Power System Inertia Estimation and Frequency Response Assessment

Power plant emissions constitute a major source of environmental pollution. This renders the gradual replacement of such power plants by renewable energy sources imperative. Changes in the quota between conventional generation and renewable energy sources introduce challenges that the modern power systems have to encounter. For example, conventional power plants are replaced by wind turbines and photovoltaics, which do not contribute to system's inertia. As a result, power system inertia decreases and frequency stability becomes a concern. Frequency stability is affected by the amount of power system inertia, along with the response of controllable frequency reserves and the amount of power imbalance. Therefore, the estimation of power system inertia, as well as the frequency response assessment is necessitated, so that appropriate actions can be taken to ensure frequency stability. The first part of this thesis focuses on power system inertia estimation. Four disturbance based inertia estimation methods are proposed. The methods accommodate the frequency and/or voltage variations that arise after a disturbance and estimate both the total inertia constant and the total power imbalance of the system. This is achieved by considering suitable functions that can approximate the voltage and frequency dependency of the loads, as well as the response from the governors. The proposed methods are applied on frequency responses from simulations of a test system under several different scenarios. The performance of the methods under lack of certain data is investigated, in order to examine if they can be employed under realistic conditions. An extensive analysis is performed, which enables the selection of the most appropriate method, depending on the information that is available. The second part of the thesis deals with frequency response assessment. First, the use of simplified dynamic equivalent models is examined. The parameters of either governor or frequency response models of the system are identified and validated by employing historic events. Data are obtained from events from the power systems of Sweden and Texas. After the identification, the frequency response model of the system can be simulated to assess frequency stability. Secondly, the thesis examines methods that deal solely with the prediction of frequency nadir. The examined methods either use neural networks or linear regression. The accuracy of the methods, as well as the uncertainty that is introduced by system non-linearities, are assessed through simulations. By proposing methods for estimating power system inertia and frequency response, this thesis attempts to provide additional solutions to the challenges that modern power systems have to face. It offers supplementary tools to increase the system awareness, in order to take appropriate actions in case of frequency events.QC 20191114