| When evaluating building evacuation processes, the inclusion of occupant response to emergency situations is essential in the overall assessment of the evacuation plan. Simulations which model the movement of people within buildings aid in the study of evacuation processes and are most often used to validate evacuation plans, optimize evacuation time, and identify any potential safety hazards. The accurate prediction of movement during a building evacuation depends greatly upon the actions and decisions individuals make at the start of the emergency. This dissertation employs a mathematical formulation which accounts for occupant behaviors during this pre-movement phase and implements a computational model to account for the situation dependent dynamic processes which determine the selection of route and exit from a building. The overall effect of these decisions and choices on the evacuation from a student center located at George Mason University (GMU) is presented.;This dissertation begins with a comprehensive review of the mathematical models most often used in computer simulations to model evacuation processes at the micro-, macro-, and mesoscopic levels of pedestrian science. The second literature survey reviews evacuation dynamics; namely, the study of occupant response to emergency situations. Evacuation dynamics encompasses a multitude of research areas and disciplines, most often seen at the intersection of psychology and sociology, mechanics and kinematics, mathematical modeling and analysis, numerical methods, parallel computing, visualization, and data collection. A final literature review highlights psychological factors commonly seen in emergency situations and summarizes evacuation data collection efforts. To better understand the physical, social, and psychological factors involved in the decision to evacuate, a continuous alarm implementation of Reneke's Evacuation Decision Model (EDM) is implemented as a sub-model to a microscopic cellular automata (CA) model.;A pedestrian flow simulation tool (PEDFLOW) is described, verified, validated, and evaluated throughout this dissertation. PEDFLOW has been in development at GMU over the past decade to numerically model the motion of pedestrians and uses a microscopic model where each pedestrian is treated individually and motion is influenced by Newtonian dynamics. As PEDFLOW was not designed specifically for evacuation, when this project began PEDFLOW did not have any pre-evacuation capabilities, nor did it contain modules to account for physiological, sociological, or psychological factors seen during evacuation. PEDFLOW now contains modules which account for pedestrian discomfort, exhaustion, social influence, and affiliative behaviors such as familiar route and exit choice. This work not only incorporated pre-movement behaviors described in literature into PEDFLOW, but also included behaviors observed during the pre-movement phase of routine fire drills on the campus of GMU. This observational study captured the activities of building occupants prior to and immediately after a routine fire drill in a student center on the campus of GMU and resulted in a very thorough newly collected set of empirical data for use by researchers in the validation of computer simulation models. The data compiled was used as input for PEDFLOW and the results obtained were compared to the field data. Furthermore, a series of trials with randomized distribution of behaviors was conducted in order to assess variability and statistical significance. |
