INFORMS Miami 2001 Annual Meeting Sponsor:
TSS

Using OR for Transit Operations & Customer Information

SA34.1 Information Critical Arcs for On-Line Routing

• Baiyu Yang; Pennsylvania State University, Dept. of Civ. & Environ. Eng., University Park, PA 16802; hxt14@psu.edu
• Elise Miller-Hooks; United Technologies Research Center, 411 Silver Lane, East Hartford, CT 06118; millered@utc.utrc.com

We present a mathematical formulation and heuristic procedure for selecting a subset of network arcs (due to monetary constraints), called information critical arcs, for which real-time travel information will be collected. A real-time routing algorithm is used in a stochastic, time-varying environment to provide path guidance given this real-time information. Solutions are evaluated.

SA34.2 Investigating Information Quality & Effectiveness of Alternative Real-Time Information Supply Strategies

• Karthik Srinivasan; Vanderbilt University, Dept. of Civil & Environ. Eng., 206B Jacobs Hall, Nashville, TN 37235; karthik.k.srinivasan@vanderbilt.edu
• Ke Chen; Vanderbilt University, Dept. of Civil & Environ. Eng., 206B Jacobs Hall, Nashville, TN 37235; ke.chen@vanderbilt.edu

We investigate the effect of alternative information supply strategies on information quality and system performance. Specifically, differences between VMS-based information and in-vehicle devices are examined using simulation experiments. The performance of the information strategies under various driver response behaviors, incident rates and recurrent congestion conditions is reported.

SA34.3 Using Current Bus Location Information in Passenger Itinerary Planning

• Mark Hickman; University of Arizona, Civil Eng. & Eng. Mechanics; mhickman@engr.arizona.edu

Many transit agencies can monitor current bus locations. It is possible to use this information to inform passengers about better paths (itineraries) from their origin to their destination. Using stochastic and time-dependent shortest paths, we present an algorithm for generating passenger itineraries that uses real-time bus location information.

SA34.4 Effects of Demand Patterns & Bus Holding on Real-Time Passenger Information

• Rabi Mishalani; Ohio State University, Dept. of Civil & Environ. Eng., Columbus, OH 43210; mishalani.1@osu.edu
• Mark McCord; Ohio State University; mccord.2@osu.edu
• Chi-Ping Lam; Ohio State University;

An evaluation methodology that quantifies the value of bus arrival information systems to passengers waiting at bus stops under various passenger arrival patterns and bus holding policies is presented. Transit demand and supply are modeled as a stochastic system with an operator providing optimal passenger information. Simulation results are discussed.

New Developments in Logistics & Transportation Planning

SA35.1 Planning Virtual Inventories

• Apostolos N. Burnetas; Case Western Reserve University, Dept. of Ops., WSOM, 10900 Euclid Ave., Cleveland, OH 44106-7235; atb4@po.cwru.edu
• Ronald H. Ballou; Case Western Reserve University, Dept. of Operations, WSOM, Cleveland, OH 44106; rhb4@po.cwru.edu

Supplying customer demand from alternate inventory locations when an item is out of stock at its primary stocking point creates a virtual inventory for that item. The expectation is that if more inventories can be drawn upon, the inventory for an item would be lower, the fill rate would be higher, or both. While generally true for safety stock, regular stocks, on the other hand, rise with such cross filling of demand...

SA35.2 Capacity & Flexibility Assessment of Transportation Systems

• Denny Cho; University of Pennsylvania, Dept. of Systems Eng., Philadelphia, PA 19104-6315; denny@seas.upenn.edu
• Edward K. Morlok; University of Pennsylvania, Dept. of Systems Eng., Philadelphia, PA 19104-6315; morlok@seas.upenn.edu
• Zhi-Long Chen; University of Pennsylvania, Dept. of Systems Eng., 220 South 33rd St., Philadelphia, PA 19104-6315; zlchen@seas.upenn.edu

We review existing models and approaches for assessing capacity and flexibility of a transportation system. New models, metrics and solution approaches based on optimization and simulation will be presented.

SA35.3 Supply Chain Networks with Competition

• Anna Nagurney; University of Massachusetts, Isenberg Sch. of Mgmt., Amherst, MA 01003; nagurney@gbfin.umass.edu
• June Dong; SUNY, Coll. of Bus., Oswego, NY 13126; dong@oswego.edu
• Ding Zhang; SUNY, Sch. of Business, Oswego, NY 13126; zhang@oswego.edu

We develop a framework for the formulation, analysis and computation of solutions to supply chain network problems with competition. We identify the network structure of the supply chain problem in equilibrium and utilize the variational inequality formulation for the prediction of the prices and shipments.

SA35.4 Vehicle Routing with Cross Docking

• Zhi-Long Chen; University of Pennsylvania, Dept. of Systems Eng., 220 South 33rd St., Philadelphia, PA 19104-6315; zlchen@seas.upenn.edu
• Hang Xu; ;

We consider vehicle routing problems where there are some cross docks available in the underlying network. An order can be either directly delivered without using a cross dock or delivered through a cross dock. We propose heuristics for solving such problems and analyze their asymptopic performance.

OR/MS Tools for Public Transit Planning

SB34.1 Miami International Airport: Automated People Mover, Passenger Flow Simulation

• Gediz Er; Lea+Elliot, Inc., 14325 Willard Rd., Ste. 200, Chantilly, VA 20151; ger@leaelliott.com
• Deanna Doan; Lea+Elliot, Inc., 14325 Willard Rd., Ste. 200, Chantilly, VA 20151; ddoan@leaelliott.com

Discrete event simulation was used system-wide to identify bottlenecks and interdependencies at the planning stage. Alternative baggage handling scenarios were evaluated. Decision criterion was Level of Service on platforms, including queue space, aisle width, and vertical circulation requirements.

SB34.2 A Hybrid Scheduling Approach for Pickup & Delivery of Para-Transit Demand

• Majid Aldaihani; University of Southern California, Dept. of ISE, Los Angeles, CA 90089-0193;
• Maged Dessouky; University of Southern California, Dept. of ISE, 3715 McClintock Ave., Los Angeles, CA 90089-0193; maged@rcf.usc.edu

Most of the scheduling literature on para-transit systems are for strictly curb-to-curb systems. With the increased demand and usage of these systems, alternative methods for transporting elderly and disabled passengers need to be developed. We present a heuristic that schedules these type of passengers that integrates a curb-to-curb system with a fixed route line.

SB34.3 An Assessment of Integrated Transit Service

• Mark Hickman; University of Arizona, Civil Eng. & Eng. Mechanics; mhickman@engr.arizona.edu

Many transit agencies are considering integrating their fixed-route and demand-responsive transit service. We present a methodology that illustrates one way to schedule such an integrated service. The technique is illustrated using an extensive case study from Houston, Texas, illustrating both passenger and agency benefits of service integration.

SB34.4 An Exact Algorithm for the Multiple Vehicle Pickup & Delivery Problem

• Quan Lu; University of Southern California, Dept. of ISE, Los Angeles, CA 90089-0193;
• Maged Dessouky; University of Southern California, Dept. of ISE, 3715 McClintock Ave., Los Angeles, CA 90089-0193; maged@rcf.usc.edu

We develop an optimal solution procedure for the multiple vehicle pickup and delivery problem. This problem has numerous applications such as the dial-a-ride problem for the transportation of elderly and disabled passengers. As opposed to other optimal solution procedures, our approach does not require either tight capacity or time window constraints.

Dynamic Vehicle Routing

SB35.1 Routing On-Line Service Orders Dynamically

• Buyang Cao; ESRI, Inc., 380 New York St., Redlands, CA 92373; bcao@esri.com
• Burcin Bozkaya; ESRI, 380 New York St., Redlands, CA 92373; bbozkaya@esri.com

We present an e-tailer system that accepts the on-line service orders, assigns the orders to the service persons and routes the service orders dynamically. The core algorithms of the system are discussed, and the system architecture is described.

SB35.2 withdrawn 10/27: Territory Planning & Vehicle Dispatching with Stochastic Custers & Demand

• Hongsheng Zhong; University of Southern California, Dept. ISE, Los Angeles, CA 90089-0193; hzhong@ups.com

SB35.3 Analysis of Day-to-Day Network Flow Evolution in Transportation Networks: Some Simulation Experiments

• Karthik Srinivasan; Vanderbilt University, Dept. of Civil & Environ. Eng., 206B Jacobs Hall, Nashville, TN 37235; karthik.k.srinivasan@vanderbilt.edu

We investigate the phenomenon day-to-day dynamics in transportation networks. In this study, realistic user-behavior rules (calibrated and validated in earlier studies) are used to simulate network evolution both within-day and day-to-day. The results are compared against conventional benchmarks: time-dependent user equilibrium and system optimal performance.

SB35.4 Dynamic Coordination for Load-Constrained Vehicle Routing with Uncertain Demands

• Alan L. Erera; Georgia Institute of Technology, Sch. of ISyE, Atlanta, GA 30332-0205; alan.erera@isye.gatech.edu
• Carlos F. Daganzo; University of California, Dept. of CEE, 109 McLaughlin Hall #1720, Berkeley, CA 94720-1720; daganzo@ce.berkeley.edu

We consider a single-period load-constrained VRP in which customer lot sizes are uncertain when planning. Operating strategies that utilize dynamic coordination can reduce costs (fleet and distance) substantially over traditional non-coordinated a priori strategies. We analyze coordinated strategies using an approximation approach, and verify results via simulation.

Information Integration for Network Traffic Management

SC34.1 Real-Time Consistent-Normative Traveler Information Supply Strategies: Applications to Network Traffic Management

• Khaled F. Abdelghany; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; kfaissal@mail.utexas.edu
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

A mathematical formulation and solution algorithm for the problem of generating real-time consistent-normative information to users are presented. The formulation seeks to provide users with predicted link travel time information which induces route choice decisions that achieve system wide objectives (better spatial and temporal traffic distribution in the network), while at the same times minimizing inconsistency between supplied and experienced travel times.

SC34.2 A Coordination Scheme of Incident Management using Telematics Devices

• Henry X. Liu; University of California, California PATH, ATMIS Ctr., Inst. of Transport Studies, Irvine, CA;
• Rachel R. He; Princeton University, Dept. of OR & Financial Eng.;
• Bin Ran; , Dept. of Civil & Environ. Eng., Madison, WI;

Currently, many incident management strategies focus on diverting traffic through roadside devices such as CMS, traffic signal and ramp control. With the fast-growing telematics market and maturing traffic information services, telematics devices provide another feasible means to manage incidents more efficiently. An integrated coordination scheme for incident management is developed and evaluated using Paramics simulation.

SC34.3 Multicriteria Network Equilibrium Modeling with Variable Weights for Decision-Making in the Information Age

• Anna Nagurney; University of Massachusetts, Isenberg Sch. of Mgmt., Amherst, MA 01003; nagurney@gbfin.umass.edu
• June Dong; SUNY, Coll. of Bus., Oswego, NY 13126; dong@oswego.edu
• Patricia L. Mokhtarian; University of California, Dept. of Civil & Environ. Eng., Davis, CA 95616; plmokhtarian@ucdavis.edu

We develop a multicriteria network equilibrium framework for modeling decision-making in the information age. The decisions take place on a network in which the links can be either physical as in the case of transportation or virtual. We apply the modeling schema to teleshopping and telecommuting.

SC34.4 An On-Line Algorithm for Characterizing Non-Recurrent Congestion

• Wei Hua Lin; University of Arizona, SIE Dept., Tucson, AZ 85721;

We consider a methodology for distinguishing non-recurrent congestion from recurrent congestion based on transformed occupancy information from loops. The method is capable of detecting the occurrence of an incident and estimating the magnitude of the incident. It requires little effort for calibrating its parameters and can be readily implemented on-line.

Dynamic Modeling of Common Carriers

SC35.1 Flight Delays, Cancellations & the Impact on Airline Passengers

• Stephane Bratu; MIT, Ctr. for Transport. Studies, Cambridge, MA 02139; sbratu@mit.edu
• Cynthia Barnhart; MIT, 77 Massachusetts Ave., Rm. 1-229, Cambridge, MA 02139; cbarnhar@mit.edu

As part of the Global Airline Industry Program at MIT, we investigate the impact of aircraft flight delays and cancellations on passenger service. Using operations and passenger booking data, we analyze the effects of schedule design and disruption, recovery strategies and policies, and load factors. We describe a case study using data from the airline service quality performance database and from major US airlines.

SC35.2 Dynamic Management of a Flexible Transit System

• Teodor Gabriel Crainic; University of Montreal, CRT, CP 6128, Succ Centre-Ville, Montreal, Quebec, H3C 3J7 , Canada; theo@crt.umontreal.ca
• Federico Malucelli; Politecnico, Milano, , Italy; malucell@elet.polimi.it
• Maddalena Nonato; University of Perugia, Perguia, , Italy; nonato@diei.unipg.it

We study a transit system where several lines may be operated to respond to dynamically posted passenger requests. In a flexible environment, buses are detourred to pick up and drop off passengers who made explicit requests. The system must determine the actual itinerary for each passenger, as well as the resulting routing and schedule. We will discuss modeling and algorithmic issues, as well as implementation challenges in a dynamic setting.

SC35.3 A Multi-Agent Approach for Stochastic Multicommodity Flow Problems with Applications in Fleet Management

• Huseyin Topaloglu; Princeton University, Dept. of OR & Financial Eng., Princeton, NJ 08544; topalglu@princeton.edu
• Warren B. Powell; Princeton University, Dept. of OR & Financial Eng., Princeton, NJ 08544; powell@princeton.edu

Problems in empty car distribution (railroads) and container management can be formulated as stochastic, integer multicommodity flow problems. We show that when we use nonlinear functional approximations in a multiagent setting, the problem reduces to sequences of pure networks. Numerical experiments demonstrate the effectiveness of the method.

OR/MS Methods in Infrastructure Management

SD34.1 Infrastructure State Transition Probability Computation using Duration Models

• Rabi Mishalani; Ohio State University, Dept. of Civil & Environ. Eng., Columbus, OH 43210; mishalani.1@osu.edu
• Samer Madanat; University of California, Dept. of Civil & Environ. Eng., 114 McLaughlin Hall, Berkeley, CA 94720; madanat@ce.berkeley.edu

State-based infrastructure deterioration models consist of transition probabilities capturing the evolution of condition over time. Current methods for determining such probabilities suffer from several serious limitations. An alternative approach is presented where time-based duration models are used. The methodology is demonstrated using a reinforced concrete bridge deck data set.

SD34.2 Determining the Value of the Air Traffic Infrastructure

• Robert B. Rovinsky; Federal Aviation Administration, 800 Independence Ave. SW, Washington, DC 20591; robert.rovinsky@faa.gov

The Civil Engineering Society recently rated the nation's air traffic control infrastructure as poor. Since airspace availability exceeds 99.9%, justifying additional funding is difficult. The FAA is connecting its infrastructure to services and determining 'derived value,' thus creating a business case to modernize and evolve its infrastructure.

SD34.3 Transportation Asset Management under Uncertainty

• Jeffrey Fine; Northwestern University, IEMS Dept., Evanston, IL 60208; j-fine@northwestern.edu
• Thanasis Ziliaskopoulos; Northwestern University, CEE Dept., Evanston, IL 60208; a-z@northwestern.edu

Transportation infrastructure problems, in particular physical assets, are modeled effectively as capital investment problems. We discuss issues of uncertainty of the value of the asset, of increasing uncertainty with respect to time, the use of assets as insurance, present and future values/utilities associated with the asset, multiple time/exercise opportunities, and risk-neutral/averse assumptions. Financial models are fundamentally concerned with long-term results and the long-term behavior of assets...

SD34.4 Optimal Infrastructure Facility Maintenance & Repair Policies under Uncertainty in Deterioration: An Adaptive Control Approach

• Pablo L. Durango-Cohen; University of California, IEOR Dept., Berkeley, CA 94720-1777; durango@ieor.berkeley.edu
• Samer Madanat; University of California, Dept. of Civil & Environ. Eng., 114 McLaughlin Hall, Berkeley, CA 94720; madanat@ce.berkeley.edu

Agencies must often develop infrastructure M&R policies with limited information about facility deterioration. We will present 2 AC models that explicitly account for the uncertainty in characterizing a facility's actual deterioration rate. These methods use infrastructure condition data obtained during operation of a facility to improve the characterization over a finite planning horizon through Bayesian updating. We show that economic benefits can be achieved.

Inventory Routing

SD35.1 Separable Pricing for Stochastic Inventory Systems

• Dan Adelman; University of Chicago, Grad. Sch. of Bus., 1101 East 58th St., Chicago, IL 60637; dan.adelman@gsb.uchicago.edu
• Anton J. Kleywegt; Georgia Institute of Technology, Sch. of ISyE, Atlanta, GA 30332-0205; anton@isye.gatech.edu

We study a general class of stochastic inventory management problems that includes the inventory routing problem. We derive and implement a math program whose optimal prices yield the 'best' (in a certain sense) separable functional approximation to the dynamic programming optimal value function.

SD35.2 The Dynamic Inventory Routing Problem

• Chi-Guhn Lee; University of Michigan, Dept. of IOE, 1205 Beal Ave., Ann Arbor, MI 48109-2117; chiguhn.lee@umich.edu
• Yavuz A. Bozer; University of Michigan, Dept. of IOE, Ann Arbor, MI 48109; yauz.bozer@umich.edu
• Chelsea C. White, III; University of Michigan, Dept. of IOE, 1205 Beal Ave., Ann Arbor, MI 48109-2117; ccwiii@umich.edu

We develop a heuristic, based on simulated annealing and LP, for a deterministic dynamic inventory routing problem involving multiple suppliers and one plant. The objective is to minimize the sum of transportation and inventory holding costs. Numerical results suggest the heuristic is computationally efficient and provides a high quality solution.

SD35.3 The Deterministic Repeatable Inventory Routing Problem

• Charles E. Noon; University of Tennessee, 613 Stokes Mgmt. Ctr., Knoxville, TN 37996; cnoon@utk.edu
• Anurag Agarwal; University of Tennessee, 2360 Cherahala Blvd., #J20, NTRC, Knoxville, TN 37932; aagarwal@utk.edu

The DRIRP seeks a minimum cost-per-unit solution of routes and delivery (pickup) amounts for servicing customers with deterministic consumption (production) rates and limited storage. Solution repeatability is achieved by constraining the initial and final vehicle locations and inventory positions to be identical. We present computational results from a cutting-plane approach.

SD35.4 The Vehicle Routing Problem with Flexible Quantities

• Ann Melissa Campbell; University of Iowa, Tippie Coll. of Bus., 108 Pappajohn Bus. Bldg., Iowa City, IA 52242-1000; ann-campbell@uiowa.edu

We examine extensions of the vehicle routing problem where flexibility exists in the delivery quantity to the customers. When delivery volume varies within a range or is time dependent, we study how research on the VRP can be modified for these new problems and the savings this additional flexibility offers.

Network Equilibrium Models I

MA34.1 withdrawn 10/31: A Random System Equilibrium Model of Location & Travel Choices

• Yong Zhao; University of Texas, Dept. of Civil Eng., ECJ 6.512, Austin, TX 78712; yzhao@mail.utexas.edu
• Kara Kockelman; University of Texas, Dept. of Civ. Eng., ECJ 6.512, Austin, TX 78712; kkockelman@mail.utexas.edu

MA34.2 A Stochastic Dynamic Trip Assignment Model for Intermodal Transportation Networks

• Khaled F. Abdelghany; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; kfaissal@mail.utexas.edu
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

We present a stochastic dynamic trip assignment model for urban intermodal transportation networks. We formulate the problem as mathematical program, for which a solution algorithm is developed. The model captures the interaction between mode choice and trip assignment through the implementation of a multi-objective shortest path algorithm.

MA34.3 Computational Performance for an Origin-Based Combined Model of Travel Demand & Route Assignment

• Hillel Bar-Gera; Ben-Gurion University of the Negev, Dept. of IEM, Beer-Sheva, 84105 , Israel; bargera@bgumail.bgu.ac.il
• David E. Boyce; University of Illinois, Dept. Civil & Materials Eng., 842 West Taylor St., Chicago, IL 60606-7023; dboyce@uic.edu

Travel forecasting models combining demand and network assignment are receiving increased attention. A fixed point formulation for general combined models is presented, and measures of solution accuracy are discussed. An origin-based algorithm for solving such models is proposed computational results demonstrate its efficiency in comparison with prevailing alternatives.

MA34.4 Implementation of a Complex Planning Model for the Southern California Association of Governments

• Michael A. Florian; INRO Solutions, 5160 Decaarie Blvd., Ste. 610, Montreal, Quebec, H3C 3J7 , Canada; mike@inro.ca
• Jia Hao Wu; INRO Solutions, 5160 Decarie Blvd., Ste. 610, Montreal, Quebec, H3C 3J7 , Canada;
• Shuguang He; INRO Solutions, 5160 Decarie Blvd., Ste. 610, Montreal, Quebec, H3C 3J7 , Canada;

An implementation of a complex transportation planning method developed by SCAG is described. The demand model has several trip purposes, hierarchical logit mode choice functions for each purpose and multiple classes of users. The network assignment is an asymmetric cost multi-class network equilibrium method.

Time-Dependent Vehicle Routing

MA35.1 A Genetic Algorithm for the Dynamic Vehicle Routing Problem with Time-Dependent Travel Times

• Ali E. Haghani; University of Maryland, Dept. of Civil Eng., College Park, MD 20742; haghani@eng.umd.edu
• Soojung Jung; University of Maryland, Dept. of Civil Eng., College Park, MD 20742;

We propose and test a GA for the time-dependent VRP. The formulation of the problem considers multiple vehicles with different capacities, pick-up or delivery demands with soft time-windows, real-time service requests and real-time variations in travel times between demand nodes. Numerical results are reported, including comparisons against lower bounds.

MA35.2 Dispatching Policies for Vehicle Routing under Different Systems Loads

• Guntram Noeth; , Bremenweg 34, Wuerzburg, Barvaria, 97084 , Germany; guntram_noeth@hotmail.com
• Randolph W. Hall; University of Southern California, Dept. of ISE, 3715 McClintock Ave., Los Angeles, CA 90089-0193; rwhall@usc.edu
• Maged Dessouky; University of Southern California, Dept. of ISE, 3715 McClintock Ave., Los Angeles, CA 90089-0193; maged@rcf.usc.edu

For simple vehicle routing environments, we analytically compare system performance of different dispatching policies under light and heavy traffic. Using simulation, we also show how these results hold under different system loads in more complex environments.

MA35.3 Reoptimization in Real-Time Routing

• Elise Miller-Hooks; United Technologies Research Center, 411 Silver Lane, East Hartford, CT 06118; millered@utc.utrc.com
• Baiyu Yang; Pennsylvania State University, Dept. of Civ. & Environ. Eng., University Park, PA 16802; hxt14@psu.edu

One of the goals of intelligent transportation systems is to provide travelers with en-route routing suggestions given real-time information. We present an efficient reoptimization approach for updating shortest paths as real-time conditions change. The approach is significantly faster than recomputing from scratch and can be used in static or dynamic, deterministic or stochastic environments.

MA35.4 An Examination of Assignment Models for Local Truckload Trucking Problems with Stochastic Service Times & Time Window Constraints

• Amelia C. Regan; University of California, Irvine, CA; aregan@uci.edu
• Xiubin Wang; American Airlines;

We examine alternative assignment models that explicitly consider stochastic handling and travel times and compare the performance of these to static assignment models which are applied in an iterative manner as the stochastic elements of the problem are realized.

Network Equilibrium Models II

MB34.1 Investigating Learning Processes in Commuter Behavior under Real-Time Information

• Karthik Srinivasan; Vanderbilt University, Dept. of Civil & Environ. Eng., 206B Jacobs Hall, Nashville, TN 37235; karthik.k.srinivasan@vanderbilt.edu
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

We provide evidence of learning processes in commuter's route and departure time choice behavior under real-time information. The evidence is obtained using longitudinal discrete choice models from behavioral data observed in interactive simulator based experiments. Specifically, evidence of discriminative and trial and error learning are reported.

MB34.2 A Space-Time Network for Telecommuting vs. Commuting Decision-Making

• Anna Nagurney; University of Massachusetts, Isenberg Sch. of Mgmt., Amherst, MA 01003; nagurney@gbfin.umass.edu
• June Dong; SUNY, Coll. of Bus., Oswego, NY 13126; dong@oswego.edu
• Patricia L. Mokhtarian; University of California, Dept. of Civil & Environ. Eng., Davis, CA 95616; plmokhtarian@ucdavis.edu

We develop a theoretical framework for the study of telecommuting vs. commuting decision-making over a fixed time horizon. We consider multicriteria decision makers and propose a network equilibrium model over a space-time network. An algorithm is given and applied to numerical examples.

MB34.3 A Model of the Pedestrian Network Equilibrium Problem with Bi-Directional Effect

• Ding Zhang; SUNY, Sch. of Business, Oswego, NY 13126; zhang@oswego.edu
• William H. K. Lam; Hong Kong Polytechnic University, Civil & Structural Eng. Dept., Hung Hom, Kowloon, , Hong Kong; cehklam@polyu.edu.hk

We present the first equilibrium model for pedestrian network assignment problems. The model incorporates explicitly the effects of bi-directional flow on increases in walking time and reduction of effective capacity by pedestrian facility. We explore qualitative properties of the model such as existence and uniqueness and provide a numerical example.

MB34.4 A Bilevel Nonlinear Programming Formulation to Estimate Dynamic Origin-Destination Flows from Traffic Counts

• Hossein Tavana; Continental Airlines, Ste. 918D, Houston, TX 77002; htavan@coair.com
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

In generalized least-square estimation of OD flows, the nonlinearity in OD flows and traffic counts is commonly ignored. Moreover, the key parameter, the link-flow proportion matrix, is not constant and itself is a function of the unknown flows. A bilevel nonlinear programming formulation is proposed to address these issues.

Time-Critical Vehicle Routing

MB35.1 Dynamic Coordination for a Deadline Vehicle Routing Problem with Uncertainty

• Alan L. Erera; Georgia Institute of Technology, Sch. of ISyE, Atlanta, GA 30332-0205; alan.erera@isye.gatech.edu
• Carlos F. Daganzo; University of California, Dept. of CEE, 109 McLaughlin Hall #1720, Berkeley, CA 94720-1720; daganzo@ce.berkeley.edu

A deadline VRP arises when a fleet must complete a set of tasks by a time deadline. We consider a deadline problem in which customer service times are uncertain when planning and analyze operating strategies that utilize dynamic coordination with approximation methods.

MB35.2 A Priori Route Construction for the Stochastic Shortest Route Problem with Delivery Deadlines

• Barrett Thomas; University of Michigan, Dept. of IOE, 1205 Beal Ave., Ann Arbor, MI 48109-1227; barrettt@umich.edu
• Chelsea C. White, III; University of Michigan, Dept. of IOE, 1205 Beal Ave., Ann Arbor, MI 48109-2117; ccwiii@umich.edu

We consider the problem of constructing an a priori route when only a subset of a known set of customers is present on any given day and each customer has a delivery deadline associated with it. We introduce a model for this problem, discuss solution techniques and present computational experience.

MB35.3 The A Priori Probabilistic Covering Tour Problem

• Hao Tang; Pennsylvania State University, Dept. of Civil & Environ. Eng., University Park, PA 16802; hxt141@psu.edu
• Elise Miller-Hooks; United Technologies Research Center, 411 Silver Lane, East Hartford, CT 06118; millered@utc.utrc.com

The PCTP seeks an a priori tour with minimum expected length. The presence of any given customer is known a priori with uncertainty. Customers are skipped if they are not present and for some customers, it is sufficient to directly visit any one of a subset of customers. Exact and heuristic procedures are proposed.

Optimization Problems in Traffic Management

MC34.1 A Combined Model for Dynamic Trip Distribution & Traffic Assignment

• Yue Irene Li; Northwestern University, Civil Eng. Dept., Evanston, IL 60208; liyue@northwestern.edu
• Thanasis Ziliaskopoulos; Northwestern University, CEE Dept., Evanston, IL 60208; a-z@northwestern.edu
• David E. Boyce; University of Illinois, Dept. Civil & Materials Eng., 842 West Taylor St., Chicago, IL 60606-7023; dboyce@uic.edu

The combined dynamic trip distribution and traffic assignment problem is formulated as a mathematical programming model. The development of a solution algorithm is discussed. The algorithm uses Dantzig-Wolfe algorithm to decompose the formulation, and the sub-problem is then solved by Lagrangian relaxation and a 3-point function interpolation algorithm.

MC34.2 A Heuristic Solution Method for the Minimum Tool Booth Problem

• Li Hui Bai; University of Florida, ISE Dept., 303 Weil Hall, Gainesville, FL 32611-6595; lbai@cao.ise.ufl.edu
• Donald W. Hearn; University of Florida, Ctr. for Applied Opt., ISE Dept., 303 Weil Hall, Gainesville, FL 32611; hearn@ise.ufl.edu
• Mehmet Bayram Yildirim; University of Florida, ISE Dept., 303 Weil Hall, PO Box 116595, Gainesville, FL 23611-6595; ybayram@cao.ise.ufl.edu

The problem of determining tolls for user-equilibrium traffic assignment so that the solution is system optimal and the number of the booths is minimal is formulated as a mixed integer problem with fixed charge constraints. A dynamic slope scaling algorithm is customized for this problem and numerical results are reported.

MC34.3 Allocating Funds to Highway Improvements under the Presence of Equity Constraints

• George Kozanidis; Northeastern University, 328 Snell Engineering Ctr., Dept. of MIME, Boston, MA 02115; gkozanid@coe.neu.edu
• Emanuel Melachrinoudis; Northeastern University, 375 Snell Engineering Ctr., Dept. of MIME, Boston, MA 02115; emelas@coe.neu.edu
• Marius M. Solomon; Northeastern University/GERAD, 314 Hayden Hall, Coll. of Bus. Admin., Boston, MA 02115; msolomon@cba.neu.edu

We present a model for allocating funds to highway improvements. Both discrete and continuous interventions are considered. Equity constraints that keep a balance on the budget amounts spent on different improvement sets are incorporated. The problem is formulated as a MIP knapsack model with special structure and interesting properties.

MC34.4 Online Adaptive Variable Message Sign Control for Traffic Network Congestion Management

• Nhan Huynh; University of Texas, Dept. of Civ. Eng., ECJ 6.2, Austin, TX 78712; nhuynh@mail.utexas.edu
• Yi-Chang Chiu; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; calmag3@yahoo.com
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

We discuss an online adaptive VMS control scheme for a closed-loop transportation system. The goal is to adaptively provide advisory information and taking into account travelers' response to the provided information so as to achieve optimal network performance.

Computational Issues in Transportation Decision Models

MC35.1 Assessment of Computational Performance of Transportation Network Modeling Software

• Xuesong Zhou; ; xszhou@mail.utexas.edu
• Nhan Huynh; University of Texas, Dept. of Civ. Eng., ECJ 6.2, Austin, TX 78712; nhuynh@mail.utexas.edu
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

When modeling large networks, the computational performance of the software becomes a critical issue. Of interest is the effect of several destination zone aggregation and of temporal and spatial resolution on the software execution speed and memory requirements. We examine these aspects through the use of DYNASMART-P, a dynamic traffic assignment-simulation program. Proposed strategies for reducing memory requirements and increasing execution speed are discussed.

MC35.2 The Dynamic Traveling Salesman Problem: An Examination of Alternative Heuristics

• Xiangwen Lu; University of California, Inst. of Transport. Studies, Irvine, CA 92697; xlu@uci.edu
• Amelia C. Regan; University of California, Irvine, CA; aregan@uci.edu
• Sandra Irani; University of California, Dept. of ICS, Irvine, CA 92697; sirani@uci.edu

We examine the performance of several algorithms for the dynamic TSP. Using results obtained from an M/G/1 queueing system, we provide bounds on and insights into the performance of these algorithms.

MC35.3 An Integrated On-Line Shortest Path & Phase Setting Algorithm

• Dasheng Lou; University of Arizona, SIE Dept., Tucson, AZ 85721;
• Pitu B. Mirchandani; University of Arizona, SIE Dept., Tucson, AZ 85721;

It is possible to jointly provide a recommended route and set signals so that an emergency vehicle travels an optimal route from an origin to a destination with least disruption to the rest of traffic. An integrated on-line algorithm is developed and analyzed it is also evaluated using simulations.

MC35.4 Time Budgets for Real-Time Decision Making

• Pitu B. Mirchandani; University of Arizona, SIE Dept., Tucson, AZ 85721;
• Dasheng Lou; University of Arizona, SIE Dept., Tucson, AZ 85721;

We focus on response time requirements for on-line algorithms that support real-time decision systems. The paradigm of time budgets for data-handling, communication and computation will be discussed and its role in design of algorithms. Some examples will be presented.

Optimization Applications in Air Transportation

MD34.1 Class Scheduling for Pilot Training

• Jonathan Bard; University of Texas, Dept. of Mech. Eng., Austin, TX 78712-1175; jbard@mail.utexas.edu
• Xiangtong Qi; University of Texas, MSIS Dept., McCombs Sch. of Bus., Austin, TX 78712; qix@mail.utexas.edu
• Gang Yu; University of Texas, Dept. of MSIS, McCombs Sch. of Bus., Austin, TX 78712; yu@uts.cc.utexas.edu

We study the pilot training class scheduling problem for a major airline. Different from most timetabling models, this is a class scheduling problem over a long time period. We will present the complexity, modeling, solution techniques as well as computational results for this problem.

MD34.2 Strong Formulations in Airline Schedule Design & Fleet Assignment

• Amr A. Farahat; MIT, OR Ctr., 77 Massachusetts Ave., Rm. E40-130, Cambridge, MA 02139; afarahat@mit.edu
• Cynthia Barnhart; MIT, 77 Massachusetts Ave., Rm. 1-229, Cambridge, MA 02139; cbarnhar@mit.edu
• Manoj Lohatepanont; American Airlines; manoj.lohatepanont@aa.com

We address the problem of determining flight times and fleet types to assign to an airline's flight network. These problems can be modeled as huge mixed-integer programs whose solution are computationally challenging. We present new formulations with tight LP relaxations and provide preliminary results based on data drawn from a major US airline.

MD34.3 Aircraft Routing under Weather Uncertainty

• Arnab Nilim; University of California, Dept. of EECS, 262M Cory Hall, Berlekey, CA 94720; nilim@robotics.eecs.berkeley.edu
• Mark Hansen; University of California, NEXTOR, Civil/Environ. Eng., 107B McLaughlin Hall, ITS, Berkeley, CA 94270; mhansen@newton.me.berkeley.edu
• Laurent El Ghaoui; University of California, Dept. of EECS, 262M Cory Hall, Berkeley, CA 94720; elghaoui@eecs.berkeley.edu
• Vu Duong; Euro Control, Ctr. Experimental Eurocontrol, Bretigny s-Orge, 15-91222 , France; vu.duong@eurocontrol.fr

Convective weather causes delay and disruption in the national airspace system, in part because of the difficulty in forecasting it. We consider the problem of optimal aircraft routing under weather uncertainty. Our approach employs Markov decision processes and a dynamic programming algorithm that will provide a routing strategy that minimizes the expected delay.

Transportation Science I

MD35.1 A Branch & Cut Algorithm for the Multiple-Depot Vehicle Scheduling Problem

• Paolo Toth; University of Bologna, DEIS, Viale Risorgimento 2, Bologna, 40136 , Italy; ptoth@deis.unibo.it
• Matteo Fischetti; University of Padova, Dept. of Electronics & CS, Via Gradenigo 6/A, Padova, 35100 , Italy; fisch@dei.unipd.it
• Andrea Lodi; University of Bologna, DEIS, Viale Risorgimento 2, Bologna, 40136 , Italy; alodi@deis.unibo.it

The MDVSP consists of determining a min-cost assignment of time-tabled trips to vehicles located in different depots. We propose new valid inequalities and the corresponding separation procedures. These results are used within a B&C scheme whose performance is evaluated throughout extensive computational results on real-world and random instances.

MD35.2 Time Constrained Dynamic VRP for E-Markets

• Monica Gentili; University di Roma 'La Sapienza', Dip. di Stat. Prob. e SA, Ple. A Moro, Rome, 00185 , Italy; gentili@pow2.sta.uniroma1.it
• Paolo Dell'Olmo; University di Roma 'La Sapienza', Dip. di Stat. Prob. e SA, Ple. A Moro, Rome, 00185 , Italy; paolo.dellolmo@uniroma1.it
• Guglielmo Lulli; University di Roma 'Tor Vergata', Dip. di Info. Sist. e Prod., Via di Tor Vergata, Rome, 00133 , Italy; glulli@disp.uniroma2.it

The e-commerce phenomena is inducing the need for technologies that support real-time decision making in the field of logistics. In particular, carriers and shippers are continually facing the problem of rapid 'reconfiguration' of the transportation systems' operations in order to satisfy the new demand and to reduce costs. In this new setting, dynamic models play an increasingly important role...

MD35.3 Stabilized Column Generation for Crew Scheduling Problems

• Jacques Desrosiers; GERAD, Ecole des HEC, Montreal, Quebec, H3T 2A7 , Canada; jacques.desrosiers@hec.ca
• Marius M. Solomon; Northeastern University/GERAD, 314 Hayden Hall, Coll. of Bus. Admin., Boston, MA 02115; msolomon@cba.neu.edu
• Julie Roy; GIRO, Montreal, Quebec, , Canada; julie.roy@giro.ca
• Hatem Ben Amor; Ecole Polytechnique, GERAD, Montreal, Quebec, , Canada; hatem@crt.umontreal.ca

The stabilized version of a problem introduces bounded surplus and slack variables. These are penalized in the objective function. We present the theoretical aspects of the stabilized column generation approach and some recent computational results for the airline crew pairing problem and the bus driver scheduling problem in urban transit.

MD35.4 Models & Algorithms for Scheduling Aircraft in the Terminal Maneuvering Area

• Lucio Bianco; University di Roma 'Tor Vergata', Info. Sistemi e Produzione, Via di Tor Vergata 110, Rome, I-00133 , Italy;
• Paolo Dell'Olmo; University di Roma 'La Sapienza', Dip. di Stat. Prob. e SA, Ple. A Moro, Rome, 00185 , Italy; paolo.dellolmo@uniroma1.it
• Stefano Giordani; University di Roma 'Tor Vergata', Info. Sistemi e Produzione, Via di Tor Vergata 110, Rome, I-00133 , Italy;

We propose scheduling model and real-time algorithms to optimize the management of the TMA. The model is capable of representing the progression of aircraft flows on all the prefixed routes in the area, different runway configurations and other procedural constraints. Performance analysis on some international airport TMAs is discussed.

Transportation Science II

TA35.1 Designing Optimal VMS Locations under Stochastic Incidents & Information Provision Scenarios

• Nhan Nuynh; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; nhuynh@mail.utexas.edu
• Yi-Chang Chiu; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; calmag3@yahoo.com
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

We formulate the VMS location problem as a 2-stage stochastic recourse program. The objective is to design VMS locations to minimize the expected recourse total system travel time in conjunction with different information provisioning scenarios. Experiment results and implications are presented and discussed.

TA35.2 Dynamic Micro-Assignment of Travel Demand with Activity/Trip Chains

• Ismail Chabini; MIT, 77 Massachusetts Ave., Rm. 1-263, Cambridge, MA 02139-4307; chabini@mit.edu
• Andrew Glenn; MIT, 77 Massachusetts Ave., Rm. 1263, Cambridge, MA 02139-4307;
• Stefano Pallottino; Universita di Pisa, Dip. di Informatica, Pisa, 56125 , Italy;
• Maria Grazia Scutella; Universita di Pisa, Dip. di Informatica, Pisa, 56125 , Italy;

We present an assignment framework for travel demand with activity/trip chains which can be used in conjunction with activity-based travel demand forecasting procedures as well as for traffic operational studies. The framework includes temporal-spatial micro-assignment models for activity/trip chaining in which trip chains are assigned over time and space.

TA35.3 Best Routing Policy Problems in Stochastic Time-Dependent Networks

• Song Gao; MIT, 77 Massachusetts Ave., Rm. 5-012, Cambridge, MA 02139; songgao@mit.edu
• Ismail Chabini; MIT, 77 Massachusetts Ave., Rm. 1-263, Cambridge, MA 02139-4307; chabini@mit.edu

We study the best routing policy problem in a stochastic and time-dependent network with various assumptions about network statistical dependency and information access. We design an exact algorithm to a variant of the problem with perfect online information. Approximation algorithms for the same problem are also presented. Computational tests are given to show the efficiency and effectiveness of the algorithms presented.

TA35.4 Re-Optimization Algorithms for Minimum-Time Path Problems in Dynamic Networks

• Ismail Chabini; MIT, 77 Massachusetts Ave., Rm. 1-263, Cambridge, MA 02139-4307; chabini@mit.edu
• Andrew Glenn; MIT, 77 Massachusetts Ave., Rm. 1263, Cambridge, MA 02139-4307;
• Stefano Pallottino; Universita di Pisa, Dip. di Informatica, Pisa, 56125 , Italy;
• Maria Grazia Scutella; Universita di Pisa, Dip. di Informatica, Pisa, 56125 , Italy;

We design re-optimization algorithms that exploit past computation results to speed up future computations in finding minimum-time paths in a time-dependent network. Numerical results are given. They show that the developed re-optimization algorithms lead to encouraging computation time savings in FIFO networks. Savings ratios in the order of 3 are obtained. Larger run time savings ratios are obtained for non-FIFO networks.

Transportation Network Simulation & Analysis I

WA43.1 Performance of a Stable Route Guidance Model for Real-Time Deployment

• Ta-Hui Yang; Purdue University, Sch. of Civil Eng., West Lafayette, IN 47907-1284;
• Srinivas Peeta; Purdue University, Sch. of Civil Eng., West Lafayette, IN 47907-1284; peeta@purdue.edu

Experiments are conducted to analyze the performance of a reactive dynamical system-based stable route guidance model for real-time deployment. The results are compared with those of a rolling horizon-based traffic assignment model and a deterministic DTA model. The proposed model can be implemented in sub-real time while ensuring performance comparable to the rolling horizon-based implementation.

WA43.2 A Hybrid Optimization-Mesoscopic Simulation Dynamic Traffic Assignment Model

• Michael A. Florian; INRO Solutions, 5160 Decaarie Blvd., Ste. 610, Montreal, Quebec, H3C 3J7 , Canada; mike@inro.ca
• Michael Mahut; University of Montreal, CRT, CP 6128, Station Centre-Ville, Montreal, Quebec, H3C 3J7 , Canada; michaelm@crt.umontreal.ca
• Nicolas Tremblay; INRO Consultants, Inc., 5160 Decarie Blvd., Ste. 610, Montreal, Quebec, H3X 2H9 , Canada; nicolas@inro.ca

We present a new dynamic traffic assignment model based on the mesoscopic space-time queue network loading method developed by Mahut. This hybrid optimization-simulation method was applied to a portion of the Stockholm, Sweden road network which consists of 220 zones, 2080 links and 4965 turns. The results are presented.

WA43.3 Calibration of Highway Micro-Simulation Models using ITS Data

• Kyu-Oh Kim; Texas A&M University, Texas Tramsportation Inst., College Station, TX 77843; k-kim@ttimail.tamu.edu
• L. R. Rilett; Texas A&M University, Dept. of Civil Eng., Texas Transportation Inst., College Station, TX 77843; rilett@tamu.edu

Our premise is that ITS data provides a unique opportunity for calibrating and validating traffic simulation models. We will focus on the automatic calibration of micro-simulation models. Three optimization methods, GA, SA and simplex, will be used to calibrate the micro-simulation parameters in CORSIM and TRANSIMS. The test bed will be a freeway corridor in Houston, TX...

WA43.4 GENOSIM: A Genetic Algorithm-Based Optimization Tool for Automating the Calibration of Microscropic Traffic Simulation Models

• Baher Abdulhai; University of Toronto, Dept. of Civil Eng., 35 Saint George St., #105, Toronto, Ontario, M5S 1A4 , Canada; baher@ecf.utoronto.ca
• Tao Ma; University of Toronto, Civil Eng., Intelligent Trans., Systems Ctr. & Testbed, Toronto, Ontario, M5S 1A4;

We introduce GENOSIM, a new optimization tool for automating the calibration of traffic micro-simulation models for faster, systematic and robust model building. GENOSIM is developed as an independent software employing state-of-the-art GAs for combinatorial parametric optimization and is integrated with a dynamic traffic microscopic simulation model using Paramics microsimulation suite...

Transportation Network Planning & Management

WA44.1 An Algorithm for the Paired Combinatorial Logit Route Choice Model

• Anthony Chen; Utah State University, Dept. of Civil & Environ. Eng., Logan, UT 84322-4110; achen@cc.usu.edu
• Panatda Kasikitwiwat; Utah State University, Dept. of Civil & Environ. Eng., Logan, UT 84322-4110; pkasikit@cc.usu.edu

Recently, there is renewed interest in the logit-based route choice model. One of these extended logit models is the paired combinatorial logit model that can overcome the overlapping problem among routes. Equivalent formulation based on the paired combinatorial logit probability has been developed. However, no efficient solution algorithm is provided. We will present an algorithm for the paired combinatorial logit route choice model.

WA44.2 Dynamic Micro-Assignment of Travel Demand with Activity/Trip Chains

• Ahmed F. Abdelghany; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; afsaissal@mail.utexas.edu
• Hani S. Mahmassani; University of Texas, Dept. of Civil Eng., ECJ 6.2, Austin, TX 78751; masmah@mail.utexas.edu

We present an assignment framework for travel demand with activity/trip chains which can be used in conjunction with activity-based travel demand forecasting procedures as well as for traffic operational studies. The framework includes temporal-spatial micro-assignment models for activity/trip chaining in which trip chains are assigned over time and space.

WA44.3 Approaches for the Implementation of Variable Speed Limits

• Thanasis Ziliaskopoulos; Northwestern University, CEE Dept., Evanston, IL 60208; a-z@northwestern.edu
• S. Travis Waller; Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208; s-waller@nwu.edu
• Huajing Shi; Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208; h-shi@nwu.edu

We will address implementation issues relating to the development and evaluation of variable speed limits under variable scenarios within the US. These scenarios often require the optimization of different objectives, i.e., safety vs. delay. The functional requirements of such a system will be discussed as well as multiple methodological approaches as they relate to varying conditions and differing objectives.

WA44.4 An Internet-Based Geographic Information System that Integrates Data, Models & Users for Transportation Algorithms

• Thanasis Ziliaskopoulos; Northwestern University, CEE Dept., Evanston, IL 60208; a-z@northwestern.edu
• S. Travis Waller; Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208; s-waller@nwu.edu

We will discuss the development of an Internet-based GIS that brings together spatio-temporal data, models and users in a single efficient framework to be used for a wide range of transportation applications - planning, engineering and operational. The functional requirements of the system are outlined, considering various enabling technologies, i.e., Internet tools, large-scale databases and distributed computing systems...

Transportation Network Simulation & Analysis II

WB43.1 Analysis of a Point-to-Point High-Coverage Transit System

• Christian Cortes; University of California, Dept. of Civil Eng., Inst. of Transport. Studies, Irvine, CA 92697; ccortes@translab.its.uci.edu
• R. Jayakrishnan; University of California, Dept. of Civil Eng., Inst. of Transport. Studies, Irvine, CA 92697; rjayakri@uci.edu
• Jun-Seok Oh; University of California, Dept. of Civil Eng., Inst. of Transport. Studies, Irvine, CA 92697; jun@translab.its.uci.edu

We propose a new design of a transit system that ensures point-to-point travel using a large number of deployed vehicles such as buses and minivans. The design is based on no more than one transfer during a trip and includes computerized dispatch and routing using probabilistic rules in a real-time management framework. The focus of the presentation is on the routing rules.

WB43.2 Estimation of Freight Origin-Destination Matrices with ITS Data

• Jose Holguin-Veras; City College of New York, Inst. for Transport Studies, 135th & Convent Ave., New York, NY 10031; jhv@ti-mail.engr.ccny.cuny.edu

We will describe a game theoretical formulation that depicts the flow of goods in urban areas as a market in Cournot-Nash equilibrium. The formulation, integrative freight market simulation, decomposes the urban good movement problem in 2 main components: the estimation of the amount of transportation service contributed to the market (in Cournot-Nash equilibrium)...

WB43.3 A Simulation Assignment Model to Evaluate Infrastructure Improvements for Freight Operations in the Chicago Area

• Thanasis Ziliaskopoulos; Northwestern University, CEE Dept., Evanston, IL 60208; a-z@northwestern.edu
• Bhuwan B. Agrawal; Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208;
• Irene Yue Li; Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208; liyue@nwu.edu
• S. Travis Waller; Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208; s-waller@nwu.edu

We analyze the truck operation in the Chicago area, its effects on both passenger and truck traffic and discuss a means for evaluating potential improvements. Numerous scenarios of infrastructure improvements on the Chicago network are evaluated. A simulation-based traffic model is developed that accounts for the simultaneous routing of passenger and truck traffic in order to develop performance measures...

WB43.4 The Life Expectancy of a Min-Path Tree: A Fast Algorithm for the Dynamic Quickest Path Problem

• Robert B. Dial; , 55 Broadway, Cambridge, MA; robertdial@email.msn.com

The algorithm finds a quickest path tree in a network whose arc traversal-times are continuous piecewise linear functions of arrival time. Using a parametric approach, the algorithm transforms a min-path tree for one epoch of departure time into the next earlier epoch at which the tree's topology changes. The local updating technique affects only a small fraction of the tree...

Traffic Control & Flow Analysis

WB44.1 A Robust Arterial Signal Progression System

• Meenakshy Vasudevan; Mitretek Systems, 600 Maryland Ave. SW, Ste. 755, Washington, DC 20024; meenakshy.vasudevan@mitretek.org
• Gang-Len Chang; University of Maryland, Dept. of Civil Eng., College Park, MD 20742; gang@eng.umd.edu

We present a robust discrete optimization technique for arterial signal progression. For real-time applications, commonly observed traffic flow variations will result in the through band not being fully realized. Hence in response to data uncertainties, this study uses robust optimization to obtain a robust solution for the progression problem.

WB44.2 Continuous Space-Time Markov Chain Formulation for Day-to-Day Flow Variation

• Srinivas Peeta; Purdue University, Sch. of Civil Eng., West Lafayette, IN 47907-1284; peeta@purdue.edu

We propose a continuous space-time Markov chain formulation to study day-to-day flow dynamics to circumvent implementation issues associated with traditional discrete formulations in this context. The discrete formulations can entail a large number of states, thereby making unwieldy simulation and control.

WB44.3 Imputation of Missing Data for Origin-Destination Estimation

• L. R. Rilett; Texas A&M University, Dept. of Civil Eng., Texas Transportation Inst., College Station, TX 77843; rilett@tamu.edu
• Yongtao Guan; Texas A&M University, Texas Transportation Inst., College Station, TX 77843;

ITS data provides a unique opportunity for improving many transportation estimation techniques. However, even in a well-maintained system, there can be serious errors in the data. We will focus on data imputation techniques for ITS that used inductance loop technology. A mean substitution imputation approach and a principal component imputation approach will be compared and contrasted...

WB44.4 Issues for the Customization of Online Traffic Control on Beowulf Clusters

• Pengcheng Zhang; Purdue University, Sch. of Civil Eng., West Lafayette, IN 47907-1284;
• Srinivas Peeta; Purdue University, Sch. of Civil Eng., West Lafayette, IN 47907-1284; peeta@purdue.edu

We explore issues for the customization of on-line traffic control models on Beowulf supercomputing environments. Aspects addressed include parallelization logic, cache, memory, file administration systems and storage space. The performance of the Beowulf cluster is compared with that of an IBM SP-2 supercomputer using a dynamic traffic simulator, dynamic traffic assignment algorithms and a time-dependent shortest path model.