INFORMS 2021 Program Book
INFORMS Anaheim 2021
TC35
2 - Algorithmic Foundations for Multi-modal Transit Systems (Helping Better Buses Make Better Cities) Samitha Samaranayake, Cornell University, School of Civil & Environmental Engineering, Ithaca, NY, 14853, United States, Siddhartha Banerjee, Chamsi Hssaine, Noemie Perivier Emerging mobility services have disrupted the urban transportation ecosystem and instilled hope that new data-driven mobility solutions can improve personal mobility for all. While these apps provide a valuable service, as evident by their popularity, there are many questions regarding their scalability, efficiency, impact on equity, and negative externalities (e.g. congestion). On the other hand, traditional public transit systems provide affordable and community-oriented access to personal mobility, but have their own operational limitations. This talk will focus on the algorithmic foundations of integrating public transit operations with agile, demand-responsive services to enable personal mobility for all. TC37 CC Room 210C In Person: Military Applications General Session Chair: Gregory S Parnell, University of Arkansas, Fayetteville, AR, 72701, United States 1 - Using Neural Hidden Markov Models to Identify Insider Threat Behavior David Elkind, Ephemer.ai Threats originating inside of an organization can take the form of an individual experiencing distress. Emotional or psychological distress has an established mutual causality with a disrupted sleep pattern. We develop an approach using firewall logs to infer an individual’s state (sleep, recreation, study, classroom instruction, etc.). Network traffic data powers a modified Hidden Markov Model that incorporates neural networks thereby creating a more flexible model not constrained by the Markov assumptions while simultaneously reflecting demographic data about individuals. 2 - Topology Optimization of Maritime Environmental Survey Operations Danielle Morey, University of Washington, Seattle, WA, United States, Randall Plate, Cherry Yu Wakayama, Zelda B. Zabinsky Topology optimization is a challenge for maritime environmental survey operations. Unmanned underwater vehicles (UUVs) are considered to collect and transport data from sensors to a centralized depot. Simulations exist to accurately model scenarios, but are computationally expensive for topology optimization. Two scenarios are defined: one uses UUVs to visit pre-defined locations for a single observation; the second involves sensors that collect data at a fixed rate to be sent to UUVs in a store-and-forward manner. We develop low-fidelity models for use in conjunction with a high-fidelity simulation to find a Pareto-optimal set of solutions with regards to latency and reliability. 3 - Optimization in Medium-term Planning for Military Aircraft Maintenance Sergio Rebouças, Brazilian Air Force, Sao Jose dos Campos, Brazil, Dennys Wallace Imbassahy, Fernando Teixeira Abrahão The assignment of military aircraft must be closely linked to the fleet maintenance plan. Optimization of maintenance activities and resources is crucial to maximize fleet availability and minimize the costs of air operations. Current optimization models generalize critical constraints that make their application in real-life difficult. This work proposes a new approach to medium-term optimization planning for military aircraft maintenance. Innovative constraints are defined and implemented through the Biased Random-Key Genetic Algorithm for optimization of the fleet maintenance plan, improving the model’s adherence to the operational context.
TC35 CC Room 210A In Person: Last-Mile Logistics for E-Commerce General Session Chair: Dipayan Banerjee, Georgia Institute of Technology, Atlanta, GA, 30318-5644, United States 1 - Marketplace Design for Crowdsourced Delivery Adam Behrendt, Georgia Tech, Atlanta, GA, United States, Martin W P Savelsbergh, He Wang Crowdsourced delivery platforms face the unique challenge of meeting dynamic customer demand using couriers not employed by the platform. As a result, the delivery capacity of the platform is uncertain. To reduce the uncertainty, the platform can offer a reward to couriers that agree to make deliveries for a specified period of time. We consider a crowdsourced courier scheduling problem in which a mix of scheduled and ad-hoc couriers is available to serve dynamically arriving orders. The platform’s objective is to determine shifts for scheduled couriers so as to minimize total courier payments and penalty costs for expired orders. We present a prescriptive machine learning method that combines simulation optimization for offline training and a neural network for online solution prescription. We validate this method using data from a crowdsourced delivery platform. 2 - Does Parking Matter in Routing Last-Mile Deliveries? Sara Reed, University of Kansas, University Of Iowa, Lawrence, KS, 52242, United States, Ann Melissa Campbell, Barrett Thomas Parking the delivery vehicle is a necessary component of traditional last-mile delivery practices but finding parking is often difficult. The Capacitated Delivery Problem with Parking (CDPP) is the problem of a delivery person needing to park the vehicle in order to service customers on foot. Unlike other models in the literature, the CDPP considers the search time for parking in the completion time of the delivery tour. We present valid inequalities and a variable reduction technique to solve this problem on realistically-sized instances. We compare the CDPP to industry practices as well as other models in the literature to understand how including the search time for parking impacts the completion time of the delivery tour. 3 - Who Has Access to E-Commerce and When? Time-Varying Service Regions in Same-Day Delivery Dipayan Banerjee, Georgia Institute of Technology, Atlanta, GA, 30318-5644, United States, Alexander Stroh, Alan Erera, Alejandro Toriello We study the tactical optimization of same-day delivery (SDD) systems under the assumption that service regions are allowed to vary over the course of each day. In most existing studies of last-mile logistics problems, service regions are assumed to be static. We use a continuous approximation approach and derive optimal dynamic service region areas and tactical vehicle dispatching policies that maximize the expected number of SDD orders served per day. We use these results to quantify the improvement in expected order fill rate when SDD service regions are allowed to vary. We discuss efficient solution algorithms, theoretical results, and issues related to equity and access within SDD systems. We illustrate and validate our models through computational studies set in the Phoenix, Arizona metropolitan area. TC36 CC Room 210B In Person: Multi-modal Transit Network Design Award Session Chair: Samitha Samaranayake, Cornell University, Ithaca, NY, 14853, United States 1 - On the Value of Demand-Responsiveness in Transit Systems Carlos Martinez Mori, Cornell University, Ithaca, NY, 14853, United States Transit systems traditionally operate fixed lines under fixed schedules. However, there is growing interest in demand-responsive transit systems, whereby operators complement their fixed services with dynamic services (e.g., microtransit) in real-time. In this work, we study fundamental benefits and limitations of demand-responsiveness on the overall performance of transit systems.
122
Made with FlippingBook Online newsletter creator