Decentralized and privacy-aware management of energy storage systems and electric vehicles in smart grids

• Main Author: Singh, Anshuman
• Other Authors: Hung Dinh Nguyen
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering; Smart grid; Electric Vehicles; Battery thermal management; Energy trading
• Online Access: https://hdl.handle.net/10356/180651
• Institution: Nanyang Technological University

Optimal operation of battery energy storage systems (ESS) and electric vehicles (EVs) is vital for integrating distributed energy resources. The conventional approach of centralized control in managing these devices encounters formidable challenges such as a large number of decision variables, communication needs, and privacy concerns. Moreover, it may not fully address device health, temperature, and economic sustainability. Decentralized control frameworks offer distinct advantages over their centralized counterparts, particularly by affording devices a degree of autonomy, ensuring safe operation and being scalable. Nonetheless, they are not devoid of their own set of challenges. Decentralized decision-making predicates upon local information which may lead to potential violations of grid constraints. Additionally, coordinating with peers can be challenging for smaller-scale ESS and EV charging stations with limited technical capabilities. One potential solution is the use of aggregator systems, but this can raise privacy concerns when the ESS or EV shares its thermal model with the aggregator. The present thesis aims to investigate various decentralized management schemes, with a particular focus on privacy-aware considerations, to optimize the operation of ESS and EVs while concurrently ensuring that the grid states are within the specified limits. The contributions in the thesis are broadly divided into three parts. The first part introduces a two-layer framework for thermal management of grid-connected ESS and optimal Microgrid (MG) operation. The first layer develops a closed-form battery optimal control model to balance fan speed and current magnitude. In the second layer, the MG operator conducts optimal power flow to minimize the entire MG's operation cost. Simulation results confirm the effectiveness of the proposed method in optimizing battery operation. The second part proposes a stress-cognizant optimal battery dispatch framework considering degradation due to partial charge-discharge cycles and the battery temperatures. The proposed physics-inspired heuristic condition facilitates the integration of the rainflow cycle counting algorithm into the optimal dispatch model. The framework is then applied to an ESS participating in both day-ahead and real-time balancing markets. To manage uncertainty in electricity prices and meet day-ahead market commitments, a model predictive control (MPC)-based framework is introduced. The numerical results indicate that the proposed framework can efficiently utilize the cooling to reduce degradation with/without modifying the market-benchmark dispatch. The third part develops a decentralized market-clearing mechanism for local energy trading among prosumers including EV charging stations. This research contributes by formulating Safe Operating Envelopes (SOEs) for the battery, ensuring it adheres to operational limits during charging while preserving EV privacy. Furthermore, an incentive model is introduced to capture the EV owner's support for Vehicle-to-Everything initiatives. Locational Marginal Price (LMP)-based grid constraint model is formulated to ensure grid state preservation. The framework was tested through simulations on the IEEE 33-bus distribution network with 5 prosumer agents, featuring solar PVs and multiple EVs. The fourth part builds upon the decentralized market mechanism introduced in the previous chapter, focusing on negotiations between EVs and EV charging stations. It introduces a novel charging comfort function for EV owners, along with probabilistic SOEs to regulate the EV's battery temperature. Additionally, a stochastic optimization model is devised for the charging station, aiming to maximize its profit amid uncertain future EV arrivals. Simulation results underscore the framework's effectiveness in optimizing EV assignment to charging stations.