2015 Informs Annual Meeting

SB72

INFORMS Philadelphia – 2015

SB69 69-Room 201C, CC

maintenance, but still allow for running traffic. Red tags are failures with require immediate intervention. The goal of the PSC is to predict when yellow tags will turn red, to be able to plan optimally maintenance and repair actions.

Routing Problems with Uncertainty I Sponsor: Transportation, Science and Logistics Sponsored Session Chair: Shu Zhang, School of Economics and Business Administration, Chongqing University, No. 174 Shazheng Street, Shapingba, Chongqing, 400030, China, zhangshu@cqu.edu.cn 1 - Technician Scheduling Problem with Experience-based Service Times We study a dynamic technician scheduling problem with experience-based service times. We use well-established models from the psychology community to model how gains in experience impact service times. We assume that customer requests are uncertain. We model the problem as a Markov decision process with the objective of minimizing the expected total service times over a finite horizon. We also propose a number of lookahead schemes and present results. 2 - Route Evaluators for the Vehicle Routing Problem with Stochastic and Correlated Travel Times Jorge E. Mendoza, Associate Professor, Polytech Tours, 64 Avenue Jean Portalis, Tours, 37200, France, jorge.mendoza@univ-tours.fr, Andres Medaglia, Andrés Sarmiento, Carlos Felipe Ruiz, Raha Akhavan-Tabatabaei Vehicle routing problems with stochastic travel times (VRPSTTs) consist in designing transportation routes of minimal expected cost over a network where times are represented by random variables. In this research we consider the case of correlated random variables. We present two methodologies, one analytical and one based on simulation, to efficiently evaluate routes. We present computational experiments showing the value of considering correlations in VRPSTTs. 3 - Revenue Driven Supply Chains under Uncertainty Nizar Zaarour, Northeastern University, Boston, MA, United States of America, n.zaarour@neu.edu, Emanuel Melachrinoudis, Marius M. Solomon We consider the discount pricing of products that are being phased-out in retail stores. This triggers a time horizon for the retailer to sell the existing inventory or sell it at a major discount in a reverse logistics move. We develop mathematical programming models to deal with this issue. The models are tested on real data provided by a national retailer. 4 - Multi-Period Orienteering with Uncertain Adoption Likelihood and Waiting at Customers Shu Zhang, School of Economics and Business Administration, Chongqing University, No. 174 Shazheng Street, Shapingba, Chongqing, 400030, China, zhangshu@cqu.edu.cn, Barrett Thomas, Jeffrey Ohlmann We consider a problem in which a traveler visits customers over a multi-period horizon to influence and observe the chance of customer adoption. Over the multiperiod horizon, each customer’s likelihood of adopting the traveler’s product stochastically evolves. We model the problem as a partially observed Markov decision process with an objective to maximize the expected sales. Each period, the traveler must decide which customers to schedule and which order to visit, knowing that due to uncertain wait times the traveler may not be able to meet customers even if they are on the schedule. SB70 70-Room 202A, CC RAS Problem Solving Competition 2015 Sponsor: Railway Applications Sponsored Session Chair: Francesco Corman, Delft University Of Technology, Mekelweg 2, Delft, Netherlands, F.Corman@tudelft.nl 1 - Railway Applications Society Problem Solving Competition Francesco Corman, Delft University Of Technology, Mekelweg 2, Delft, Netherlands, F.Corman@tudelft.nl This session is reserved for the finalists of the RAS Problem Solving Competition (PSC). The presenters and their abstracts won’t be determined until we finish the judging process, which happens around mid-October. The selection committee will identify the top three teams who will present their results during the session. PSC asks participants to build a model to predict track failures on a railway network. Data about recorded traffic over a railway network is given, together with yellow and red tags. Yellow tags are measured deviations from the railway regulations for track layout, alignment and condition, which would require Xi Chen, The University of Iowa, Iowa City IA 52242 xi-chen-3@uiowa.edu, Michael Hewitt, Barrett Thomas

SB71 71-Room 202B, CC Traffic State Estimation Methods and Data Sponsor: TSL/Urban Transportation Sponsored Session

Chair: Kai Yin, Nomis Solutions, 1111 Bayhill Drive, Suite 230, San Bruno, CA, 94066, United States of America, yinkai1000@gmail.com 1 - Trip Splitting Approximation to Link Travel Time Estimation on a Transportation Network Kai Yin, Nomis Solutions, 1111 Bayhill Drive, Suite 230, San Bruno, CA, 94066, United States of America, yinkai1000@gmail.com, Bruce Wang, Wen Wang, Teresa Adams We study link travel time estimation using time stamps of trips on a transportation network. Assuming that each link associates with a random travel time, a statistical inference method, trip splitting approximation, is proposed to deal with a general link travel time distribution and also address the case that the routes of some trip observations are unknown. We explore an iterative procedure analogous to the Expectation-Maximization (EM) algorithm and apply the Bayesian method for solutions. The properties of solutions and the effect of data quality are examined. 2 - Map Inference using Probe Vehicle GPS Data with Low Reporting Frequency Wen Jin, Tsinghua University, 519A Shunde Building, Beijing, 100084, China, jinw11@mails.tsinghua.edu.cn, Hai Jiang The accuracy of road maps is critical to providing quality navigation services. As rapid urbanization takes place in Chinese cities, there are frequent changes to road networks due to the expansion of the city to suburban areas as well as urban redevelopment in developed districts. Most, if not all, existing literature on map inference problems requires GPS data with high reporting frequency (>= 1 Hz). In this talk, we present a network optimization based approach to solve the map inference problem when the reporting frequency is low. Our algorithm is validated using real data from a leading navigation service provider in China. 3 - Predicting the Status of Traffic Signals using GPS Data from Probe Vehicles It has been recognized that intersection delay accounts for over 1/3 of the total travel time for a trip in the city. To provide quality route guidance to drivers, it is therefore critical to be able to estimate the delays caused by traffic signals. Since intersection delays are dependent on the status of traffic signals, in this talk we develop models to estimate key operating parameters of intersection signals: their cycle lengths as well as the corresponding signal timings. We test our models in the city of Beijing and results show that satisfactory results can be obtained in a wide variety of scenarios. SB72 72-Room 203A, CC QSR Student Introduction and Interaction and Best Student Poster Competition Sponsor: Quality, Statistics and Reliability Sponsored Session Chair: Kaibo Wang, Associate Professor, Tsinghua University, Department of Industrial Engineering, Beijing, China, kbwang@tsinghua.edu.cn Co-Chair: Hui Yang, Associate Professor, Pennsylvania State University, 310 Leonhard Building, Industrial and Manufacturing Eng., State College, PA, 16801, United States of America, huy25@psu.edu 1 - QSR Student Introduction and Interaction and Best Student Poster Competition Xin Qi, Department of Industrial Engineering, Tsinghua University, Shunde Building 519A,Tsinghua University, Beijing, China, qixin19900808@126.com, Hai Jiang

Kaibo Wang, Associate Professor, Tsinghua University, Department of Industrial Engineering, Beijing, China, kbwang@tsinghua.edu.cn

Student poster exhibition starts at 10:00am in the session room; all are welcomed to arrive earlier to interact with the students. The Introduction&Interaction session formally starts at 11:00am. Each of the student members presents a 2-min elevator speech first; interactions with invited guests are then followed. A lunch gathering in the same room for student members and invited guests with tickets starts at 12:30pm. The student poster exhibition continuous until 1:00pm.

89