Collaborative energy management systems: design and evaluation for intelligent buildings, electric vehicles and street lighting

Abstract

This thesis is devoted to the design of collaborative energy management systems (CEMS): for coordinating large numbers of distributed energy resources (DER) and balancing the uncertainty arising from the unpredictability of DER user behavior. There is currently no field of research on holistic solutions to CEMS problems since their nature is very diverse for different applications. This thesis addresses the design requirements for various CEMS according to the objectives of energy efficiency, flexibility and robustness. The first case study is a fully decentralized CEMS for adaptive street lighting system aiming at energy savings. LED technology permits frequent dimming of luminaires, so real-time street lighting automation based on vehicle sensing was investigated, to provide only the required level of illumination at the right time and in the right place. The designed system is robust against changes to infrastructure and is flexible for adding or removing smart luminaires.The second case study is decentralized control of autonomous batteries for a group of households. The objective is to use the electric vehicle (EV) batteries to supply the households to minimize the cost of electricity bought from the grid. The smart batteries collaborate in a fully decentralized peer-to-peer architecture, exchanging all relevant information and reacting to electricity price. They execute the same algorithms to achieve electricity cost savings at the level of the group of households taking into account the unpredictable availability of each EV at the parking lot. The third CEMS case investigates additional possibilities for using EVs in the context of V2B (Vehicle-to-Building) at a large prosumer building with local photovoltaic (PV) generation. Excess solar generation is stored to the EV batteries and later used to cover the load of the building at times when the load exceeds the PV generation. The profitability of the approach is assessed, taking into consideration battery degradation costs. The EV notifies the prosumer of its willingness to participate in the CEMS, the departure time and the intended State of Charge (SOC). A centralized agent-based design of the CEMS was investigated to ensure that the prosumer is able to achieve the intended SOC for all EVs at the announced departure time, at the same time coping with early departures of some EVs. The last line of research was devoted to the questions about how the quantitative benefits from the previous use cases would change with more accurate battery models and optimal variable control of discharge current. The high-fidelity battery model was developed and variable current control schemes were investigated. It is identified that at a certain SOC, the battery stops working, and this SOC depends on the discharging current that was used. An optimal discharging strategy is presented to pursue the conflicting objectives of achieving a low SOC while using a high discharge current.