TY - CONF

T1 - Modelling user preferences in one-way electric carsharing systems

AU - Boyacı, Burak

AU - Zografos, K. G.

PY - 2018/9/11

Y1 - 2018/9/11

N2 - Carsharing is an advanced car rental system that allows its users to rent vehicles for a short period with increased flexibility. Depending on their properties, carsharing systems can be categorised in various ways. In this research, we are dealing with the operational decisions in one-way station-based electric carsharing systems with dynamic relocations. In these systems, the users are not restricted to return the electric vehicles to their origin stations and a group of personnel relocate vehicles during the system is in operation to balance vehicle distribution among stations. In this research, we explore the effect of providing alternative origin and destination stations, and pick-up times with discounted prices to the users different than they requested, to the system. Our aim is to develop an operational framework that maximises the profit of the operator while providing cheaper alternatives to the users.We model the main problem as a network flow problem on the time-space networks of vehicles and each personnel shifts. In order to decrease the number of arcs in the networks, we first cluster the stations. Then the operation optimisation model finds the flows of vehicles and personnel on these networks. The next mathematical model creates feasible assignments for each personnel. The last model takes the output of the two former models as input and finds the vehicle assignment with minimum infeasible charging levels. If the charging levels of all vehicles drop never below the threshold, we release the solution. Otherwise, we add the necessary constraints to the operation optimisation model and repeat the steps until we find a charging level feasible solution.We applied this framework to an electric carsharing system operating in Nice, France. Preliminary results show that, the model is efficient enough to solve operational planning problem of real carsharing systems.

AB - Carsharing is an advanced car rental system that allows its users to rent vehicles for a short period with increased flexibility. Depending on their properties, carsharing systems can be categorised in various ways. In this research, we are dealing with the operational decisions in one-way station-based electric carsharing systems with dynamic relocations. In these systems, the users are not restricted to return the electric vehicles to their origin stations and a group of personnel relocate vehicles during the system is in operation to balance vehicle distribution among stations. In this research, we explore the effect of providing alternative origin and destination stations, and pick-up times with discounted prices to the users different than they requested, to the system. Our aim is to develop an operational framework that maximises the profit of the operator while providing cheaper alternatives to the users.We model the main problem as a network flow problem on the time-space networks of vehicles and each personnel shifts. In order to decrease the number of arcs in the networks, we first cluster the stations. Then the operation optimisation model finds the flows of vehicles and personnel on these networks. The next mathematical model creates feasible assignments for each personnel. The last model takes the output of the two former models as input and finds the vehicle assignment with minimum infeasible charging levels. If the charging levels of all vehicles drop never below the threshold, we release the solution. Otherwise, we add the necessary constraints to the operation optimisation model and repeat the steps until we find a charging level feasible solution.We applied this framework to an electric carsharing system operating in Nice, France. Preliminary results show that, the model is efficient enough to solve operational planning problem of real carsharing systems.

M3 - Abstract

T2 - 60th Annual Conference of the Operational Research Society

Y2 - 11 September 2018 through 13 September 2018

ER -