Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TA72

2 - Time-expanded Network for Space Mission Planning Koki Ho, University of Illinois at Urbana-Champaign, 104 S. Wright Street, Talbot Laboratory, Urbana, IL, 61801, United States This presentation develops a technique based on time-expanded network for human space mission planning and scheduling. As we explore distant destinations like Mars or asteroids, we need a series of flight missions to accomplish the objectives. Logistics and mission planning considerations are critical for those applications. This research will provide a unique solution to this problem using time-expanded network optimization. 3 - Branch-cut-and-price for the Vehicle Routing Problem with Time Windows and Convex Node Costs Two critical yet frequently conflicting objectives for transportation service companies are improving customer satisfaction and reducing transportation cost. Given a network of customer requests with preferred service times, it is challenging to find vehicle routes and service schedules simultaneously that respect all operating constraints and minimize the total transportation and customers’ inconvenience costs. We introduce the Vehicle Routing Problem with Time Windows and Convex Node Costs, in which we model a customer’s inconvenience cost as a convex function of its service start time. We propose a branch-cut-and-price algorithm to solve the problem with general convex cost functions. 4 - The Non-Homogeneous Time CIRP Felipe Lagos, Georgia Institute of Technology, H. Milton Stewart School, 755 Ferst Drive NW, Atlanta, GA, 30332, United States, Natasha Boland, Martin Savelsbergh We study a continuous time variant of the Inventory Routing Problem (CIRP). The CIRP is the problem of routing vehicles in space and time to deliver product to customers whose demand is a continuous function of time so that customers are never short of product and so that the total delivery cost is minimized. We propose new integer programming models to find good primal and dual bounds for this problem, using these to develop a Dynamic Discretization Discovery algorithm. We prove finite convergence of the algorithm and present computational results, which illustrate the behaviour of the algorithm under different instance characteristics. n TA72 West Bldg 211A Joint Session INFORMS Prize/Practice: 2018 INFORMS Prize Emerging Topic: INFORMS Prize Emerging Topic Session Chair: Tarun Mohan Lal, Mayo Clinic, 200 1st Street Sw, Rochester, MN, 55905, United States 1 - BNSF -2018 INFORMS Prize Winner Nathaniel O. Richmond, BNSF Railway, The Colony, TX, 75056, United States BNSF Railway has more than 41,000 employees working to move freight across its sprawling 32,500-mile rail network. To support its complex operational challenges and drive efficiency, BNSF has embraced a data-driven mentality throughout its business. BNSF was awarded the 2018 INFORMS Prize for pioneering and integrating operations research and analytics programs into its organization. This presentation will highlight several innovative applications of these techniques in areas such as capacity expansion, crew management, and equipment maintenance at BNSF. Qie He, University of Minnesota, 111 Church Street SE, Minneapolis, MN, 55455, United States, Stefan Irnich, Yongjia Song

n TA70 West Bldg 106B Estimation Issues in Frontier Estimation Emerging Topic: Productivity, Efficiency and Data Envelopment Analysis Emerging Topic Session Chair: John Ruggiero, University of Dayton, Dayton, OH, 45469-2251, United States 1 - A New Family of Copulas, with Application to Estimation of a Production Frontier System Artem Prokhorov, University of Sydney Business School, Sydney, Australia We consider a system of equations where one equation is a production function and the others are the first order conditions for cost minimization. The production function equation contains a one-sided error that represents technical inefficiency. The cost minimization equations contain errors that represent allocative inefficiency. If technical and allocative inefficiency are not independent, we encounter the issue that common copulas do not capture the type of dependence that the economic model implies. We design a new copula family that does capture the right type of dependence and illustrate how to use it in practice. This is joint work with C Amsler and P Schmidt. 2 - Hinging Hyperplanes as a Nonparametric Estimator Ole Bent Olesen, University of Southern Denmark, Campusvej 55, Odense M, DK-5230, Denmark, John Ruggiero Nonparametric estimators with shape constraints of production functions/frontiers have recently received attention in the literature. In this paper we focus on the hinging hyperplane approach suggested by Breiman in 1993. An unknown function is approximated by a sum hinge functions, where each hinge function can be expressed as a sum of two hyperplanes. The intersection of the hyperplanes is denoted a hinge. Using simulation we show how well the hinging hyperplane approach performs compared to other nonparametric estimators. We define several types of hinges and discuss the possibility of using hinge finding John Ruggiero, University of Dayton, Dayton, OH, United States In this paper, we analyze an error structure defined by the convolution of a normal distribution and an half normal distribution. The resulting skewed normal distribution has been widely used in production economics, perhaps for mathematical convenience. The normal and half-normal distributions allow modeling of statistical noise and production efficiency, respectively. We will focus on estimation of this error structure given its relationship to other distributions. 4 - Using DEA to Eliminate Unnecessary Greenhouse Gas Emissions and Save The Planet Thomas R. Sexton, Stony Brook University, Stony Brook, NY, United States, Christine Pitocco We use DEA to analyze the greenhouse gas emissions (GHG) of 172 countries using total GHG as an input, gross domestic product as an output, and human development index and population as site characteristics. Our preliminary analysis finds that 20 countries lie on the efficient frontier and that all countries can reduce their total greenhouse gases by approximately 37%. n TA71 West Bldg 106C Time-Expanded Integer Programming Sponsored: Computing Sponsored Session Chair: Natashia Boland, Georgia Institute of Technology, Atlanta, GA 1 - Solving Time-Dependent Path Problems with Dynamic Discretization Discovery Natashia Boland, Georgia Institute of Technology, Ferst Drive, Atlanta, GA, 30332, United States, Edward He, George L. Nemhauser, Martin W. P. Savelsbergh Routing in congested networks prompts the need to find paths and associated departure times so as to minimize path duration, where arc travel time depends on the time at which the arc is traversed. We develop an exact dynamic discretization discovery method for the case of piecewise linear travel time functions. Dynamic discretization enables these problems to be solved to optimality by exploring only a small fraction of the breakpoints. Computational experiments show that this fraction decreases as key instance parameters increase, making the algorithm highly scalable. We also generalize the algorithm to minimize total travel time, and demonstrate that it has similar advantages. algorithms to decide which types of hinges to use in a grid search. 3 - The Half-Normal Normal Stochastic Frontier Model: An Improved Understanding

269