Development of a Mesoscopic Evacuation Model for the Identification of Evacuation Bottleneck Areas in Mega Buildings

Project: Research

View graph of relations


In the case of a fire outbreak in a building, an inefficient evacuation system could leadto high levels of fatality by exposing the evacuees to an excessively smoke-filledenvironment. The evacuation system should be designed to allow the evacuees to reach aplace of safety (e.g., a protected staircase) before their compartments become untenablydangerous. Therefore, evacuation bottleneck areas should be avoided in the spatialdesign of a building. Currently, macroscopic and microscopic evacuation models arewidely used to simulate the movements of individual evacuees. Macroscopic models use anetwork graph approach to compute the evacuation times for different areas and for thewhole building. The computational speed of macroscopic models is generally higher thanthat of microscopic models. However, these macroscopic models do not consider theevacuees’ movement patterns. These models are therefore difficult to determine thebottleneck areas in the building layout. Microscopic models use an agent-based approachto simulate the movements of the evacuees and their interactions with each other insidethe building environment. These models provide a very useful tool for discerning thedetails of evacuation patterns and allow building designers to visually identify thebottleneck areas. The designers can then adjust the building layout to improve theoverall efficiency of building evacuation. The major shortcoming of these microscopicmodels is the length of time they require, as they involve extensive computation topredict the detailed movements of all evacuees.We propose to develop a grid-based mesoscopic evacuation model. The grid size of thismodel will be larger than that of the microscopic models (e.g., cellular automata models)because it will compute the flow of evacuee density instead of the individual movementsof each evacuee. A density flow model based on the density-speed correlation will bedeveloped. This proposed mesoscopic model will also allow local grid refinement for areasthat require detailed analysis. When the grid size is reduced to 0.43 m x 0.43 m, themesoscopic model will become a general cellular automata model (i.e., a microscopicmodel). When the grid size is enlarged to occupy a whole room or corridor, themesoscopic model will become a general node-and-link model (i.e., a macroscopic model).The proposed mesoscopic evacuation model will run faster than the traditionalmicroscopic models, and provide more local detail in terms of pedestrian movement thanthe traditional macroscopic model.?


Project number9042182
Grant typeGRF
Effective start/end date1/01/1611/06/19