Enhancing the performance of one-way electric carsharing systems through the optimum deployment of fast chargers

Abstract

Abstract One-way electric carsharing systems (OWECS) provide environmentally friendly mobility that enables users to commence and terminate their trips at a preselected station within a region. However, the operations of OWECS are complicated mainly due to: (i) unbalanced spatial and temporal distribution of demand which causes shortage or surplus of vehicles at stations, and (ii) excessive battery charging requirements that can reach up to 8\xa0h. To ensure vehicle availability to their customers, carsharing companies hire personnel to relocate vehicles to restore the demand–supply balance, and increase the number of trips served. On the other hand, fast charging technologies can reduce the charging time drastically and can help carsharing companies to cope with the inefficiencies arising from excessive battery charging times. Therefore, fast charging technologies have the potential to enhance the operational performance of OWECS. In this study, we propose an integer programming model to determine the number and location of fast chargers to be implemented in OWECS while considering vehicle relocation and battery availability. We propose a time–space–battery level network model which allows to track battery levels of each vehicle. As the number of stations increases, the number of relocation variables created increases polynomially which makes the model intractable for problem instances found in real world OWECS. Therefore, we are introducing three heuristics, two of which are based on the concept of reduction of vehicle relocation variables, while the third heuristic is based on station grouping. The heuristics that are categorized as relocation reduction type heuristics generate only a fraction of all possible relocation arcs significantly reducing the number of relocation variables. The third heuristic reduces the number of variables by grouping the stations and solves the problem initially at an aggregate level using groups of stations instead of individual stations, while subsequently optimizes the location of chargers within each group of stations. We tested different approaches combining heuristics and tested them on smaller instances with exact solutions to identify the approach that is both accurate and efficient. We applied the selected heuristic approach on real-life instances taken from an OWECS based in Nice, France. The results suggest that the use of the proposed modeling and solution framework leads to a configuration of the network of fast and conventional chargers that improves the profitability and the number of trips served by the OWECS system.