The term "monitoring" is comprehended as the measurement of basic parameters related to the transformer's operation. In monitoring, the integration and data acquisition from sensors are performed to detect evidence of faults and, if necessary, emit alerts that any of the transformer's parameters magnitude is at an abnormal level.
The system is only active if the captured parameters trigger some action such as activating a relay, alarm, or reporting to a supervisory system about this data. On the other hand, the term "diagnosis" indicates the incorporation of advanced analysis technologies capable of reliably assessing the equipment's condition and suggesting actions to be taken.
The basic task of a diagnostic system is to act as a data capture supervisory system and transform this data into useful actions. In diagnosis, signal processing techniques play an unique role. In power transformers, diagnostics are applied with the following objectives:
Identify increased degradation;
The reasons for degradation;
If possible, take necessary corrective actions;
Indicate trend curves for future failures.
The preferred features of monitoring and diagnostic systems are detailed below. It is evident that untreated or minimally processed data result in alerts for asset monitoring, while there is a greater demand for analysis and correlation to diagnose an asset.
In order to better understand the failure structure of a power transformer, one can analyze the following figure, which presents the fault tree of a power transformer. This figure highlights that components classified as active: windings and core, can undergo accelerated degradation if any auxiliary system such as the tank and cooling system fails, making the fault origin analysis more complex and difficult.
Therefore, this is the first element of understanding to structure an adequate monitoring system and subsequent diagnosis. This shows that it is not enough to monitor the formation of partial discharges in the windings or the temperature directly through optical fiber sensors if there is no measurement of oil level, presence of moisture, or a simple temperature measurement of the metallic parts, as these accessories could trigger a critical failure. This type of simpler measurement and monitoring generally has a lower cost than the previously mentioned one.
The following figure makes evident the components that can be monitored and the strict correlation between them, thus allocating sensors correctly tends to diagnose a failure early and when possible detectable by more than one sensor element.
Was this article helpful to you?
Don't forget to share and subscribe to our newsletter to receive more content from the electrical sector.