2015 Informs Annual Meeting

WC69

INFORMS Philadelphia – 2015

WC70 70-Room 202A, CC Vehicle Routing II Contributed Session

4 - Potential Effectiveness of Liability Rules and Automated Vehicles in Reducing Rear-ending Crashes Indrajit Chatterjee, University of Minnesota, Twin Cities, 500 Pillsbury Drive SE, Minneapolis, MN, 55455, United States of America, chat0123@umn.edu, Gary Davis This research focuses on understanding the behavior of drivers involved in rear- ending crashes on congested freeways, and using this understanding to evaluate (a) the safety implications of changes in liability policies where individual drivers are penalized based on degree of causal contribution to the crash and (b) the safety implications of mixtures of human-operated and automated vehicles in the future traffic streams.

Chair: Jiahong Zhao, School of Business,Guangzhou University, NO.230,Guangzhou Daxuecheng Waihuanxilu, Guangzhou, China 1 - An Optimization-Based Heuristic for the School Bus Problems with Regret Minimization Mehmet Ivgin, Lecturer, Turkish Military Academy, Devlet Mahallesi Kara Harp Okulu Caddesi, Ankara, Turkey, mivgin@kho.edu.tr, Elif Rabia Karadeniz We present in this study an application of the School Bus problem in practice. Based on an integer programming formulation of the School Bus problem, we implemented a heuristic using column generation to solve a real-life instance provided by a public school in Ankara. Our results show that our approach yields a big improvement in terms of customer satisfaction when compared to the previously used scheduling approaches. 2 - A Simulated Annealing Approach to Solve Large-Scale VRP: An Application at GE Appliances & Lighting Ehsan Khodabandeh, University of Louisville, Speed School of A simulated annealing approach with a network shrinking heuristic is considered to solve a vehicle routing problem with time windows where routes have limited duration. The objective is to minimize traveled time and total number of vehicles required. Implementation of this algorithm for solving large-scale problems has positively impacted GE Appliances & Lighting’s operation by reducing delivery time from three to two days and by reducing the number of required trucks by half in some instances. 3 - A Branch-and-Price-and-Cut Algorithm for the Generalized Vehicle Routing Problem Engineering, Department of Industrial Engineering, Louisville, KY, 40292, United States of America, ehsan.khodabandeh@louisville.edu, Sunderesh Heragu, Lihui Bai, Gerald Evans We examine the Generalized Vehicle Routing Problem, a VRP variant where customers are partitioned into mutually exclusive clusters, each with a specific demand. We propose a branch-and-price-and-cut algorithm that takes advantage of the characteristics of the problem and treats a GVRP instance with n customers and m clusters nearly as a VRP instance with m customers. Our computational study reports encouraging results. 4 - Rail Freight Service Design with Consideration of Consolidation and Heterogenous Demand Xiao Lin, PhD Candidate, Tsinghua University, Beijing, China, lin-x12@mails.tsinghua.edu.cn, Tianhu Deng, Simin Huang Rail carriers are forced to serve more time-sensitive customers in nowaday China. To serve the new market the China Railway Company need to provide faster transportation service and design differentiated service products. For a consolidation carrier like the rail, faster service means less stops, which further results in less consolidation of freight. This study provides a differentiated service design model that maximize revenue for consolidation carriers facing heterogeneous demand. In solving this problem, we proposed two heuristic methods which can solve the problem of real size and compared its performance with lagrangian relaxation method. 5 - A Multi-Depot Vehicle-Routing Model for the Explosive Waste Recycling Jiahong Zhao, School of Business, Guanzhou University, No. 230 in Daxuecheng Waihuanxilu, Guangzhou, Guangdong, 510006, China, zhaojiahong1@126.com The explosive waste recycling is a significant concern because it has immense impacts on economy and safety during its transportation among multiple depots. We develop a multi-depot vehicle-routing model with the minimizations of cost and risk. It is formulated through the two-commodity flow formulation, and characterized by simultaneously planning tours, vehicle acquisitions and return- trips. To solve this model, a modified lexicographic weighted Tchebycheff method is also proposed. Mohammad Reihaneh, Isenberg school of management, University of Massachusetts Amherst, 121 Presidents Dr., Amherst, MA, 01002, United States of America, mreihaneh@som.umass.edu, Ahmed Ghoniem

WC69 69-Room 201C, CC

Innovative Data Sources in Transportation Sponsor: TSL/Intelligent Transportation Systems (ITS) Sponsored Session

Chair: Xianyuan Zhan, PhD Candidate, Purdue University, 550 Stadium Mall Drive, School of Civil Engineering, West Lafayette, In, 47906, United States of America, zhanxianyuan@purdue.edu 1 - Spatial-temporal Traffic Pattern Identification in a Large-Scale Urban Network Zhenhua Zhang, The University at Buffalo, NY, United States of America, zhenhuaz@buffalo.edu, Qing He, Jizhan Gou, Xiaoling Li We employ the dictionary-based compression method to identify the regional traffic pattern within a large-scale urban network and aim to quantify the traffic pattern fluctuations from different time and space perspectives. Studies unveil characteristics of the geographic pattern distribution, time-of-day pattern fluctuations and the day-to-day pattern differences etc. It can even quantify the influences of the large municipal works on its surrounding traffic. 2 - A Bayesian Mixture Model for Short-term Average Link Travel Time Estimation using Large-scale Limited Information Trip-based Data Xianyuan Zhan, PhD Candidate, Purdue University, 550 Stadium Mall Drive, School of Civil Engineering, West Lafayette, In, 47906, United States of America, zhanxianyuan@purdue.edu, Chao Yang, Satish V. Ukkusuri This paper develops a Bayesian mixture model to estimate the urban link travel times utilizing large-scale limited trip-based data without trajectory information. The model estimates the mean and variance of the average link travel times. A transition model is also introduced as an informative prior to capture the temporal and spatial dependencies of link travel times. An efficient solution approach based on expectation-optimization (EM) algorithm is proposed to solve the problem. The model is tested using a large-scale taxi trip data from New York City. 3 - Exploring Behavior during Hurricane Sandy Arif Mohaimin Sadri, Purdue University, 149 Arnold Drive, Apt 12, West Lafayette, IN, 47906, United States of America, asadri@purdue.edu, Satish V. Ukkusuri Individuals, being subjected to different personal constraints and environments, may want to evacuate or not during a major hurricane. Evacuation decision can also be influenced by the social network partners. In this study, data has been obtained by interviewing individuals from high storm surge areas of New York and New Jersey. Individuals’ social network information were obtained by using an ego-centric approach. A mixed logit model is developed to explain the effects of individual, household and social network characteristics on evacuation decision making. 4 - Temporal-Spatial Domain Trajectory Matching Algorithm Ali Arian, Graduate Research Assistant, The University of Arizona, 1209 E. Second Street, Room 206A-1, Tucson, AZ, 85721, United States of America, arian@email.arizona.edu, Yi-Chang Chiu This talk presents a Longest Common Sequence (LCS) based algorithm to compare and match two distinct trajectories for degree similarity based on temporal-spatial domain information matching. Applications of the presented algorithm in passive multi-person carpool matching and day-to-day route set building are presented.

450