ULTRA-SCALE COMPUTING FOR EMERGENCY EVACUATION
BUDHENDRA BHADURI, JAMES NUTARO, CHENG LIU, AND THOMAS ZACHARIA
Oak Ridge National Laboratory, Oak Ridge, Tennessee
Disasters impose a high level of risk on human lives within a physiographic space and evacuation is the physical movement of the population at risk to safer locations. In a broader sense of that definition which includes situations relevant to homeland and national security, emergency evacuations may occur in a variety of physiographic spaces such as passenger vessels in accidents (airplanes, trains, buses, and ships), buildings (and facilities) on fire, as well as large geographic areas (multiple counties) impacted by natural (hurricanes, floods, volcanic eruptions) or technological disasters (atmospheric dispersions of harmful gaseous agents). In all of those situations, modeling the evacuation process involves accounting for the number of people, the available evacuation routes and transportation modes (pedestrian, vehicular), and, most significantly, the behavioral characteristics of the evacuating population. High resolution databases are becoming available at an increasing rate which in turn facilitates development and incorporation of detailed behavioral processes in evacuation models. Computational complexity can originate from increasing resolutions in data (population and transportation), which increase the number of entities to be modeled, or from increasing complexity in the evacuation behavior ...