Abstract: This paper presents an overview of the ongoing research into the development of an agent-based model to enable simulations to be performed of agents evacuating from a multi-floor building with a complex layout, including staircases. Specifically, a flow field of navigation objects is constructed pre-computation, which stores the directions and shortest distances to all exits and staircases. Using the flow field, a navigation method is proposed for agents familiar with the environment to identify and follow the shortest route to a chosen exit. Preliminary simulations have been performed to investigate the effect on evacuation time of (i) exit configurations and (ii) familiarity of agents with the building layout. In assessing the effect of exit configurations, results show that the location of the main entrance has a significant influence on evacuation time. In addition, having more exits does not necessarily lead to a shorter evacuation time. In terms of the effect of familiarity of agents, having more agents with a greater level of familiarity does not significantly reduce evacuation time in most cases.

Abstract: Underground space like subway stations is prone to be flooded which can lead to severe and unpredictable damage and even threaten human lives. In this paper, four groups of contrastive simulation of urban waterlogging at two subway stations' entry-exit platforms in heavy rainstorm are conducted, and emergency response strategies are suggested. A waterlogging simulation method named UPFLOOD based on shallow water equations is proposed considering complex topography. It has been found that the waterlogging at subway station's entry-exit platforms is easily influenced by several factors and the site selection of the subway stations is very important. A disaster process construction method based on PN model is proposed and it has been found that the response strategies including plugging, drainage and evacuation are important for disaster mitigation. This study helps decision makers to response quickly to meet the emergency of the waterlogging disaster at subway stations caused by heavy rainstorm.

Abstract: Crisis and disaster management is a field that requires the understanding and application of tools and knowledge from multiple disciplines. Hurricanes Katrina and Rita in 2005 have proven that U.S. petroleum infrastructure is vulnerable to major supply disruptions as a direct result of disasters. Due to the structure of U.S. oil supply chain, primary oil production centers (i.e. PADD* 3) are geographically separated from primary demand centers (i.e. PADD 1), which creates a natural dependency between those districts. To better understand the extent of those dependencies and downstream impacts of supply disruptions, a multi-disciplinary research approach is necessary. The cross-disciplines in this research include disaster management, critical infrastructure and oil supply chain management, and the utilization of geographic information systems (GIS) and systems simulation. This paper specifically focuses on the framework for integrating GIS and systems simulation as analysis tools in this research.

Abstract: This paper presents a dynamic data driven approach of describing driving variability in microscopic traffic simulations for both normal and emergency situations. A four-layer DGIT (Decision, Games, Individual and Transform) framework provides the capability of describing the driving variability among different scenarios, vehicles, time and models. A four-step CCAR (Capture, Calibration, Analysis and Refactor) procedure captures the driving behaviors from mass real-time data to calibrate and analyze the driving variability. Combining the DGIT framework and the CCAR procedure, the system can carry out adaptive simulation in both normal and emergency situations, so that be able to provide more accurate prediction of traffic scenarios and help for decision-making support. A preliminary experiment is performed on a major urban road, and the results verified the feasibility and capability of providing prediction and decision-making support.

Abstract: Simulation exercises are particularly popular for training in emergency situations. Exercises can vary in their degree of realism, complexity and level of stress, but they all try to reproduce a scenario of a real emergency so that each participant simulates the actions carried out for the role they should play. They not only support effective and situated learning, but they can also serve to improve the plan by allowing the identification of weak points and potential drawbacks in it. To facilitate the design and implementation of 3D virtual environments in which training exercises can be conducted, in this paper we propose to use the Cross-Impact Analysis technique in combination with an educational game platform called GRE. We also present a Simulation Authoring Tool that allows the designer to carry out the integration of the knowledge captured by means of Cross-Impact into the game designs that GRE can interpret.

Abstract: This study attempted to evaluate the utility of the Incident Command System (ICS) in varying disaster contexts. ICS is mandated in the United States and practitioners assert that it is an effective organizing system for emergency management. However, researchers contend that the utility of ICS is conflated with inter-team familiarity gained during ICS exercises. A military team-in-the-loop simulator was customized to represent the problems, resources, and command structures found in civilian led disaster management teams. A modified single case design drawn from behavioral psychology was used to explore possible casual relationships between changes team heterogeneity and performance. The design also allowed for the evaluation of improvisation on performance. Further, psychological factors that may underpin improvisation were explored. In addition to some preliminary empirical findings, the successes and difficulties in adapting the DDD simulator are briefly discussed as part of an effort to achieved greater interdisciplinary integration.