2015 Informs Annual Meeting

MD69

INFORMS Philadelphia – 2015

MD69 69-Room 201C, CC

2 - Diagnostic Method to Measure the Impact of Railway Traffic Heterogeneity from Field Operations Data Mei-Cheng Shih, University of Illinois, U-C, 205 North Mathew Ave, Urbana, IL, 61801, United States of America, mshih2@illinois.edu, C. Tyler Dick Growing demand for freight rail transportation of crude oil and domestic intermodal traffic has increased both the volume and disparity in types of trains operated on many mainlines. Differences in train characteristics and priority, has become one of the major causes of congestion. Based on actual train operations data from a Class 1 railroad, this study develops two indices that help railroads measure the degree to which traffic heterogeneity impacts observed operating performance. 3 - Evaluating Track Maintenance Aggregation in Extended Work Windows on Freight Railroad Lines Alexander Lovett, Graduate Research Assistant, University of Illinois at Urbana-Champaign, 205 N Mathews Ave, B118 Newmark Lab, Urbana, IL, 61801, United States of America, alovett2@illinois.edu, Christopher Barkan, C. Tyler Dick Combining track maintenance activities into extended work windows can improve maintenance efficiency. Efficiencies come from reduced set-up time and efforts common to multiple activities at the cost of being more disruptive to train operations. Aggregating activities requires adjusting maintenance cycles which may increase accident risk or maintenance frequency. A methodology for evaluating the costs of aggregating track maintenance into an extended work window is presented. MD71 71-Room 202B, CC Transportation Network Modeling and Design Sponsor: TSL/Urban Transportation Sponsored Session Chair: Antoine Petit, UIUC, Urbana, IL, United States of America, apetit@illinois.edu 1 - Generalized Statistical Traffic Assignment: Methodology, Properties and Variance Analysis Sean Qian, Carnegie Mellon University, Pittsburgh, PA, 15213, United States of America, seanqian@cmu.edu, Wei Ma The traffic assignment problem has been typically used to estimate deterministic network traffic volumes. Day-to-day variations of link and path flow are unfortunately overlooked, but often times they are required for modeling network uncertainty and resilience. We propose a generalized statistical traffic assignment model where variance of demand, route choices and individual perception are considered. We show some statistical properties of this model and discuss the variance analysis. 2 - Construction Schedule Design Considering Cost, Traffic Delay and Energy Consumption Laura E. Ghosh, UIUC, Urbana, IL, United States of America, vingerh1@illinois.edu, Khaled El-Rayes, Ahmed Z. Abdelmohsen, Yanfeng Ouyang A model that determines the optimal freeway maintenance schedule over a construction season is presented. It optimally selects work-zone length, workday start time and duration, lane and shoulder widths, temporary lane construction and access strategy in order to (i) minimize the costs of construction-related traffic delay and corresponding mitigation measures and (ii) minimize construction related energy consumption. 3 - Flexible Transit Network Design with and Without Branching under Spatially Heterogeneous Demand Antoine Petit, UIUC, Urbana, IL, United States of America, apetit@illinois.edu, Yanfeng Ouyang While public transportation systems are usually designed with fixed routes, this work presents an alternative flexible-route transit system where vehicles travel within predetermined areas in response to trip demand to provide door-to-door service. Two additional features will be considered in the network design, hybrid structure and local branching.

Arc Routing Problems and Applications Sponsor: Transportation, Science and Logistics Sponsored Session

Chair: Mehmet Basdere, Northwestern University, 2145 Sheridan Road, Tech Institute C210, Evanston, IL, 60208, United States of America, mehmetbasdere2016@u.northwestern.edu 1 - The Windy Rural Postman Problem with a Time-dependent Zigzag Option

Rui Zhang, University of Maryland, College Park, MD, ruizhang@rhsmith.umd.edu, Erwin Pesch, Jenny Nossack, Bruce Golden

We focus on the windy rural postman problem with an additional zigzag option. It combines two classes of arc routing problems known before: those with zigzag options and those with time dependencies. We present two MIP formulations and suggest exact solution approaches. Furthermore, we analyze the effects of zigzag and time window options on the objective value. 2 - Arc Routing Problems to Restore Connectivity of a Road Network after a Disaster Vahid Akbari, Koc University, College of Engineering, Sariyer, Istanbul, 34450, Turkey, vakbarighadkolaei@ku.edu.tr, Sibel Salman Routes should be generated to reconnect a disconnected road network in the shortest time by opening blocked roads. We study two versions: 1) minimize the time to reconnect the network, 2) maximize the total prize of components reconnected within a time limit. We develop MIP formulations and heuristic algorithms. 3 - Arc Routing, Vehicle Routing and Turn Penalties Thibaut Vidal, Professor, Pontifícia Universidade Católica do Rio We introduce a structural decomposition for arc routing problems, in which all decisions about driving lanes, turns, edge traversal and service orientations are addressed via dynamic programming. We show that a neighborhood based on moves on the sequences of services with optimal traversal decisions can be explored in O(1) per move. The approach is integrated into two classical metaheuristic frameworks, leading to remarkable results for many arc routing variants, with possible turn penalties. 4 - The Lock Free Arc Touring Problem with an Application to Marathon Course Design Mehmet Basdere, Northwestern University, 2145 Sheridan Road, de Janeiro, R. Marquìs de São Vicente, 225 - Gávea, Rio de Janeiro, 22451-900, Brazil, vidalt@inf.puc-rio.br

Tech Institute C210, Evanston, IL, 60208, United States of America, mehmetbasdere2016@u.northwestern.edu, Karen Smilowitz, Sanjay Mehrotra

In this talk, we present a new type of arc routing problem in the marathon course design setting. The aim is to find a valid marathon course that minimizes the average distance to the medical facilities within the region of interest without preventing the public access to those facilities while visiting a predetermined subset of landmark streets. A novel solution approach which utilizes visit restrictions and new valid inequalities are introduced.

MD70 70-Room 202A, CC Railway Analytics Sponsor: Railway Applications Sponsored Session

Chair: Qing He, Assistant Professor, SUNY Buffalo, 313 Bell Hall, Buffalo, NY, 14051, United States of America, qinghe@buffalo.edu 1 - Multi-task Learning for Joint Prediction of Failure Time and Failure Types of Train Wheels Weixin Wang, University at Buffalo, 1357 Millersport Hwy Apt 8, Buffalo, NY, 14221, United States of America, weixinwa@buffalo.edu, Zhiguo Li, Qing He The failures of train wheels account for half of all train derailments. Both failure time and failure types of wheels are critical for wheel maintenance. Failure time prediction is a regression task, whereas failure type is a classification task. In this work, we propose a multi-task learning approach to jointly predict these two tasks by using a common input space to achieve more desirable results.

254