The Bike-sharing allocation and rebalancing problem is the problem of determining
the initial daily allocation of bikes to stations in a bike-sharing system composed of
one depot and multiple capacitated stations, in which bikes can be rebalanced at a
point in time during the day. Due to the uncertain demand in each station, we propose
a two-stage stochastic programming formulation, where the allocation is made at the
first stage and the recourse decisions related to rebalancing are made at the second
stage. The impact of the stochastic demand on the problem solution is examined,
showing the benefits of the proposed methodology with respect to the deterministic
formulation. A procedure for solving the stochastic program using information from
the deterministic solution is derived showing a significant reduction in solution time
without losing solution quality. The proposed approach is finally benchmarked on the
real bike-sharing system of the city of San Francisco.
Citation
Submitted for evaluation on Sept. 2023