Grey-box Modeling of Hydropower Plants for Improved Frequency Regulation : Evaluation of Double-Regulated Hydropower Turbines for Fulfillment of the New FCR-requirements

Abstract: Over the last decades, the frequency on the Nordic electrical power grid has deteriorated. Therefore, new stricter requirements are developed for the hydropower delivering regulating active power on the Frequency Containment Reserve market (FCR). This thesis aims to investigate the possibility of modeling two double-regulated hydropower plants, referred to as Unit 1 and 2, to evaluate their compliance with the new FCR-requirements.  By modeling the hydropower plants, the first goal was to find a model structure that captures the essential dynamics of the systems. A second goal was to evaluate whether the two units currently fulfill the new FCR-requirements, and investigate how the turbine governors’ settings could be optimized to fulfill the new requirements. Data obtained from FCR-tests was used in MATLAB to evaluate the two stations’ dynamic stability and performance requirements. Through system identification in MATLAB, grey-box modeling was used to create linear and non-linear turbine and waterways models for Unit 1 and Unit 2. The non-linear turbine and waterways models were implemented in Simulink, together with corresponding turbine governors, to find optimal parameter settings to fulfill the FCR-requirements.  The evaluation of the new FCR-requirements shows that none of the two units fulfills the dynamic stability requirement. However, Unit 1 fulfills the performance requirement. The results imply that double-regulated turbines will most likely have difficulties fulfilling the new requirements, which will cause major consequences in improving the frequency regulation quality. The results from the grey-box modeling present that the linear models are not validated with the step response data, due to not capturing the system dynamics when compared with provided data from the units. On the other hand, the non-linear models are validated with step response data as the model captures the system dynamics more accurately. However, the non- linear Simulink models cannot capture the dynamics of the hydropower systems for sinusoidal signals with varying frequencies which are used in the new FCR-requirement test. Consequently, the thesis has no result of the optimal parameter setting of the turbine governors to fulfill the new FCR-requirements. In conclusion, the grey-box models, with the level of detail presented in this thesis, are inadequate in capturing the system’s dynamics to evaluate the new FCR-requirements. Thus, the thesis contributes to filling a knowledge gap within the area of modeling for frequency regulation.