Abstract

The batteries of electric vehicles can be used as Virtual Power Plants to balance out frequency deviations in the electricity grid. Carsharing fleet owners have the options to charge an electric vehicle's battery, discharge an electric vehicle's battery, or keep an electric vehicle idle for potential rentals. Charging and discharging can be used to provide reliable operating reserves. We develop an analytical model that manages carsharing fleets. On the one hand, the energy in the batteries of an electric vehicle can be made available to the grid as operating reserves. On the other hand, the electric vehicle can be made available for rental, where the location within the city matters: drivers want a car to be close to their place of departure or arrival. The model can also be used by Transportation Network Companies such as Uber to preposition their vehicles conveniently in a city or optimize zonal pricing. To validate our model we develop a Discrete Event Simulation platform. We calibrate this simulation with locational information (GPS), rental, and charging transactions of 1,500 electric vehicles from the carsharing services Car2Go (Daimler) and DriveNow (BMW) over several years. We investigate the influence of the charging infrastructure density, battery technology, and rental demand for vehicles on the payoff for the carsharing operator and make an international comparison between the USA, Germany, the Netherlands, and Denmark. We show that Virtual Power Plants of electric vehicles create sustainable revenue streams for electric vehicle carsharing companies without compromising their rental business.

Abstract

The introduction of electric vehicles (EVs) will have an important impact on global power systems, in particular on distribution networks. Several approaches can be used to schedule the charge and discharge of EVs in coordination with the other distributed energy resources connected on the network operated by the distribution system operator (DSO). The aggregators, as virtual power plants (VPPs), can help the system operator in the management of these distributed resources taking into account the network characteristics. In the present work, an innovative hybrid methodology using deterministic and the elitist nondominated sorting genetic algorithm (NSGA-II) for the EV scheduling problem is proposed. The main goal is to test this method with two conflicting functions (cost and greenhouse gas (GHG) emissions minimization) and performing a comparison with a deterministic approach. The proposed method shows clear advantages in relation to the deterministic method, namely concerning the execution time (takes only 2% of the time) without impacting substantially the obtained results in both objectives (less than 5%). Document type: Article