Scheduling a Multi-Cable Electric Vehicle Charging Facility

We consider scheduling electric vehicles in a charging facility where customers arrive dynamically and tend to park longer than their charge time. In this setting, it is reasonable and technologically feasible to have charging docks with multiple cables, although such docks do not currently exist in practice. Assuming such a dock design, we study three information conditions: we know the number of electric vehicles at each dock, we know stochastic information about arrival and charging requirements, and we are able to observe exact charging requirements for vehicles in the system. We formulate a continuous-time Markov decision process (CTMDP) to optimize the system performance under the first two conditions and demonstrate that it does not scale to realisticsize problems with multiple docks. However, a single-dock version of the CTMDP is tractable. We propose and numerically evaluate a number of admission and scheduling schemes building on both the single-dock CTMDP and approaches from the scheduling literature under each of the three information conditions. Our results demonstrate (i) the value of a multi-cable dock, (ii) the importance of obtaining actual charging requirement information, and (iii) the integral role of admission and scheduling policies based on available information to improve performance.