Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MC32

n MC32 North Bldg 222B

n MC33 North Bldg 222C Joint Session ORAM/QSR/Practice Curated: Advancements in Spatial Data Analytics Methods for Manufacturing Quality Control Emerging Topic: OR and Advanced Manufacturing Emerging Topic Session Chair: Mohammed Shafae, University of Arizona, Tucson, AZ, 24061, United States Co-Chair: Wenmeng Tian, Mississippi State University, Mississippi State, MS, 39762, United States 1 - Spatial Statistics and Scan Statistics for SPC: A Comparison Romina Dastoorian, Western Michigan University, Kalamazoo, MI, 49009, United States, Lee Wells Modern manufacturing systems contain numerous high-density spatial data-sets that are obtained from advanced measurement technologies. Statistical Process Control (SPC) techniques that have been developed for these data-sets can be separated into two primary categories, namely spatial statistics and scan statistics. Unfortunately, the performances of proposed SPC techniques developed within these two areas are rarely (if ever) been compared. In response, this work presents a fair and balanced simulation-based performance comparison between several spatial statistics and scan statistics SPC approaches across a variety of process shift scenarios for image data. 2 - Layer-wise Anomaly Detection for Spatially-correlated Thermal Images in Laser-based Additive Manufacturing Ahmed Aziz Ezzat, Texas A&M University, College Station, TX, United States We propose a layer-wise anomaly detection method for laser-based additive manufacturing as a step towards enabling real-time process control. Using in- process thermal images, process anomalies are signaled through a multi-step data-driven procedure that includes segmentation, spatial modeling, and classification analysis. We validate the proposed method using a case study on a commercial laser powder bed fusion system instrumented with a dual-wavelength imaging pyrometer for capturing the thermal signature during fabrication. 3 - Scene Understanding and Obstacle Detection for Off-road Autonomous Driving Based on Real-time LiDAR Sensing Wenmeng Tian, Mississippi State University, P.O. Box 9542, Mississippi State, MS, 39762, United States Scene understanding and autonomous navigation methodologies have been developing rapidly with the help of machine learning and artificial intelligence technologies. Moreover, advanced sensing techniques such as LiDAR systems can generate high volume of point cloud data for scene understanding. In this work, we propose a novel scene understanding and obstacle detection method based on a modified version of the Adaptive Generalized Likelihood Ratio (AGLR) method, originally proposed for surface inspection in manufacturing applications. A case study based on a LiDAR simulator is used to illustrate the effectiveness of the proposed approach. 4 - Generalized Heterogenous Recurrence Analysis of Spatial Data Cheng-Bang Chen, The Pennsylvania State University, State College, PA, 16801, United States, Hui Yang, Soundar Kumara Prior research has shown that the heterogeneous recurrence method is effective in the characterization and quantification of heterogeneous recurrence behaviors in the temporal data. With the use of fractal representation, recurrence patterns are investigated in multiple scales. However, very little has been done to investigate heterogeneous recurrence behaviors in spatial data. This research extends the heterogeneous recurrence analysis to the high dimensional spatial data. We develop a novel structure to define the heterogeneous states, then provide a new fractal representation. Our method provides a novel platform for monitoring and control of dynamic recurrences of spatial data.

Intermodal Transportation and Ports Sponsored: TSL/Freight Transportation & Logistics Sponsored Session Chair: Rob A. Zuidwijk, Erasmus University-Rotterdam,3000 DR, Netherlands 1 - Transportation Network Resilience, Supply Chain Resilience and Economic Impacts John F. Betak, Managing Member, Collaborative Solutions LLC, Dept. of Information, Risk & Operations Mgmt, McCombs School of Business, The University of Texas-Austin, David R. Fletcher We discuss relationships between supply chain approaches to resiliency, risk management and public sector activities involving risk-based transportation asset management and resilience management by: 1) public sector infrastructure decisions impacting certain supply chain systems and networks, including military materiel supply chains; and 2) incorporating specific reliability and resilience requirements of these supply chains into public investment decision-making processes. We show how the interconnected decisions impact international Intermodal container movements in the supply chain. Alternative modeling approaches are explored and research recommendations are offered. 2 - A Comprehensive Overview of Hinterland Transport as a System of Interacting Actors: A Supply Chain Perspective Volkan Gumuskaya, PhD Candidate, Eindhoven University of Technology, Eindhoven, Netherlands, Willem van Jaarsveld, Remco Dijkman, Albert Veenstra, Paul Grefen Hinterland transport is the total of import and export activities between a port and its hinterland, involving physical and planning processes, which are characterized by the interaction of actors and the contractual relationships. In practice, the actors act according to their own goals resulting in coordination issues and system wide inefficiencies. These inefficiencies emerge in the form of a lack of information exchange and a lack of collaboration leading to operational problems. To address these problems and enhance future research directions, we create a comprehensive overview of the hinterland transport focusing on the interaction of the actors and identify the coordination issues. 3 - Enhancing Resilience through Port Collaboration in Maritime Freight Networks Wenjie Li, George Mason University, Fairfax, VA, 22030, United States, Ali Asadabadi, Elise Miller-Hooks Reliable port services are key to the performance of maritime freight transport systems that are vulnerable to disaster events of anthropogenic or natural cause. Strategies involving capacity expansion and protective cross-port investments, as well as collaborations that enable capacity sharing between ports, are proposed. These strategies aim to reduce the impact of such disruptive events on goods movements through maritime freight transport networks. This collaborative port protection and investment problem is formulated as a bi-level program with multiple players (leaders) in the upper level and a common liner shipping problem (the followers) in the lower level. 4 - Planning of Heterogeneous Capacity in Intermodal Networks with Uncertain Demand Hobbs White, Erasmus University Rotterdam, Rotterdam, Netherlands, Bart Wiegmans, Rob A. Zuidwijk Faced with uncertain demand, an intermodal network operator must source transportation capacity from a pool of carriers and service providers, heterogeneous in characteristics such as mode and availability. The operator will seek to hedge against the risks of inefficient asset utilization and costly recourse by selecting optimally from a set of potential agreements with capacity suppliers, given ranges of duration, lead-time, and levels of purchase commitment. To support interpretation, we extend the formulation of this network design problem as a minimum cost stochastic program by studying an analytically tractable Alberto Giudici, Erasmus University Rotterdam, Rotterdam, Netherlands, Tao Lu, Clemens Thielen, Rob A. Zuidwijk Evidence from practice shows that the goal of improving service quality of port- to-door container transport can motivate operators to cooperate. Once cost savings are generated by sharing orders or transport capacity, companies will selfishly bargain for the cost allocation. Following Nash Program, we consider the Shapley value as the resulting non-cooperative cost allocation and test its coalition rationality. Basing the cooperative game on a parametric min-cost flow problem, we analytically study the dependency of the Shapley value and core on penalty costs for late deliveries. Our results show that when the amount of orders shared is close to the total capacity pooled, cooperation is stable. approach centered on a multidimensional newsvendor model. 5 - A Parametric Analysis of Collaborative Hinterland Container Transport

196