Multi agent control based energy optimisation of a prosumer household and a community with bidirectional electric vehicles

Kurzfassung

Residential sectors have experienced an increase in solar photovoltaic installations in the last decades, due to which the prices of installation have reduced significantly, however, the feed-in tariff has also been lowered considerably. Many people have already invested in an electric vehicle and the numbers are increasing as accessibility and reliability have increased whereas the cost has decreased as compared to previous years. The problem emerges of overloading the local grid with high penetration when the photovoltaic systems are generating energy or when there is high peak demand when those cars are charging as there is a mismatch in time of generation and demand. This study presents how bidirectional electric vehicles can optimize the self-consumption of solar photovoltaic and increase the self-sufficiency of the loads in a household and a community, which are a group of households, by using controlled charging strategies. The optimization is performed using controllers in three different hierarchical levels: car, household and community. The car controller is a bidirectional charging station where the electric vehicle is connected, it takes user preferences that are used by the household controller to perform energy optimization by handling the mismatch between the generation and demand. The community level controller performs an on-the-top optimization along with the household and car controllers which curtail the power flow between the community and the electricity grid. The impact on the self-consumption and self-sufficiency of the community with a community charging station, where all the electric vehicles of the community are parked at one place for charging, is also studied in this thesis. The effect of the proposed control framework is investigated on a reference distribution grid (MONA grid, type 5) simulated in the software package DIgSILENT PowerFactory whereas the control framework is developed in Python. Real load and photovoltaic profiles were used to execute the simulations. The results show that there was a 40% and 36% increase in the self-consumption and self-sufficiency on a household level whereas a 51% and 31% increase on the community level when a coordinated control system was implemented.