An integrated optimization-simulation framework for vehicle and personnel relocations of electric carsharing systems with reservations

Abstract

a b s t r a c tOne-way electric vehicle carsharing systems are receiving increasing attention due to their mobility, environmental, and societal benefits. One of the major issues faced by the operators of these systems is the optimization of the relocation operations of personnel and vehicles. These relocation operations are essential in order to ensure that vehicles are available for use at the right place at the right time. Vehicle availability is a key indicator expressing the level of service offered to customers. However, the relocation operations, that ensure this availability, constitute a major cost component for the provision of these services. Therefore, clearly there is a trade-off between the cost of vehicle and personnel relocation and the level of service offered. In this paper we are developing, solving, and applying, in a real world context, an integrated multi-objective mixed integer linear programming (MMILP) optimization and discrete event simulation framework to optimize operational decisions for vehicle and personnel relocation in a carsharing system with reservations. We are using a clustering procedure to cope with the dimensionality of the operational problem without compromising on the quality of the obtained results. The optimization framework involves three mathematical models: (i) station clustering, (ii) operations optimization and (iii) personnel flow. The output of the optimization is used by the simulation in order to test the feasibility of the optimization outcome in terms of vehicle recharging requirements. The optimization model is solved iteratively considering the new constraints restricting the vehicles that require further charging to stay in the station until the results of the simulation are feasible in terms of electric vehicles' battery charging levels. The application of the proposed framework using data from a real world system operating in Nice, France sheds light to trade-offs existing between the level of service offered, resource utilization, and certainty of fulfilling a trip reservation.

Performance Analysis of a Forecasting Relocation Model for One-Way Carsharing

“…To guarantee the vehicle availability to the customer, we used a strict policy for reservation, i.e., the car is reserved until it is picked-up by the customer. Previous research (Boyaci et al [18]) used dynamic relocation (relocate during the day) and safety gap (i.e., 30 min of additional time, starting at the end time reservation to guarantee the space at the destination station; and avoid the car being reserved and relocated until the end of the safety gap duration) [18]. In our study, static relocation is used as we only focus on the performance of the forecasting relocation model at the end of the day.…”

“…Boyaci et al have proposed an integrated optimization-simulation framework for one-way electric carsharing system [18]. The study considered station clustering in order to reduce the computation time for the optimization model.…”

“…Based on the departure distribution of each group, customers mostly departed from Seongbuk-gu and Gyeongido-paju (38% and 22%, respectively), and the detailed departure distribution is shown in Figure 3d. However, the clustering based on the district is not the only solution, as other studies (Boyaci et al [18]) used different techniques to group the stations. The k-medoid based algorithm was used to cluster the stations.…”

“…A long booking time will reduce utilization (i.e., reduce the number of accepted requests), particularly if the operator has a high number of reservations and a low number of operational cars. However, a previous study showed that a long-term reservation system can produce better results, i.e., it can serve more demands from customers [18]. In a previous study, a different operational model was applied as the reservations are performed at least one day before (all demand is known in advance), thus the operator can perform relocation to serve these reservations.…”

“…In addition, Boyaci et al [18] and Repoux et al [19] have presented integrated optimizationsimulation framework to evaluate the performance of long-term and last-minute one-way reservation systems. In long-term reservation, the one-way system can be optimized as all the demand is known in advance (reservations are performed at least a day before) thus it can serve a higher number of requests compared to last-minute reservation.…”