Dynamic matching of uncertain demand with uncertain supply for bike sharing systems
Journal
International Journal of Industrial Engineering : Theory Applications and Practice
Journal Volume
26
Journal Issue
6
Pages
866-877
Date Issued
2019
Author(s)
Abstract
Most production and inventory models address either demand or supply uncertainty. When both uncertainties are involved, intermediary inventory stock usually serves as a buffering mechanism so that they are dealt with separately. In this paper, we address a new problem of directly matching uncertain demand with uncertain supply that arises in the dynamic bike balancing problem of bike sharing systems. We show that it is not optimal for the bike stock in any station to be less than a certain floor threshold or more than a certain upper threshold when the dual performance measures of bike utilization and lost customers are considered. We next construct a threshold-based Integer Programming model for dynamic balancing. Through numerical examples, we find that the problem is characterized by many multiple optimal solutions, which lead to dispersed transfers of bikes, but this imperfection is resolved by ranking and re-sequencing candidate transfers in an enhancing step. By using random numerical cases, we compare the merits of the model with a mean-value model, and assess its capability of prepositioning bikes to hedge the uncertainty. This paper contributes to the methodology of matching uncertain demand with uncertain supply. © INTERNATIONAL JOURNAL OF INDUSTRIAL ENGINEERING.
Subjects
Bike sharing systems; Dynamic bike balancing; Matching uncertain demand and supply
Other Subjects
Integer programming; Uncertainty analysis; Buffering mechanisms; Integer programming models; Mean-value modeling; Multiple optimal solutions; Performance measure; Production and inventory; Sharing systems; Uncertain demand and supplies; Bicycles
Type
journal article