Real-Time Energy Management System for Optimal Scheduling of EV Charging in Renewable Microgrids

Coordinating electric-vehicle (EV) charging with renewable generation can reduce diesel dispatch, but the added value of predictive optimization depends on the comparator, forecast error, and economic assumptions. This paper evaluates a linear-programming receding-horizon energy management system (EMS) for an islanded photovoltaic–wind–diesel microgrid with optional vehicle-to-grid (V2G) operation. A reproducible synthetic benchmark includes eight deterministic cases, 100 uncertainty realizations shared by four controllers, renewable profile sensitivity, reoptimized peak-penalty sensitivity, and offline optimization references. Under nominal inputs, the EMS reduces total diesel energy by 1.52% and EV-attributable diesel energy by 70.09% relative to uncontrolled charging. Its additional benefit over renewable-aware charging is 11.18 kWh of diesel and 61.48 USD in the daily surrogate operating cost. Under uncertainty, the EMS meets all EV targets in every run and reduces diesel energy by a mean 1009.94 kWh relative to uncontrolled charging. Relative to renewable-aware charging, however, it uses 219.85 kWh more diesel on average while reducing peak dispatch by 52.42 kW and the surrogate cost by 91.29 USD. V2G reduces the deterministic event peak by 385 kW; its economic valuation depends strongly on the assumed peak penalty. The results establish an energy–peak–service tradeoff for the evaluated model rather than universal controller superiority or hardware realtime validation.