2016 INFORMS Annual Meeting Program

MA71

INFORMS Nashville – 2016

4 - A Vessel Scheduling Transportation-inventory Problem With Stochastic Demands Hossein M Soroush, Kuwait University, Department of Statistics & Operations Research, P.O. Box 5969, Safat, 13060, Kuwait, h.soroush@ku.edu.kw Salem Al-Yakoob We study a vessel scheduling transportation-inventory problem to transport a product from a source to a destination where demands are stochastic and penalties are imposed on the shortages/excesses in storage levels. The goal is to schedule a set of heterogeneous fleet to meet the demands with acceptable level of reliabilities while minimizing the expected total cost.

reflect a range of possible defender interceptor inventories. The ADP policy provides high-quality decisions for a substantial proportion of the state space, achieving a 7.7 percent mean optimality gap in the baseline scenario. 3 - Containing The Mess Of Optional Meals Via Approximate Dynamic Programming Sandra Jackson, Instructor, United States Military Academy, West Point, NY, United States, sandra.jackson@usma.edu, Keith DeGregory, Matthew Fletcher On any given day, the United States Military Academy Mess Hall provides three meals to the Corps of Cadets, approximately 4,400 people. These meals are simultaneously served meaning the entire Corps arrives, eats, and departs at the same time. In recent years, the Academy has allowed cadets to option out of formally mandatory meals thus moving what was a consistent demand to a stochastic one. As a result, the Academy Mess Hall experiences a stochastic demand similar to dining facilities in the active Army. Variable demand opens the door to food waste and the question of how to make sequential resourcing decisions under uncertainty, a problem for which approximate dynamic programming is suited to solve. 4 - Heterogeneous Surface-to-air Defense Battery Location: A Game Theoretic Approach We examine a game theoretic model for the location of air defense batteries having different interceptor capabilities, and we find high quality solutions using the game tree search technique Double Oracle, within which we embed either of two alternative heuristics to solve an important subproblem for the attacker. We test and compare these solution methods to solve a designed set of 52 instances having parametric variations. Enhancing the solution methods with alternative initialization strategies, our superlative methodology attains the optimal solution for 75% of the instances tested and solutions within 2.12% of optimal, on average, for the remaining 25% of the instances. MA70 Acoustic- Omni Transportation, Maritime I Contributed Session Chair: Hossein M Soroush, Kuwait University, Department of Statistics & Operations Research, PO Box 5969, Safat, 13060, Kuwait, h.soroush@ku.edu.kw 1 - Understanding Vehicle Movement Patterns With Artificial Neural Networks Burak Cankaya, Lamar University, 13960 Hillcroft St. Apt 724, Houston, TX, 77085, United States, mbcankaya@gmail.com Geographical Identification System (GIS) is utilized by most of the vehicles and cellphones in recent years. This research proposes an alternative type of methodology to understand vehicle movement patterns with historic geospatial data. This research investigates the vehicle movement patterns with artificial neural networks and compares the results with other machine learning methodologies including decision trees and random forest algorithm. The methodology will be applied on a case study, which is strategic Gulf of Mexico Ports’ vessel traffic data. The result of the study will explain the question “Can we understand vessel movement patterns and optimize the vessel traffic? 2 - A Discrete Simulation Of A New Container Terminal – The Case Of Hamad Port Of Qatar Ghaith Rabadi, Professor, Old Dominion University, Engineering Management Systems Engineering, Engineering Systems Building, Room 2102, Norfolk, VA, 23529, United States, grabadi@odu.edu Mariam Kotachi, Moahmed K Msakni, Mohammad Al-Salem, Ali Diabat A discrete even simulation is developed for a future container terminal of Hamad’s new port of Qatar. The simulation models vessel arrivals, ship to shore crane operation, container movement from vessels to yard via yard trucks and the operations in the opposite direction from the yard to the vessels. Furthermore, external trucks dropping off and picking up containers from the yard are also modeled. Berth allocation and crane assignment methods are embedded in the simulation. Preliminary analysis and scenarios are presented. 3 - Modeling The Service Network In Container Terminals Considering Process Integration And Decomposition Qingcheng Zeng, Professor, Dalian Maritime University, School of Transportation Management, Dalian, 116026, China, qzeng@dlmu.edu.cn Process integration/decomposition and process variation are pair of critical decision variables in service network design. In this paper, the variation of each service process of container terminals is analyzed. A cyclic queue network model is developed. Principles of integration or decomposition, methods of stabilizing the processes are proposed. Brian Lunday, Air Force Institute of Technology, brian.lunday@afit.edu Nicholas Boardman, Matthew Jd Robbins

MA71 Electric- Omni Supply Chain, Shipping I

Contributed Session Chair: Tao Lu, Erasmus University, Rotterdam, Netherlands, lutao0927@hotmail.com 1 - Shipping Peak Demand For Online Sellers Ju Myung (J.M.) Song, Rutgers Business School, PhD Program, Online retailing is changing the landscape of retail industry in countries as Amazon’s market cap has recently doubled that of Wal-Mart in the US. Different from brick and mortar, online sellers rely on 3rd party logistics for the delivery of the goods but the hugely spiked demand during holiday seasons (Christmas in the US, Singles’ day in China) poses a substantial challenge for the 3PLs to deliver on time. To better manage demand, 3PLs such as UPS, require the sellers to make reservation and to pay a surcharge for extra work. In this paper, we discuss how these shipping arrangements may affect the online sellers’ inventory decisions, how to coordinate the channel for the sellers and shippers to win-win. 2 - Explosive Storage Location Assignment Problem For Amazon Class Internet Fulfillment Warehouses Sanchoy Das, New Jersey Institute of Technology, University Heights, Newark, NJ, 07102, United States, das@njit.edu, Jingran Zhang We establish a storage assignment heuristic for Internet Fulfillment Warehouses. xSLAP is based on an explosive policy, whereby the same item is stored simultaneously in small lots in a large number of locations. Compared with classical storage policies used in traditional warehouses, an explosive policy leverages demand correlated storage assignment by commingling SKUs in the same bin, bins in close proximity, or bins in the same zone. The xSLAP heuristic optimizes the downstream picking processes by finding the optimal assignment of items in order to meet quick customer orders fulfillment. 3 - Analysis Of Hub Ports In Southeast Asia And Northeast Asia Washington Park, Room 430C, Newark, NJ, 07102, United States, jumyungsong@gmail.com, Yao Zhao Major seaports in Southeast Asia and Northeast Asia have experienced significant growth in the last two decades. Several ports are known as “hub” ports due to the dominant volume of transshipments. This paper will present descriptive statistics for the major ports in Asia among the Top 50 Container Ports in the world. A secondary analysis will focus on a smaller sample of the top hub ports to compare the growth of container volume for those ports with higher transshipment volumes. The differences between hub ports and other ports will also be discussed. 4 - Approximate Dynamic Programming For An Empty Container Repositioning Problem In A Cyclic Route Shaorui Zhou, Assistant Professor, Sun Yat-sen University, Guangzhou, China, zshaorui@gmail.com, Fan Wang In this work, we study an empty container repositioning problem in a cyclic route with uncertain demands. We formulate it as a stochastic dynamic programming problem. We study two special cases: in case 1, the route covers only 2 ports and we propose a optimal t ld policy due to the separability of the value function; In case 2, the route covers 3 ports, and the optimal policy can be characterized by state-dependent threshold points. For general case, in order to overcome curse of dimensionality, we propose an approximate dynamic programming algorithm. We compare the performance with heuristics. Numerical results demonstrate the efficiency of the algorithm. Richard W Monroe, Longwood University, 7413 Nicklaus Cir, Farmville, VA, 23909, United States, monroerw@longwood.edu

145