Informs Annual Meeting 2017

MB59

INFORMS Houston – 2017

2 - Evaluation Model of Traffic Network Utilizing Geographical Divergence Degree of Multiple Routes Motoki Tajima, The University of Tokyo, Tokyo, Japan, m-tajima@iis.u-tokyo.ac.jp, Yudai Honma In the research, we try to evaluate the traffic network in terms of geographical tolerance. Especially, we consider multiple alternate routes, and focus on divergence of them to calculate the robustness of networks. We firstly search alternate routes by adopting K shortest path routing algorithm. Next, we define several local factors which show the robustness quantitatively, then expand them to the whole network. 3 - Robust Design of Refueling Station Locations for Alternative Fuel Vehicles with Driving Range Constraint Samuel Labi, Purdue University, 550 Stadium Mall Drive, West Lafayette, IN, 47907, United States, labi@purdue.edu, Mohammad Miralinaghi his study develops a multi-period framework to investigate the optimal refueling locations for alternative fuel vehicles (AFVs) under travel demand uncertainty and driving range limitations. AFVs are classified by their driving range, and the uncertainty of travel demand in each AFV class is duly factored through a multi- class robust discrete optimization model. An algorithm is developed to solve the decision model for a large network. 362D Parallel Simulation Optimization Sponsored: Simulation Sponsored Session Chair: Jie Xu, George Mason University, Fairfax, VA, 22030, United States, jxu13@gmu.edu 1 - An Efficient Fully Sequential Selection Procedure Guaranteeing Probably Approximately Correct Selection Sijia Ma, PhD Student, Cornell University, 292 Rhodes Hall, Cornell University, Ithaca, NY, 14853, United States, sm2462@cornell.edu, Shane Henderson Ranking and Selection procedures are designed for selecting the best among a finite set of systems using stochastic simulation. Instead of assuming a known lower bound on the difference between the best and others, we consider the probably approximately correct selection (PAC) formulation. We present a new fully sequential selection procedure, called the Envelope Procedure (EP), which delivers PAC guarantee. A particular sampling rule that achieves good efficiency is proposed. We compare the efficiency of the EP with some existing procedures in numerical experiments, and the results show that the EP saves considerable computational effort in many problem configurations. 2 - Challenges in Parallel Ranking and Selection Susan R. Hunter, Purdue University, School of Industrial Engineering, Grissom Hall, 315 N. Grant Street, West Lafayette, IN, 47907-2023, United States, susanhunter@purdue.edu, Barry L. Nelson We discuss the current state of the art in R&S for large problems solved on parallel computing platforms. We provide a new stylized model for representing parallel R&S procedures, and discuss the mathematical and computational formulations of existing R&S procedures under the stylized model. We also speculate on the future of parallel R&S procedures. 3 - Simulation Based Optimization Towards High Performance Computing Haobin Li, Institute of High Performance Computing, MB57 This talk provides a brief introduction on how to speed up the simulation based optimization towards high computational performance. Specifically, three areas are to be explored, namely 1) designing a Pareto-efficient configuration for various computational infrastructures; 2) adapting the optimization algorithms to the parallel environment, e.g., OCBA algorithm with GPU computing; and 3) developing a modelling paradigm that could systematically distribute the computational effort to parallel cores, such as the modularized O2DES framework. The motivation for the three areas of studies is to be discussed, and preliminary results will be illustrated to show the potential in each area. A*STAR.Singapore, 1 Fusionopolis Way, #16-16 Connexis, Singapore, 138632, Singapore, lihb@ihpc.a-star.edu.sg, Giulia Pedrielli, Loo Hay Lee

4 - A Structured Approach for Parallel Simulation Optimization Implementation Ji Xu, George Mason University, Fairfax, VA, 22205, United States, jxu13@gmu.edu, Scott Rosen In this talk we present our framework for performing parallel simulation optimization that involves a library of simulation optimization techniques and a guidance system to assist the general analyst in using the library. We also present parallel adaptations of two simulation optimization algorithms: Industrial Strength COMPASS and Empirical Stochastic Branch and Bound. After presenting these algorithms we provide numerical performance results under a case study involving a large-scale, real-world simulation. In addition performance of these algorithms, with respect to speed and quality of the final solution, is compared to a metamodel-based optimization procedure. 362E Spatial Analysis in Transportation and Logistics Sponsored: Transportation Science & Logistics Sponsored Session Chair: EunSu Lee, PhD, New Jersey City University, Harborside 2, 200 Hudson Sttreet, Suite #234H, Jersey City, NJ, 07311, United States, ELee3@njcu.edu 1 - Forecasting Growth by Transit Oriented Developments using Spatial Analysis Nimish Dharmadhikari, INCOG, 2 W. 2nd street Suite 800, Tulsa, OK, 74103, United States, ndharmadhikari@incog.org Many cities are planning Transit Oriented Developments (TODs) in conjunction with the new transit projects. They expect the TODs to be the future growth corridors. This research proposes a methodology to forecast the growth happening with TODs. We use spatial analysis techniques to study the current land use and the future potential. This method will be useful for the cities to study the potential development. It will also generate a map of the growth centers and potential growth corridors to focus. 2 - Hyperloop Spatial Economic Analysis Dapeng Zhang, PhD, Hyperloop One, Los Angeles, CA, United States, zhangdapeng@live.com Hyperloop is a new way to move people and things at the airline speed with price of a bus ticket. Hyperloop can greatly contribution to spatial interaction which benefit for agglomeration benefits. This presentation will talk about what Hyperloop economic analysis is doing. MB58

MB59

362F Joint session RAS/Practice: Railroad OR Applications Showcase Sponsored: Railway Applications Sponsored Session

Chair: Kamalesh Somani, CSX Transportation, CSX Transportation, Jacksonville, FL, 32202, United States, kamalesh_somani@csx.com 1 - Railroad OR Applications Showcase Kamalesh Somani, CSX Transportation, CSX Transportation, Jacksonville, FL, 32202, United States, kamalesh_somani@csx.com Join us for an interactive look at the substantial use of advanced OR techniques in the railroad industry. Four major North American rail carriers (BNSF, CSX, NS, and Amtrak) will be onsite to give in-depth practical demonstrations of OR tools. Join us to learn how railroads implement robust solutions to complex business problems. The demonstrations will focus on the practical implementation of advanced OR models within companywide systems including the core software and technologies used, large scale data constraints, production level deployment, and business alignment.

185