Online Pricing Mechanisms for Electric Vehicle Management at Workplace Charging Facilities

In this paper, we design posted price mechanisms for assigning electric vehicles (EVs) to electric vehicle supply equipment (EVSEs) to maximize the smart charging potential of a workplace parking facility. Our approach can accommodate diverse smart charging objectives for the facility, e.g. minimize electricity costs from time-of-use rates, maximize behind-the-meter solar integration, or provide ancillary services. We also accommodate individual users' preferences for specific charging spots. To optimize the assignment of EVs to EVSEs, we develop three different pricing heuristics that allow each arriving EV to solve for its own optimal (i.e., cost minimizing) EVSE assignment. The first approach is based on the use of online convex optimization (OCO) to learn the EVSE prices and performs price updates after each arrival. We discuss performance guarantees for the OCO approach compared to the optimal offline solution. The second and third approaches make use of Lagrangian relaxation to decompose a modified multi-user problem into many single-user problems of small dimension. These two approaches provide EVSE prices that are only updated once per day. The second approach uses a stochastic supergradient based method to converge to optimal daily EVSE prices and can only accommodate stationary charge statistics. The third approach makes use of a one-shot learning technique to learn the prices each day, which allows for non-stationary charge request statistics.