Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SD26

3 - Dynamic Cloud Capacity Planning Jon M. Stauffer, Texas A&M University, 4217 TAMU, College Station, TX, 77843, United States, Aly Megahed, Chelliah Sriskandarajah, Ahmed Nazeem Many companies are shifting from on-premise IT environments to cloud computing. The computing application resources (instances) need to be adjusted according to the demand (queries) over time. Determining the optimal number of instances needed in a given planning horizon is challenging due to the combinatorial nature of the optimization problem involved. Deploying too many instances results in unnecessary deployment/capacity costs, while deploying too few instances results in penalties for not being able to process all incoming queries on-time. We propose heuristics to solve the offline scheduling problem and illustrate their effectiveness in a real-world case study. Undergraduate Operations Research Prize II Emerging Topic: Undergraduate Operations Research Prize Emerging Topic Session Chair: Jennifer A. Pazour, Rensselaer Polytechnic Institute, 110 8th street, CII 5217, Troy, NY, 12180, United States 1 - Evolving Contact Network Algorithm: A New Simulation Method for Modeling HIV, a Disease with Low Prevalence but a Critical Public Health Issue in the US Matthew Eden, University of Massachusetts-Amherst, Amherst, MA, United States, Buyannemekh Munkhbat, Chaitra Gopalappa, Hari Balasubramanian Agent-based network modeling (ABNM) is suitable for simulating multiple interacting events, related to contact network structures, behavior, and care access, for predicting epidemic trajectories. However, ABNM is infeasible to use for HIV due to disproportionate HIV prevalence across population groups in the US. We present a new modeling technique, evolving contact network algorithm (ECNA). Numerical results comparing ECNA with ABNM disease predictions are promising for low prevalence diseases. 2- Limited-memory Kelley’s Method Converges for Composite Convex and Submodular Objectives Swati Gupta, Georgia Institute of Technology, Atlanta, GA, United States Original simplicial method (OSM) has been shown to converge to the minimizer of composite convex and submodular objectives, though no rate of convergence was known. Moreover, OSM is required to solve subproblems in each iteration whose size grows linearly in the number of iterations. We propose L-KM, a novel limited memory version of OSM that maintains convergence to the optimum. We show that the dual method of L-KM is a special case of the Fully-Corrective Frank-Wolfe (FCFW) method with approximate correction, thereby deriving a limited memory version of FCFW (holds for general polytopes) and proving a rate of convergence L-KM. 3 - Assortment Optimization for a Multi-Stage Choice Model Yunzong Xu, Tsinghua University, Beijing, China, Zizhuo Wang Motivated by several practical selling scenarios that require previous purchases to unlock future options, we consider a multi-stage assortment optimization problem, where the seller makes sequential assortment decisions with commitment, and the customer makes sequential choices to maximize her expected utility. We show that this problem is polynomial-time solvable when the customer is fully myopic or fully forward-looking. The optimal policy entails that the assortment in each stage is revenue-ordered and a product with higher revenue always leads to a wider range of future options. We also show that the problem is NP-hard in general and give efficient algorithms for various scenarios. n SD27 North Bldg 132B Student-centered Experiential Education: Learning by Doing Sponsored: Education (INFORMED) Sponsored Session Chair: Chun-Miin Chen, Bucknell University, Lewisburg, PA, 17837, United States 1 - The Capstone Experience in Business Analytics Programs Andrew Urbaczewski, University of Denver, Daniels College of Business 593, 2101 S. University Boulevard, Denver, CO, 80208, United States n SD26 North Bldg 132A

This session will show experiential learning opportunities at both the undergraduate and graduate levels of the Business Analytics Programs at the University of Denver. Recruitment, NDA, delivery, and feedback will all be discussed. 2 - Introduction to Reverse Auctions: The BucknellAuto Game Chun-Miin Chen, Bucknell University, 95 Sunnyside Dr., Lewisburg, PA, 17837, United States This article presents a spreadsheet-based game, the BucknellAuto game, that simulates the bidding process in a reverse auction. The BucknellAuto game serves as a pedagogical tool for efficiently and effectively introducing the reverse auction to undergraduate students in a fun and interactive way. 3 - Optimization scheduling for Bobact Pizza: An Optimization Experiential Learning for Service Management Iddrisu Awudu, Quinnipiac University, Hamden, CT, 06514, United States, David Boison In this experiential learning experience,students consider a pizza making place (Bobcat Pizza) on a university campus. Students develop a simple optimization scheduling approach using excel as the main driver. The aim of this optimization scheduling is to enhance worker schedules to maintain a competitive urge for Bobcat Pizza over its competitors in the areas of local flavor and quality ingredients while implementing a system that minimizes wasted resources. Students collect data over a period of when the class was taught and proffered solutions for the said organization. n SD28 North Bldg 221A RAS Roundtable: Horsepower-hour in Today’s Railway Sponsored: Railway Applications Sponsored Session Chair: Michael F. Gorman, University of Dayton, Dayton, OH, 45469, United States Co-Chair: Shantih Marie Spanton, CSX, CSX, Jacksonville, FL, 32204, United States Railroads may define locomotive work performed over time in terms of horsepower-hours. This term is primarily used in the context of the work of locomotives shared between railroads, wherein the horsepower-hours of a locomotive used by another railroad must be reimbursed to the owner. The balancing the horsepower-hours owed to a foreign carrier with usage of a railroad’s own locomotive power is an interesting problem that raises both tactical and strategic questions. In this panel experts from several leading US rail carriers will discuss the impact of balancing horsepower-hours in today’s cost sensitive environment, including issues of fleet sizing, fleet mix (foreign or system), tactical/seasonal balancing strategies, rules and regulations around horsepower balancing in the US system, and more. Panelists Grant Metcalf, Kansas City Southern RR, Kansas city, MO, United States Kim Archer, CSX, CSX, Jacksonville, FL, 32201, United States Jermaine Wilkinson, Norfolk Southern Corporation, Norfolk n SD29 North Bldg 221B Demand and Revenue Management in Urban Transportation Sponsored: TSL/Urban Transportation Sponsored Session Chair: Xinchang Wang, Mississippi State University, Mississippi State, MS, 39762, United States 1 - Management of Shuttle Systems with Modular Vehicles under Oversaturated Demand Xiaopeng Li, University of South Florida, Eng 207, Tampa, FL, 33620, United States, Zhiwei Chen, Xuesong Zhou In this paper, we investigate the joint design problem of dispatch headway and vehicle capacity for one to one shuttle systems considering oversaturated demand. The first model we proposed is a mixed integer linear programming model that 1 - Roundtable: Horsepower-hour in Today’s Railway Shantih Marie Spanton, CSX, Jacksonville, FL, 32204, United States Southern Corporation, Norfolk, VA, United States Ted Smith, BNSF, BNSF, Ft. Worth, TX, United States

100