|Figure 1||Figure 2|
Large emergencies present an even greater challenge for disaster planning. Large emergencies often require assets from the entire city, or beyond the city, acting as a kind of surge capacity. Large events also may change the basis upon which the plans were made, such as when roads, bridges, and tunnels become inaccessible.
The basic problem with this type of emergency plan is that it is brittle. Cities change. Economic development and major real estate projects rearrange the shape and distribution of activities and people in the city. New areas that open for development--- such as waterfronts, hillsides, and limited-access neighborhoods-- are often, by their nature, harder to respond to in emergencies.
In many cities, emergency response is less uniform across neighborhoods because of these kinds of change. More frequent re planning is problematic, because the real and political costs of reallocating response assets are very high.
D4S2 provides an independent laboratory for testing how the type and scale of an event, situational variables, and command decisions affect responders' efficiency and effectiveness in dealing with disasters. D4S2 seamlessly integrates commercially available off-the-shelf components, including ArcGis 9.2, Rockwell Automation's Arena discrete event simulation with a custom-built rule-based decision modeling system, and a control interface that mirrors an emergency operations center.
The City of Pittsburgh provides the development case for the construction and validation of the system. Many insights were gleaned as a result of applying D4S2 to specific disaster scenarios in Pittsburgh.
D4S2 contains more than 100 layers of geographic, asset, and other geo-referenced information. The geographic data describe infrastructure and physical details, such as roads, waterways, and topography. Information about emergency response resources -- such as fire police, and EMS units – is stored in the geo-database and associated with the infrastructure. Real-time environmental data such as weather and traffic conditions are also part of the system.
When used with the simulation model, the geographic information system (GIS) can feed data to the simulator and make the simulation more realistic and robust. Keeping the geographic-related data in an independent GIS system simplifies the system deployment process. The disaster simulation system also can be quickly implemented in any area that has the appropriate GIS data.
The simulation model allows us to create any number, type, and size of emergency events. In essence, the system “reads the map” and forms a simulation model. The simulation model uses discrete event simulation as the main construct and models the emergency response system as a transportation network. Important street intersections are chosen as network nodes. The response vehicles are the entities moving along the network and performing various response tasks. The entities are built in different layers, such as cars on the roads, trains on the railways, boats in the rivers, and helicopters in the air. D4S2 uses an innovative method to model other pieces of the system to reduce computational efforts.
In addition to discrete event simulation, D4S2 uses agent-based simulation techniques to incorporate more realistic and flexible entity operations and interactions. Agent-based modeling originated from artificial intelligence. A computer agent is an autonomously controlled entity that can perceive its own operations as well as the surrounding environment, compile the predefined rules to make operational decisions, and act based on these decisions. The individual agents operate on their own but are affected by other agents and the environment.
The behavior and interaction of the agents are defined by rules derived from the industry standards, training, best practices, exercises and research on first responders, emergency managers, dispatchers, the public, terrorists, other actors, and environmental factors. All of the rules are formulated to the if-then format to state explicitly the conditions and consequences. A software module – the knowledge engine – uses the rules in conjunctions with the system's data and user input to make decisions. These decisions may change the simulation, move assets within the GIS, or cause other actions to be taken.
Each component continuously informs the others of decisions and other changes as the events unfold. The model provides an independent laboratory for testing how the type and scale of an event, situational variables, and command decisions affect the responders' efficiency and effectiveness in dealing with disasters.
Users interact with the system through interfaces modeled after existing emergency operations center interfaces. The figures above show two interface displays. Figure 1 is the event specification interface. This allows the user to define, in detail, the characteristics of an emergency (location,type, time, casualties, etc.). The system builds in event profiles from the 15 emergency event scenarios created by the Department of Homeland Security. In addition, casualty distributions from the experience of the military are available to help the user describe an event. The user is completely free to override these aids.