Enhancing Residential Building Safety: A Numerical Study of Attached Safe Rooms for Bushfires

Abstract

Early evacuation during bushfires remains the safest strategy; however, in many realistic scenarios, timely evacuation is challenging, making safe sheltering a last-resort option to reduce risk compared to late evacuation attempts. However, most Australian homes in bushfire-prone areas are neither designed nor retrofitted to provide adequate protection against extreme bushfires, raising safety concerns. This study addresses this gap by investigating the concept of retrofitting a part of the residential buildings as attached safe rooms for sheltering and protection of valuables, providing a potential last-resort solution for bushfire-prone communities. Numerical simulations were conducted using the Fire Dynamics Simulator to assess heat transfer and internal temperature conditions in a representative residential building under bushfire exposure conditions. The study investigated the impact of the placement of the safe room relative to the fire front direction, failure of vulnerable building components, and the effectiveness of steel shutters in response to internal temperatures. The results showed that the strategic placement of safe rooms inside the building, along with adequate protective measures for windows, can substantially reduce internal temperatures. The findings emphasised the importance of maintaining the integrity of openings and the external building envelope, demonstrating the potential of retrofitted attached safe rooms as a last-resort solution for existing residential buildings in bushfire-prone areas where the entire building was not constructed to withstand bushfire conditions.

1. Introduction

Bushfires/wildfires often result in significant life and property losses [1,2,3] and have become a cause of concern for many countries. In Australia, bushfires are responsible for the fourth-highest number of fatalities and the third-highest number of property losses caused by natural disasters [4,5,6]. Even though early evacuation during bushfires is the safest option for survival, a number of information, social, protection, and operational factors influence its timely execution [7]. However, late evacuations during bushfires are life-threatening; therefore, there were many occurrences in recent bushfire events that required people to shelter in place as safe evacuation was no longer feasible. Therefore, sheltering during bushfires is crucial in ensuring the safety and survival of individuals and communities [8].

The common sheltering locations people have used during past bushfires include residential buildings, commercial buildings, safe rooms, schools, vehicles, and water bodies, out of which the majority of people have selected residential buildings [9]. Residential buildings provide a layer of protection from the immediate dangers of a bushfire, such as intense radiant heat, ember attacks, flame contact, and smoke. They serve as a barrier that helps prevent direct exposure to flames. The structure of a building, including its walls and roof, can help shield occupants from external risks and reduce the inhalation of smoke and harmful gases, given that they have been built to resist bushfire exposure.

However, 90% of buildings in bushfire-prone areas in Australia are not built according to current bushfire standards [10]; thus, their safety during bushfires is questionable. Furthermore, post-bushfire surveys on previous bushfire events have shown that a significant number of people died while sheltering inside buildings, increasing the risks and concerns about sheltering during bushfires [9]. The causes for these fatalities included the failure of the shelters they used (residential buildings), which necessitated their relocation from one location to another during fires. Therefore, the requirement to regulate the safety of sheltering options has significant importance. However, retrofitting the entire building to resist bushfires is costly and not a feasible option for many people living in bushfire-prone areas. Therefore, the possibility of retrofitting a part of the building as an attached safe room is identified as a potential last resort option [11], as this enables some occupants to shelter while others can actively defend against the fires.

Despite this, very limited experimental and numerical studies have been conducted on bushfire safe rooms for bushfire conditions [11,12,13,14], and they mainly focused on detached safe rooms. These are stand-alone, purpose-built structures that are separated from the main residential building and are used as a safe storage or a sheltering location during bushfires. The performance standard for private bushfire shelters in Australia [15], currently the only available standard for bushfire safe rooms, only allows detached safe rooms. The standard identifies that attached safe rooms may pose far greater risks, as in addition to bushfire attacks, attached safe rooms may be subject to additional fire exposure if the residential building catches fire. Therefore, it advises against the use of attached safe rooms. However, when analysing past sheltering data, sheltering has often occurred in parts of residential buildings. This highlights the necessity to investigate the factors that influence the performance, risks, and viability of attached safe rooms.

Therefore, this study focuses on investigating the performance of attached bushfire safe rooms (or retrofitting a part of a building as a sheltering location). Considering the large scale of experimental investigations and the sophisticated testing facilities required, full-scale experimental studies are challenging for this application. In this study, a numerical investigation was conducted for a selected building in a bushfire-prone area in Australia to identify the critical parameters affecting its bushfire performance. The selected building is surrounded by thick vegetation (moist to dry eucalyptus woodland on coastal lowlands and ranges), as shown in Figure 1. There is only one access road to the location that goes through the forest; therefore, in the event of a bushfire, timely evacuation is challenging.

A review of the literature on people’s decision-making under pressurised situations, such as bushfires, when selecting a location for sheltering was used as a basis for selecting a part of the building to conduct a detailed numerical investigation. Then, the study investigated the performance of attached safe rooms during bushfires, their placement compared to the bushfire direction, and the influence of the room being attached to the residential building. It further numerically investigated the influence of openings, such as windows, and their potential failures on the overall sheltering experience. The performance of these attached safe rooms was then compared with the results of previous experimental and numerical investigations conducted for detached (stand-alone) bushfire safe rooms to understand the comparative risks of the two options.

The Fire Dynamics Simulator (FDS) Version 6.7.6 was used as the simulation tool in this numerical analysis. The FDS is a computational fluid dynamics (CFD)-based modelling tool used in a vast range of fire-related applications, from small-scale building elements [16,17,18] to large-scale buildings [13,19,20,21]. Past studies of these different scales of applications have shown good agreement with respective experimental results, emphasising the capabilities of this modelling tool. For example, the performance of external wall systems in bushfire conditions has been investigated using small-scale FDS models, and the model predictions were validated using experimental results for different bushfire radiant heat and flame exposure conditions [22]. Coupled CFD and finite element (FE) analyses have shown the utilisation of FDS in predicting structural performance during fires [23,24,25,26]. Furthermore, cinema fire and pool fire applications have demonstrated the internal fire spread in buildings and smoke transport [27,28]. Full-scale CFD models of detached bushfire safe rooms have shown the capabilities of predicting heat transfer through the building envelope from the external side to the internal side [11,13]. Evacuation models coupled with FDS have predicted the evacuation behaviours of groups of people in the case of fires [29,30]. This vast range of related studies emphasises the suitability of FDS in the numerical modelling application of this study.

Therefore, this study investigates the potential of retrofitting a part of residential buildings with attached safe rooms and evaluates temperature-related risks through numerical modelling using the Fire Dynamics Simulator (FDS). Using case studies on a selected residential building, it investigates the potential, challenges, and priority considerations of retrofitted attached safe rooms as a last-resort solution for existing residential buildings, focusing on the placement of the safe room relative to the bushfire front direction, the failure of vulnerable building components, and the effectiveness of steel shutters in response to maintaining internal temperatures.

2. Selection of Sheltering Locations Inside Residential Buildings

Robinson, in his study, shows that one of the main reasons for unfavourable human behaviour during disasters is the tendency of individuals to deny or reject the existence of risk [31]. This phenomenon was observed in previous bushfire situations, including the 2009 Black Saturday bushfires [31,32]. Denial of risk, along with many other factors (such as unknown fire paths, fire direction changes, and damage to exit roads), adversely contribute to safe early evacuations during bushfires, leading them to shelter in place [33]. Even though sheltering without previous planning is dangerous, a study [34,35] based on a survey conducted on a Code Red day (catastrophic fire danger in the current scale used in Australia) revealed that only 1.5% of respondents evacuated early due to bushfire danger. Furthermore, only 15.8% expressed an intention to stay and defend if there was a bushfire in the area. In contrast, a significant proportion (75.4%) planned to evacuate only if a fire started in the area. This mindset is dangerous, particularly during fast-moving bushfires, as it leaves individuals without a predetermined course of action, potentially forcing them into unprepared sheltering situations. Therefore, it is most likely that they will need to seek shelter inside their houses. While detailed studies of human behaviour in such bushfire situations are limited, the key insights of the available studies emphasise that individuals lacking prior exposure to smoke exhibit diminished emotional stability and are prone to distress at a rate three times faster than those familiar with smoke [36]. Therefore, the integrity of sheltering locations becomes a significant factor in ensuring the safety of these individuals, who will most likely shelter passively.

The majority of the population, who do not have previous experience with bushfires, may proceed with common beliefs when determining sheltering locations. One such example is the belief that bathrooms represent the safest sheltering locations [37]. As an example, the results of a survey showed that people (mainly passive sheltering occupants) identified bathrooms as the safest option for sheltering during bushfires [9,38]. Even though it is advised to shelter in small rooms without big windows, such as bathrooms, during cyclones [39], the safety level offered by a bathroom during bushfires is contingent on its specific location within a building. Small rooms can be dangerous, considering bushfire smoke and rising air temperatures [13]. Therefore, it is required to conduct a comprehensive analysis of residential buildings to identify relatively secure areas within the building to be used in the event of a bushfire and the utilisation of computational tools helps to quantify these effects. All household members should be well-acquainted with these designated safe locations [40], with careful consideration given to determining the most direct pathways to/from these areas connecting the rest of the building and the outdoors. It is recommended that these locations be easily accessible to everyone, have limited openings (as windows are comparatively weaker in heat transfer), and have unobstructed escape routes in case of the failure of the shelter. Adequate retrofitting of these locations for use as safe rooms during emergency conditions will enhance the occupants’ safety.

Furthermore, previous studies have explored the habitability of private bushfire shelters, with a focus on identifying the temperature and humidity thresholds that individuals can endure for up to an hour, which corresponds to the duration of a typical bushfire [41,42]. Additionally, virtual reality-based studies have sought to model human behaviour within bushfire safe rooms, although comprehensive details regarding these investigations remain limited. However, very limited attempts have been made to conduct physics-based CFD simulations of bushfire safe rooms, and the available studies focus on detached safe rooms [11,12,13,14]. With the insights gathered from psychological studies to yield actionable outcomes for enhancing bushfire resilience, it is necessary to investigate the performance of attached safe rooms during bushfires. Therefore, in this study, the most likely sheltering location of the selected building was identified, and a CFD-based numerical investigation was conducted to determine the performance during bushfire conditions.

3. Case Studies Using Heat Transfer Modelling

The selected building (Figure 1) was built many years ago, and the plans or models of the existing state of the building were not available. Therefore, the methodology followed by Hendawitharana et al. [2] was adapted when developing the heat transfer model of the building. A point cloud obtained by both hand and airborne LiDAR sensors was used to create a 3D model of the selected building with near-accurate architectural features. The developed heat transfer models were used to conduct the case studies.

Figure 2 shows the external and internal views of the building. After evaluating the building arrangement, the selected section (in yellow), as shown in Figure 2, was identified as the most likely location in the building for sheltering. This compartment consists of a bathroom and kitchen; therefore, is considered to satisfy the common conception of sheltering locations as well. Based on the interior arrangement of the building, this location is convenient to access from anywhere in the building. This part of the building has both internal doors connecting it to the rest of the building and external doors facilitating easy escape in the case of an emergency.

3.1. Model Details

Figure 3a shows the developed 3D model of the complete building. Then, the selected attached safe room was isolated from the full-scale model and considered for further analysis (Figure 3b). Figure 3b shows two external doors (Doors 1 and 3) and one internal door (Door 2) connecting the safe room to the building. It also shows five glass windows (refer to Figure 3) opening into the safe room (W1–W5). Additionally, there are two glass windows on the wall connecting the attached safe room to the main dwelling unit (residential building) (W6 and W7). However, they do not open into the safe room.

The original building considered in this study is an elevated building with walls externally lined with timber cladding [2,11]. However, timber is a combustible material and can char when exposed to bushfire radiant heat and flame conditions. Combustible materials are not recommended for use in bushfire-prone areas as they increase fire risks, leading to bushfire-induced building fires. As a part of the building is selected for a specialised application for sheltering, this part of the building was selected to be retrofitted with non-combustible, bushfire-resistant external walls based on the findings of previous research studies. Experimental and numerical studies have shown that AAC (autoclaved aerated concrete)-lined wall systems perform well in bushfire conditions [11,13,14]. Therefore, a similar wall configuration was used for external walls, which included light gauge steel frames (LSF) lined with 75 mm-thick AAC panels on the external side and two layers of 16 mm gypsum plasterboards on the internal side. The stud cavity was insulated using glass fibre insulation. The roof consisted of corrugated steel sheets on the external side and 10 mm gypsum plasterboard on the internal side with a 90 mm cavity being insulated using glass fibre insulation. All three doors had the same configuration, where 3.6 mm-thick plywood layers were present on both sides of the 29 mm-thick insulated core, with a 1 mm steel sheet present on the external side in front of the plywood layer. Figure 4 shows the configurations of these building elements. Temperature-dependent thermal properties of the materials were used in the numerical models and are shown in Figure 5. The windows consisted of tempered glass panels which experienced cracking at 332 °C (thermal conductivity 0.96 W/(m·K), specific heat 0.84 kJ/(kg·K), and density 2700 kg/m³). All the components were simulated as one-cell-thick elements with their actual thickness provided as a surface property for solid phase heat transfer calculation. The gap between the safe room floor and the ground was removed as part of the modification process to use this section of the building as a safe room, as elevated floor systems increase risks in bushfire conditions.

The mesh size used for this model was 0.1 m, similar to a previous study conducted for a detached bushfire safe room [13]. Further mesh refinements showed significantly high computational cost while producing similar outcomes. The fire was introduced as “heater” surfaces on the boundaries, and the ground was referred to as “INERT”. All other boundaries remained “OPEN”. Solid phase temperature sensors were used to output both external and internal temperature data of each building component. Wall, door, and window surface temperatures were recorded at 1.3 m height, and controls based on surface temperature were used to simulate glass breakage. Roof temperatures were measured at every 1 m distance along the mid-span. Temperature sensors were located on the W6 and W7 windows and on the wall connecting the safe room to the building at a height of 3.77 m. Incident heat flux was measured at 1.3 m height on the bushfire-exposed side of the safe room. Internal air temperatures were recorded at 0.3, 1.3, and 2.3 m elevations. Furthermore, 2D gas phase temperature, velocity, and pressure distribution spectra and surface temperature distributions were obtained throughout the simulation time. It was assumed that the safe room was fully sealed around the door, window frames, and roof perimeter. The total computational domain of 7 (width) × 10 (length) × 4.3 m (height) was divided into five meshes, which ran in parallel, and high-performance computing facilities were used to reduce the run-time of these models.

These simulations involved heat transfer in both gas and solid phases. In gas phase simulations, the model solved Navier–Stokes equations for buoyancy-driven flows, used large Eddy simulations (LES) to incorporate turbulence, and incorporated the finite volume method (FVM) for radiation transport. In solid phase heat transfer, which represents heat transfer through the building envelope, 1D heat conduction was simulated with temperature-dependent material properties.

3.2. Model Validation

The simulation approach used in this study builds upon our previously validated full-scale heat transfer models of detached safe rooms, which used similar material configurations and were validated against a full-scale experiment [13]. Similar fire exposure conditions were applied to simulate approaching bushfires in this study. In addition, previous studies extensively validated the performance of building elements exposed to the standard fire curve (Figure 6a) [11]. The current study adopts comparable fire exposure profiles, builds upon these validated scenarios, and extends the validated cases to evaluate the performance of attached safe rooms.

3.3. Fire Exposure Conditions

As mentioned previously, based on the location of the building and its surroundings, there are two main directions from which bushfires can approach. The fire exposure conditions are different for these cases and are shown in Figure 7.

• Case 1 (Figure 7a): In the event of a fire approaching in the direction shown in Figure 7a, the building protects the safe room from radiant heat and direct flame contact from the approaching bushfire. However, once the bushfire front reaches the building, building fires are initiated, and the safe room is exposed to building fire conditions. Therefore, in the heat transfer model, only the effect of the building fire burning adjacent to the safe room is simulated. Based on the performance standard for private bushfire shelters (which is the only available standard for these specialised applications), the effect of a nearby building fire in the safe room is simulated as a 10.3 kW/m² constant heat flux for 30 min. However, that value is based on the assumption that there is a 10 m separation distance between the building and the safe room. However, when the safe room is attached to the building, this fire simulation is no longer valid. Therefore, the effect of the building fire is simulated using the standard fire curve (ISO fire curve used when testing for building fires in Australian Standards, AS 1530.4 [45] and AS 1530.8.2 [43]) for 30 min. The fire curve is shown in Figure 6a. The temperatures are recorded for a total duration of 90 min (60 min after the fire is terminated).
• Case 2 (Figure 7b): In the event of a bushfire approaching in the direction shown in Figure 7b, the building does not provide any shield to the safe room from the radiant heat from the approaching bushfire. Therefore, the safe room is directly exposed to the approaching bushfire. This exposure includes the pre-bushfire radiant heat, flame contact, and the post-bushfire radiant heat on the exposed components of the building, and is simulated as shown in Figure 6b. This fire curve provides a realistic simulation of the bushfire front and has been used in several previous studies [14,22,44,46,47]. In the areas attached to the building, the standard fire curve simulates the effect of a building fire (Figure 6a). Building fire exposure starts at the time of flame immersion in the bushfire exposure curve. The temperature data are recorded for 2 h (120 min) in this model.
• Case 3 (Figure 7b): Case 3 is similar to Case 2, considering the fire direction, with the key difference being the inclusion of steel shutters on the windows.

Figure 7. The directions of a bushfire’s approach and the respective bushfire exposure conditions in Cases (a) 1, (b) 2, and 3.

Figure 7. The directions of a bushfire’s approach and the respective bushfire exposure conditions in Cases (a) 1, (b) 2, and 3.

4. Results and Discussion

The internal and external surface and air temperature values were evaluated to determine the bushfire performance of the attached safe room. Based on the outcomes, an additional case, Case 3, was considered later on during this analysis. Case 3 was similar to Case 2 except for the steel shutters introduced to all the external windows as an improvement to the bushfire performance. This enabled us to quantify the impact of bushfire shutter systems on the internal building environment during bushfires. The obtained results of the three cases are discussed in this section.

4.1. Case 1

Figure 8 shows the external surface temperature profiles of the safe room in Case 1 at 30 min (i.e., just after 30 min of standard fire exposure). As shown in the figure (black circle), the surface temperatures on the unexposed side (Side A in Figure 4) experienced a negligible temperature increment as the building itself acted as a radiation shield to the safe room. Similarly, the fire-exposed side wall (Side C) provided a shield to the roof. The Side C wall and the door (Door 2) reached maximum temperatures of 741 and 773 °C on the external side.

The two windows above the roof on the fire side (W6 and W7) cracked after 11 min. However, since they were not opening into the safe room, they did not contribute to increasing the internal temperatures. The windows opening to the external side of the safe room were not exposed to intense radiant heat, as the building provided a radiation shield. The internal surface temperatures on the Side C wall and the door reached up to 30 °C and 123 °C, respectively. The internal air temperatures reached a peak temperature of 36 °C (at 34.7 min), which was a 16 °C temperature increment from the initial environmental temperature of 20 °C. During a bushfire event, it is expected to maintain the internal surface temperatures of safe rooms below 70 °C [15] and internal air temperatures below 45 °C (assuming 50% RH) [15]. These values are based on the performance standard for private bushfire shelters in Australia [15]. However, this standard only focuses on detached shelters/safe rooms, and no standard or guideline is currently available for attached safe rooms. Therefore, these acceptable temperature limits were used to compare the performance of the attached safe rooms in this study. Based on that, the surface temperatures of Door 2 exceeded the acceptable limits. Furthermore, the temperature increment on the internal surface of the door may have contributed to the increment in the internal air temperatures, emphasising the requirement of using higher insulated doors where the safe room connects to the rest of the building.

During hot summer days, initial environmental temperatures can be significantly higher than 20 °C and in those scenarios, a 16 °C temperature increment may not be acceptable if the maximum temperature reaches above 45 °C.

4.2. Case 2

Figure 9 shows the external surface temperatures of the attached safe room in Case 2 at 30 min (i.e., during the flame immersion from the external side) and 60 min (i.e., just after the building fire exposure). The maximum external wall surface temperature reached up to 746 °C on the bushfire-exposed sides (Side A and parts of Sides B and D) during the flame immersion phase. All the glass windows that were exposed to the bushfire attack cracked between 27.2 and 28.3 min when the incident radiant heat flux values were at 26.5 to 35 kW/m². The breaking of glass windows created huge openings on the external envelope, leading to a rapid increase in the internal air temperatures during the flame immersion. This led to an increase in the internal air temperature to 100 °C (an increment of 80 °C), which was well above the maximum acceptable level for safe rooms (45 °C). Furthermore, it was identified that one of the main causes of bushfire-related building loss is bushfire-induced building fires caused by burning embers. Glass breakage, similar to Case 2, creates large openings on the external building envelope and thereby creates spaces for burning embers to collect inside the building, making the building highly vulnerable.

Figure 10 shows the internal and external surface temperatures of the Side C wall at different locations. It shows a uniform temperature distribution on the surfaces. The maximum external and internal wall temperatures of the building fire-exposed side wall (Side C) were 746 °C (at 60 min) and 90 °C (at 30 min), respectively. This shows that the internal surface temperatures have reached their peak well before the respective external side of the wall achieves its peak. Therefore, the main contributor to the internal wall temperature increment was not the solid phase heat conduction through the wall, but the direct radiation received after breaking the glass windows on the opposite wall.

A comparison of internal door temperatures, as shown in Figure 11, indicates that Doors 1 and 3 did not exhibit the sudden temperature rise at the time of the glass break, and they gradually reached their peak at 68.5 min. However, Door 2 experienced a local maxima of 46 °C at the time of the glass break but continued to increase and peaked at 136 °C after 68.5 min. This is due to the location of Door 2 on the directly opposite side of the windows, exposing it to sudden radiation exposure, unlike Doors 1 and 3. The figure further compares the internal surface temperatures of the four walls (the locations of the thermocouples are provided in the figure). The internal surface temperature distributions of the walls showed significant variations compared to door temperatures when the pattern and peak temperatures were considered. The temperature increment on the internal door surface was comparatively low compared to the wall due to the location of the door in the corner of the safe room, with comparatively less area of direct radiation on the door from the broken glass (Figure 11). However, the continued increment in internal surface temperatures emphasises the contribution of solid phase heat transfer through the door during the building fire exposure. Furthermore, C1 and D2 thermocouples (Figure 11) recorded the maximum internal surface temperature at the time of the glass break, and this was caused by their position aligning with a window on the opposite wall. This was further confirmed by A1, as it recorded the minimum peak internal surface temperature during the glass break.

Overall, when the obtained results were compared with the maximum acceptable temperature values provided in the “Performance Standard for Private Bushfire Shelter” [15], the selected part of the building was not suitable for human tenability or even as a safe storage room during bushfire flame zone conditions. i.e., this room is not suitable for tenability as the maximum internal surface temperatures of the doors exceeded 70 °C and the maximum internal air temperature exceeded 45 °C (assuming 50% RH). It is also not suitable for storage as the integrity of the external envelope was not maintained and the large openings were created during the fire. The outcome of Case 2 emphasised the significant impact of glass breakage on the performance of the safe room, placing it as a higher priority to retrofit. Therefore, Case 3 was investigated by introducing bushfire shutter systems to all the windows.

4.3. Case 3

Considering the outcomes of Case 2, a third case was considered by introducing a simple bushfire shutter system in front of the glass windows. These shutters consisted of a 1 mm-thick steel sheet with an air gap of 0.1 m from the glass pane. The performance of the modified attached safe room was then analysed for Case 2 fire conditions. Figure 12 shows a comparison of the glass surface temperatures with and without shutter systems. It shows that the glass temperatures reached the breaking temperature at 26 min without shutter systems, and no glass breakage was observed when the shutters were present. Furthermore, the maximum temperature recorded on the glass was below 250 °C, and this was recorded at the peak flame exposure. Therefore, no premature failure was observed, and all the windows remained intact with no openings on the external envelope.

The maximum recorded internal air temperature in Case 3 was 42 °C (less than the maximum acceptable temperature of 45 °C) (Figure 13 and Figure 14), which was a 58 °C temperature reduction when compared to Case 2 with unprotected glass windows without shutters. Furthermore, no openings were observed on the external building envelope, increasing the protection against embers. Recent experimental and numerical studies have also demonstrated the effectiveness of different shutter systems exposed to bushfire radiant heat and flame conditions. Possible alternative shutter systems included mineral wool insulation sandwiched between 1.2 mm steel sheets, which reported no warping during flame exposure [48]. Other shutter options considered in the study included plasterboard sandwiched between steel sheets, ceramic fibre blankets with aluminium foil, and ceramic fibre blankets with woven glass fibre fabric blankets. These systems recorded significantly lower temperatures on the glass compared to single steel sheets in this study, indicating improved performance on internal temperatures. Therefore, it shows that incorporating even a simple bushfire shutter system is highly beneficial in enhancing the bushfire resistance of attached safe rooms. Detailed studies on other potential options for bushfire shutter systems are identified as important in ensuring the bushfire resistance of attached safe rooms and buildings in general.

Figure 13 shows that at 20 °C, the safe room performance was barely adequate, and if the initial temperatures were high, the maximum temperature exceeded 45 °C. That means, when the initial environmental temperatures are high (above 20 °C), the performance of this attached safe room can become unsatisfactory when the maximum acceptable internal air temperature is considered. As discussed in Cases 1–3, the heat transfer analyses showed that the main contributing factor for this was not heat transfer through walls, but a combined effect of heat transfer through doors and windows. Therefore, further improvements, such as improving door and shutter systems to demonstrate lower thermal conductivity and high thermal shock resistance, must be considered as a solution to improve internal temperatures. However, previous studies on detached safe rooms [13] have shown that internal safe room temperatures increased beyond tenable levels when the initial environmental temperature exceeded 45 °C. The improvements to door and window systems will not be adequate at these elevated ambient temperatures; therefore, the safe room is not recommended for human tenability in flame zone exposure conditions.

4.4. Comparison of the Performance of Attached and Detached Safe Rooms

Furthermore, the obtained results in Case 2 were compared with a previous study [11,13] conducted on a detached purpose-built safe room using a similar external wall and door configuration (the room did not have windows). The results showed that the detached safe room recorded significantly low internal air temperatures (Figure 15). The main reasons for this behaviour include additional features, such as windows and multiple doors, in the attached safe room compared to the detached purpose-built safe room. In addition, attached safe rooms are exposed to a much higher heat load due to the potential ignition of the main building in comparison to detached safe rooms. Therefore, it shows that attached safe rooms demonstrate significantly higher risks than stand-alone/detached purpose-built safe rooms when exposed to bushfire conditions. This study emphasises the need for further studies on developing guidelines and acceptable conditions for attached safe rooms and highlights the additional risks associated.

4.5. Limitations and Future Work

The study focuses on temperature-related risks in attached safe rooms, which is an essential, under-explored aspect of safe room performance. The results are limited to the previously described wall, roof, and door configurations. Further studies are required on other building materials and building designs to develop guidelines and design considerations for attached safe rooms, as they are most commonly used as sheltering locations. Importantly, further studies are required on bushfire-resistant shutter systems and door systems. Other factors that impact sheltering, such as toxic gases, oxygen depletion, occupancy considerations, and human behaviour, were beyond the scope of this study and should be explored in future research. Future work should build upon these findings to explore additional requirements and real-world validations before considering the concept of attached safe rooms for future regulatory frameworks.

5. Conclusions

This study evaluated temperature-related risks of retrofitting a part of residential buildings with attached safe rooms using numerical modelling with a Fire Dynamics Simulator (FDS). Case studies were conducted on a representative residential building located in a bushfire-prone area in Australia. The specific findings of the study are listed as follows:

• Placement of the safe room: The location of the safe room within the building is important to maintain acceptable temperatures within the safe room. Locating it on the fire-exposed side of the building is likely to compromise its performance due to the potential failure of the external envelope, such as windows. In contrast, when the safe room is on the side of the building away from the fire-approaching direction, it improves the internal temperature, even when a structural fire has started.
• Failure of vulnerable building components: The performance of the weakest elements of the building connected to the safe room has a direct impact on the internal safe room temperature.

  -

  The temperature increments from the door connecting the safe room to the rest of the building noticeably contribute to the internal air temperature.

  -

  The failure of windows during fire exposure contributes to rapid temperature increments in the safe room and is referred to the failure initiation location.

• Effectiveness of steel shutters: The introduction of steel shutters for fire-exposed side windows as a solution prevents external envelope failure. Therefore, the results support the use of bushfire shutter systems to enhance building integrity during bushfires.
• Attachment to the residential building: Attached safe rooms are more vulnerable than detached safe rooms due to additional openings and increased heat exposure from potential fires in the residential building. Even with the introduction of a shutter system, internal temperatures are higher than those in the detached safe room.

Overall, our findings highlight the importance of the optimal placement of safe rooms in residential buildings and the requirement for proactive measures for vulnerable building elements to minimise the additional risks posed by attached safe rooms when used as a last-resort sheltering solution in existing residential buildings.