Fire safety should be a major concern of building designers, engineers, and governments. Previous fire experience has made us understand the importance of acquiring fire-ground information to facilitate firefighting operations, evacuation processes, rescues, etc. The rapid advancement in Information Technology, Data Analytics, and other detection and monitoring systems has provided the basis for fire safety researchers to re-think fire safety strategies in the built environment. Amongst all fire safety studies, evacuation in tall buildings, including elevator evacuations, has attracted much attention. IoT-aided building fire evacuation is a new concept of the building evacuation mode, which improves the building evacuation process by making decisions of escape based on the real-time fire-ground information, such as the fire environment and occupant situations.
Furthermore, in terms of occupants' situations during the evacuation process, chaos and disorder are also a concern [55]. Tall buildings encompass a large number of occupants working and living together. In fire emergencies, the convergence of the crowd in the stairwells may cause mass blockage and panic, so it is rather challenging to evacuate all these occupants simultaneously. Phased evacuation control has been proven more suitable for tall building fire evacuation [56]. The phased evacuation strategy conducts a controlled, area-by-area evacuation to decrease the queuing time and density of people. Still, its effectiveness must be strongly built on the timely information feedback of the overall evacuation progress [57]. Therefore, real-time occupant count data of different building areas should be collected to optimize the evacuation strategy and assist the building evacuation.
Finally, evacuation signals and the implemented evacuation strategy need to be delivered to occupants in time in a fire emergency. The existing fire protection systems, such as fire alarm systems, fire emergency lighting systems, and PA systems serve this purpose. Therefore, it is necessary to monitor the state of these systems to avoid a malfunction in emergencies. Besides, due to the importance of providing evacuation guidance in tall building fire evacuation, smart devices that can promptly guide the occupants to escape should be used. The most direct way to deliver clear evacuation guidance in tall buildings is through voice and visual indicators. Therefore, remote control and operation of the PA system and the “smart exit signs” need to be built.
To monitor the environmental fire information in a building, various physical sensors are mainly applied. Thermocouples are the most widely used sensor devices to enable the measurement of temperatures in an enclosure space. The range of the measurements varies from 0 to 1700 ℃ by type [58]. To measure the indoor carbon monoxide (CO) concentration levels, CO gas detectors can be used [59]. By installing these sensors in spaces and corridors throughout the whole building, identification of the fire origin and its affecting areas can be achieved. In addition, the state of various fire protection systems can be monitored by additional sensor sources. For example, door lock sensors can be installed to monitor fire doors’ open or close status in buildings. The state condition of firewater systems can be monitored by water flow or pressure sensors.
Due to the fact that there is a large amount of environmental and equipment state sensor sources and they are installed everywhere inside buildings, flexibility and simplicity may be one of the major concerns. Data communication via wireless networks shows its advantages in this aspect. The transmission of these sensor data runs in a small volume of data and low refresh rate requirements, so narrowband wireless networks for communication, such as ZigBee, NB-IoT, LTE-M, and Lora, can fit their needs. At the same time, narrowband wireless networks also support communications among massive low throughput IoT devices at a relatively low cost and low device power consumption. However, the prerequisite of using these wireless communication networks for sensor data transmission is that they should be resistant to high temperatures and dense smoke in fire environments. Former tests carried out by Schubert and Scholz [60] showed that sensor data transmitted under 2.4 GHz signals (e.g., Wi-Fi, ZigBee, Bluetooth) might not be affected much by the dense smoke and high fire temperature within their test regime. A similar conclusion was obtained by Hofmann et al. [61], also showing that smoke and fire do not influence much on communication in the 2.4 GHz frequency band. Therefore, using narrowband wireless networks working in a lower frequency band with stronger penetration, these proposed narrowband networks are believed to support sensor data communication under fire conditions.
To obtain the occupant count and localization information inside buildings, various occupant counting techniques can be employed. The mainly used technologies are radio frequency (RF), infrared, video cameras, or network connecting signals such as GPS, cellular data, wireless local area network (WLAN), and Bluetooth [62]. The benefits and drawbacks of these approaches were reviewed by Yang et al. [62]. In order to cause minimal influence on occupants’ normal lives and privacy, multiple applicable techniques are applied to count occupants in building fire evacuation control.
First, a turning gate and swipe card system can be used at the building entrances to count the total entering and leaving occupants. The data count can be directly sent to the local server and connected to the IoT platform with a wired connection. Since the turn gate system only allows the authorized occupants to gain entry, manual recording should be kept for additional visitor entry. The building manager can record visitors’ information with a web-based spreadsheet through the Wi-Fi network in buildings.
Additionally, it is considered that a turn gate system may not be suitable for people counting in residential or commercial buildings, which much affects occupants’ normal life, and it hardly tells the specific number of occupants on each floor. Therefore, it is suggested to use visual-based video counters in these areas. The visual-based video counter is a video camera combined with computer vision-based algorithms to count people [63]. It can be used by direct mounting above the building/floor entrances and conducting the counting. This approach causes minimal influence on building occupants and has high recognition accuracy. Due to the high bandwidth and high reliability needs for streaming video data transmission, it is better to use wired connections for local data transmission. By connecting with the existing backbone cable of the building, the real-time local video can be directly sent to the local server and connected to the IoT service platform. An alternative approach is to link these visual-based video counters to a gateway (placed elsewhere) with cables and then connect to the IoT service platform with broadband cellular networks. The connectivity in this way not only makes it easier for the service platform to be placed on the cloud but also can ensure the reliability of streaming video data transmission in local fire environments. For the broadband cellular networks, 4G LTE and the latest 5G NR can be employed.
However, it is worth noting that people counting using visual-based video counters may fail when the building floor is full of smoke, so a thermal counter can be used at the exits of each floor as an alternative option in case the thick smoke obscures the visual-based video counters. The thermal counter is a low-resolution infrared camera combined with the people counting algorithm to enable the counting of people in low light conditions [45]. In practical usage, the thermal counter can be put inside a protective box to isolate it from the effects of fire temperature and smoke. For the data transmission of the thermal imaging data of the thermal counter, wired connections to the service platform of this IoT system are better applied.
Finally, occupant counting needs to be carried out at the room level. For corridors and public areas in buildings, the visual-based approach is applicable. However, the visual-based approach seems inappropriate for individual rooms and private use areas because privacy protection is the major concern in these places. As a result, it is recommended to use infrared beam counters. The infrared beam counter mainly consists of Passive Infra-Red (PIR) sensors and their communication modules. The PIR sensors can detect general movement by identifying infrared radiation emitted by or reflected from occupants without collecting any other personal information. They have been widely used in occupancy counting in buildings [64-67]. The major disadvantage for infrared beam counters is that when two or more persons walk side by side into the area of PIR sensor detection, the infrared beam counter may count only one, so it is not suitable for rooms with large front entrances. Fortunately, the width of the doors for most individual rooms in buildings only allows one or two persons to pass at a time. At the same time, the test by Schubert [60] also showed that fire conditions have no direct influence on infrared sensor uses. Therefore, it is applicable to use the PIR sensor for counting occupants at room level. Since the infrared beam counter transmits only the counted number of occupants, narrowband wireless networks in low power consumption can be applied for data communication between the sensors and the IoT platform, the same as those sensors and devices for fire hazard monitoring.
The evacuation guiding system needs to maintain functionality at all times. Therefore, monitoring of the state of various emergency evacuation systems and remote control of the public address system and the “smart exit signs” system is needed. Like the fire protection systems, the state monitoring of the emergency evacuation systems can be achieved by detecting the states and transmitting the data through narrowband wireless networks. For the exit and directional sign systems, narrowband wireless communications can be used to directly build the connection between the IoT devices and the system platform due to its low transmission of data. For public address and voice alarm systems, considering that voice messages or later video streaming may even be used to aid the occupants in building fire evacuation, it is suggested to use cables to build the local connectivity and then link these systems to the IoT system platform on the cloud through a route or gateway with wired or wireless broadband networks. In this way, the voice or video guidance of the evacuation can be easily delivered by ICs from the exterior of the incident premises.
In recent years, Building Information Modeling (BIM), as well as Virtual Design and construction (VDC), has gained increasing impact in the Architecture, Engineering, and Construction (AEC) Industry. The development of BIM/VDC provides a virtual design and construction platform to facilitate the design and construction as well as the project management process. The tool permits the display of building plans in three dimensions; people can easily understand the building geometries and the added information right into the building components. Recently, there have been some research works concerning the use of BIM for building fire safety management and evacuations [29,47,49,68,69].
For the proposed IoT-aided building fire evacuation control system, it is also important to integrate the collected sensor data into a BIM model and establishing visualization for the users. Visualized information can make it easier for building managers and ICs to guide the evacuation and manage various emergency evacuation facilities and equipment. The visualized information includes both static and dynamic information. The static information contains geometries of the building and compartments, locations of various fire protection and evacuation equipment, and maintenance information. The dynamic information is data collected from the local sensors, which can contain people counts, states of various facilities, and detections of the indoor environment. By attaching the dynamic information to the corresponding equipment or spaces inside the BIM model, browsing the real-time sensor data from a 3D interface is able to be achieved. Apart from visualizing these monitoring data, the BIM model also creates a platform for exchanging data with the fire risk assessment module and evacuation decision supporting module to provide more useful services. The approaches and functions of these two modules are introduced next.
The major function of the fire risk assessment module is to help identify the hazardous areas of the fire event so that occupants can bypass the area and evacuate smoothly. By collecting the ambient status from various environmental sensors, monitoring of the indoor fire environment is enabled. Afterwards, by comparing these monitoring data with the criteria for occupants to safely evacuate the building, such as at 2.0 m height of the enclosure space, air temperature below 100 ℃, concentrations of carbon monoxide below 2800 ppm, etc. [70], the areas at risk can be identified. Finally, by linking these identified hazardous areas to the BIM platform, hazardous areas can be directly viewed on BIM models.
In addition to identifying the hazardous areas in a fire, the collected data can also be used to mine more necessary information regarding fire characteristics, such as fire origins, fire sizes, stages of fire development, etc. This sort of problem is called a fire inverse problem, which involves using the observed sensor data to infer the causal factors that produced it. Many studies have explored the inference of fire origins, fire sizes, and stages of fire development from temperature observations [71–74]. With the machine learning approach to find out the relationship between the fire characteristics and a large volume of simulated fire data, dynamic inference of the various fire characteristics from a series of on-site sensor data collection can be enabled. Understanding the characteristics of the ongoing fire provides the necessary information basis for building managers and ICs to predict the risks in the evacuation process, so it is very important for evacuation decisions.The evacuation decision support module works for the major purpose of evacuation control, including the planning of evacuation strategies and subsequently the evacuation guidance. For the planning of evacuation strategies, evacuation simulation provides a possible approach. Evacuation simulation is a computer-based approach to determine the total egress time based on the simulation of crowd dynamics and pedestrian motion. The existing software or modeling approach include FDS+Evac [75], SGEM [76], agent-based modeling (ABM) [77], and cellular automata (CA) [78]. By determining the input parameters, such as building geometries, initial occupants’ locations, evacuation strategies, and individual behaviors and interactions, evacuation simulation can predict the evacuation process and evaluate the evacuation with some indicators, including total egress time, travel distance, door flow rate, etc. From these indicators, whether or not the evacuation strategy is acceptable can be determined. In the currently proposed system, the data collection of people counting for each floor is dynamically linked with the simulation model. In addition, the hazardous areas and characteristics of the ongoing fire analyzed in the fire risk assessment module are also connected to the evacuation model. The evacuation simulation may become more adaptable to the real fire situation on-site by considering these dynamic parameter inputs. With such dynamic evacuation modeling to assess the effectiveness of different evacuation strategies, such as evacuating occupants with different sequences, whether evacuating occupants with the help of elevators, etc., safety and effectiveness of implementing the building evacuation can be significantly improved.
As for the evacuation guidance part, it is meant to make the planned evacuation strategies come into action. With the control of the exit and directional signs, the evacuation decision support module can deliver the planned evacuation route to evacuees. Moreover, the state and condition of each exit and directional sign can be monitored and viewed in the BIM-based platform. Furthermore, considering that it may need to control the sequence of evacuation for each floor, the evacuation decision support module also enables sending voice alarms and instructions to the designated areas or floors via PA systems.By summarizing the information sources, means of data transmission, and potential services and applications in building fire evacuation control, a preliminary design of an IoT-aided building fire evacuation control system is proposed. The system architecture is demonstrated in Figure 2. With the sensors and devices to collect various information on the fire-ground, a variety of means of data transmission are applied to build the communication between sensors and the IoT platform. In order to provide more diversified services and applications to help the building evacuation, promising technologies including BIM, fire inverse, and evacuation simulation are embraced in the system. Furthermore, considering the importance of data protection and system security, the system is also equipped with an authentication and encryption layer to protect the endpoint devices so that malicious attacks and data breaches can be avoided. In addition to protecting the security of endpoint devices, data security in the IoT platform can be enabled by firewalls and strong user authentication processes. With these measures, the proposed system can be secure and only used for its intended uses (i.e., fire-ground situational awareness, evacuation planning, or evacuation guidance).
Figure 2. IoT-aided building fire evacuation control system.
The advantages of the proposed system are summarized as follows:
IoT technology provides a possible approach to link the cyber and physical environment together in building fire evacuation control. By establishing the perception layer and network layer of the proposed IoT system, the ubiquitous sensors and various communication networks can be integrated to enable the monitoring of the fire environment and occupants’ situations in buildings. Compared with traditional fire evacuation, the IoT-aided fire evacuation highly improves the information collection and update on-site, ensuring safety in evacuations. Moreover, the collected people count information is also useful for building energy management in regular time. Therefore, it is applicable to implement such a system in buildings.
By exchanging the monitoring data from various sensors and devices through the Internet, IoT enables the aggregation of all the on-site information into one place. Thus, combined with BIM, the visualization and management of various evacuation equipment and facilities within the BIM model can be achieved. With the BIM model to show the locations of various evacuation facilities and hazardous areas, a direct understanding of the relationship between the building geometry and the fire situation can be built by building managers and ICs, making it easier to guide occupants’ evacuations. As for the information regarding the state of various firefighting equipment and evacuation facilities, it consists of its historical data and real-time monitoring data. By attaching the state information to the specific equipment and facilities in the BIM model, browsing and control of these installations can be achieved.