Minority Issues Forum Student Poster Presenters
INFORMS 2004,
special acknowledgement to NSF (DMI-0440785)
Todd P. Coleman
Electrical Engineering and Computer Science
Massachusetts Institute of Technology
__________________
Optimal Control of Traffic Lights: Stochastic and Fluid Model Analysis
Optimal control of traffic lights is a problem that has been studied in some detail since the 1950s. Our focus here is on optimal static control of the timing schemes in high traffic settings. The objective which we have chosen to focus on is the expected waiting time of a car in the system. First, we examine the simple single intersection and focus on the underlying stochastic process. Using a unique approach whereby we treat the system as a periodic, preemptive queue with two classes, we are able to compute in closed form the expected waiting time as a function of the cycle period and fraction of green light time per cycle. We next consider a fluid model for the system in the hopes of attacking general networks. Again, we start with the simplest case of a single intersection, and compute in closed form the optimal fraction of red-green time. Then, we extend this to a general network model and formulate this as an optimization problem which turns out to be combinatorial in nature. We discuss some solution techniques for this problem and finish with some simulation results comparing these different approaches. This is joint work with David Brown.
Edward Gonzalez
Computational and Applied Mathematics
Rice University
__________________
Algorithms for the Graph Minimum Bisection Problems Based on a Combination of the Rank-2 Relaxation, Multilevel and Frontiering Approaches
The minimum bisection problem, more commonly identified as the Min–bisection problem, is arguably the most popular graph partitioning problem for its wide range of real-world applications and its role in generating other types of partitions. Since graph partitioning problems are usually NP-hard for which exact solutions are generally difficult to obtain, algorithmic research in this area has been focused on heuristic-based approaches. Extensive research shows that so far the most successful heuristic algorithms are based on the multilevel approach, as exemplified by the popular graph partitioning software package METIS. However, the more recent rank-2 relaxation approach to the Min–bisection problem, which relaxes the discrete problem to a continuous optimization problem, has shown the capability to produce high quality results. A serious drawback to this more recent approach is its relatively poor scalability to very large problems, thus making it inappropriate in larger applications. In this project, we propose to investigate ways to combine the rank-2 relaxation approach with multilevel and frontiering strategies. The overall goal is to produce practical, high-quality partitioning algorithms and software tools to enhance and complement existing state-of-art algorithms and software.
Philippe E. Gouel
Industrial and Operations Engineering
University of Michigan
__________________
Problem Management in Complex Product Development Projects
In the development of complex engineering products or systems problems often become visible only late in the product development (PD) process when the design information is converted into manufacturing information. We investigate two technology management processes in this type of PD system: problem detection and problem treatment. With examples from the auto industry, we develop a categorization scheme to classify problem types and problem treatment strategies.
Marcos Goycoolea
Industrial and Systems Engineering
Georgia Institute of Technology
__________________
Scheduling an Air Taxi Service
Air taxi service is an emerging business opportunity that will use a new class of small, relatively inexpensive, jets to give travelers the chance to fly to and from small or midsize U.S. airports on 4-8 seat planes at a much lower cost than chartering. Scheduling such a service is a 2-stage problem. In the first stage requests for are either accepted or rejected, which is largely a feasibility problem that has to be solved online, and in the second stage the accepted requests are scheduled by minimizing cost. This poster focuses on the cost optimization problem. This is joint work with Mo Bazaraa, Daniel Espinoza, Renan Garcia, George Nemhauser, and Martin Savelsbergh.
Robert Hampshire
Operations Research and Financial Engineering
Princeton University
__________________
The Impact of Options on Supply Chain Management
This paper explores the impact of options on supply chain performance when consumer demand is uncertain. The analysis is based on an environment involving a single supplier and multiple retailers. Each retailer can either buy product directly from the supplier, or purchase options on product. An option gives the retailer the right, but not the obligation, to buy an additional unit of product. As such, the retailer will exercise an option if and only if that unit of product is needed to meet demand. The retailer must balance the reduced uncertainty that options afford against the price premium that must be paid to procure product with options.
Industrial Engineering
Texas A&M University
__________________
Strategies for Anonymous Communication Protocols
The internet can be used to perform tasks that range from finding directions, managing bank accounts, and sending emails to paying bills or casting votes. For many of these and other tasks, varying levels of security and privacy may be desired. There have been a number of systems designed to mask the identity of the system users and the effectiveness of these approaches depends on the manner in which the information is transmitted through the system. This poster presents results from a paper by Guan et al. on the subject. The paper discusses formulating an optimal route selection strategy to maximize the probability of successfully masking the sender of a message through a system. The authors define an anonymity degree which is maximized over all possible path lengths to determine an optimal path length distribution. It is with this metric that one can measure the ability to successfully mask the sender of the information through the system. The poster will also discuss relevant future work the presenter will consider as a foundation for a doctoral thesis
Luis Otero
Industrial and Management Systems Engineering
University of South Florida
__________________
Alleviating the Parking Problem in Schedule-Based Demand Systems
Parking availability is directly affected by the drivers’ scheduled activities (i.e., work shift, appointments, classes, flights, or production). Typically, the schedulers of these activities overlook parking capacity, creating peak periods for parking space demand. In this study, an integer programming approach that will optimize the schedule during the planning phase is presented in order to alleviate the parking problem. The methodology is tested in a university parking scenario. The model can be adapted to different parking systems with similar characteristics.
Eric Smith
Systems and Industrial Engineering
University of Arizona
__________________
Trade-Off Study Tool
In order to facilitate the dissemination of knowledge about the effectiveness of Trade-Off Studies in Human Decision Making, a web tool with an easily used interface was created. The tool, which is accompanied by information pertinent to Trade-Off Studies, allows the easy use of Wymorian Scoring Functions - smooth normalizing scoring functions. Requirement and sub-requirement related input is provided to scoring functions, with the normalized output easily combinable with various combining functions. TradeOffStudyTool is freely available at
http://rayser.sie.arizona.edu:8080/TradeOffStudyTool.
This research is supported by AFOSR/MURI F49620-03-1-0377.
Warren Sutton
Industrial and Operations Engineering
University of Michigan
Recovering from Delays: A DEA Study of United States Airport Efficiency
The operational efficiency of airlines has been analyzed through numerous studies. Most of these studies focus on the individual airlines and how the operations of the airlines are organized and planned out. These types of studies provide an in-depth analysis of airline scheduling looking for ways to optimize airline efficiency. However, this line of research pays little attention to the critical role that the airports play in the operational efficiency of airlines. The airport and airlines are, in fact, dependent entities that rely heavily upon each other to succeed in daily operations.
This study will attempt to measure the effect that airplane delays have on the efficiency of 33 U.S. airports using data envelopment analysis (DEA). The study will make a comparison between the efficiency of airports that are designated as hubs by a major U.S. airline and those which are not. Knowing that delays are only one factor that impacts the efficiency of an airport this study also examines the number of employees, number of loading bridges, and percentage of on-time arrivals, as well as the number of departures from an airport and percentage of on-time departures as part of the efficiency equation. The results of the study indicate that there is no significant difference between the efficiency of hub airports and non-hub airports; however ‘large’ non-hub airports are generally inefficient.
Kendra Taylor
Industrial and Systems Engineering
Georgia Institute of Technology
A Model of Forward and Reverse Sequential Auctions
Electronic auctions allow for economic discovery in business-to-business (B2B) transactions. Motivated by industries with seasonal cycles, such as the pulp and paper industry, we aid procurement management in analyzing the profitability of online auctions while exploring the less-studied area of auction theory where individual buyers and suppliers initiate auctions sequentially and are able to alternate roles of bidder and auctioneer. We characterize sequential auction processes to determine the advantage received from the ability to initiate an auction in a period of a sequence. We start with a two-period sequence and formalize our results with an N-period sequence. The standard assumptions of the benchmark model are held for each single period. In addition, we assume that there are a small number of buyers and a large number of suppliers and their valuations for future periods are unknown and follow a uniform distribution. This analysis leads toward the overall goal, which is to understand how to use sequential forward and reverse auctions to potentially mitigate risks currently associated with using uninterrupted sequences of reverse auctions or forward auctions.
Sarah K. Womack
Industrial and Operations Engineering
University of Michigan
__________________
Modeling the Relationship between Modern Management Systems and Musculoskeletal Disorders: Does Lean Manufacturing Lead to Poor Ergonomics?
The purpose of this research is to investigate the relationship between lean production practices and work-related musculoskeletal disorder (WMSD) risk exposure.
Since the late 1980’s, American manufacturing firms, particularly in the automotive sector, have begun to widely adopt the Japanese Management System (e.g. Toyota’s Lean Production System) because of its superior quality and productivity measured against the traditional manufacturing systems of American and European firms. The proliferation of implementation of the lean production system within American firms gave rise to the debate over the extent to which this management system has adverse physical and psychological health effects on its labor force. A basic regression model of the relationship between physical WMSD risk exposure and lean production was developed and the results show a significant relationship. This model is limited in that the proxy for lean, value-added work content, does not capture the complexities of the independent variable, lean.