Informs Annual Meeting 2017

TA59

INFORMS Houston – 2017

4 - Cost-effective Interpretable Treatment Regimes Cynthia Rudin, Duke University, Durham, NC, 27708, United States, cynthia@cs.duke.edu, Himabindu Lakkaraju I will present work on cost-effective interpretable treatment regimes. This is a policy design problem that involves estimating counterfactuals and using tools from interpretable machine learning.

2 - A Junction Traffic Coordination in the Large Scale Act Network Ek Peng Chew, National University of Singapore, 10 Kent Ridge Crescent, Dept of Industrial and Systems Engineering, Singapore, 119260, Singapore, isecep@nus.edu.sg In the mega transshipment automated container terminal (ACT), there are many potential conflicts and congestion occur in the junctions because of interactions among traffic flows from different direction. We present a scheme for control of AGV traffic flow and coordination of junctions to improve the operation performance in terms of container throughput. 3 - Module Based AGV Routing Algorithms for Automated Container Terminal Loo Hay Lee, National University of Singapore, 10 Kent Ridge Cresent, Industrial and Systems Engineering, Singapore, 119260, Singapore, iseleelh@nus.edu.sg, Ek Peng Chew, Qitong Zhao In this talk, we will present a module based routing algorithms for AGV for Automated Container Terminal. This routing algorithm consists of two level planning. The high level planning will determine the rough routing for each AGV by considering congestion, while the low level planning looks into the inner module control which avoids the deadlock for a small section of road. 4 - Developing Reshuffling Free Container Stacking Operations Amir Gharehgozli, Texas A.& M.University, P.O. Box 1675, Galveston, TX, 77553, United States, gharehga@tamug.edu, Nima Zaerpour This paper addresses a highly-researched area, the reshuffling problem in container terminals. Container stacking and reshuffling operations can cause ship delays and additional risk. In deep-sea terminals, outbound containers are tightly stacked according to the retrieval sequence. Due to lack of space, terminals stack containers in multiple tiers. This means any delay in the arrival of a ship can impose extra handlings and reshuffling of containers delaying future cargo handling. To address this issue, this study proposes an alternative stacking policy allowing different container types to share the same pile. At the same time, we avoid reshuffling. We aim to minimize the total retrieval time. 362F Joint Session RAS/Practice: Passenger and High Density Railway Corridor OR/MS Sponsored: Railway Applications Sponsored Session Chair: Shane Wu, National Railroad Passenger Corporation (Amtrak), Washington, DC, 20002, United States, SWu@alumni.pitt.edu Co-Chair: Steven Harrod, Technical University of Denmark, Kgs. Lyngby, 2800, Denmark, stehar@dtu.dk 1 - CO2REOPT: Real-time Train Dispatching on the Iron Ore Line Lukas Bach, SINTEF, Forskningsveien 1, 0314 Oslo, Oslo, 0314, Norway, lukas_bach@hotmail.com, Carlo Mannino On a congested critical railway corridor, the Iron Ore line between Narvik (Norway) and Lluleå (Sweden), we investigate if optimized train dispatching can increase the capacity of the corridor such that more trains can be operated while reducing delays. This is achieved by applying an exact mathematical model as real-time decision support to the dispatchers, given a railway infrastructure and its status, a set of trains and their current position, we find a route for every train from its current position to the destination and a disposition timetable to minimize the cost function. 2 - CO2REOPT: Integrated Planning for Multimodal Networks with Stochastic Demand and Customer Service Requirements Ioannis Fragkos, Assistant Professor, Rotterdam School of Management, Kralingse Plaslaan 120 bgg, Rottredam, Netherlands, Fragkos@rsm.nl, Joris Wagenaar, Rob A. Zuidwijk Multimodal networks use multiple modes of transport to carry freight from its origin to its destination. Such networks are more cost efficient than their single- mode counterparts, but their planning is more challenging, as resources have to be coordinated across several modes of transport. Motivated by the operations of a Dutch provider, we integrate strategic, tactical and operational decisions in a single framework, which, on its original form, is intractable. By utilizing a careful reformulation, we develop decomposition heuristics and lower bounding procedures. Our experiments demonstrate that good solutions for realistic instances can be obtained within reasonable time. TA59

TA57

362D Simulation Optimization and Ranking and Selection Sponsored: Simulation Sponsored Session Chair: Demet Batur, University of Nebraska-Lincoln, dbatur@unl.edu 1 - Ranking and Selection as Stochastic Control Yijie Peng, George Mason University, Fairfax, VA, United States, ypeng10@gmu.edu, Edwin P.Chong, Chun-Hung Chen, Michael Fu We propose a new framework for ranking and selection (R&S). The speaker shows an unexpected phenomenon that many popular methods in R&S decrease the probability of correct selection (PCS), which is caused by the imperfect theoretical foundation of these methods. To address the problem, R&S is formulated as a stochastic control problem and efficiently solved in an approximate dynamic programming (ADP) paradigm. Two ADP approaches are provided. One using a single feature of the value function sequentially achieves an asymptotically optimal sampling ratio that cannot be achieved by the many existing methods. Another ADP approach using two features avoids the non- monotonicity of the PCS. 2 - Quantile Based Selection Demet Batur, University of Nebraska-Lincoln, Department of Management, CBA 209, Lincoln, NE, 68588-0491, United States, dbatur@unl.edu, Fred Choobineh We study comparison of stochastic simulated systems based on a quantile performance measure. An indifference-zone quantile selection procedure is presented. 3 - Tractable Sampling Strategies for Ordinal Optimization Dongwook Shin, Hong Kong University of Science and Technology, Kowloon, Hong Kong, dwshin@ust.hk Dongwook Shin, Columbia University, New York, NY, 10027, United States, dwshin@ust.hk, Mark Nathan Broadie, Assaf Zeevi We consider a problem of ordinal optimization when the objective is selecting the best of several competing alternatives (“systems”), where probability distributions governing each system’s performance are not known, but can be learned via sampling. The objective is to dynamically allocate a finite sampling budget to minimize the probability of selecting a system that is not the best. The aforementioned objective does not possess an analytically tractable solution. We introduce a family of practically implementable sampling policies and show that the performance exhibits (asymptotically) near-optimal performance. 362E Shipping and Port Operations of the Future Sponsored: TSL, Facility Logistics Sponsored Session Chair: Amir Gharehgozli, Texas A&M University at Galveston, Galveston, TX, 77553, United States, gharehga@tamug.edu 1 - Integrated Planning of Ship Waterway Traversals and Berthing at Maritime Ports using Resource-constrained Project Scheduling Alessandro Hill, Universidad Adolfo Ibanez, Santiago, Chile, alessandro.hill@uai.cl, Marcos Goycoolea, Eduardo Lalla-Ruiz, Stefan Voss We consider short-term scheduling of ship traversals of waterways that connect the seacoast with the actual port area in accordance with arrival and departure times. Side constraints arising from the port’s geographical situation, traffic and the influence of the tides make this problem challenging when minimizing the total ships’ turnaround times. We extend this problem for including ship berthing operations, which have to be completed within pre-assigned time windows, and we show that this problem can be expressed as a multi-mode resource- constrained project scheduling problem. Using mathematical programming, we improve previous results for real-world instances. TA58

291