Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WD28

n WD28 North Bldg 221A Practice- Transportation-Operations IV Contributed Session

n WD29 North Bldg 221B Urban Public Transport Sponsored: TSL/Urban Transportation Sponsored Session Chair: Yanshuo Sun, Florida State University, Tallahassee, FL, 20742, United States 1 - Multi-period Line Planning in Urban Public Transportation Guvenc Sahin, Sabanci University, Faculty of Engineering and Natural Sciences, Istanbul, 34956, Turkey, Amin Ahmadi, Ralf Borndörfer, Malte Ranken, Thomas Schlechte Line planning in urban transportation is involved with determining the number and routes of the lines along with their frequencies given an underlying physical transportation network. In traditional line planning, demand is considered as static while fleet-related constraints are secondary. We solve a multi-period line planning problem with integer programming formulations that consider dynamicity of the demand. Multi-period solutions outweigh the single-period solutions with respect to system-wide service levels. Indeed, single period solutions when combined with each other may not even be feasible due to suboptimal allocations of vehicles among lines. 2 - Regulation of Public Transit Monopoly under Asymmetric Information Yanshuo Sun, Florida State University, Tallahassee, FL, 20742, United States, Qianwen Guo Contrary to the dominant assumption, we study how to regulate a monopolistic transit operator with unknown cost parameters. A benchmark model which assumes complete cost information is proposed and solved analytically. Then government’s decisions based on prior estimates of cost parameters are evaluated and the welfare loss due to the incomplete information is thus identified. Next, an incentive-compatibility regulatory policy which induces the operator to report its true parameter is proposed. Finally, the derivation of the optimal policy is provided and its properties are analyzed. Thus, this study provides an effective tool for designing regulatory policies and evaluating practices. 3 - Uncertainty Quantification of Dynamic Urban Transit System State Estimates in Big Data Applications: An Information Space Projection Approach Jiangtao Liu, Arizona State University, Tempe, AZ, 85282, United States, Xuesong Zhou This talk aims at a modeling framework to systemically account for the multi- source sensor information in urban transit systems to quantify the estimated state uncertainty. By developing a system of linear equations and inequalities, the information space is generated based on the available heterogeneous sensor data. Then, a number of projection functions are introduced to match the relation between the unique information space and different system states of interest. Finally, different state estimate uncertainties are quantified by calculating its maximum state range. 4 - Passenger Trajectory Generation and Characterization using Modern Datasets Xiaoyan Xie, Cole des Ponts ParisTech, Champs-sur-Marne, France, Fabien Leurent Passenger trajectory generation and characterization along a line on a Public Transport (PT) network was still a difficult task, as it must capture not only the complexity of trip building and route choice on the network; but also the heterogeneity in walking and waiting behaviors related to individual characteristics and the topological complexity of each station. Fortunately, over the last decade, the emerging modern data from automatic observation systems improved this task. However, traditional data modeling approaches were challenged. Based on the state-of-the-art review about passenger trajectory rebuilding and characterizing approaches, this paper proposed an approach to generate passenger trajectory based on smartcard data on a PT network. It is applied to a real case study in Paris. Passenger trajectory is characterized using the model output. Passenger in-station and on-network value of times (VoT) are appraised. 5 - A Cumulative Service State Representation for the Pickup and Delivery Problem with Synchronized Transfers Monirehalsadat Mahmoudi, Arizona State University, Tempe, AZ, United States, Tie Shi, Yongxiang Zhang, Xuesong Zhou In this research, in order to solve the pickup and delivery problem with time windows and synchronized transfers, we construct multi-dimensional state-space- time networks and apply a dynamic programming to solve a shortest path problem on the constructed networks. In order to handle a large set of passengers, we develop the traditional cluster-first, route-second approach. We also develop a continuous time approximation approach using cumulative arrival, departure, and on-board count diagrams to effectively assess the dynamic system performance and guide the search.

Chair: Zhen Tan, Cornell University, Ithaca, NY, 14850, United States 1 - Multi-methodological Approach for Risk Mitigation Strategy Selection in the Trucking Industry: A Truck Drivers Perspective Krishna Kumar Dadsena, Indian Institute of Technology Kharagpur, Kharagpur, Vs Hall, B. 243, IIT Kharagpur, Kharagpur, 721302, India, Sarada Prasad Sarmah, V. N. A. Naikan This study aims to identify the operational risks induced by the truck drivers’ job satisfaction criteria, and to select the most suitable risk mitigation strategy on managing the operational efficiency of the trucking industry. The approach based on the systematic application of survey-based analysis and fuzzy theory is used to develop a novel algorithm to support the strategy selection process. 2 - Dynamic Subsidy-pricing Models for Multi-modal Transportation Integration: The One Belt-one Road Strategic Context Tanmoy Kundu, National Taiwan University, Taipei, Taiwan, Jiuh-Biing Sheu One Belt-One Road (OBOR) is a regional/international developmental initiative recently initiated by the Chinese government. One of the challenges being faced in the OBOR operations is the ineffective utilization of the multiple modes of transportation. Huge subsidy to one mode of transportation is leading to the cannibalization of the other. Hence, this work offers various dynamic subsidy- pricing models associated with the dynamic multimodal transportation networks along the OBOR corridors. 3 - An Efficient Iterative Algorithm for the Integrated Optimization of Train Timetabling and Maintenance Task Scheduling Yongxiang Zhang, Southwest Jiaotong University, Chengdu, 610031, China, Qiyuan Peng, Andrea D’Ariano, Bisheng He Track maintenance tasks need to be scheduled to retain the railway tracks in an appropriate state. However, the planning of track maintenance activities is interrelated with the train timetabling process. In this work, train timetabling is optimized together with maintenance task scheduling to solve potential planning conflicts and to allocate the available railway capacity more efficiently to the scheduled services. A new mixed integer linear programming model is formulated and an efficient iterative algorithm is proposed to find near-optimal solutions within a short computation time. The experimental results show that the proposed algorithm outperforms previously proposed approaches. 4 - The Pareto-improving Hybrid Fare Scheme with Heterogeneity in Commuter’s Scheduling Flexibility Yili Tang, Hong Kong University of Science and Technology, Academic Building, Room 3595, Hong Kong, China, Hai Yang This paper proposes a hybrid fare scheme (HFS) combining a fare-reward scheme (H-FRS) and a uniform fare scheme (H-UFS) by considering the heterogeneous commuter’s scheduling flexibility in transit bottleneck model. It aims at reducing peak-hour congestions with alternative options catered for various commuters. In H-FRS, a commuter is rewarded with a free ride during shoulder periods after taking a number of paid rides during the central period in peak hours. The H-UFS determines a uniform fare. The preliminary results demonstrate that the HFS is not only revenue-preserving but also Pareto-improving. An optimally designed hybrid scheme can achieve a reduction in total time costs by at least 25%. 5 - Effect of Information Delay on Real-time Routing and a Potential Remedy Zhen Tan, Nottingham University Business School (China), Ningbo, 14850, China, Jamol Pender, H. Oliver Gao, Xiaoning Zhang In dynamic routing, travel time information is often delayed and hence inaccurate because of challenges in data collection and sensor working principal. This inaccuracy can misguide motorists and result in unstable traffic patterns that exacerbate congestion. To alleviate this negative effect, we analyzed the potential of providing drivers with real-time en-route air pollution information (in addition to travel time) using a new queueing model. Results of our theoretical and numerical analysis indicate that provision of real-time air pollution information can help stabilize traffic. We verified this benefit by traffic simulation of the George Washington Bridge based on real-world data.

504