Informs Annual Meeting 2017

MC59

INFORMS Houston – 2017

MC57

repositioning costs are equivalent and proportional to the distance, then the optimal policy is non-repositioning and quasi-myopic. The strategies suggested are aimed at a centralized taxi dispatching system. For instance, an autonomous taxi system. 3 - Package Delivery with Trucks and UAVS Siyuan Song, University of Southern California, 11657 Kerwwod Pl, El Monte, CA, 91732, United States, siyuanso@usc.edu We determine the efficiency of a delivery system in which an unmanned aerial vehicle (UAV), or a fleet of UAVs, provides service to customers while making return trips to a truck that is itself moving. Although the hardware for such systems already exists, it is not yet understood to what extent such an approach can actually provide a significantly improved quality of service. In this paper, a mathematical formulation of this problem is given and followed by an asymptotic result based on probabilistic analysis. We then put forward a heuristic algorithm and give some numerical experiments based on it. In addition to that, some variants of this problem that may happen in practice are also discussed. 4 - Rich Trailer Routing Optimization for Chemicals and Liquid Bulk Ashesh Kumar Sinha, Assistant Professor, Kansas State University, Manhatan, KS, 66502, United States, ashesh.sinha24@gmail.com We consider a rich trailer routing problem with stochastic transit times for chemicals and liquid bulk shipments. A route of a trailer comprises of pickup and delivery of restricted sequence of shipments while undergoing specialized tank wash operations. Using stochastic models and efficient parallel computing, we determine optimal trailer, shipments, and washes combinations while meeting the expected service level and cost. 5 - Joint Electric Vehicle Sharing and Vehicle2grid Service System Operations Dongfang Zhao, Doctor, University of South Florida, 4202 E. Fowler Avenue ENG 216, TAMPA, FL, 33617, United States, dongfangzhao@mail.usf.edu, Xiaopeng Li This study proposes a new model for designing a new (electric vehicle) EV sharing and V2G service system. This model considers the operations of a fleet of EVs that may opt to serve a transportation system when transportation demand is high or return electricity to the grid when the power price shoots high. An integer programming model is proposed to determine the optimal allocation of the fleet’s time over these two markets. This model is solved with an efficient dynamic programming algorithm. Real-world data about the configurations of these two markets are used to build numerical examples. Sensitivity studies are constructed to draw insights into the feasibility of this joint service system. 6 - Online Vehicle Routing the Edge of Optimization at Scale Sebastien Martin, Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA, 02139, United States, semartin@mit.edu, Dimitris Bertsimas, Patrick Jaillet Developing vehicle routing optimization algorithms that are capable of solving very large real-time problems involving tens of thousands of customers per hour is becoming more and more important and impactful. We introduce generalizable algorithms to scale mixed integer optimization to these practical problem sizes. A novel backbone algorithm allows us to dispatch in real time thousands of taxis serving more than 25,000 customers per hour in New York City. Historical simulations shows that our algorithms improve upon the performance of existing heuristics in real-world settings. 362F Joint Session RAS/Practice: Roundtable A: State of Driverless Technology – Trucking and Railways Sponsored: Railway Applications Sponsored Session Chair: Dharma Acharya, GE Transportation, GE Transportation, Ponte Vedra, FL, 32081, United States, dharma.acharya@ge.com 1 - The Road to an Automated Transportation Industry Douglas Mattenburg, J.B. Hunt, Lowell, AZ, United States, Douglas.Mettenburg@jbhunt.com Adaptation is a key part of business success, and the transportation industry is set for a big shake-up. How will autonomous vehicles affect the transportation industry? Adjusting to this change will make-or-break many of the trucking companies out there today. Come hear how J.B. Hunt is preparing for this modernization of the industry. J.B. Hunt is a Fortune 500 supply chain management company that provides technology-driven solutions through a variety of services, including intermodal, dedicated, refrigerated, truckload, less- than-truckload, flatbed, single source, final mile, and more. MC59

362D Simulation in Healthcare Sponsored: Simulation Sponsored Session

Chair: Tahir Ekin, Texas State University, t_e18@txstate.edu 1 - Minimizing Patient Waiting Time at a Pure Walk-in Clinic Eduardo Perez, Texas State University, Roy F. Mitte Complex, 749 N. Comanche St., San Marcos, TX, 78666, United States, eduardopr@txstate.edu, Vivekanand Anandhan Walk-in clinics have grown in popularity in the United States as a substitute for traditional medical care delivered in primary care clinics and emergency rooms. The open access nature and lack of patient scheduling can lead to long wait times for patients or long periods of idle time for providers. In this talk, we derive a discrete event simulation model to study pure walk-in clinics where patients are served without appointments. A case study is discussed that consider a walk-in clinic located in central Texas. The computational study provides useful insights that are applicable to any walk-in health care facility. 2 - A JaamSim Model of Medical Evacuation and Treatment in Combat Lawrence Fulton, Texas Tech University, 911 Stoney Ridge Road, Bulverde, TX, 78163, United States, lfulton159@gmail.com, Nathaniel D. Bastian A JaamSim discrete event simulation (DES) of medical treatment and hospitalization requirements for a simulated, large-scale combat operation is presented. Simulation excursions illustrate the inherent trade-offs among the number of ground / air evacuation assets, the location of treatment / evacuation assets, and the type of treatment provided by location. The DES model is compared to a simpler Monte Carlo simulation of the same scenario. Results of these models are being used to support military force structure decisions. 3 - Multivariate Spatial Models for Substance Abuse Related Incidences Rasim M.Musal, Texas State University, School of Business, 601 University Drive, San Marcos, TX, 78666, United States, mmusal@gmail.com We investigate the relationship between correlated substance abuse incidences and socio-economic covariates within the counties of New York State. 4 - Medical Fraud and Abuse Assessment using Prescriber-level Medicare Data Babak Zafari, Babson College, Babson Hall 218C, Babson Park, MA, 02457, United States, bzafari@babson.edu, Tahir Ekin The existence and magnitude of fraud in health insurance programs, requires the utilization of fraud prevention and detection procedures. Data Mining methods are used to uncover odd billing patterns in large databases of health care claims history. In this work, we propose data-driven methods to find the anomalies in the medical claims data using publicly available CMS Part D Prescriber dataset. In so doing, we introduce a set of new indicators that can better separate providers with unusual billing patterns.

MC58

362E Route Optimization for Vehicle Fleet Sponsored: Transportation Science & Logistics Sponsored Session

Chair: Sebastien Martin, Massachusetts Institute of Technology, 70 Pacific Street, Apt 462A, Cambridge, MA, 02139, United States, semartin@mit.edu 1 - Clustering Algorithms for Vehicle Routing Problems Castaing Jeremy, Applied Research Scientist, LLamasoft, Ann Arbor, MI, United States, jeremy.castaing@llamasoft.com As the typical size of vehicle routing problems (VRP) solved in the industry keeps increasing, we need to develop ways to partition the data into smaller sub- problems. We describe several variants of clustering algorithms adapted to VRP that are used in the Transportation Optimization engine of Supply Chain Guru, developed by LLamasoft. These clustering methods allow to reduce the size of the problem with minimal loss of solution quality. We consider several VRP features such as different asset types, time windows or continuous pick-ups and deliveries. 2 - Dispatching Strategy for Autonomous Taxi Fleets Barry R. Cobb, Associate Professor, Missouri State University, Department of Marketing, 216 Glass Hall, Springfield, MO, 65897, United States, barrycobb@missouristate.edu, Linda Li We build a real time dispactching model for taxi fleets, assuming total backlogging if the customers not immediately served. We show if the assignment and

223