Robust energy management for multi-mode charging stations equipped with batteries

Increasing penetration of electric vehicles may pose dramatic challenges for existing infrastructures. Indeed, current fast (and upcoming ultra -fast) charging technologies allow charging at high power levels which, together with the unpredictable and high volatile pattern of charging demand, may lead to serious operational problems if charging infrastructures are not properly planned and operated. In this context, planning and operational tools for modern charging stations become a valuable contribution for station owners and network operators. This paper focuses on day -ahead and real-time energy management of multi -mode (fast and semi -fast) charging stations, equipped with photovoltaic generators and eventually large-scale stationary battery systems. The developed management model casts as a two -stage model in which day -ahead decisions are taken in the first level, where uncertainties in charging demand and local generation are modelled using robust optimization. This way, day -ahead operational decisions are taken under worst -case realization of uncertainties. Secondly, a fast and always -feasible optimal routine is proposed for real-time management in order to correct possible imbalances caused by deviations of actual realization of uncertainties. To properly model charging demand into the proposed robust framework, a novel methodology is proposed based on synthetic demand scenarios, from which the expected and maximum demand can be derived for different charging modes. The day -ahead robust model is posed as a bi-level optimization framework, which is further reduced to a master -slave structure solved using the well-known Column & Constraint Generation Algorithm. The resulting mixed -integer -linear programming model is validated on a benchmark charging infrastructure for different battery capacities, showing their capabilities and usefulness. The results show that uncertainty in demand notably impacts the total operational cost, incrementing it by 85 %, while the uncertainty in renewable generation has a more limited impact. On the other hand, it is shown that the performance of batteries mainly depends on the tariff scheme adopted. Moreover, it is highlighted that although importing energy supposes the main expenditure of the station (>80 % of total cost), exporting surplus energy may improve the economy of the system, reducing by 16 % the total energy costs. Finally, a scalability analysis is conducted showing that the proposed methodology scales well with the system size.