System Architecture for Smart Fire Evacuation in High-Rise Buildings

Abstract

The Grenfell Tower fire in London, UK, shocked the world in June 2017. The fire spread across multiple floors, causing 80 deaths and over 70 injuries. One of the major reasons leading to the great loss of life was considered to be the delayed evacuation operation. The current high-rise building fire evacuation follows a prescriptive emergency fire evacuation plan, which is unable to be dynamically adjusted as emergency situations worsen. Thus, it is prone to lead to an unsafe and less efficient evacuation operation once the fire evacuation plan is incompatible with the current fire and evacuation situation. In order to improve such a situation, it is strongly recommended that building fire evacuation embraces IoT and other related technologies. The future building fire evacuation should enhance situational awareness (SA) by enabling comprehensive on-site information collection and facilitating more scientific and appropriate evacuation strategy decision-making. A new mode of building evacuation called “smart building fire evacuation” should be fulfilled.

In order to support the development of such a “smart building fire evacuation”, system architecture for smart fire evacuation in high-rise buildings is proposed in this thesis. The system architecture follows an IoT structure to enable ubiquitous information collection by considering information needs, information sources and data transmission, and potential services and applications throughout the fire evacuation process in high-rise buildings. Moreover, three major service functions and their enabling modeling techniques for smart fire evacuation are proposed. First is a fire hazard identification module, which aims to improve on-site SA by providing an in-depth understanding of more valuable fire situation information. By combining with machine learning (ML) approaches, the service module identifies the origin of the fire source and stages of fire development based on a series of on-site temperature sensor measurements. Afterward, a horizontal evacuation guidance module is proposed. This service module facilitates safe evacuation route planning by predicting the fire-influenced hazardous areas in real-time fire emergencies. Deep learning (DL) modeling methods are used to build the mapping relation between the on-site sensor temperature measurements and predicted fire characteristics. Finally, a vertical evacuation strategy planning module is proposed. This proposed module introduces a smart elevator-aided building fire evacuation (SEABFE) scheme to improve vertical evacuation in high-rise buildings. By taking into account the fire situation and occupant count information on the scene, customized elevator-aided evacuation (EAE) strategies can be planned for different situations to ensure evacuation safety and efficiency. Following the designed system architecture, a simplified system prototype is implemented. The availability of the prototype is exemplified by running several tests. The test indicates a successful collection of the concerned characteristics with the smart sensors. Furthermore, analyses of the optimal EAE strategy based on the real-time occupant count are successfully presented. Therefore, it is believed that the designed system architecture for smart fire evacuation in high-rise buildings is feasible.

This study develops an innovative system architecture for smart fire evacuation in high-rise buildings, which improves the current building fire evacuation to a more advanced and intelligent level. In addition, by embracing the techniques of IoT, ML methods, as well as fire and evacuation simulations, this study provides more insight into the combination of firefighting and building evacuation with promising new technologies.

Date of Award	4 Aug 2023
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Siu Ming LO (Supervisor)