INFORMS 2021 Program Book

INFORMS Anaheim 2021

MD08

2 - Comparing the Effectiveness of Different Train Following Control Algorithms for Freight Trains in Moving Block Geordie Roscoe, University of Illinois at Urbana-Champaign, IL, 61801-2350, United States, Tyler Dick To create capacity for a projected increase in freight rail transportation demand of 24% by 2045, advanced train traffic control systems utilizing Positive Train Control (PTC) infrastructure, including moving block systems, may be developed and utilized. Moving block systems increase network capacity through decreasing train spacing compared to existing fixed block systems. However, running trains very near to their individual PTC braking distances requires quick and precise control actions to avoid both PTC penalty braking applications and uncontrolled oscillatory action between closely following trains. This research develops and evaluates several train following control algorithms that are derived from highway vehicle following literature as well as first principles. 3 - How Positive Train Control Can Help a Railroad Run on Schedule With Precision, Safety, Efficiency, and Profitability and Please its Customers Steven R. Ditmeyer, Principal, Transportation Technology and Economics, Alexandria, VA, United States PTC systems, comprised of digital radios, GPS receivers, sensors, and on-board and control center computers, generate continuous, real-time information on the location and speed of trains. This enables dispatchers and their computers to plan improved meets and passes of trains, thus shortening, and improving the predictability of, their running times. This, in turn, increases the capacity of the tracks while improving the utilization of assets locomotives, cars, and crews. Railroads are using the continuous, real-time information on train location and speed to improve the safety of train operations, but as yet are not using it to improve the efficiency of train operations. 4 - Designing and Rolling Out Movement Planners for European and North American Operations: A Comparison Leonardo Lamorgese, Head of R&D, Optrail, Rome, Italy Numerous projects and undertakings around the world indicate how the rail industry is increasingly aware of the potential of movement planning technology and the opportunity it presents in terms of increasing efficiency in operations. In this talk we discuss European and North American railroading: similarities, differences and the mathematical challenges faced in modelling them within a movement planner. We also present our first-hand experience in tackling these challenges and deploying state-of-the-art movement planners in both contexts. MD07 CC Room 303B In Person: Social Media Analytics: New Techniques and Applications General Session Chair: Tung Cu, Northeastern Illinois University, Chicago, IL, 60625- 4625, United States 1 - Abuse of the Fact-checking Hashtag on Twitter Yuchen Liu, University of Washington, Seattle, WA, 98105, United States, Elina Hwang Twitter uses social technology such as hashtags to enable users to exchange information. However, how does the fact-checking hashtag affect the cascade of true and false information remains unknown. Studying the COVID-19 pandemic, we empirically investigate the effect of the fact-checking hashtag on the lifespan of true and false tweets. We find that tweets with the fact-checking hashtag have an 8% lower hazard of stopping cascading. However, there is no significant difference between the effects of the hashtag of the hazard stopping the true tweets and the false tweets cascading. Our results caution the existence of the abuse of the fact-checking hashtag in large online social platforms. 2 - Video Sponsorship in Social Media: A Video Content Analysis Shahryar Doosti, Chapman University, Argyros School Of Business And Economics, CA, 92866-1005, United States With the growth of social media contents, video sponsorship has become a popular form of advertising for brands. It is imperative to understand how various marketing strategies affect user engagement with sponsored contents. Using a rich dataset of sponsored videos on a popular social media platform, we implement video content analysis (VCA) to extract key information from the sponsored video. We investigate the role of sponsorship strategies in content dissemination on the social network.

MD08 CC Room 303C Wagner Session Award Session

MD08 1 - Data-Driven Optimization for Atlanta Police Zone Design Shixiang Zhu, Georgia Institute of Technology, Marietta, GA, 30067, United States, He Wang, Yao Xie We present a data-driven optimization framework for redesigning police patrol zones in an urban environment. The objectives are to rebalance police workload among geographical areas and to reduce response time to emergency calls. Our proposed design was implemented by the Atlanta Police Department in March 2019. By analyzing data before and after the zone redesign, we show that the new design has reduced the response time to high priority 911 calls by 5.8% and the imbalance of police workload among different zones by 43%. 2 - Solving the Ride-sharing Productivity Paradox: Priority Dispatch and Optimal Priority Sets Garrett J. van Ryzin; Amazon Corporate LLC, USA. Transportation Network Companies are often tasked with balancing rideshare markets as they fluctuate between different marketplace balance conditions. During the Covid-19 pandemic, a particularly salient problem has been ‘oversup- ply’, where there is an excess of active drivers relative to the stream of passenger requests. The Priority Mode Controller is a novel solution to this problem, granti- ng a certain subset of drivers access to prioritized dispatch. In this paper, we dis- cuss the algorithmic and logical foundations of this product, rolled out successful- ly by Lyft throughout North America. 3 - Increasing Chip Availability Through a New After-Sales Service Supply Concept at ASML Douniel Lamghari-Idrissi, ASML and Eindhoven University of Technology, Netherlands, 2Eindhoven University of Technology, Netherlands, Rob Basten, Geert-Jan van Houtum ASML is the market leader in the photolithography sector. A joint project of ASML and Eindhoven University of Technology began in January 2017 to reform ASML’s service supply concept driven by the increased focus of its customers on extreme long downtimes. Changes were made to the service measure and the planning approach. They resulted in a new service concept rolled out to ASML’s complete installed base worldwide after successful pilots. The new concept result- ed in a yearly benefit of 1.5 Billion euros across the semiconductor industry and increased sustainability. 4 - Collaborating with Local and Federal Law Enforcement for Disrupting Sex Trafficking Networks Nickolas K. Freeman, Burcu B. Keskin, Gregory J. Bott; University of Alabama, USA. In the U.S., 95% of human trafficking cases are associated with sex trafficking, and the majority of these cases were facilitated by the use of the internet, in the form of classified sexual service ads. Collaborating with the West Alabama Human Trafficking Task Force, Department of Homeland Security, and Long Island Police Department, UA researchers developed an analytical tool to surveil multiple websites, differentiate fake and real ads, detect sex trafficking networks, and plan interdiction operations. The tool used in undercover stings and led to identifying victims and several arrests. 5 - Interpretable OR for High-Stakes Decisions: Designing the Greek COVID-19 Testing System Hamsa Sridhar Bastani, Wharton School, USA, Kimon Drakopoulos, Vishal Gupta After the first wave of the COVID-19 pandemic abated, countries sought to ease restrictions on non-essential travel while still safeguarding public health. In col- laboration with Greece, we designed and deployed a national-scale targeted test- ing system to dynamically allocate Greece’s limiting testing resources to screen visitors at the border for COVID-19. The system combines responsive supply- chain design, a novel empirical Bayes estimation strategy, and a new batched, multi-armed bandit algorithm with constrained actions and delayed feedback. We detail the method and impact of our system.

75