Two-phase optimization model for ride-sharing with transfers in short-notice evacuations
Introduction
Natural disasters (such as hurricanes, floods, earthquakes, etc.) or man-made disasters (for instance, nuclear power plant accidents, chemical spills, terrorist attacks, etc.) caused unfortunate events from time to time. Such an extreme event may necessitate the short-notice or no-notice evacuation of a large population from the stricken area. Generally, compared with alerted emergency events, a no-notice evacuation takes place immediately after the occurrence of an unexpected disaster, and entails pre-disaster planning and preparedness for a more efficient response to the disaster. Examples of such disasters include earthquakes, terror attacks, and hazardous material release (Hsu and Peeta, 2013). Short-notice disasters, such as hurricanes or flooding, typically provide a lead-time from 24 to 72 h (Wolshon, 2002). The lead-time can allow both the evacuees and emergency management agencies to be more prepared for an evacuation. With a focus on the evacuation period, this research presents a two-phase model to optimize ride-sharing matching with transfers during short-notice evacuation trip planning and operations.
Carless groups and fuel shortage are two major concerns in a short-notice evacuation (Renne at al., 2008). Before Hurricane Katrina hit in 2005, an estimate of 100,000–200,000 New Orleans residents had no access to reliable personal transportation (White et al., 2008). In recent Hurricane Irma in Florida, the gasoline-shortage situation was worst in Gainesville, where more than 63% of gas stations suffered from an acute shortage of fuel, according to Gas Buddy (USA Today News, 2017). In Miami, at least 62% of stations had no gasoline (Sayyady, 2007). To mitigate some of these problems, making use of public transit systems in evacuation has been a popular topic in the past 10 years. But the public transit system was often restricted to fixed routes, limited pickup locations, fixed shelter locations and non-on-demand services. In contrast, car sharing mode can not only reduce congestion on roads, mitigate the fuel shortage issue, help the carless groups but also provide flexible routes and on-demand schedules. Especially, an evacuation provides potential rationales for car sharing: (1) Many routes of evacuees are similar or overlapping; (2) The evacuation trip departure times from the disaster zone are highly distributed within 3 or 4 days when the threat approaches; and (3) Most shelters are located along evacuation routes. Therefore, when evacuees driving on their own vehicles with vacant seats are willing to share their spare seats with other evacuees who need a ride and are frequently willing to pay a fee for the ride, then a possible ride-sharing matching could be achieved through necessary communication efforts such as an online platform or the social media. Through an online survey from 645 individuals impacted by Hurricane Irma in 2017, Wong et al. (2019) found that there existed spare capacity of vehicles in the process of evacuation and a moderate willingness of evacuees to share their private car resources.
In practice, ride-sharing services can be provided by volunteers to evacuate people who have no access to transportation under an evacuation. Individuals who operate vehicles, such as police officers, emergency response staff, along with other government affiliated and civil agencies may also contribute to evacuations (Naoum-Sawaya and Yu, 2017). What’s more, some car-sharing platforms/brokers like Uber, Lyft, and Didi can be used to collect and post the requests of the drivers and riders. Since a simple carpooling is restricted to pick up and drop off a passenger on the driver’s way (with little or no detouring), it is generally difficult to satisfy the trip demands for all people sharing the same ride. Under the space-time network modeling framework, we introduce a carpooling system with certain detour tolerance and necessary transfer connections to increase the success rate of matching for both drivers and riders with flexible trip windows. In addition, through imposing a penalty on transfer connections, defining maximum parking time, and presenting the ride states with the IDs of the vehicles on which the riders are dwelling, this paper aims to simultaneously maximizing the number of carless evacuees served, while reducing the transfer connections and controlling the vehicle parking time for transfers.
Section snippets
Mobility challenges for carless groups
In the wake of Hurricanes Katrina and Rita, numerous articles and studies were published, discussing the inadequacy of current evacuation planning for carless populations and calling for better planning. In the study by Hess and Gotham (2007), the authors found that most evacuation plans did not ponder multimodal evacuation planning, and the majority of written emergency planning information for the public rarely mentioned alternatives to private vehicles. A study in 2014 stated that 9.22% of
Assumptions
Similar to other related researchers on DARPT, we also make the following assumptions: (1) driver information and requests of carless evacuees are collected in advance; (2) drivers and riders must comply with the arrangement generated by the proposed method; and (3) network nodes will have enough parking space to pick up or drop off evacuees.
Description of basic information
We formulate the DARPT on a transportation network which is represented by a directed graph , where is the set of nodes and is the set of links
Optimization model for vehicle-space-time network flow
Based on the constructed vehicle-space-time network involving essential arrivals, departures, and pickup and delivery time window information, we now start constructing a multi-rider multi-driver network flow programming model for the DARPT. A binary decision variable is hereby introduced for the model. If is equal to 1, person can travel from node at time with vehicle to node at time with vehicle. Note that: (1) each driver must start a trip at a
Results from medium-scale transportation networks
The two-phase method described in this paper were coded in Python-GAMS platforms using a Dell XPS15 9560 laptop with 16 GB RAM. In this section, we examine our proposed method on the Chicago traffic network with 933 nodes and 2967 links, which is shown in Fig. 7. We assume that a disaster has an influence on the southeast of Chicago, and the safety area including shelters are located in the west. Based on the proposed methods, for drivers and riders who share their trip requests in a
Conclusion
Around the media question “Where Is the App for Escaping a Hurricane?” doubting the fact that many empty car seats just escaped South Florida in the last few hours but some carless evacuees had not been transited, this paper provides a two-phase method for the ride-sharing optimization with transfers in short-notice evacuations by formulating the DARPT through space-time networks. The proposed method is suitable for offline applications, or small batch processes after partitioning large-scale
Author contributions
The authors confirm contribution to the paper as follows: study conception and design: Weike Lu, Lan Liu; data collection: Weike Lu; analysis and interpretation of results: Weike Lu, Lan Liu, Feng Wang, and Xuesong Zhou; draft manuscript preparation: Weike Lu and Guojing Hu. All authors reviewed the results and approved the final version of the manuscript.
Acknowledgments
This work is supported by National Natural Science Foundation of China (General Program, Grant No. 61873216). The fourth author is partially funded by National Science Foundation–United States under NSF Grant No. CMMI 1663657. “Real-time Management of Large Fleets of Self-Driving Vehicles Using Virtual Cyber Tracks”.
References (51)
- et al.
Optimization for dynamic ride-sharing: a review
Eur. J. Oper. Res.
(2012) Optimization models for large scale network evacuation planning and management: a literature review
Surv. Oper. Res. Manage. Sci.
(2016)- et al.
Optimization models in emergency logistics: a literature review
Socio-economic Plan. Sci.
(2012) - et al.
The pickup and delivery problem with transfers: formulation and a branch-and-cut solution method
Eur. J. Oper. Res.
(2010) - et al.
Ridesharing: the state-of-the-art and future directions
Transport. Res. Part B: Methodol.
(2013) - et al.
A two-stage robustness approach to evacuation planning with buses
Transport. Res. Part B: Methodol.
(2015) - et al.
The share-a-ride problem: people and parcels sharing taxis
Eur. J. Oper. Res.
(2014) - et al.
Capacitated transit service network design with boundedly rational agents
Transport. Res. Part B: Methodol.
(2016) - et al.
Eco-system optimal time-dependent flow assignment in a congested network
Transport. Res. Part B: Methodol.
(2016) - et al.
Finding optimal solutions for vehicle routing problem with pickup and delivery services with time windows: a dynamic programming approach based on state–space–time network representations
Transport. Res. Part B: Methodol.
(2016)
The dial-a-ride problem with transfers
Comput. Oper. Res.
Evacuation transportation modeling: an overview of research, development, and practice
Transport. Res. Part C: Emerg. Technol.
Demand variations and evacuation route flexibility in short-notice bus-based evacuation planning
IATSS Res.
Analysis of hurricane evacuee mode choice behavior
Transport. Res. Part C: Emerg. Technol.
Hurricane evacuation planning using public transportation
Socio-Econ. Plan. Sci.
Logistics of hurricane evacuation in Hurricanes Katrina and Rita
Transport. Res. Part F: Traffic Psychol. Behav.
Optimizing on-time arrival probability and percentile travel time for elementary path finding in time-dependent transportation networks: Linear mixed integer programming reformulations
Transport. Res. Part B: Methodol.
Dynamic passenger demand oriented metro train scheduling with energy-efficiency and waiting time minimization: mixed-integer linear programming approaches
Transport. Res. Part B Methodol.
Planning for a bus-based evacuation
OR Spectrum
The continuous-time service network design problem
Oper. Res.
A branch-and-cut algorithm for the dial-a-ride problem
Oper. Res.
Cited by (13)
Ridesharing evacuation models of disaster response
2023, Computers and Industrial EngineeringA multi-trip electric bus routing model considering equity during short-notice evacuations
2022, Transportation Research Part D: Transport and EnvironmentCitation Excerpt :Studies on equity can be divided into four fields: disaster relief equity, mobility services equity, food distribution equity, and hazardous material transportation equity (Balcik et al., 2010). A few studies used equity as an objective to assess post-disaster emergency relief performance (Li et al., 2019; Lu et al., 2020). Zhu et al. (2019) addressed the relative deprivation cost of victims as an equity objective in an emergency relief decision.
Many-to-one stable matching for taxi-sharing service with selfish players
2022, Transportation Research Part A: Policy and PracticeCitation Excerpt :They considered future demand and maximized the cumulative number of matchings made in the time horizon. Lu et al. (2020) built a two-phase model with transfer consideration. The total routing cost was minimized.