility changes with oil temperature, load cycles move readings within a day, and every instrument drifts. Compare readings taken under similar conditions, use averages over a defined window, and record any calibration or maintenance of the monitor next to the data. Separating recent change from old gas follows the same logic for laboratory samples.
Why the laboratory still decides
A monitor and a laboratory do not measure the same way. The extraction method, the calibration and the uncertainty differ, and CIGRE TB 783 (2019), the brochure on DGA monitoring systems, treats the comparison with laboratory results as part of verifying a monitor. A difference between the two is expected; the question is whether it is larger than the expected band and whether it changes the diagnosis. When the monitor and the laboratory disagree walks through that check.
The laboratory also measures what the monitor does not: the complete gas set at reference conditions, oil quality under IEC 60422:2024, and furans for the paper. Before an outage is planned on the strength of a monitor alarm, a laboratory sample confirms the gas profile and gives the interpretation a defensible basis.
From monitoring data to a maintenance trigger
Condition-based maintenance means acting when evidence crosses a defined condition, instead of on a calendar. CIGRE TB 630 (2015), the guide on transformer intelligent condition monitoring, frames the work as moving from data to information to a decision. A trigger is useful only when each part is written down:
| Signal | Confirmation | Typical next step |
|---|---|---|
| Hydrogen rising faster than the rate set for that unit | Laboratory sample with full DGA | Shorter sampling interval, fault-type interpretation under IEC 60599:2022 |
| Any increase in acetylene | Laboratory confirmation of the value and of the sampling | Engineering review; IEEE C57.104-2019 gives acetylene particular weight even at low concentration |
| High relative saturation of moisture at low oil temperature | Oil test and an estimate of paper moisture | Loading review and a drying decision |
| Hot-spot above the limit planned for that load | Check of load, cooling stages and ambient temperature | Load management and cooling inspection |
Every trigger also needs an owner and a response time. An alarm without a defined action becomes noise, and teams learn to ignore it. Reducing monitoring false alarms without losing evidence and owning the trigger in condition-based maintenance cover that side.
Reading monitors and laboratory reports together
The value appears when monitor data and laboratory results are read on the same asset, with the same identifiers, units and dates. That is a data problem before it is a diagnostic one: see multi-source data for transformer monitoring and importing laboratory files without losing meaning. To compare one monitor reading with a laboratory result gas by gas, the free Online DGA Monitor vs Lab Check applies the CIGRE TB 783 approach in your browser.
Sources
- IEEE C57.143-2012, Guide for Application for Monitoring Equipment to Liquid-Immersed Transformers and Components.
- IEEE C57.104-2019, Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers, Tables 3 and 4.
- IEC 60599:2022, Mineral oil-filled electrical equipment in service: guidance on the interpretation of dissolved and free gases analysis.
- IEC 60076-7:2018, Loading guide for mineral-oil-immersed power transformers.
- IEC 60422:2024, Mineral insulating oils in electrical equipment: supervision and maintenance guidance.
- CIGRE TB 630 (2015), Guide on Transformer Intelligent Condition Monitoring (TICM) Systems.
- CIGRE TB 783 (2019), DGA Monitoring Systems.
