Decentralized charging decisions for the smart grid

Abstract

This thesis implements a decentralized smart charging strategy and V2G simulation for EVs and PHEVs within the large scale transport simulation framework MATSim [2]. The charging decisions of all vehicles aim to reach a maximum load flattening effect and can be completed with minimal information remotely in individual on-board processing units. This reduces the need for communication and infrastructure intensive systems. The decentralized smart charging algorithm relies on linear programming to optimize the charging durations for each parking interval and uses probability density functions, indicating the distribution of charging slots over the simulated day, to guide the exact time choices. In the V2G simulation, every vehicle estimates its required contribution to regulate the grid from the current total V2G need and an estimation of the number of connected vehicles available for V2G. Then, each vehicle makes an economic decision, if V2G regulation is provided dependent on the agent’s state, his next plans and the opportunity to reschedule. The decentralized smart charging algorithm proves to be a powerful method to shift charging times according to the distribution of free charging slots. Two suggestions to improve the methodology are made to mitigate grid violations completely. It is found that increases in battery size can significantly improve the performance of EVs and avoid CO2 emissions. The ratio of EVs in the system has no influence on the charging behaviour of agents and the gas price has only a small impact on the total charging costs. In the proposed V2G setup the maximum capacity of agents to provide regulation is relatively low and the potential revenues are unattractive. To make V2G regulation a feasible and economical concept, it is proposed to offer capacity payments and to limit V2G to PHEVs enforcing uneconomic but reliable V2G regulation decisions.