2016 INFORMS Annual Meeting Program

MB61

INFORMS Nashville – 2016

MB61 Cumberland 3- Omni Practical Steps Towards Shipment & Network Capacity Management Sponsored: Railway Applications Sponsored Session Chair: Carl D Van Dyke, TransNetOpt, 6 Snowbird Ct, West Windsor, NJ, 08550, United States, carl@cvdzone.com 1 - Managing Train Scheduling To Optimize Network Capacity Dharma Acharya, Transport Consultant, acharya.dharma@gmail.com To move North American rail freight efficiently and reliably, an option for the railroads is to lock down on running of all scheduled and unscheduled trains over their rail network at least a few days in advance. This way railways will be able to line up resources at the right place at right time and avoid/minimize any resource waste and train delays. Railroads will also be able to better plan when trains could meet and pass on their predominantly single track corridors and be able to better predict train/shipment ETAs. We will also discuss what kind of changes in railroad’s operational practice will be needed from their traditional philosophy of running trains whenever there is enough tonnage/shipments. 2 - Managing Intermodal Capacity Through Differentiated Service Products And Load Acceptance Carl Van Dyke, TransNetOpt, carl@cvdzone.com As intermodal grows in sophistication and volume, it becomes important to employ effective means to manage capacity to ensure customer service expectations are met, and total revenue & profits are maximized. Currently this is being done by providing differentiated service, and adjusting the underlying terminal and train operations to both match the promised service and better balance variations of traffic volumes between services. An attitude of unlimited capacity is giving way to adopting specific constraints on capacity & the introduction of load acceptance processes. These capacity management strategies, plus some potential new ones that could be adopted in the near future, will be explored. 3 - Unit Train Management System Bob Golbasi, CSX Transportation, Jacksonville, FL, United States, Hakan_Golbasi@csx.com, Robert Gutman Unit Train Management System (UTMS) is a comprehensive system that was developed at CSX Transportation to bring all the relevant unit trains information together into one planning tool. A unit train is a special order train of only one commodity type that is not on a fixed scheduled. Shippers, Receivers and CSXT Unit Train Managers all work together in UTMS to ensure full visibility and alignment of upcoming trains. UTMS includes an optimization model to accept, modify or reject reservations in the selected time period based present business conditions and current and predicted availability of right cars at the right place at the right time. 4 - Connected Driver Advisory System: Cost Efficient Way For Improving Rail Traffic Management Per Leander, Transrail Sweden AB, per.leander@transrail.se Trains and Traffic Management in co-operation. The presentation will explain the concepts of C-DAS (Connected Driver Advisory System) and C-Cruise (Connected Cruise) for punctual and eco-efficient operation of trains and improved Traffic Management at low cost. These are concepts currently developed and deployed in Europe in order to improve punctuality, capacity and sustainability and to reduce costs. The algorithm developed by Transrail Sweden AB may be used in all types rail operations and the cost function be tuned to the specific needs of the operation. The algorithm may also be used as a powerful engine for future eco- efficient and interoperable ATO.

4 - Integrated Mode Choice And Assignment-simulation Framework With Automated Transit Vehicles Omer Verbas, Northwestern University, Transportation Center, Evanston, IL, 60208, United States, omer@northwestern.edu Hani S Mahmassani, Michael F. Hyland With the advent of automated and connected transportation systems, car and bike sharing, ride sourcing, and on-demand transit services, as well as the increasing availability of real-time traffic and transit information; travelers have the opportunity to evaluate their multiple routing options and make better-informed decisions. This study proposes an integrated mode-choice and a path finding- assignment-simulation framework that evaluates the system performance and traveler behavior under the existence of automated transit vehicles.

MB60 Cumberland 2- Omni Latest Advances in Last Mile Distribution Sponsored: TSL, Urban Transportation Sponsored Session

Chair: Mathias A Klapp, PhD Candidate, Georgia Institute of Technology, 765 Ferst Dr NW, Atlanta, GA, 30318, United States, maklapp@gatech.edu 1 - Complexity Of Dynamic Delivery Problems With Release Dates And Deadlines Damian Reyes, Georgia Institute Of Technology, Atlanta, GA, United States, ldrr3@gatech.edu, Alan Erera, Martin W P Savelsbergh Motivated by a case-study in food delivery operations, we investigate the complexity of dynamic delivery problems with release times and service guarantees. At the heart of these problems, there is a trade-off between waiting to consolidate more orders - enabling cost-effective delivery routes - and dispatching a vehicle earlier - in order to complete some orders while preserving capacity for others released later in the operating period. We introduce polynomial-time algorithms for some deterministic variants on a 1-dimensional geometry. 2 - Consolidating Last-mile Delivery Flows Niels Agatz, Erasmus University, Rotterdam, Netherlands, nagatz@rsm.nl, Joydeep Paul, Remy Spliet Most multi-channel retailers offer in-store pickup to their online customers. Pickup orders are typically shipped from a dedicated e-fulfilment warehouse while store replenishment takes place from another warehouse. In this contribution, we study the opportunity to use the excess capacity in the replenishment routes to accommodate some of the in-store pickup demand. We develop a heuristic to support the consolidation decisions and present numerical Felipe Lagos, Georgia Institute Of Technology, falg3@gatech.edu We study a probabilistic VRP in which a customer’s appearance is uncertain. Customers are divided into routes within which the vehicle may skip customers that do not appear, and the objective is minimizing expected routing cost. We propose a column generation algorithm that uses successively tighter cost approximations to solve the problem within any given numerical tolerance. We also provide an a priori guarantee on the number of iterations needed to satisfy any tolerance, which can be calculated from problem parameters. We embed the column generation framework into an exact branch-and-price algorithm, and test our methods on instances adapted from the literature. 4 - Cost Efficiency Versus Customer Service In The Dynamic Dispatch Waves Problem Mathias A Klapp, PhD Candidate, Georgia Institute of Technology, 755 Ferst Drive NW, Atlanta, GA, 30332, United States, maklapp@gatech.edu, Mathias A Klapp, PhD Candidate, Pontificia Universidad Católica de Chile, Vicuña Mackenna 4860, Macul, Santiago, Chile, maklapp@gatech.edu, Alan Erera, Alejandro Toriello We study the Dynamic Dispatch Waves Problem that models the trade-offs between vehicle dispatch, route sequencing, and request selection in same-day delivery systems with dynamic disclosure of orders. The objective is to minimize vehicle travel time (efficiency) and penalties for unserved requests (service). We provide an optimal solution to the deterministic and a priori problems, and design two heuristic dynamic policies. Our computational experiments indicate that in the efficient frontier there is a decreasing marginal rate of substitution between efficiency and service, and that frequency and structure of vehicle dispatches significantly change between these two objectives. experiments based on artificial and real-world data. 3 - Branch-and-Price For Probabilistic VRP

169