2016 INFORMS Annual Meeting Program

SA60

INFORMS Nashville – 2016

3 - Modeling Plug-in Electric Vehicles Driving And Charging Behavior Using Real-world Connected Vehicles Data Kuilin Zhang, Assistant Professor, Michigan Technological University, Houghton, MI, 49931, United States, klzhang@mtu.edu Shuaidong Zhao We propose to investigate driving and charging behavior of Plug-in Electric Vehicles using real-world connected vehicles data. We use a data-driven approach to estimating 24-hour activity-travel dynamics of individual drivers from connected vehicles data collected in real-world. Based on this real-world driver’s activity and mobility pattern, we formulate an optimization model to address driving and charging behavior of Plug-in Electric Vehicles to better understand the battery performance of electric vehicles under real-world conditions. 4 - Operation Of Electricity And Transportation Networks With Ev Wireless Charging Mohammad Khodayar, Southern Methodist University, mkhodayar@smu.edu, Saeed D Manshadi, Khaled Abdelghany, Halit Uster This research presents the coordinated operation of wireless electric vehicle charging stations (WECS) in electricity and transportation networks. The traffic flow pattern in transportation network is assumed to follow the user equilibrium (UE) traffic assignment, where the cost of utilized electricity is incorporated in the total traveling cost. The presented formulation leverages consensus optimization to address the unit commitment in the electricity network as well as user equilibrium traffic assignment in the transportation network. SA60 Cumberland 2- Omni Topics on Shared Public Transportation Systems Sponsored: TSL, Urban Transportation Sponsored Session Chair: Hai Wang, Singapore Management University, Singapore Management University, Singapore, Singapore, Singapore, haiwang@smu.edu.sg 1 - Matching Problem For A Stochastic And Dynamic Online Vehicle Sharing System Hai Wang, Singapore Management University, haiwang@smu.edu.sg, Chiwei Yan We study a stochastic and dynamic matching problem for online vehicle sharing platform: match the spatial and temporal changing demand (ride request) with supply (vehicle). We propose an algorithm to determine the pairings of drivers to riders’ requests. At any decision epoch, we consider the set of known available drivers and potential available drivers, as well as the set of known existing passengers and potential passengers. We use an iterative procedure which calls a static and deterministic matching problem as a sub-routine. The objective is to minimize the average waiting time until picked-up for ride requests. We demonstrate the advantages of our algorithm by testing in real world data sets. 2 - Estimating Primary Demand In One-way Vehicle Sharing Systems Chiwei Yan, Massachusetts Institute of Technology, Cambridge, MA, United States, chiwei@mit.edu, Chong Yang Goh Observed trip data for one-way vehicle sharing systems do not always correspond to true demands for the service due to varying vehicle and parking availability. For example, in bike sharing systems, passengers arriving at an empty pickup station may either leave the system or spill over to nearby stations. We propose efficient methods to estimate the true origin-destination demands in a one-way vehicle sharing system using observed trip data. Our approach models a customer’s station substitution behavior based on a ranking-based choice model. We demonstrate the effectiveness of our approach using data from a bike-sharing system in Boston. 3 - The Learning Curve Of Taxi Drivers In an Urban Area: An Empirical Analysis Youngsoo Kim, Singapore Management University, Singapore, Singapore, yskim@smu.edu.sg This study aims to better understand the dynamic change of individual taxi drivers’ performance on both an aggregated output level (e.g., revenue and trips) and process level (e.g., occupancy rate and zone selection decision). We also conduct counterfactual policy experiments that capture the change derived through knowledge sharing of demanding zone on both individual and company levels. The implications of our findings for both theory and practice are discussed.

4 - Stochastic Ride-matching In Peer-to-peer Ridesharing Systems Neda Masoud, University of Michigan, 2350 Hayward St., 2124 GG Brown Bldg., Ann Arbor, MI, 48109, United States, nmasoud@umich.edu, R. Jayakrishnan We formulate the multi-hop peer-to-peer stochastic ride-matching problem as a binary program, and propose an efficient algorithm to solve the problem. We use a forecast of passenger arrivals, and take into consideration the possible future states of the ridesharing system when routing drivers. The multi-hop property of the system allows passengers to transfer between different vehicles/modes of transportation.systems. SA61 Cumberland 3- Omni RAS Student Paper Award Sponsored Session Chair: Steven Harrod, Technical University of Denmark, KGS. Lyngby, Denmark, stehar@transport.dtu.dk Rail Applications Section (RAS) sponsored a student research paper contest on analytics and decision making in railway applications. Papers must advance the application or theory of OR/MS for improvement of freight or passenger railway transportation, and it must represent original research that has not been published elsewhere by the time it is submitted. Authors of the First, Second and Third Place award winning papers will present their papers in this session. SA62 Cumberland 4- Omni Aviation Applications Section: Best Student Presentation Competition Sponsored: Aviation Applications Sponsored Session Chair: Lavanya Marla, University of Illinois, lavanyam@illinois.edu Chair: Jennifer A Pazour, Rensselaer Polytechnic Institute, 110 8th street, CII 5217, Troy, NY, 12180, United States, pazouj@rpi.edu 1 - Facility-level Item Allocation Problem In Ship-from-Store Environment Seyed Shahab Mofidi, Rensselaer Polytechnic Institute, 110 8th Street, RPI ISE Department, CII 5015, Troy, NY, 12180, United States, mofids@rpi.edu, Jennifer A Pazour Leading retailers are using their brick-and-mortar stores to fulfill online order requests, which results in ambidextrous stores that use inventory to serve both in-store and on-line shoppers. We develop a novel multi-product optimization model that captures the tradeoff between applying resources in advance when the demand is unknown or applying resources after the demand realizes. A case study illustrates how our model can be used to recommend item allocation policies for omni-channel supply chains. 2 - Parallel Algorithms For Large Assignment Problems On Graphics Processing Unit Clusters Rakesh Nagi, University of Illinois, Urbana-Champai, 117 Transportation Building, MC-238, 104 South Mathews Avenue, Urbana, IL, 61801, United States, nagi@illinois.edu, Ketan Date We discuss efficient parallel algorithms for solving large instances of the Linear Assignment Problem (LAP) and the Quadratic Assignment Problem (QAP). Our parallel architecture is comprised of CUDA enabled NVIDIA Graphics Processing Units (GPUs) on a computational cluster. We propose novel parallelization of the Hungarian algorithm on the GPUs, which shows excellent parallel speedup for large LAPs. We also propose a novel parallel Dual Ascent algorithm on the GPUs, which is used for solving the RLT2 linearization of the QAP, which also utilizes our parallel Hungarian algorithm. We show that this GPU-accelerated approach is extremely valuable in a branch-and-bound scheme. SA63 Cumberland 5- Omni Facility Logistics Sponsored: TSL, Facility Logistics Sponsored Session

38