Fleet management for Electric Vehicles sharing system under uncertain demand

In this paper, the problem of fleet management in electric vehicles (EVs) sharing service under demand uncertainty is studied. To solve this problem, we propose two robust methodologies to generate robust fleets reposition plans that mitigate rental demand uncertainty and imbalances. More specifically, an adjustable robust optimization model (AROM) and joint chance constraints model (CCM) are developed for dynamically assigning EVs and redistribute traffic flow problems with uncertainty. These two model both utilize the information (i.e., partial distributed information in CCM and all information in AROM) about the realized demand from the previous periods to make decision for future stages in an adjustable way. Computational test shows that necessity of accounting for uncertainty, as the total cost of nominal solution increase significantly even when only a small percentage of uncertainty is in place. Moreover, solutions generated by our models outperform deterministic solutions, not only on average objective value, but also in the gap with ideal solution.