Optimising EV Charging Profiles by Price Steering

Due to the increasing number of electric vehicles (EVs), profile flattening control methods are needed to prevent that large peaks in electricity usage overload the grid. Although several methods for flattening the charging profile of EVs already exist, they usually require perfect information. In this work, we present a price steering approach. To study its effectiveness, we employ a game theoretic model. Theory guarantees that our model results in profiles that approximate the optimal solution by a factor of at most 1.457. This bound is tight, as worst-case examples with this objective exist. To complement this, we perform a simulation with data from a real-life parking lot to evaluate performance in real-world instances. Results indicate that this results in a performance within 1% of the optimal solution. As our pricing scheme sets prices dynamically according to the actual load in each time step, it does not require any information of the instance at hand. However, the analysis requires standard rationality assumptions, which imply perfect anticipation from the users. To alleviate this, we also study a version of the pricing mechanism that gives users fixed and personalized prices at the time of arrival. Although this results in a 21% loss of efficiency compared to the optimal solution, it still outperforms uncontrolled EV-charging by a factor of 3 in our test case.