This project thesis has investigated the potential for condition monitoring of a generator cooling system in a hydropower plant, using physical models on data from a SCADA system. The work has been supervised by Voith Hydro AS.
SCADA data from Brattset hydropower plant, made available by Voith Hydro, was used throughout the work with the project. The scope of the project focused in the generator cooling system, with a particular focus on the generator air coolers. Several approaches to condition monitoring have been investigated during the project. The data that was used were mostly analogue air temperature measurements, but measurement of power output were also used. The temperatures for the cold air of the coolers and the measurements of the warm air from the generator, were used to determine the condition of the coolers. The difference in the air temperature over the air coolers were believed to be influenced by the condition of the coolers. The coolers of both generators were investigated individually and collectively.
Several analyses were conducted, using Python and associated libraries, to develop three condition monitoring concepts. One concept monitors the cold air of the air coolers individually, and may function as an early warning system following a leakage of the water pipes in the cooler. The second concept may monitor the cooling effect of each cooler individually or all the coolers belonging to one generator at the same time by monitoring the difference in air temperature over time. The last concept predicts the temperature given a specific power output of the generator.
The developed concepts show that SCADA data has the necessary sampling frequency and accuracy to be used for condition monitoring. The concepts also show that the air coolers in a hydropower plant may be suited for condition monitoring provided that you have enough data.