EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
1163. On Static and Dynamic Inventory Rebalancing in Bike Sharing
Invited abstract in session MA-49: Lot-sizing with joint replenishment and routing decisions, stream Lot Sizing, Lot Scheduling and Production Planning.
Monday, 8:30-10:00Room: M1 (building: 101)
Authors (first author is the speaker)
| 1. | Haoxiang Wang
|
| Department of Informatics, King's College London | |
| 2. | Dimitrios Letsios
|
| Department of Informatics, King's College London | |
| 3. | Xinyi Ye
|
| Department of Informatics, King's College London |
Abstract
We elaborate on inventory rebalancing in bike sharing. The goal is to decide how to reposition bikes between the stations of a bike sharing network so as to optimize service loss, based on the dynamic lot sizing model. Previous work computationally demonstrates the benefits of dynamic compared to static rebalancing. We propose new optimal algorithms for both problem variants. On the negative side, we show that static rebalancing results in O(1/epsilon) proportion of successful requests. On the positive side, we prove that dynamic rebalancing achieves an Omega(1-epsilon) performance guarantee. To our knowledge, this is the first theoretical characterization of the dynamic rebalancing benefits compared to static rebalancing. Next, we consider the more general problem of optimizing service loss with bounded rebalancing costs. For both the static and dynamic variants, we develop new scalable mixed-integer approaches to practically relevant instances. We computationally substantiate the proposed approaches using data from the London bike sharing network.
Keywords
- Inventory
- Algorithms
- Programming, Mixed-Integer
Status: accepted
Back to the list of papers