| Summary: | Background: Parameters of the hydroelectric power plant controllers are typically tuned at the nominal operating conditions such as nominal head and single unit operation. Water level variations in reservoir and/or tailwater, and the presence of other active units sharing the penstock are common disturbances to the nominal assumption. Methods: This article proposes two adaptive add-ons, namely gain scheduling and model reference adaptive control, to the existing speed controllers to improve grid synchronization performance when the site conditions are not nominal. The add-ons were designed and tested on a validated dynamic model of a power plant unit by using a software-in-the-loop simulation setup. An off-season scenario is simulated, in which the original controller of the unit cannot bring the turbine to synchronize with the grid due to low gross head. Then, the add-ons were implemented on-site and experiments were performed under similar conditions. The parameter sets used in gain scheduling for different operation bands are determined off-line with the help of operational experience. The model reference adaptive control add-on requires a reference model and a learning rate. A description of the turbine speed-up profile at nominal operating conditions is sufficient to be used as the reference model. The proposed piecewise linear reference model favors stability over speed in settling to the nominal speed. Results: It is experimentally shown that the proposed add-ons compensate the negative effect of head loss in grid synchronization, and perform similar to the ideal performance at the nominal head. Conclusion: Both add-ons can be implemented on the available off-the-shelf speed governor controllers. They are suitable for use in all hydroelectric power plants, especially in unmanned ones, for automatic synchronization with less waste water.
|
|---|
