Efficient optimization of building emergency evacuation considering social bond of evacuees

• Published in: 2009 IEEE International Conference on Automation Science and Engineering pp. 250 - 255
• Main Authors: Peng Wang, Luh, P.B., Shi-Chung Chang, Marsh, K.L.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2009
• Subjects: Automation; Bonding; Buildings; Dynamic programming; Fires; Lagrangian functions; Optimization methods; Psychology; Testing; Time factors
• ISBN: 1424445787; 9781424445783
• ISSN: 2161-8070
• DOI: 10.1109/COASE.2009.5234131

In building emergency evacuation, the social bond of evacuees is a critical feature related to blocking in egress, and effective crowd guidance should consider this feature to prevent or mitigate such undesired blocking. Besides, the time-criticality of emergencies also requires an efficient computation method, and how to efficiently optimize crowd guidance considering the social bond effect is then identified as a problem to be addressed. By extending the egress model of Wang et al., 2008, this paper parameterizes the relation between the social bond of evacuees and the blocking event in egress. The optimization problem is then formulated to optimize crowd guidance with such advance of modeling. To efficiently optimize the guidance, our original computation method is improved in several aspects: reducing decision variables by infrequent update of guidance, reducing the decision space based on the prior information on the building layout, and approximating the state transition probability by discretizing probability measures. An example in the literature is tested to demonstrate improvement on our modeling and methods.