2016 INFORMS Annual Meeting Program

WD59

INFORMS Nashville – 2016

WD59 Cumberland 1- Omni

3 - A Partially Time-expanded Network-based Solution Approach For The Traveling Salesman Problem With Time Windows And Its Variants Minh Duc Vu, Quinlan School of Business, Loyola University Chicago, 16 E. Pearson, Chicago, IL, 60611, United States, dvu3@luc.edu, Natashia Boland, Mike Hewitt, Martin Savelsbergh We present an exact approach for solving the Traveling Salesman Problem with Time Windows (TSPTW) that is based on repeatedly solving a mixed integer programming formulation of the problem over partially time-expanded networks. We show that the same algorithmic framework can be applied to different objectives for the problem. We develop new speed-up techniques for the algorithm and with an extensive computational study we are able to outperform many of the state-of-the-art approaches. Finally, we show how the algorithm can be adapted to variants of the TSPTW wherein the makespan is the objective or travel times are time dependent. 4 - Minimizing Routing Costs In Blood Supply Chain Okan Orsan Ozener, Ozyegin University, orsan.ozener@ozyegin.edu.tr We study the routing of blood collection vehicles for managing the platelet supply in blood supply chain. In order to extract platelets, donated blood has to be processed at a central processing facility within six hours of donation time. Blood collection organizations have to dispatch collection vehicles and schedule pickups from the donation sites so that the donated units can be used in platelet production. The objective is to collect and process a pre-specified number of donations for platelet production via a limited number of vehicles at minimum transportation cost. In our analysis, motivated by the practices in real-life, we cluster the donation sites so that a single vehicle serves the donation sites in each cluster. We propose a solution method based on route generation/selection and test the performances of the proposed method on randomly generated instances. WD61 Cumberland 3- Omni Scheduling in Passenger-Oriented Railway Networks Sponsored: Railway Applications Sponsored Session Chair: Nikola Besinovic, Delft University of Technology, Mekelweg 2, Deflt, 2628, Netherlands, N.Besinovic@tudelft.nl 1 - Functional Relationship Between Primary And Secondary Delays On Railway Lines Steven Harrod, Technical University of Denmark, stehar@dtu.dk, Fabrizio Cerreto, Otto Anker Nielsen Performance, robustness, and punctuality of railway lines are most commonly evaluated with simulation software. This research proposes a cubic relationship between aggregate train delays over the planning horizon and identifiable primary delays. This relationship provides a heuristic for reducing the number of simulation instances. The method is presented in the context of ongoing research into robustness and punctuality on the Danish railways under the IPTOP research project. In addition, the presentation will discuss the identification of cause and effect between primary and secondary delays in a very large data set. 2 - Passenger Train Service Network Design Based On A State-space-time Network Lingyun Meng, Beijing Jiaotong University, menglingyun2001@hotmail.com, Xuesong Zhou We propose a solution framework for matching passenger travel demand to vehicle transport services and vehicles to infrastructure in a railway system based on a triple-layer (state-space-time) network. Following a method for carefully constructing the triple-layer network, we present an integer programming formulation and a lagrangian relaxation solution framework. Numerical experiments are conducted for validating the proposed approach. 3 - Towards Operationally Feasible And Robust Railway Timetables Nikola Besinovic, Delft University of Technology, Stevinweg 1, Delft, Netherlands, n.besinovic@tudelft.nl, Rob M.P. Goverde Railway timetabling has been a focus of research over the years and multiple approaches have been proposed. However, most of the existing models for timetabling focus on a macroscopic level of detail, taking stations as nodes and tracks between as arcs, and disregard changes in operational aspects. We propose an optimization approach that integrates a macroscopic model with microscopic details. The objective is to find a robust timetable that is also operationally feasible and ready to be implemented in practice. In addition, the model allows for optimal use of infrastructure capacity and reduces wasted resources. We present computational results on real-life cases from the Dutch railways.

Facility Logistics - Potpouri Sponsored: TSL, Facility Logistics Sponsored Session Chair: Debjit Roy, Indian Institute of Management - Ahmedabad, 34355, India, debzitt@gmail.com 1 - Retrieving Multiple Loads In Very High Density Storage Systems Masoud Mirzaei, Erasmus University, mirzaei@rsm.nl, Masoud MirzaeiB.M. De Koster, Nima Zaerpour Compact storage systems aim to utilize the available space more efficiently than conventional aisle-based storage systems. The most compact system has only one open space available to perform the retrieval and storage requests. More open locations, reduce space usage, but can reduce the retrieval time. However, in both situations, much space and time can be saved by performing multiple-load retrieval using a limited number of open locations. Current models for such systems study retrieving loads one at a time. We develop optimal models to do this for two loads and propose heuristics for more loads. Results show savings of 20% can be achieved compared to single-load retrieval. 2 - Loading And Unloading Trains And Trucks At Container Terminals Debjit Roy, Indian Institute of Management - Ahmedabad, We first develop a closed queuing network with a fixed number of Automated Guided Vehicles that continuously circulate in the network during train loading or unloading process and interact with truck arrivals at the stacking cranes. We develop exact solutions for the case with one automated stacking crane by using an imbedded Markov chain approach. We then use these results to develop a semi-open queuing network model to analyse and approximate the expected throughput times for handling containers that arrive via trains (bulk arrivals) and trucks (single arrivals). 3 - A Fluid Model For Optimal Lot Selection And Dispatching Policies In Semiconductor Wafer Fabs Dima Nazzal, Georgia Institute of Technology, Atlanta, GA, 30332, United States, dima.nazzal@gatech.edu, Shabbir Ahmed, Kelly Bartlett, Haejoong Kim, Junho Lee, George L Nemhauser, Matias Siebert Sandoval, Joel Sokol In semiconductor manufacturing, congestion in an automated material handling system may negatively impact production throughput. For a prototype facility, we identify optimal lot selection and dispatching policies for various WIP levels and workload conditions. We propose a fluid-model lot selection policy that iteratively optimizes selection based on current WIP distribution. WD60 Cumberland 2- Omni Time-Sensitive Delivery Planning and Routing Sponsored: TSL, Freight Transportation & Logistics Sponsored Session Chair: Okan Orsan Ozener, Ozyegin University, Istanbul, Turkey, orsan.ozener@ozyegin.edu.tr 1 - A Lagrangian-based Approach For Consolidating Freight Of Perishable Products Christine Vi Nguyen, Northern Illinois University, DeKalb, IL, United States, cnguyen@niu.edu, Alejandro Toriello, Maged M Dessouky We study a supply chain with small suppliers of perishable products, where shipping via a shared consolidation center allows for more economical FTL rates versus LTL or courier rates. We propose a Lagrangean relaxation of the center’s inventory capacity to decompose the problem by destination, and develop a related heuristic that balances consolidated shipping costs and inventory costs. 2 - Same-day Delivery Planning With Store Fulfillment Ming Ni, University at Buffalo, mingni@buffalo.edu, Qing He, Jose Walteros, Xuan Liu, Arun Hampapur This study presents a same-day delivery planning with store fulfillment problem, which can be characterized by three main factors, the store selection, the delivery fleet sizing, and the multi-commodity distribution for online shoppers. It develops a mix integer programming model to represent the problem and use a Benders decomposition based approach to solve it efficiently. The numerical examples are derived from same day delivery from a real-world retailer store network. Ahmedabad, 380015, India, debjit@iima.ac.in, Jan-Kees Van Ommeren, Amir Gharehgozli

474