Informs 2015

117 / 552

INFORMS Philadelphia – 2015
115
4 - Obtaining Engineering Design Targets in the Presence
of Uncertainty
Robert Bordley, Expert Systems Engr Professional,
Booz-Allen-Hamilton, 525 Choice Court, Troy, Mi, 48085, United
States of America, , Steve Pollock
A typical systems engineering process selects a design concept to meet the
customer’s stated objectives. To implement this concept, design targets are
assigned to various engineering teams. It is important that target assignments
explicitly recognize uncertainty about what is technically feasible and about the
customer’s objectives. This paper addresses the problem of determining optimal
design targets (and the associated design margins) using a decision analytic
approach.
SC66
66-Room 113C, CC
Aviation Applications Section: Best Student
Presentation Competition 2
Sponsor: Aviation Applications
Sponsored Session
Chair: Bo Zou, University of Illinois at Chicago, 2095 Engineering
Research Facility, 842 W. Taylor Street (M/C 246), Chicago, IL,
60607-7023, United States of America,
1 - Quantifying Delay Propagation through Crew Connections
Keji Wei, Dartmouth College, Thayer School of Engineering,
Hanover, NH, United States of America,
Vikrant Vaze
The purpose of this study is to investigate delay propagation through airline crew
pairings. In order to solve the crew pairing sub-problem, we develop a heuristic
that combines column generation with branch-and-bound for computational
speedup and solution accuracy. Based on multiple criteria that serve as a proxy
for the extent of delay propagation through crew pairing, we build learning
hyper-models to generate crew pairings that are similar to those in the real world
crew pairing samples.
2 - A Strategic Prioritization Approach to Airline Scheduling
during Disruptions
Prateek Srivastava, Graduate Student, University of Texas at
Austin, 204 E. Dean Keeton Street, Stop C2200 ET, ETC II 5.160,
Austin, TX, United States of America,
Whenever the arrival capacity of an airport reduces due to bad weather, the FAA
uses a Ration By Schedule approach to allocate arrival slots to airlines. The major
drawback of this approach is that it does not take the airline operations into
account. In our study, we consider a framework in which airlines can prioritize
their flights at different airports based on their needs. We show through
simulations on historical data that significant cost benefits are achieved using this
framework.
3 - Characterizing the Tradeoff between Planned and Operational
Costs in Robust Crew Scheduling
David Antunes, University of Coimbra, Portugal, ,
Antonio Antunes, Vikrant Vaze
To mitigate disturbances created by delays in an airline’s network, we develop a
robust crew scheduling model. Input delay profiles were built using real-world
data. To avoid the implementation and customization challenges associated with
the well-established crew pairing models, and tractability challenges associated
with adding robustness features to them, we use an integer programming model
that’s easy to solve using commercial software and reduces the need for
calibrations and fine-tuning.
SC67
67-Room 201A, CC
Advances in Network Design
Sponsor: TSL/Freight Transportation & Logistics
Sponsored Session
Chair: Michael Hewitt, Loyola University Chicago, 820 N. Michigan
Ave, Chicago, IL, 60611, United States of America,
1 - The Price of Discretizing Time in Service Network Design
Luke Marshall, Georgia Institute of Technology,
North Ave NW, Atlanta, GA, United States of America,
Natashia Boland,
Martin Savelsbergh, Michael Hewitt
When solving transportation problems using time-expanded networks, the choice
of discretization has a strong impact on solution quality; the choice trades off
solution accuracy to solution time. We empirically investigate this trade-off to be
able to make a more informed choice.
2 - Real-time Dynamic Load Planning for Less-than-truckload
Motor Carriers
Belgacem Bouzaiene-ayari, Research Staff, Princeton, 113A
Sherrerd Hall, Princeton University, Princeton, NJ, 08540, United
States of America, Warren Powell
LTL motor carriers need to plan the flows of shipments and trailers over time to
minimize costs while meeting service commitments. These decisions need to be
made dynamically, responding to current and forecasted shipments. We present a
novel formulation that allows us to solve large problems while obtaining high
quality solutions. This approach enables us to optimize balance in a network
where loaded movements might be handled with company drivers, purchased
transportation and intermodal.
3 - Solving Large-scale Service Network Design Problems
Michael Hewitt, Loyola University Chicago, 820 N. Michigan Ave,
Chicago, IL, 60611, United States of America, ,
Martin Savelsbergh
In this talk we will discuss how aggregation techniques can be used to solve large-
scale service network design problems. We will discuss methods for two
dimensions of the service network design problem that contribute to its size: (1)
the precision with which time is modeled, and, (2) the number of shipments to be
delivered. We will present the results of an extensive computational study based
on instances derived from a large, U.S. Less-than-truckload transportation
company.
4 - Dynamic Shortest-path Interdiction
Jorge A Sefair, Arizona State University, 699 S. Mill Ave. BYENG
330, P.O. Box 878809, Tempe, AZ, 85287-8809, United States of
America, Cole Smith
We study a dynamic network game between an attacker and a user. The user
seeks a shortest path between a pair of nodes, and the attacker seeks to interdict a
subset of arcs to maximize the user’s shortest-path cost. The attacker can interdict
arcs any time while the user travels the network, and the user can respond by
altering its chosen path. We propose an optimal dynamic-programming algorithm
as well as upper and lower bounds based on interdiction and robust optimization.
SC68
68-Room 201B, CC
Joint Session TSL/Public Sector: Transportation
Disruption Management
Sponsor: Transportation, Science and Logistics & Public Sector
Sponsored Session
Chair: Kash Barker, Associate Professor, University of Oklahoma, 202
W Boyd St, Rm. 124, Norman, Ok, 73019, United States of America,
1 - Management of Water and Transportation Networks during
Flood Disasters
Mehdi Nourinejad, University of Toronto, Civil Engineering
Department, University of Toronto, Toronto, On, M5S 1A4,
Canada,
Flood disaster death tolls can be reduced through in-time evacuation planning.
An optimal evacuation schedule takes into account the route and intensity of the
water flow. Evacuation priority should be given to regions prone are being
overflown faster than others. To enhance evacuation planning, water flow can be
managed through appropriate reservoir design to create a time lag for evacuation.
A model is developed that simultaneously considers evacuation planning and
reservoir design.
2 - Freight Transportation Network Recovery Based on
Interdependent Impact
Mohamad Darayi, , Kash Barker,
Nazanin Morshedlou
Freight transportation networks, considered a means to enable the flow of
commodities and to facilitate economic productivity, are prone to natural and
human-made hazards. This research pursues an approach to improve restoration
order decision making based on the broader perspective of their impact to
multiple industries and multiple regions.
3 - Stochastic Resilience Modeling with Bayesian Kernel Methods:
Application to Inland Waterway Networks
Hiba Baroud, Vanderbilt University, 400 24th Avenue South,
Nashville, TN, 37205, United States of America,
Kash Barker
This work applies a Bayesian kernel approach to model the resilience of
infrastructure systems. The approach quantifies the resilience of transportation
systems under the uncertainty of disruptive events given data describing the
characteristics of the infrastructure system and disruption scenario. The model is
deployed in an application to an inland waterway transportation network for
which the recovery of disrupted links represented by sections of the river is
SC68