WSC 2007 Final Abstracts

Panel: Agent-based Modeling of Mass Egress and Evacuations
Douglas A Samuelson (Serco), Matt Parker (ANSER), Austin Zimmerman (Homeland Security Institute), Stephen Guerin, Joshua Thorp, and Owen Densmore (Redfish Group) and Pat McCormick and Tom McCormick (Alpha Informatics Ltd.)

Abstract:
We will discuss several recent advances in agent-based modeling, with applications to mass egress and wide-area evacuations following disasters. These advances include: efficiency improvements in specifying people's identification and selection of exit routes, making it possible to handle up to 70,000 people-agents; incorporation of new crowd movement depiction ("Continuum Crowds") for greatly increased realism; efficient real-time visualization depiction of the resulting behaviors; and including the effects of communication and direction, on scales ranging from individual facilities to metropolitan areas.

Simulation of Passenger Check-in at a Medium-sized US Airport
Rajan Batta, Li Lin, Colin Drury, and Simone Appelt (University at Buffalo (SUNY))

Abstract:
Delays in the check-in system at an airport vary with times of the day, day of the week, and types of check-in modes chosen by the passengers. Extensive data collection of the check-in system can be used to build a simulation that helps predict these delays. This paper explains the data collection process, simulation modeling, and scenario analysis for the check-in procedure at the Buffalo Niagara International Airport. Results from this study can be linked to other processes (security checkpoint and parking) in order to obtain information on a passenger’s experience at the airport. The goal of this study is to identify delays and create scenarios that will improve the efficiency.

Advanced National Airspace Traffic Flow Management
George Hunter, Benjamin Boisvert, and Kris Ramamoorthy (Sensis Corporation)

Abstract:
Traffic flow management in the National Airspace is an important problem in our air transportation system. We have developed ProbTFM, a traffic flow management evaluation platform and algorithmic solution. ProbTFM works with existing traffic flow management tools and provides probabilistic data modeling and decision making. ProbTFM forecasts airport and airspace capacity and demand; and airport, airspace, and route congestion. ProbTFM creates a list of high congestion, "critical" flights and recommends delays or reroutes for specific flights. ProbTFM can be used as an evaluation platform for advanced traffic flow management concepts, and to model today's National Airspace System. In this paper we report on validation results and how ProbTFM can be used to understand operational tradeoffs and inform policy decisions.

IRS Post-filing Processes Simulation Modeling: A Comparison of DES with Econometric Microsimulation in Tax Administration
Arnold Greenland (IBM Corporation), David Connors (TranSystems), John Guyton (Internal Revenue Service), Erica Layne Morrison (IBM Corporation) and Michael Sebastiani (Internal Revenue Service)

Abstract:
IRS Office of Research Headquarters measures and models taxpayer burden, defined as expenditures of time and money by taxpayers to comply with the federal tax system. In this research activity, IRS created two microsimulation models using econometric techniques to enable the Service to produce annual estimates of taxpayer compliance burden for individual and small business populations. Additionally, a Discrete Event Simulation model was developed to represent taxpayer activities and IRS administration in post-filing processes. This paper discusses the development of the DES Post-filing Model and compares microsimulation and DES approaches from the perspectives of policy measurement, flexibility and reporting by IRS analysts. The main strengths of microsimulation are robust segmentation of results and the ability to support representation of imbedded, joint distributions in a complex, structural model. The strengths of using DES are queuing capability and increased flexibility to update the granularity of both the data and process changes.

Agent-based Modeling and Simulation of Wildland Fire Suppression
Xiaolin Hu and Yi Sun (Georgia State University)

Abstract:
Simulation of wildland fire suppression is useful to evaluate deployment plans of firefighting resources and to experiment different fire suppression strategies and tactics. Previous work of fire suppression simulation uses analytical models based on a continuous space. This paper presents a design of fire suppression simulation using a discrete event agent model based on a discrete cellular space. We present a framework of wildland fire suppression simulation and describe how firefighting agents in direct attack, parallel attack, and indirect attack are modeled. Experiment results are provided to demonstrate the agent models and to compare them in different fire suppression scenarios.

Modeling and Simulation of Group Behavior in E-government Implementation
Jiang Wu and Bin Hu (Huazhong University of Science and Technology)

Abstract:
This study proposes a multi-agent modeling and simulation approach using EGGBM (E-Government Group Behavior Model) to research complex group behavior in E-government implementation. A multi-agent simulation decision system based on Java-REPAST is developed for qualitative validation to show that EGGBM is consistent with common sense. We give an example of EGGBM application to show that EGGBM method can help decision-makers choose appropriate decisions to improve the level of accepting information technology (LAIT) of groups. Finally, we conclude that this approach could provide a new attempt for the research of group behavior in E-government organization.

Emergency Departments Nurse Allocation to Face a Pandemic Influenza Outbreak
Florentino Rico, Ehsan Salari, and Grisselle Centeno (University of South Florida)

Abstract:
This study proposes a nurse allocation policy to manage patient overflow during a pandemic influenza outbreak. The objective is to minimize the number of patients wait-ing in queue to be treated for the virus while maximizing patient flow. The model is built using ARENA simulation software and OptQuest heuristic optimization to propose various combinations for the number of nurses needed for healthcare delivery. Results are compared with a basic setting that closely emulates the resources and components in a Veteran’s Hospital. The proposed method significantly reduces the number of patients waiting in queue (between 4 to 37 percent on average) for the simulated zones. ARENA process analyzer was used to evaluate various scenarios for nurse availability. Sensitivity on the results for these changes was tested by increasing the flow of patients through the system.