Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SD31

3 - Integrated Pickup and Delivery Operations using Trucks and Drones Nawin Yanpirat, Auburn University, Auburn, AL, United States, Daniel F. Silva, Alice E. Smith We study optimal routing to perform both delivery and pickup operations with a single truck, outfitted with a single UAV, that it can deploy en-route to reach some drone-enabled customers. On each sortie, the drone can perform a single delivery operation, a single pickup or a delivery followed by a pickup. We model the problem as a mixed integer program and perform numerical experiments to measure the benefit to total route time of incorporating pickup operations. We also discuss the effects of battery life, drone speed and ratio of pickup to delivery customers on the total route time. 4 - The Multi Visit Drone Routing Problem Stefan Poikonen, Assistant Professor, University of Colorado Denver, 1475 Lawrence Street, Denver, CO, 80202, United States Many papers have considered hybrid truck-and-drone delivery models. Often they assume there is a single drone per truck, each package is homogeneous, the drone is capable of carrying one package at a time, the set of feasible launch locations is restricted to customer locations, and the drone has a fixed time battery life. In the Multi-visit Drone Routing Problem, we consider the case of a heterogeneous set of packages, a drone that is capable of carrying multiple packages at a time, a more flexible launch site set, and a user-defined energy drain function, which specifies the energy expenditure required by a drone to carry a given set of packages. Heuristics tractable for large instances will be presented. 5 - Travelling Salesman Problem with Drones Mehdi Behroozi, Northeastern University, Department of Mech. & Ind. Engineering, 334 Snell Engineering Center, Boston, MA, 02115, United States, Dinghao Ma, Reyhaneh Mohammadi Unmanned Aerial Vehicles (UAVs), commonly known as drones, have opened up their way into the massive industry of parcel delivery. In this paper, we study a travelling salesman problem with a drone in which a drone is dispatched from a delivery truck and while the truck is delivering the packages, the drone can also deliver a package and return to the truck after each delivery to pick up the next package. We present optimization models and heuristics to solve this problem. We show that the combination of a truck and a drone provides a much more efficient last-mile delivery service when compared to the truck-only delivery option under reasonable assumptions for the ratio of the speeds of the drone and the truck. n SD31 North Bldg 222A Innovative Transportation Data Sponsored: TSL/Intelligent Transportation Systems (ITS) Sponsored Session Chair: Zhen Qian, Carnegie Mellon University, China Co-Chair: Shanjiang Zhu, George Mason University, Fairfax, VA, 22030, United States 1 - Understanding Ride-sourcing Drivers’ Customer Search Behaviors Zhengtian Xu, University of Michigan, 2489 Stone Rd, Ann Arbor, MI, 48105-2540, United States, Yafeng Yin Ride-sourcing services have become increasingly important in meeting travel needs in metropolitan areas. Even though extensive studies were conducted to advance our understanding for such an emerging service, few of them paid attention to drivers’ behaviors, partially due to the lack of related datasets disclosed by ride-sourcing companies. By leveraging the empirical evidences from Didi Chuxing, this study aims to comprehensively survey ride-sourcing drivers’ decision making on customer search. We will investigate how drivers’ search behaviors are interacted with different factors, and provide insights for ride- sourcing companies in managing drivers’ labor supply. 2 - Measuring and Optimizing the Network Disequilibrium Levels through Ridesourced Vehicle Data Wei Ma, Carnegie Mellon University, Pittsburgh, PA, 15213, United States, Zhen Qian With the boom of the mobile internet, on-demand ride sourcing services such as Uber, are becoming an indispensable component of urban transportation systems. This research proposes a novel measure of network disequilibrium level with ridesourced vehicle data. It also proposes an information disclosure scheme such that the transportation networking companies are able to disclose the traffic information to benefit the public society without leaking the commercial confidential. A real-time NDL based routing method under the proposed information disclosure scheme to reduce total network congestion. Two large- scale real networks with disaggregated and aggregated ridesourced vehicle data.

can yield exact solutions to the optimal design with a customized dynamic programming algorithm, where a series of valid inequalities based on the relationship between passenger demand and vehicle capacity are applied to expedite the solution speed. The second model is a continuum approximation (CA) model that presents a macroscopic view of the system and yields simple analytical rules into the optimal design. 2 - Optimal Rebalancing for Bike Sharing Systems with Information Assisted Riders Mohammad Javad Feizollahi, Assistant Professor, Georgia State University, 35 Broad St., Room 408, Atlanta, GA, 30303, United States, Xinchang Wang We consider a bike sharing system with riders who are well informed of the number of available bikes and dockers at each docking station. Meanwhile, this information affects the likelihood of riders choosing a station to pick up or drop off bikes. A fleet of homogeneous trucks is employed to reposition bikes between the stations. The objective is to minimize the sum of the bike rebalancing cost and the lost demand penalty cost. We focus on the static version of the problem and formulate it as a mixed-integer nonlinear program. To solve the program, we develop solution approaches leveraging techniques from both dynamic programming and discrete optimization. Our solutions are tested with numerical studies. 3 - Optimal Capacity Sizing of Park-and-ride Lots when Parking Availability Information is Publicized Xinchang Wang, Mississippi State University, Marketing Department, 324C McCool Hall; Mailstop: 9582, Mississippi State, MS, 39762, United Statess, Qie He We study the optimal capacity sizing of parking lots for park-and-ride commuters when parking availability information is publicized. Commuters’ choice of parking lots follows a multinomial logit model accounting for the effects of road congestion and parking availability information. The objective is to maximize the total social welfare of all park-and-ride commuters. The problem is formulated as a stylized non-convex optimization model with the choice model. We provide a characterization of the optimal capacities with homogeneous routes. For the general case, we develop an efficient search algorithm to solve the model. 4 - A Multiplayer Parking Pricing Problem for Emerging On-line Markets Hossein Fotouhi, George Mason University, 4630 Buckhorn Ridge, Fairfax, VA, 22030, United States, Elise Miller-Hooks A bi-level, multi-player, parking pricing problem is presented. In the upper level, competing parking operators seek to maximize their revenue, while in the lower level users seek to minimize their parking choice disutility. The problem is formulated as an Equilibrium Problem with Equilibrium Constraints (EPEC). A heuristic algorithm is presented based on a concept of Nash domination. Drone-assisted Logistics Sponsored: TSL/Facility Logistics Sponsored Session Chair: Mehdi Behroozi, Northeastern University, Boston, MA, United States 1 - On Travelling Salesman Problems with Drone-truck Synchronizations Minh Hoang H , PhD, University of Engineering and Technology, Vietnam National University, Hanoi, Viet Nam In this talk, we will study two variants of travelling salesman problems with drone-truck synchronizations. In the first variant, while visiting a customer to provide a delivery, truck can launch a drone to service other customers, then continue to visit other customers before retrieving the drone at another customer location. In the second variant, truck arrives at a location and launches several drones to service customers. It has to stay there to retrieve all the drones before visiting other locations. Mathematical formulations, solution approaches and result discussions are presented. 2 - A Multi-modal Drone Delivery System for Urban Logistics Mohammad Moshref-Javadi, Massachusetts Institute of Technology, 1 Amherst St, E40-247A, Cambridge, MA, 02142, United States, Ahmad Hemmati, Matthias Winkenbach We present a mathematical formulation and heuristic solution approach for the optimal planning of last-mile delivery routes in a customer-centric, multi-modal urban logistics system combining truck and UAV-based delivery operations. The mathematical model and heuristic solution approach are applied to several generated and adapted problem instances from the literature. Further, we consider a real-world case study for e-commerce deliveries in Brazil. n SD30 North Bldg 221C

101