Abstract

In recent decades, due to the carbon-free energy policies of power utilities, small and medium-scale renewable energy sources (RES), such as roof-top solar photovoltaic panels (PV), have been widely deployed across the distribution systems.However, due to intermittency and variability in their output, RES are often interfaced with energy storage systems, particularly electrochemical batteries, capable of supplying and absorbing energy in a matter of seconds or a couple of cycles. As controllable, flexible, and fast-response distributed energy resources (DER), battery energy storage systems (BESS) can be used in a wide range of grid and customer applications. However, storage technology is costly and a single application may not recover the costs, while a combination of multiple services might justify the expenses. Particularly, behind-the-meter (BTM) BESS is only used for the owner's energy bill reduction and part of their capacity often remains idle or underutilized, while their aggregated underutilized capacities can be used to provide grid services through an advanced control mechanism. Nonetheless, integration of the growing number of aforementioned DERs, especially those located behind the customers' meters, significantly impacts the size and complexity of the network operation problem and is a challenge to a central DER management system (DERMS). In this thesis, a decentralized agent-based control framework is proposed for a DER management system that facilitates the scalable operation of DERs. The focus is on improving the utilization of the BESS to provide multiple services in order to justify their costs. In this framework, initially, the prosumer agents schedule their resources for daily cost management based on time-of-use energy price and demand charge, considering the day-ahead demand and PV forecasts. Thereafter, a local customer aggregator (LCA) agent performs day-ahead active and reactive power control on the underutilized BTM capacities for providing multiple grid services, including network bottleneck congestion relief and grid-edge voltage support.In order to deal with the uncertainties in the demand and generation forecasts, a novel near-real-time (NRT) approach is proposed to utilize the utility-owned medium-scale BESS to provide area power regulation. The proposed approach improves the utilization of the BESS by using their remaining capacity for the secondary service of congestion relief. A novel algorithm is developed based on a modified $k$-means clustering method for grouping the customers to be assigned to LCAs.In this work, a linear programming (LP) model is used to formulate various optimization problems associated with different BESS applications and scheduling stages. A linearized power flow model is developed based on multivariate linear regression applied to a synthetic dataset, in order to simplify the inclusion of power flow constraints in the linear optimization problems.The proposed framework is applied to case studies and the simulation results are compared to those of a central control framework in order to show the effectiveness of the proposed framework.