The objective of this exploratory research is to investigate the potential of dynamic pricing to avoid unbalanced inventory in bicycle sharing systems, and therefore the needs for manual rebalancing of bicycles by truck. The proposed idea of dynamic pricing has been not used in the current practice, and the idea’s potential is unknown to the service providers. In this exploratory research, I will investigate the idea’s potential to eliminate manual rebalancing using an optimization model with demand learning process. The model will be validated with real data. This proposal addresses one of UTRC’s focus areas, Focus Area 5: Promoting livable and sustainable communities through quality of life improvements and diverse transportation development, and responds to the US DOT goals of Livable Communities and Environmental Sustainability.

Shared-mobility has attracted attentions from many researchers and policy-makers to help build livable and sustainable communities. Especially, bicycle sharing programs have been very successful in many places in the world. In the United States, there are several major cities, including Denver, Chicago, San Antonio, Houston, Minneapolis, and Washington D.C., that operate bicycle sharing programs. To the interest of the UTRC, New York City is launching a program called “City Bike” in July, 2012 with 600 stations and 10,000 bikes. This proposal will make contributions to the transportation needs in Region 2 and meet the mission of UTRC by providing knowledges for efficient and effective operations of bicycle sharing programs.

Such bicycle sharing systems are growing in popularity because of their ability to decrease the number of automobile trips, reduce traffic congestion and the need for parking in crowded downtown areas, as well as providing environmental benefits and health benefits for the riders.

Because these systems encourage one-way rentals, there exists the possibility that over time random fluctuations in demand will cause the bicycles to become concentrated in some areas of the system, causing shortages elsewhere. This may prevent the system from effectively meeting demand, if for example potential customers are unable to find a bicycle at the stations where they wish to start their trips. While direct intervention may be taken in such a case (e.g. by having an employee of the system operator reposition a number of bicycles with a pickup truck (Tse, 2011)), it should also be possible to have the customers reposition the bicycles by providing a price incentive (discount) for trips originating at stations with a surplus of bicycles or terminating at stations with a deficit. This is conceptually similar to dynamic toll pricing to encourage motorists to take different routes or travel at different times in order to ease traffic congestion. Here we also give an incentive for customers to change their routes, but both our motivation for doing so and the operational details are different.

There is not an extensive literature on the rebalancing problem, but it has been studied by some authors. Uesugi et al. (2005) suggest a scheme for rebalancing cars in a car sharing system, where cars may be repositioned by having two traveling companions use separate cars, or by having people who are not together share cars. However they do not discuss how this behavior is to be induced (by an incentive, by a mandate, etc.). Benchimol et al. (2011) discuss the rebalancing-by-truck problem as a form of Traveling Salesman Problem. Lin and Yang (2011) model the design of a bicycle sharing system as a multiple-allocation hub problem, but do not consider the possibility that one-way trips may result in inefficient bicycle positioning. Some mention the possibility of using price incentives to motivate customers to reposition inventories, but there is not very much research on this question. One system employs a simple kind of price incentive: the Vélib’ system gives customers a 15-minute credit for depositing bicycles at stations which are significantly uphill from the point of origin (DeMaio, 2009).