Researchers from Politecnica Salesiana University in Cuenca, Ecuador have proposed a new diagnostic system for hydroelectric power plants (HPPs). The newly developed tool makes it possible to speed up fault detection for most equipment components by up to 90%, which significantly reduces the risks of long-term downtimes. This is especially important for Ecuador, since about 78% of the country’s electricity is produced by HPPs.
Key to the operation of HPPs is the automatic turbine speed control system (Governor). It is responsible for maintaining a stable current frequency, synchronizing the generator with the power grid, controlling equipment start-up and shutdown, and ensuring stable operation when load changes or emergencies occur. System failures can lead to breakdowns in power generation and significant time losses.
As part of their study, the scientists analyzed the operation of the Governor system at the Alazán HPP, which forms part of the Mazar Dudas project. Using technical documentation and expert evaluations, they identified 15 main types of failures and measured the average time spent on their elimination through a conventional approach based on visual inspection and manual verification of signals. On average, it took from 60 to 180 minutes to detect and eliminate each malfunction. Some failures, such as speed sensor failures or malfunctions of deflector valves, were impossible to diagnose without complete equipment shutdowns.
To solve this problem, the researchers used a Petri net model, a tool that makes it possible to describe the system as a set of logical states, events and transitions between them. This model clearly displays the structure and logic of the interaction of components, from initialization to activation of the mechanisms. This allowed the researchers to simulate the entire process of Governor operation, taking into account its behavior during various failures, and accurately determine where the failure occurs.
The results were impressive. The speed of diagnostics for electronic and control components increased by an average of 75–92%. For instance, troubleshooting an injector, a position sensor or a solenoid now takes only 10 to 15 minutes instead of the usual one and a half to two hours thanks to the new model. For mechanical components, improvements can reach 50% due to the need for physical access and inspection. Overall, the use of Petri nets has reduced diagnostic time for nearly every item. Moreover, the model has made it possible to correctly handle situations with multiple or parallel failures, which is especially important in real-life production conditions.
The researchers are currently proposing the implementation of this model in the HPPs’ automated control and monitoring system (SCADA) in order to track the state of Governor components in real time, automatically detecting deviations and notifying personnel thereof. This will pave the way for predictive maintenance, reduce equipment downtime and improve the overall reliability of HPP operation.
