2016 INFORMS Annual Meeting Program

MB62

INFORMS Nashville – 2016

MB62 Cumberland 4- Omni Data Mining in Air Transportation Sponsored: Aviation Applications Sponsored Session Chair: Yi Liu, University of California, 107 McLaughlin Hall, Berkeley, CA, 94720, United States, liuyi.feier@gmail.com 1 - Using Historical Data To Support Traffic Management Initiative Decisions Alexander Estes, University of Maryland-College Park, 3117 A.V. Williams, University of Maryland-College Park, College Park, MD, United States, aestes@math.umd.edu, David J Lovell, Michael O Ball There is a large amount of data collected about traffic management initiatives that have been taken in the past by the Federal Aviation Administration. While this information could be helpful to decision-makers that are attempting to plan traffic management initiatives, this data is currently not very accessible. We propose unsupervised learning methods that identify relevant data and present it to the decision-makers. 2 - Identifying Similar Days To Guide Traffic Management Decision Making Sreeta Gorripaty, UC Berkeley, Berkeley, CA, 94703, United States, gorripaty@berkeley.edu, Mark M Hansen The experience of traffic flow management specialists is crucial in managing airport operations efficiently. Historical airport operations data can assist decision- making by augmenting controller experience with a systematic record of past traffic management actions under similar conditions and their consequences. A decision-support tool that finds historical days similar to a query day can guide day-of-operation decisions and assess past performance. Using machine-learning algorithms, we learn a similarity measure between two days based on weather and demand data. This measure is assessed for accuracy using airport operational outcomes and traffic management initiatives (TMI) data. 3 - Impacts Of Airline Mergers On Consumer Welfare Vikrant Vaze, Assistant Professor, Dartmouth College, 14 Engineering Drive, Hanover, NH, 03755, United States, Vikrant.S.Vaze@dartmouth.edu, Tian Luo We used publicly available passenger flows data and a discrete choice modeling framework to examine welfare changes due to the five major airline mergers in the past decade in the United States. We found that consolidations of legacy airlines with significantly overlapping markets generally increased the passenger welfare. However, overall passenger welfare in small communities declined after the two mergers whose small community markets data sets are sufficiently large for our analyses. We also found that welfare of passengers, traveling to or from hub airports of the primary merging airline, increased significantly. 4 - Hourly Ground Delay Program Prediction With Local And Convective Weather Variables Mark M Hansen, University of California - Berkeley, 114 McLaughlin Hall, Berkeley, CA, 94720, United States, mhansen@ce.berkeley.edu, Yi Liu, Danqing Zhang, Alexey Pozdnukhov We propose a method to predict ground delay program status for each hour using data mining techniques. We use local and convective weather variables as our predictors. We apply the method to 5 top-traffic US airports. The results include prediction performance, variable importance analysis and convective weather weight map. MB63 Cumberland 5- Omni HUB Location Sponsored: Location Analysis Sponsored Session Chair: Sibel Alumur Alev, University of Waterloo, 200 University Avenue West, Waterloo, ON, N2L 3G1, Canada, sibel.alumur@uwaterloo.ca 1 - Modeling Congestion And Service Time In Hub Location Problems Stefan Nickel, Karlsruhe Institute of Technology, stefan.nickel@kit.edu, Sibel Alumur Alev, Brita Rohrbeck, Francisco Saldanha-da-Gama In this paper, we present a modeling framework for hub location problems with a service time limit considering congestion at hubs. Service time is modeled taking the traveling time on the hub network as well as the handling time and the delay caused by congestion at hubs into account. We develop mixed-integer linear

programming formulations for the single and multiple allocation versions of this problem. We further extend the multiple allocation model with a possibility of direct shipments. We test our models on the well-known AP data set and analyze the effects of congestion and service time on costs and hub network design. 2 - An Enhanced Milp Model For Stochastic Multi-period Multiple Allocation Hub Location Francisco Saldanha-da-Gama, University of Lisbon, Lisbon, Portugal, fsgama@ciencias.ulisboa.pt Isabel Correia, Stefan Nickel A two-stage stochastic programming modeling framework is proposed for a pure phase-in multi-period multiple allocation hub location problem. Stochasticity is associated with the demands. Assuming a finite support for the underlying random vector, a compact formulation can be derived for the extensive form of the deterministic equivalent, which leads to a large-scale mixed-integer linear optimization problem. By considering 4 sets of valid inequalities, the model is enhanced, which makes it possible to solve to optimality by means of a general solver, instances that could not be tackled when the original formulation was considered. Results obtained using the CAB data set are reported. 3 - Modeling Hub Location Problems James F Campbell, University of Missouri-St Louis, campbell@umsl.edu, Sibel Alumur Alev In this talk, we focus on some of the key features of hub location models such as demands, costs, economies of scale, capacity and service level constraints, and network topologies including single and multiple allocation and complete and incomplete inter-hub networks. We discuss some of the possible implications of how these features are modeled on the hub locations, network design and performance measures, and emphasize the characteristics within the context of different applications. Additionally, we identify some distinguishing properties of the CAB, AP, and TR data sets commonly used in hub location studies. We aim to provide a road map for future hub location research. MB64 Cumberland 6- Omni MCDM for System Design: No Camels Allowed Sponsored: Multiple Criteria Decision Making Sponsored Session Chair: Stephen Henry, Sandia National Laboratories, P.O. Box 5800, Albuquerque, NM, 87185, United States, smhenry@sandia.gov 1 - Multiple Criteria Decision Making For The United States Army’s Robotic Pack Mule Design Lucas Waddell, Sandia National Laboratories, lawadde@sandia.gov Stephen Henry The U.S. Army has a strong interest in the development and deployment of unmanned ground systems (UGSs) to provide soldiers with many unique battlefield advantages. As with any new, complex system, UGSs present a vast array of design tradeoffs, interdependencies, and competing stakeholder goals. There is an extremely fine line between design solutions that represent fair compromises, and solutions that nobody is willing to accept. We present lessons learned from an in-depth trade study performed for the U.S. Army on the Squad Multipurpose Equipment Transport (SMET) UGS. 2 - Infrastructure Equipment Optimization For United States Military Contingency Base Designs Alexander Dessanti, Sandia National Laboratories, Albuquerque, NM, United States, adessan@sandia.gov Contingency bases provide temporary facilities from which deployed forces can operate overseas. These bases are large-scale complex systems with many interrelated functions, making it difficult to identify the best equipment to utilize when designing a new one without an optimization approach. The Whole System Trades Analysis Tool (WSTAT), a multi-objective optimization capability, is being applied to this problem for the U.S. Army to enable more efficient and affordable future contingency base designs. Focus will be on the unique technical challenges presented by a problem of this magnitude. 3 - Ultra-high Dimensional Optimization For Military Systems Requirements Negotiation Stephen Henry, Sandia National Laboratories, smhenry@sandia.gov Complex military system design begins long before the welding of metal. A key early step is the development of a set of requirements - performance levels in various categories that must be achieved by the new system. These requirements are typically drafted by separate communities of experts and lack an analytic mechanism to address incompatibilities between the individual requirements - often leading to program cancellation due to unmet requirements. In this talk, we present a new tool for ensuring holistically feasible requirements. We discuss the mathematical challenges of high-dimensional optimization (>30 objectives) as well as the human challenges of real-time requirements negotiation.

170