Online Dynamic Scheduling for Charging PHEVs in V2G

One of the major challenges of the twenty-first century is developing technologies that reduce greenhouse gas emissions. One technology with the potential to address these challenges is plug-in (hybrid) electric vehicles (PHEVs). PHEVs derive much of their energy from the electric power grid rather than gasoline. If the projections of large PHEV penetration are true, they will put considerable additional stress onto existing power grids. It has been proposed that the appropriate scheduling of PHEV charging can reduce this stress through demand response. The PHEV charging scheduling has multiple facets. The focus of this paper is to develop algorithmic approaches to deal with the uncertainties associated with PHEV charging. The scheduling problem is modeled as a multi-stage online decision problem where input parameters of future charging requests and power grid status are not revealed when current charging decisions are made. We present two algorithms: consensus and expectation that use predictions about the future to make scheduling decisions. The performance of these algorithms is compared with algorithms that do not account for future uncertainty.