Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SC52

4 - Packaged Gas Supply Chain Planning with SKU Rationalization: Customer-selection and Incentivization-based Methodologies Ethan Malinowski, University at Buffalo (SUNY)Buffalo, NY, 14216, United States, Mark Henry Karwan, Lei Sun We present a multi-component project, applying SKU rationalization (in the form of a variant of product substitution) towards a traditional supply chain planning problem including production, allocation, and distribution decisions. A customer- selection heuristic is developed and shown to perform very well compared to a full MILP formulation, which includes customer selection for substitution as an explicit decision. Lastly, incentivization strategies are studied in an effort to reduce lost demand via product substitution, while simultaneously ensuring adequate business profit. n SC50 North Bldg 231A Joint Session Practice/Practice Curated: Edelman Reprise I Sponsored: INFORMS Section on Practice (formerly CPMS) Sponsored Session Chair: Anne G. Robinson, Verizon Wireless, Basking Ridge, NJ, 07920, United States Co-Chair: Carrie Beam, University of Arkansas, Walnut Creek, CA, 94596, United States 1 - China National Petroleum Zuo-Jun Max Shen, University of California Berkeley, 4141 Etcheverry Hall, Mail code 1777, Berkeley, CA, 94720-1777, United States Abstract not available. 2 - Turner Blazes a Trail for Audience Targeting on Television with Operations Research and Advanced Analytics J. Antonio Carbajal, Turner Broadcasting System, Inc., Atlanta, GA, USA Turner has designed and implemented innovative and integrated forecasting and optimization models that power audience targeting solutions disrupting decades- old paradigms and business processes in the media industry, and producing significant sales and advertisement efficiencies for Turner and its clients. Turner is on track to sell 50 percent of its inventory through audience targeting by 2020, representing billions in ad revenue. 3 - Europcar Integrates Forecast, Simulation and Optimization Techniques in a Capacity and Revenue Management System Europcar, Y, Aruba. Europcar, the leading European car rental company, partnered with ACT Operations Research to create Opticar, a complex decision support system. Opticar features forecasts, discrete event simulations and optimization techniques providing an integrated approach to revenue and capacity management. Opticar anticipates future demand for Europcar’s vehicles fleet, up to six months in advance, improving capacity management. In addition, Opticar enables Europcar to optimize its approach to revenue management and rental pricing, taking into account competitors information, the currently available fleet and expected demand for vehicles. Opticar provides a shared mathematical approach used as a starting point for all daily operations to nine Europcar’s corporate countries. n SC51 North Bldg 231B Joint Session OMS/Practice Curated: Applications in Scheduling Emerging Topic: Project Management and Scheduling, in Memory of Joe Leung, Emerging Topic Session Chair: Rodrigo A. Carrasco, Ph.D., Universidad Adolfo Ibáñez, Santiago, Chile 1 - A Scheduling Problem Motivated by Cybersecurity and Adaptive Machine Learning

also find another interpretation that applies this problem to adaptive machine learning. We motivate and define the problem, give some preliminary complexity results, and discuss practical (in)tractability. 2 - The Value of Flexibility in Bottling Operations in the Wine Industry under Sequence Dependent Setup Times Alejandro Francisco Mac Cawley, Pontificia Universidad Catolica de Chile, Vicuna Mackenna 4860, Santiago, 7820436, Chile, Sergio Maturana, Mauricio Varas Wineries must optimize their bottling process and deal with many products to be processed, high-demand variability and sequence dependent setup times. Under these conditions, managers must generate scheduling plans. We look at flexible strategies such as the postponement of the labeling of bottled wines; to gain productivity in the process. We study the performance impact of implementing flexible strategies by developing a multi-stage mixed-integer stochastic programming model with full recourse for demand scenarios. Results show benefits of implementing flexible strategies under given capacity of the system, demand variability and setup times. 3 - Scheduling Surgeries with Variable Times: The Value of More Data Rodrigo A. Carrasco, Diagonal Las Torres 2640, Of. 532, Edicifio C, Santiago, 7941169, Chile Although operating room scheduling has been studied since the early 60’s, dealing with surgery variability has been one of the main difficulties when implementing scheduling tools. In this work we present a way of incorporating such variability by using chance constraints to improve schedule performance. We develop specific constraints, through data analysis of real instances, which improve the schedule, reducing the need for overtime but without affecting the utilization significantly.. 4 - Bicriteria Job Scheduling with Split Lots Rasaratnam Logendran, Professor, Oregon State University, School of Mech lndustrial & Mfgr Engr, Rogers Hall Rm 204, Corvallis, OR, 97331-6001, United States, Ayush R. Aryal An algorithm, which incorporates tabu search into the framework of path relinking, is presented to solve a job shop scheduling problem, consisting of M = {1,?., m} machines. Each job j e J = {1, ?, n} consists of i(j) ordered operations. During an operation, a job is split or merged into ki(j) split-lots. Each split-lot operation can be performed by a subset of machines in M. The production is discontinuous within a two-week planning horizon. The objective function focuses on minimizing a linear combination of weighted flowtime and weighted tardiness. Three initial solution finding mechanisms are used and their comparative performance in the algorithm is evaluated using various problem instances. n SC52 North Bldg 231C Section Leaders & Research on Opinions and Facts Emerging Topic: Social Media Analytics Emerging Topic Session Chair: Theodore T. Allen, Ohio State University, Columbus, OH, 43210- 1271, United States 1 - Determining Twitter Users Opinions using Research Evan Munson, Air Force Institute of Technology, Wright Patterson AFB, OH, 45385, United States, Christopher M. Smith The rise in popularity of social media has changed the World Wide Web from a static repository to a dynamic forum for anyone to voice theiropinion across the globe. People have become increasingly used to sharing their opinions on social media platforms. Harvested data, analyzed for opinions andsentiment can provide insight into a population. This research utilizes Twitter data to examine the sentiment associated with tweets. An approach utilizing LatentDirichlet Allocation topic modeling was utilized to differentiate between tweet topics. A lexicographical dictionary was then utilized to classify sentiment. Thismethod provides insight into the sentiment contained within the Twitter data. 2 - Comparative Analysis of Information Flow and Rumor Debunking and Spreading through Twitter During Hurricanes Harvey and Irma Kyle Hunt, University at Buffalo, SUNY, 317 Bell Hall, Buffalo, NY, 14260, United States, Jun Zhuang The topic of the comparative analysis of information flow and rumor debunking is studied both empirically and theoretically.

Nourhan Sakr, Columbia University, New York, NY, 10027, United States, Clifford Stein, Ojas Parekh, Cynthia Phillips, Vladlena Powers

We consider a multiple-machine problem where each machine is associated with its predetermined sequence of jobs. A machine may start a new job if it is given a signal (a “take”) to do so. A take is global across all machines. Given a fixed budget of takes, we schedule takes to minimize total idle time. This problem comes from a stochastic-programming exploration of a cybersecurity game. We

79