Project Description

Carsharing is a transportation option with great potential, allowing people to use a private vehicle without having to own it. Nowadays carsharing systems are implemented almost all over the world, being the round-trip system the most predominant. However, the implementation of one-way systems has been expanded considerably since 2008. Despite the proliferation of carsharing, there are few analytical studies addressing its management. One-way systems offer users more flexibility.

Contrarily to the typical round-trip systems, one-way systems allow users to pick up the vehicles at one station and deliver them to a different station from the one where they were picked up. This flexibility poses added management complexities which have never been solved completely. Most of the existing studies cannot accurately represent the reality of these systems, due to algorithmic, structural, and functional complexities. Therefore, carsharing is an interesting topic to be addressed under an operations research perspective.

This thesis has two main objectives. The first objective consists in assisting one-way carsharing companies to plan and manage their systems in a more profitable way while at the same time offering users a good quality of service. The second objective targets helping round-trip carsharing companies to start offering one-way trips, taking full advantage of idle fleets and offering users a transportation option for other trips than just shopping or leisure. Three optimization models as well as a simulation model have been developed to reach the previously defined objectives.

In this thesis, we consider the perspectives of two main stakeholders: the carsharing provider and the carsharing user. It is our belief that the improvement of carsharing systems from these two perspectives is also relevant to transportation authorities and to local and central governments, since it provides a complement to existing private and public transportation modes.

The greatest problem of one-way carsharing companies is associated with the imbalance of vehicles across stations. This is due to the imbalance inherent to trip patterns in most of the cities worldwide. In this work, the two approaches proposed to mitigate this problem are: vehicle relocation between stations through a staff of drivers, and changing the price of the trips according to its effect on system balance. With respect to relocation operations, both an optimization model and a simulation model are developed. For the second approach, only an optimization model is proposed. Given the complexity inherent to this optimization model, we resorted to an iterated local search meta-heuristic algorithm. All of the developed models aim at maximizing the profit of the carsharing companies taking into consideration the revenue obtained through the trips paid by carsharing users, and the costs involved, namely fleet, stations and relocation operations costs. For the trip price changing approach, costs associated with relocation operations are not considered since vehicle relocation is not considered.

As one-way carsharing systems present greater flexibility in terms of trip purposes (for example, commute, shopping, and leisure), round-trip companies have started to consider providing one-way trips. For instance, in 2014, Zipcar began accepting some one-way trips. Notwithstanding, as it was previously said, one-way trips are more difficult to manage than the round-trip ones. Thus, there is the need to study the integration of both carsharing types, which is another objective of this thesis. For this purpose, an optimization model is developed with the same goal as the previous, the maximization of carsharing companies’ profit.

All the models are applied to realistic case studies. For the approaches studied to balance the vehicle stocks in one-way carsharing systems, the municipality of Lisbon, in Portugal, is used. For the integration of both round-trip and one-way carsharing, the Zipcar round-trip carsharing service in Boston, Massachusetts, USA, is considered. The developed methodological approaches are able to deal with the size and complexity of the case studies considered, as all the applications reached satisfactory results and some of them achieved optimal results. Results show the usefulness of these methodologies as viable tools to help carsharing companies plan and manage their systems, improving the level of service offered to the users.

Research Team

CITTA

• Diana Jorge
• Gonçalo Correia (supervisor)

MIT

• Cynthia Barnhart (supervisor)

Financial Support

• FCT

Stage of Progress

• Concluded in 2014