Informs Annual Meeting 2017

WC67

INFORMS Houston – 2017

2 - Robust Optimization for Service Network Design and Hub Location Problem under Uncertain Capacities and Demand Mengtong Wang, Tsinghua University, Zhongguancun Street, Beijing, 231567, China, wmt15@mails.tsinghua.edu.cn We propose a novel two-stage robust optimization model for the service network design and hub location problem with stochastic capacities and demand, in which a freight operator needs to determine the locations and capacities of hubs, allocation of non-hub nodes to hubs, and the number of vehicles to purchase and rent for multi-modal network with minimum total cost. A two-level cutting plane based method is developed, which includes an algorithm to generate lower bound inequalities in the outer level, and a hybrid algorithm in the inner level. A number of numerical experiments are offered to test the efficiency of the proposed solution algorithm and the robustness of the novel robust mode. 371B Data Science for Reliability and Quality Assurance Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Mingyang Li, Tampa, FL, 33647, United States, mingyangli@usf.edu 1 - Multi-state Reliability Demonstration Tests Suiyao Chen, University of South Florida, 4202 E. Fowler Avenue, Tampa, FL, 33620, United States, suiyaochen@mail.usf.edu, Lu Lu, Mingyang Li Reliability demonstration tests (RDTs) are important reliability assurance activities to ensure product quality over time and safeguard market competitiveness. Conventional binomial RDTs become inadequate to meet diverse reliability requirements of the consumers. This work proposes multi-state RDTs to demonstrate product reliability at multiple time periods or involving multiple failure modes. The design strategies allow the incorporation of prior knowledge to reduce test sample sizes and realize simultaneous demonstration of multiple objectives to ensure customer satisfaction. Two case studies are used to demonstrate the effectiveness of the proposed methods. 2 - Weighted Band Depth for Functional Data Outlier Detection Xudong Zhang, University of Iowa, Iowa City, IA, United States, xudong-zhang@uiowa.edu, Yuxing Hou, Yong Chen The use of depth has shown great usefulness in robust analysis of functional data, such as producing robust location estimators of sample curves, or detecting various outliers. In this paper, we focus on using depth for detecting outliers in functional data, which are considered as coming from a different process compared to normal curves. We propose a new depth concept for function data called weighted band depth (WBD), which improves the performance of the band depth proposed in the literature. Numerical studies based on simulated signals and application to a real data example are used to evaluate the performance of proposed depth methods and other existing depth notions. 3 - Projection Based Outlier Detection in Functional Data Nan Chen, National University of Singapore, Department of WC67 To detect outliers in functional data, we developed a procedure based on a high- breakdown mean function estimator. The robust estimator is obtained from a clean subset of observations, excluding potential outliers, by minimizing the least trimmed squares of the projection coefficients after functional principal component analysis. A threshold rule is constructed based on the asymptotic distribution of the functional-score-based distance. The thresholding robustly controls the false positive rate and detects outliers effectively. Further power improvement is proposed by adding a one-step reweighting procedure. 4 - Sequential Test Planning for Polymer Composites Yili Hong, Professor, Virginia Tech, Blacksburg, VA, United States, yilihong@vt.edu, Ichen Lee, Sheng-Tsaing Tseng, Tirthankar Dasgupta Polymer composite materials are widely used in areas such as aerospace and alternative energy industries, due to their lightweight and comparable levels of strength and endurance. To ensure that the material can last long enough in the field, accelerated cyclic fatigue tests are commonly used to collect data and then make predictions for the field performance. In this paper, we propose a sequential strategy for test planning, and use a Bayesian framework for the sequential model updating. We also use extensive simulations to evaluate the properties of the proposed strategy. Our results show that the proposed strategy is more robust and efficient, as compared to traditional optimum designs. Industrial and System Eng, 1 Engineering Drive 2, Singapore, 117576, Singapore, isecn@nus.edu.sg

destroy/repair operators are designed. We then study a number of real-world problems: the split delivery with multi-dimensional demand, the time-uncritical delivery with T + n possibility, the urban delivery with density consideration and the generalized VRP with multiple available locations for each node. Those tests illustrate its effectiveness and applicability to practical industrial situations, e.g., 20 thousands orders can be solved in 5 minutes. 2 - Traveling Salesman Problems with Multiple-product Mixing German Paredes-Belmar, Assistant Professor, Universidad Andres Bello, Quillota 980, Viña del Mar, 2390457, Chile, german.paredes@unab.cl, Armin Lüer-Villagra, Vladimir Marianov The traveling salesman problems with multiple-product mixing are generalizations of the known Traveling salesman problem (TSP), where a set of products needs being transported between depots and customers at minimum cost. Those products can be carried in the same vehicle. The mix of two products can become a third product, which has an impact on the routes and on the objective function (e.g., cost, profit, risk). We propose an integer formulation to solve different instances of TSP with multiple-product mixing. 3 - A Hybrid Algorithm for Class of Vehicle Routing Problems with Loading Constraints David Alvarez Martinez, Assistant Professor, Universidad de Los Andes-Colombia, Cra 1 No 18A- 12, Bogota, Colombia, d.alvarezm@uniandes.edu.co, Luis Miguel Escobar In this paper we present a hybrid algorithm for a class of Vehicle Routing Problems with Loading Constraints. The routing problem is solved through sequential Set Partitioning models, the columns correspond to routes found through an ILS algorithm, which can interact with the MIP solution process. It was implemented a reactive GRASP algorithm in order to verify all the packing constraints established and in this way only valid columns are incorporated into the solution process. The algorithm was tested using instances of the following variants: (i) CVRP (ii) 2D Loading Problem and (iii) 3D Loading Problem. The results obtained were quite competitive in comparison with those found by the previous works 4 - Scheduling Periodic Deliveries with Heterogeneous Vehicles and Driver Consistency Soheyl Zehtabiyan, The University of Alabama, 317 Bidgood Hall, Tuscaloosa, AL, 35401, United States, zehtabiyan@gmail.com, Mesut Yavuz This study is concerned with minimizing the total fleet cost in a direct-shipment setting with a single supplier and many customers. The customers require perfectly periodic deliveries, as well as a minimally acceptable vehicle type and / or driver consistency. We provide an IP model covering four versions of the problem and devise a constructive heuristic to solve them. 5 - Asymptotic Analysis of the Generalized TSP Xiangfei Meng, USC, 2801 Orchard Ave, Apt 10, Los Angeles, CA, 90007, United States, xiangfem@usc.edu In this work, we will analyze the asymptotic behavior on a generalized version of the TSP, which we call the Generalized Travelling Salesman Problem (GTSP), in which the goal is to select one point each from multiple sets of points and come up with a tour with the minimum length. Two different limiting cases are examined: one is the case where the number of point sets goes to infinity, and the other is the case where the number of points in each set goes to infinity. Numerical simulations confirm that our analysis is valid when applied to simulations in the Euclidean plane and on a road network. 371A Robust and Resilient Facility Location Sponsored: Transportation Science & Logistics Sponsored Session Chair: Mengtong Wang, Tsinghua University, Zhongguancun Street, Beijing, 100084, China, mengtong321@163.com 1 - Secure Transportation in Dangerous Zones Azar Sadeghnejad Barkousaraie, University at Buffalo, 93 Pepper Tree Dr, Apt 4, Amherst, NY, 14228, United States, azarsade@buffalo.edu In order to secure transportation of goods and passengers in dangerous areas usage of escort vehicles is sometimes necessary and inevitable. However due to resource limitation the assignment of security vehicles to each transportation unit in a timely manner is not always applicable, it may result in either long waiting time for customers or overusing of resources. In this research we study the effect of resting areas for limited number of escort vehicles to secure the transportation of customers in order to minimize the response time to the transportation units in dangerous areas. WC66

525