Integrated network approach of evacuation simulation for large complex buildings

Publication Type:

Journal Article


Fire Safety Journal, Elsevier Ltd., Volume 44, Issue 2, p.266-275 (2009)



Evacuation system; Integrated network approach; Large complex buildings; Evacuation simulation


Safe evacuation is the most fundamental requirement of fire safety measures in buildings. Traditional building evacuation design is governed by prescriptive building codes which neither have a clear statement of design objectives, nor proper consideration of the interactions between different components of the evacuation system. With rapid developments in performance-based fire engineering, computer evacuation models have been widely used by fire engineers and government officials in fire safety design of buildings. Two approaches are usually used to represent the buildings in evacuation models: fine and coarse networks [Gwynne et al., A review of the methodologies used in the computer simulation of evacuation from the built environment, Building and Environment 34 (1999) 741–749]. For large and complex buildings, such as high-rise or assembly buildings, owing to their spatio-temporal complexity, neither of the 2 network approaches in isolation is satisfactory. The fine network model requires extensive computational power while the coarse network model may not be adequate in representing details of the evacuation process. This paper presents an approach that integrates both network approaches for efficient and detailed assessment of evacuation in large and complex buildings. To demonstrate the advantages of using the integrated approach, this paper presents the results of evacuation simulation for 2 representative buildings. It has been found that the fine network approach (FNA) is inadequate to simulate evacuation of large complex buildings on a normal PC in terms of computational resources. In contrast, the integrated approach can not only perform evacuation analysis for the whole building but is also able to give detailed movement pattern in places of interest without consuming a large amount of CPU time, which makes it possible to directly identify the potential “bottlenecks” in the building.