A power transformer is the most expensive single asset in most substations, and its insulation fails through several mechanisms at once: electrical stress produces partial discharge, thermal stress accelerates ageing, and both leave chemical traces in the oil. The consequence is that no single measurement gives the full picture. The established practice is a diagnostic strategy that combines three families of signals – dissolved gas analysis (DGA), partial discharge monitoring and temperature sensing. This article explains how the three fit together.
The three signal families
DGA – the chemical layer. The insulating oil decomposes under electrical and thermal stress, releasing characteristic gases: hydrogen and acetylene from arcing and high-energy discharges, methane and ethane from thermal decomposition, carbon oxides from paper ageing. Periodic oil samples give a chemical fingerprint of what is happening inside the tank.
Partial discharge – the electrical layer. PD sensors measure the electrical activity of the discharge itself: the UHF signal inside the tank (300-1500 MHz), the HFCT pulse on the bushing and neutral connections, and the acoustic signature on the tank wall. PD monitoring shows the activity in real time, between oil samples.
Temperature – the thermal layer. Winding and oil temperature drives insulation ageing. A transformer running hot ages faster, and a local hotspot can indicate a cooling fault or a winding problem. Temperature monitoring quantifies the stress that the other two layers detect as a consequence.
How the layers interact
The three layers are not redundant; they explain each other:
- A rising DGA hydrogen trend with no PD signal may be a thermal or low-energy phenomenon; a simultaneous PD rise confirms an electrical defect and its location.
- A PD reading that appears only at high load points to a mechanical or thermal mechanism; the temperature channel shows the load dependence.
- A temperature rise with stable PD and DGA points to a cooling problem – a different failure path that the other layers would miss.
Used together, the three layers separate the failure modes that any one of them alone would leave ambiguous.
Practical sensor layout for a transformer
A typical HUWOR transformer monitoring installation:
- UHF sensors on the dielectric ports, or acoustic sensors on the tank wall, for internal PD
- HFCT sensors (HMJ1010-HCT) on the bushing and neutral connections for the conducted pulse
- Temperature sensors on the top oil and the winding indication points
- An IoT gateway (HMJ1020-WJM485) aggregating the signals to the monitoring platform
The platform displays the three layers together, so the operator sees the chemical trend, the electrical activity and the thermal stress on one screen.
From monitoring to decision
The diagnostic strategy supports the asset decisions:
- Stable DGA, no PD, normal temperature – continue the schedule
- Rising PD with hydrogen – take an oil sample, plan an outage, prepare for internal inspection
- High temperature with normal electrical signals – check the cooling system, the fans and the oil pumps
- Several signals moving together – high priority: the transformer is telling a consistent story
This is how a monitoring investment becomes a maintenance decision: each layer narrows the possibilities, and the combination reaches a level of confidence that a single measurement cannot.
Summary
Transformer insulation fails through electrical, thermal and chemical mechanisms, and the diagnostic strategy that works uses all three signal families: DGA for the chemical fingerprint, PD monitoring for real-time electrical activity, and temperature for the thermal stress. HUWOR supplies the PD sensors, HFCT, temperature sensing and gateways for this architecture. Contact the HUWOR sales team for a transformer monitoring proposal.