Impact of Distributed Battery Energy Storage on Electric Power Transmission and Distribution Systems

Abstract

The penetration of Renewable Energy Sources (RES) in electricity grids has increased worldwide over the past decade because of their decreasing costs, especially of Photovoltaic (PV) and wind generation resources with government support for their deployment to counteract global warming effects. Indeed, nowadays, not only utility-scale, but small-scale RES connected at the distribution level are being installed by residential and industrial customers to improve their energy supply and costs. In this context, Energy Storage Systems (ESSs) can be used to facilitate the integration of RES into the grid; Battery Energy Storage Systems (BESSs) being a relatively matured and suitable storage technology for such applications. In particular, distribution systems in some jurisdictions are experiencing an increasing number of new installations of Distributed Energy Resources (DERs), including PV generation accompanied by BESSs, thus, transforming the traditionally passive utility grids into Active Distribution Networks (ADNs), whose operation has the potential to influence the transmission system upstream. Some issues associated with large quantities of DERs connected to ADNs are reduction of transmission level flexibility to accommodate changes at the distribution system, larger frequency deviations due to reduction of system inertia, and various other grid stability issues associated with DER converter interfaces. BESSs can help address some of these problems by providing grid services such as voltage control, oscillation damping, frequency regulation, and active and reactive power control. As a result, appropriate assessment of the integration of distributed DERs on transmission grids, particularly BESSs, is necessary. In this thesis, the impacts of grid-scale and distributed BESSs connected at the distribution system level, on the transmission grid are studied, for which suitable models for steady-state and dynamic analyses are proposed. Thus, first, a dynamic average BESS model is proposed, which comprises a simplified representation of the battery cells to allow simulating the effects of battery degradation, a bidirectional buck-boost converter (dc-to-dc), a Voltage Source Converter (VSC), an ac filter, and associated controls. The decoupled dq-current control of the VSC enables independent control of the BESSs’ active and reactive power injections, thus allowing their operation in several modes studied and improved in this work, namely, constant active and reactive power, constant power factor, voltage regulation, frequency regulation, oscillation damping, and a combination of the last two. The BESS average model is included within a commercial-grade software for power system analysis, validated against a detailed model that considers the high-frequency switching in the converters, and tested for different contingencies when connected to a benchmark system to demonstrate the effectiveness of a grid-scale BESS to provide the services stated earlier.

In the second part of the thesis, in order to investigate the effects of distributed BESSs connected to ADNs, on the transmission grid, for dynamic electrical studies, an aggregated black-box BESS model at the boundary bus of the transmission and ADN is proposed. ADN measurements of the aggregated response of the BESSs at the boundary bus with the transmission system are used to develop the aggregated black-box model, which is based on two Neural Networks (NNs), one for active power and the other for reactive power, with their optimal topology obtained using a Genetic Algorithm (GA). Detailed simulations are performed considering multiple BESSs connected to a CIGRE benchmark and located at a load bus of the 9-bus WSCC benchmark transmission network to generate training data for the NNs. Then, the test ADN is replaced by the proposed black-box model, with aggregated models of the loads and PV generation, demonstrating that the model can accurately reproduce the results obtained for trained and untrained events. The main conclusions of this work are that the inclusion of the proposed controllers for the BESS can significantly improve the contribution of both grid-scale and distribute BESSs to the stability of transmission grids. In addition, the need of including the dc-to-dc converter in the BESS model for dynamic studies is demonstrated, especially when degraded batteries are considered, due to the limitations this operating condition creates on the dc-to-dc operation and its associated controller. Finally, the proposed methodology used to develop the black-box model to represent the aggregated response of BEESs is proved to be robust, since this model is shown to accurately reproduce the behavior the aggregated response of the battery systems providing various grid services, not only for the events and associated data used to train the proposed NN-based models, but also for contingencies for which the models were not trained.