Numerical modeling of external light gauge steel framed wall systems exposed to bushfire flame zone conditions

Abstract

Bushfire-related building losses cause adverse economic impacts to countries prone to bushfires. Building materials and components play a vital role in reducing these impacts. However, due to high costs of experimental studies and lack of numerical studies, the heat transfer behavior of building’s external components in bushfire-prone areas has not been adequately investigated. Often large-scale heat transfer models are developed using Computational Fluid Dynamics (CFD) tools, and the availability of CFD models for heat transfer in building components improves the understanding of the behavior of systems and systems of systems. Therefore, this paper uses a numerical modeling approach to investigate the bushfire/wildfire resistance of external Light gauge Steel Framed (LSF) wall systems. Both full-scale and small-scale heat transfer models were developed for the LSF wall systems. Experimental results of six internal and external LSF wall systems with varying plasterboard thickness and cladding material were used to validate the developed models. The study was then extended to investigate the bushfire resistance of seven external wall systems under two different bushfire flame zone conditions. The results illustrate the significant effects of fire curves, LSF wall components and configuration on the heat transfer across the walls. They have shown 1) the favorable performance of steel cladding and Autoclaved Aerated Concrete (AAC) panels when used on the external side of wall systems and 2) the adequacy of thin-walled steel studs’ load-bearing capacity during bushfire exposures. This study has shown that most of the investigated external LSF walls could be reused with cost-effective retrofitting such as replacing the Fire Side (FS) steel cladding after bushfire exposures. Overall, this study has advanced the understanding of the behavior of external light steel framed walls under bushfire flame zone conditions.

Graphical abstract

Keywords

bushfires / flame zone / external light gauge steel framed walls / heat transfer analyses / numerical models

1 Introduction

Bushfire/Wildfire-related building damage has been a significant problem in many countries over many decades. Australia, the United States of America, and Europe witness significant building losses due to bushfires every year and the losses are increasing over time [1]. The need to minimise building losses in bushfires has been identified as significant in minimising bushfire-related economic and social costs. Among the many strategies identified to minimise bushfire building losses, protecting the external building envelope is considered an important solution since it eliminates bushfire-induced building fires [2–5]. In a building, external walls form the largest area exposed to the external environment. Thus, they play a critical role in ensuring the safety of buildings in bushfires.

Light steel and timber framed wall systems are widely used in house constructions in many countries due to the associated low cost and fast construction [6–8]. However, the limited understanding of their performance with different wall lining materials in bushfires can lead to devastating outcomes. Thus, it is necessary to investigate and understand the bushfire resistance of these commonly used cladding materials and external wall systems. It is often recommended to use non-combustible cladding for these wall systems to avoid ignition of cladding. However, relying only on this criterion is insufficient. Understanding whether these non-combustible cladding systems can withstand bushfire conditions to protect the building’s external envelope is also required. Therefore, this study uses numerical modeling to investigate the bushfire resistance of external Light gauge Steel Framed (LSF) wall systems lined with different non-combustible cladding materials.

Although many experimental and numerical studies have been conducted on the fire resistance of internal wall systems in building fires [9–15], very limited studies have been conducted on the fire resistance of external wall systems exposed to external fires (such as bushfires or external building fires) [16–20]. In building fires, it is assumed that the fires start inside the building and then propagate, whereas, in external fires, the fires attack the building from outside making the external building envelope more vulnerable. Further, due to the differences in cladding materials and configurations, the behavior of external wall systems exposed to external fires is different compared to internal wall systems exposed to internal fires. For instance, unlike internal wall systems, external wall systems are generally asymmetric across the stud depth with different cladding materials placed on either side [16]. Additionally, internal framed walls are generally lined with gypsum plasterboards on either side, whereas in external walls, the external surface is generally lined with different materials including Autoclaved Aerated Concrete (AAC) panels, fiber cement boards, bricks and blocks, etc. [17].

Thus, detailed studies are required to evaluate the performance of external wall systems exposed to external fires (bushfires for this research).

In general, external cladding materials are fixed to the thin-walled steel studs or the internal plasterboard layers, via thin steel battens. This creates two internal cavities within the wall systems. In literature, the heat transfer through two internal cavities is rarely investigated. When an external wall is exposed to bushfires, two prominent heat transfer mechanisms are identified: 1) conduction through solid materials and, 2) radiation through cavities [21,22]. Due to the close spacing of wall lining materials with one stud and negligible air movement inside the cavity, the influence of convection heat transfer is negligible.

Experimental studies are commonly used in literature to study heat transfer in wall systems. However, associated high time and monetary costs make it not feasible to conduct repeated experiments with different wall configurations. Thus, numerical modeling is widely used in research studies as an alternative approach. However, the modeling and simulation tools must be selected carefully to meet the requirements of the research since these tools are based on different concepts, such as the Finite Element Method (FEM), Finite Difference Method Finite Volume Method (FVM), Computational Fluid Dynamics (CFD), etc. [23–32]. When simulating the heat transfer in large-scale fire models such as rooms/compartments, houses or multiple buildings, CFD tools are frequently used [26,32–34]. Availability of CFD models of the building components (such as walls) helps in understanding their behavior and to improve the outcomes of large-scale models which is a system of these systems (components). Therefore, this study focused on utilizing Fire Dynamics Simulator (FDS), a CFD and FVM-based tool specialized for fire modeling applications to simulate the heat transfer in wall systems. Furthermore, in the validation phase, the model predictions were compared to the predictions obtained from FEM models and experimental results to ensure the validity of predictions.

In this study, numerical analyses of external LSF wall systems were conducted to understand their behavior when exposed to bushfire flame zone conditions. First, a CFD heat transfer model of an internal wall was developed using FDS and verified to predict the heat transfer through cavities followed by mesh sensitivity analysis to understand the mesh dependencies on model predictions. The modeling results were then validated using six experimental results (both internal and external wall configurations), and the experimental and model uncertainties were discussed. Then the internal wall systems were re-configured to be used as seven external wall systems in a numerical parametric study. The heat transfer across the selected external wall systems was then evaluated in terms of time–temperature curves as per the requirement of testing of building elements for bushfire flame zone conditions (AS 15030.8.2) [2] and a realistic bushfire time–temperature curve [35]. Several measures were taken to consider the effect of asymmetry and the presence of different materials and cavities in the external wall systems, and they include using temperature-dependent material properties for all the different materials in the wall system, using actual material thicknesses and cavity depths, allowing heat transfer from Fire Side (FS) to Ambient Side (AS). Effects of fire curves, wall components and configuration on the performance and bushfire resistance of external wall systems were investigated and the results are presented.

2 Summary of the experimental study

Due to the limited availability of experimental data for external wall systems exposed to external fires, the results of three experimental studies [10,12,16] in the literature were used in verifying and validating the model, which included both internal and external LSF wall systems exposed to standard fire conditions. A summary of the test wall configurations used in this study is given in Tab.1. The first four tests (T1–T4) [10] were internal wall configurations, while the rest (T5 [12] and T6 [16]) were external wall configurations. Furthermore, these test wall configurations with or without cavity insulation included different cladding materials, stud spacing, wall thicknesses and specimen sizes, resulting in a diverse set of configuration parameters.

Test 1 (T1). 3 m × 3 m internal LSF wall specimen with a steel frame lined with one layer of 16 mm gypsum plasterboards on both sides of the wall. 92 mm × 35 mm cold-formed steel lipped channel studs (0.55 mm thick) were used at 600 mm spacing with suitable tracks placed on top and bottom of the frame. No noggings were used in the frame. Gypsum plasterboards were screwed to the steel frame at 200 mm staggered spacing along the plasterboard joints and at 300 mm spacing along the studs. No cavity insulation was used in this test. Fig.1 illustrates the fire test set-up and the LSF wall test frame for T1.

Test 2 (T2). Similar to T1, however, LSF wall is lined with 13 mm thick gypsum plasterboards.

Test 3 (T3). Similar to T1, however, LSF wall cavity was insulated with 75 mm thick 11 kg/m³ glass fiber insulation.

Test 4 (T4). Similar to T1, however, LSF wall is lined with two layers of 16 mm thick gypsum plasterboards.

Test 5 (T5). It is a 1.4 m × 1.2 m external LSF wall specimen with 9 mm thick fiber cement boards lined on both sides. The 90 mm studs were spaced at 450 mm spacing and no cavity insulation was used.

Test 6 (T6). It is a 3 m × 3 m external LSF wall specimen with 75 mm thick AAC panels fixed to 25 mm battens which were then screwed to the steel frame. Glass fiber insulation (24 kg/m³) was used in the cavity. The AS of the wall consisted of one layer of 16 mm thick gypsum plasterboards. Studs were placed at 600 mm spacing while the noggings were placed at 1 m spacing.

During the experiments, the constructed walls were placed in front of a gas furnace (inside a laboratory) and then exposed to the standard fire curve. A similar arrangement as shown in Fig.1 was used for T1–T6 walls except for T5 for which a smaller furnace was used during the experiment. The temperature readings along and across the wall specimen (FS, Fire Cavity (FC), AS, Ambient Cavity (AC), and steel hot and cold flanges) were measured and they were used to compare with numerical modeling results.

3 Model development

3.1 Numerical method

FDS is mainly utilized to simulate thermally driven flows within buildings. However, when the fires approach from the external side of the building (during bushfires or large settlement fires etc.), the heat transfer across the external building envelope plays an important role in building’s safety. Thus, it is necessary to investigate the heat transfer predictions from FDS through external walls. Radiation heat transfer through wall cavities and conduction through solids are significant contributors to this. Radiation heat transfer is included in the governing equations of FDS as {\dot{q}}_{\mathrm{r}}^{‴} [36]. Equation (1) shows the energy transport equation and the terms {\dot{q}}^{‴}, {\dot{q}}_{\mathrm{b}}^{‴}, {\dot{q}}_{\mathrm{r}}^{″} denote the heat transfer rate per unit volume, the energy transferred to sub grid-scale droplets and particles and heat flux (conductive, diffusive and radiative), respectively.

Equation (2) provides the equation for {\dot{q}}^{″} and the three components denote the heat flux values of conductivity, diffusivity and radiation, respectively. Equation (3) shows the contribution from radiation in the energy equation. A detailed description is provided in the FDS technical reference guide [36]. The radiation equation is solved using an FVM for convective transport.

(1)\frac{\mathrm{\partial }\left(\rho h_{\mathrm{s}}\right)}{\mathrm{\partial }t}+\mathrm{\nabla }\cdot \left(\rho h_{\mathrm{s}}\mathit{u}\right)=\frac{\mathrm{D}\overline{p}}{\mathrm{D}t}+{\dot{q}}^{‴}+{{\dot{q}}^{‴}}_{\mathrm{b}}-\mathrm{\nabla }\cdot {\dot{q}}^{″},

(2){\dot{\mathit{q}}}^{″}=-k\mathrm{\nabla }T-\sum _{\alpha }{h}_{\mathrm{s},\alpha }\rho D_{\alpha }\mathrm{\nabla }{Z}_{\alpha }+{\dot{\mathit{q}}}_{\mathbf{r}}^{″},

(3)\begin{array}{rl}& {\dot{q}}_{\mathrm{r}}^{‴}\equiv -\mathrm{\nabla }\cdot {{\dot{q}}^{″}}_{\mathrm{r}}\left(x\right)=\kappa \left(x\right)\left[U\left(x\right)-4\text{π}{I}_{\mathrm{b}}\left(\mathit{x}\right)\right];\\ & U\left(x\right)={\int }_{4\text{π}}I\left(x,s^{\prime }\right)\mathrm{d}{s}^{\prime },\end{array}

where ρ, h_s, u, p, k, T, D_α, Z_α, κ(x), and I_b(x) refer to density, sensible enthalpy, velocity components (u,v,w), background pressure, thermal conductivity, temperature, diffusivity of α species, mass fraction of α, absorption coefficient, and source term, respectively.

3.2 Model details

Using CFD tools to conduct detailed studies of heat transfer in building components such as walls provide the opportunity to understand the effects of heat transfer in single models compared to buildings. Therefore, FDS, a CFD-based fire modeling tool, was used in this numerical study. Many previous studies have used FDS for a wide range of applications [26,32–34], and facade fire spread and heat transfer in wall systems are some of them [25,37,38].

FDS version 6.7.6 was used in this study. The external walls considered in the experimental study were 3 m × 3 m. In this study, we investigated the effect of model size when compared to the actual walls considered as it is beneficial to develop simplified numerical models resembling the large-scale models’ behavior to save computational cost. To develop and verify a simplified model, the T1 wall system was used (Fig.2). The initial model height was selected as 3 m (the height of the tested wall) and the model width was 600 mm (the spacing between two studs). Then the model size was systematically changed while keeping the model outcomes unchanged. The components of the wall system (studs, battens and wall lining materials were modeled as one-cell thick obstructions with backing conditions as “EXPOSED.” This is to ensure solid phase heat transfer across the wall thickness. The actual thicknesses of the wall components were given as a surface property to ensure that the correct thickness is accounted for in the solid phase heat conduction calculations. A previous study on heat transfer through a long-span composite floor beam in a compartment showed that the use of one-dimensional (1D) heat conduction model showed significant deviations compared to three-dimensional (3D) heat conduction model for steel member temperatures [39]. However, in this study, all the steel members are thin (1 mm thickness) compared to the thick steel members considered in that study. 1D heat conduction is prominent in thin steel members and this was observed in the model predictions as well. Therefore, the 1D solid phase heat conduction model was used to investigate these wall systems.

Furthermore, it should be noted that the significantly smaller thicknesses of the plasterboard and studs compared to the wall specimen sizes can result in a trade-off between the model accuracy and the computational complexity. As an example, using a finer mesh can result in a better representation of individual wall components, but the associated run-time and computational requirements will be much higher when modeling a large-scale (3 m × 3 m) wall specimen. On the other hand, using a coarser mesh could result in quicker results, but the models will be more error prone. To overcome these limitations, the symmetry concept is generally used. However, in FDS (the numerical modeling tool used in the study), careful consideration needs to be given when applying symmetry due to the use of Large Eddy Simulations in FDS calculations. Therefore, the following procedure was used to develop a small-scale model, which provides the results of a large-scale model while reducing the computational cost.

Tab.2 shows the LSF wall models (B1 to B6) that were developed during the simplification process, with model dimensions and boundary conditions. Due to the arrangement of the steel frame of the previously discussed specimen (refer to Section 1), the cavities inside the wall are limited to the volumes surrounded by the studs and tracks (3000 mm × 600 mm in height and width, respectively). Hence, the above dimensions were selected for the initial model (B1). Model B1 consisted of two 16 mm thick plasterboards placed parallel to each other at a distance of 92 and a 3000 mm steel stud was placed at the center of the cavity. The boundary condition of the sides of the cavity was set to “MIRROR” and the top and bottom model boundaries were set to “ADIABATIC.” The wall system was exposed to the standard fire curve from one side using a “HEATER/COOLER” surface and the AS boundary was kept “OPEN.” The results obtained from the numerical analysis of model B1 were then compared with the rest of the model configurations given in Tab.2 (B2 to B6), which were the simplified configurations of B1. Fig.3 presents a visual explanation of the simplification process highlighting the model height (y-axis), width (x-axis) and the steel studs (shown in blue color) in each configuration.

Fig.4 shows the temperature recordings across the wall thickness at mid-height. Even though the reduction of the model height from 3 to 1 m or 0.2 m did not show any changes in surface temperatures, a slight reduction of the top and bottom edge temperatures was observed while maintaining the mid-height temperatures constant. This is further described in Subsection 3.4. Furthermore, this observation confirms that radiation is the prominent method of heat transfer inside the wall cavities as convection and conduction heat transfer through the air is negligible as discussed earlier.

However, when the model width was reduced from 0.6 to 0.2 m, it resulted in a change in temperature readings (see the results of B2 and B3 models in Fig.4). Model B3 showed a significant deviation from the original model. In general, the radiation heat within the cavity is absorbed by the steel stud and emitted to the surroundings. However, when the stud location was changed due to the change in the model width, the radiation fields altered the temperature readings. This observation was further demonstrated with the models B4, B5, and B6. In B4, the model height and width were 0.2 and 0.6 m, respectively, and the surface temperatures were similar to B1. In B5, the model width was increased to 1.2 m to have two steel studs spaced at 0.6 m spacing, and the temperatures across the wall thickness showed similar results to B1. Model B6 in which the studs are located at the two edges of the model with a center-to-center spacing of 0.6 m also showed similar temperature readings to model B1 (Fig.4). Furthermore, the presence of the steel stud highly affects the cavity surface temperatures than the fire or AS surface temperatures. Therefore, considering the above observations, 0.6 m (width) × 0.2 m (height) model size was selected (Model B4).

3.3 Mesh sensitivity

Model B4 was initially selected for the mesh sensitivity analysis with three different mesh sizes: 15, 10, and 5 mm. The largest cell size was determined to be 15 mm since any mesh sizes larger than 15 mm cannot represent the model’s cross-sectional features accurately. Tab.3 presents the relationship between the mesh size and the computational cost. During the mesh sensitivity analysis, the distance from the fire source to the wall (30 mm) and the cavity depth (90 mm) were kept constant in all three models. Due to the smaller mesh size, the 5 mm mesh model resulted in a significantly higher cell count than the other two models and showed the highest computational cost and simulation time. The 10 mm model showed a better geometric representation of the wall than the 15 mm model. The temperature readings on the surfaces across the thickness of the wall were used to compare the three models as shown in Fig.5. The observed temperatures at each point of interest (FS, FC, AC, and AS) were approximately similar (negligible differences in temperature readings). This highlights that the selected mesh sizes had little to no impact on the temperature profiles. However, since the mesh size is inversely proportional to the computational cost, coarser mesh sizes are preferred, given that the model accuracy is not compromised. Therefore, considering the computational cost and accurate wall representation, the mesh size of 10 mm was selected for further studies.

3.4 Sensitivity of radiation parameters

Radiation can be identified as the prominent heat transfer method from the furnace to the FS wall surface and from the FC to the AC. Therefore, the sensitivity of the radiation through the cavities was studied using a simplified model. Two hollow cuboid models with different widths (width represents the cavity depths) were developed. In each cuboid model, one of the surfaces was set as a “HEATER” surface (at 1290 °C) and the rest as “INERT” (Fig.6). The heat flux values on the wall opposite the heater surface were measured using devices that were located along the diagonal and at mid-height along the model length (600 mm) (refer to the red lines in Fig.6).

The analytical solution for the above case was calculated considering the configuration factor between a differential area element to a finite area (in this case, the finite area was 200 mm × 600 mm) as specified [40] and recommended [41] in past studies. dA₁ is an isothermal diffuse element on a hot surface and A₂ is the finite area (cold surface) which exchanges energy with dA₁. A₁ and A₂ surfaces are assumed to be isothermal and at T1 and T2 temperatures, respectively.

The following equations were solved to obtain the configuration factors, and they were plotted against the distance along the model length as shown in Fig.7.

(4)F_{\mathrm{d}1-2}=\frac{{\int }_{A2}{I}_1(\mathrm{c}\mathrm{o}\mathrm{s}{\theta }_1\mathrm{c}\mathrm{o}\mathrm{s}{\theta }_2\mathrm{d}{A}_1/S^2)\mathrm{d}{A}_2}{\text{π}{I}_1\mathrm{d}{A}_1},

(5)F_{\mathrm{d}1-2}={\int }_{A2}\frac{\mathrm{c}\mathrm{o}\mathrm{s}{\theta }_1\mathrm{c}\mathrm{o}\mathrm{s}{\theta }_2}{\text{π}{S}^2}\mathrm{d}{A}_2,

for parallel planes, and

(6)\mathrm{c}\mathrm{o}\mathrm{s}{\theta }_1=\mathrm{c}\mathrm{o}\mathrm{s}{\theta }_2.

Therefore,

(7)F_{\mathrm{d}1-2}={\int }_{A2}\frac{{\mathrm{c}\mathrm{o}\mathrm{s}}^2\theta }{\text{π}{S}^2}\mathrm{d}{A}_2.

The calculation was repeated for two different mesh sizes (10 and 15 mm) and two different numbers of radiation angles (100 and 200). Fig.7(a) and Fig.7(b) show a good agreement between the model calculated values and the analytical values. Both mid-height and diagonal heat flux values showed good agreement with the theoretical calculations in both cases. However, slight deviations from the theoretical calculations were observed toward the edges of mid-height and the diagonal lines and the amount of variation was reduced in the 90 mm cavity model compared to the 30 mm cavity model. Furthermore, increasing the number of radiation angles did not visibly affect the heat flux readings for the above cavity depths. Furthermore, a reduction in mesh size from 15 to 10 mm showed a better agreement with the analytical results; especially at the edges of the diagonal. Therefore, overall results show that the near field radiation in the model (for both cavity depths) agreed well with the analytical solution for this application.

Then the boundary conditions of these cavities were changed to represent the actual scenario in the wall system. The “INERT” wall was changed to “gypsum” and the model boundaries were changed to “MIRROR” on the sides and “ADIABATIC” at the top and bottom. This resulted in an equal distribution of the incident heat flux along the mid-height of the model (the points of interest) as shown in Fig.7(c), which is the required radiation distribution to represent the continuity of the wall beyond the boundary. The radiation heat transfer across the cavities and the model sizes were verified.

3.5 Elevated temperature thermal properties

The heat transfer through the solid components in the model is determined by the thermophysical properties of the materials. Therefore, temperature-dependent thermal properties were used for gypsum plasterboard, cold-formed steel studs, fiber cement boards, AAC and glass fiber insulation as shown in Fig.8 [12,16,25,42,43]. The default convective heat transfer coefficients were used. The emissivity of steel, gypsum plasterboard and all other materials were considered as 0.7, 0.88, and 0.9, respectively. These values are similar to those used in previous research studies on heat transfer models [26,39,44,45].

4 Model validation

Fig.9 shows the developed numerical model in FDS. The mid-height temperatures of cladding/board and steel stud surfaces obtained from the model for different internal and external LSF wall systems (refer to Section 3) were compared with the corresponding experimental values. These comparisons for both internal and external LSF wall systems with different cladding materials, stud spacing, thicknesses and with/without cavity insulation enabled the validation of the developed FDS models for a range of LSF wall systems. The model predictions agreed reasonably well with the experimental results and were similar to previous research studies conducted using Finite Element (FE) models [43,46,47]. A comparison of these results with FE model predictions (using ABAQUS) is provided for T1 wall configuration in Fig.10 [21].

4.1 Internal walls

Fig.10 and Fig.11 show the temperature predictions for each internal wall test in Tab.1 (T1−T4) for surfaces across the wall thickness. They showed a good agreement between the model predictions and the experimental results. Even though the model predictions for the T1 wall system (refer to Section 3 and see Fig.10(a) and Fig.10(b)) exhibited a similar temperature pattern to the experiments, the model predicted slightly higher temperatures than the experiments for plasterboard surface temperatures after one hour into the test, resulting in a conservatively predicted early insulation failure (at 79.2 min in the modeled wall system compared to 94 min in the experiment). The insulation failure time was based on when the average AS surface temperature exceeded the room temperature by 140 °C.

The model predictions for the wall system T2 (Fig.10(c) and Fig.10(d)) showed a very good agreement with the experimental values. Furthermore, the insulation failure time was 56 min in the experiment, whereas the model predicted insulation failure time was 59 min, agreeing well with the experimental failure time. The model predictions for the wall system T3 (Fig.11(a) and Fig.11(b)) showed a reasonable agreement with the experimental temperature values and the failure time. The plasterboard joints opened up during the experiment, which increased the AC temperatures. The wall failed at 106 min in the experiment and the model predicted insulation failure time was 108 min. Furthermore, the cavity surface temperatures were affected by the deterioration of the insulation as well.

The model predictions for the wall system T4 (Fig.11(c) and Fig.11(d) showed a better agreement in the cladding surface temperatures for up to about 3.5 h, after which the plasterboard fall-off happened. However, after 190 min, the FS pasteboard layer had fallen off on most of the locations and the internal layers’ temperatures had increased rapidly. The experimental values shown in Fig.11(c) are from a surface location where the plasterboard fall-off did not occur. Therefore, the predictions showed a good agreement with the experimental results up to 200 min. The sudden increment in the AS surface temperature at 190 min can also be explained as an effect of plasterboard fall-off. However, in the experiment, the wall experienced an insulation failure at 197 min due to the maximum temperature criterion and the model did not predict the failure, which can be addressed by the manual accommodation of the plasterboard fall-off in the model.

4.2 External walls

Fig.12 shows the time–temperature curves for the external wall systems in Tab.1 (T5 and T6). The model predictions for the wall system T5 (Fig.12(a) and Fig.12(b)) showed a good agreement with the experimental results up to 26 min. However, the test was terminated at 26 min due to an insulation failure of the wall specimen at 23 min. The model predictions agreed well with the experimental results until the insulation failure temperature was reached. The fiber cement board lined LSF wall (i.e., Model T5) failed much earlier than other wall configurations.

The wall system T6 did not reach insulation failure either during the experiment or in the model. The temperature distributions across the wall thickness showed a reasonable agreement with the experimental results (Fig.12(c) and Fig.12(d)).

4.3 Summary of results

Tab.4 summarizes the insulation failure times of the six LSF wall systems from experiments and the FDS models. The maximum and time-averaged model predicted results for the temperature of each surface through thickness were plotted against the respective experimental values as shown in Fig.13. The combined uncertainty (measurement uncertainty and propagated input uncertainty) for the gas and solid phase temperatures measured through thermocouples was assumed to be 0.07 based on the FDS validation guide. The calculated model uncertainty was 0.10 with a bias factor of 0.99 for maximum temperature, and for the average temperatures, the model uncertainty was 0.086 with a bias factor of 0.98. The maximum temperature uncertainty values were lower than the previous studies on surface temperatures using FDS [48]. Furthermore, the average temperatures showed a significantly lower model uncertainty. However, it should be noted that the thermal conductivity of the materials could only be measured up to 500 °C and the apparent thermal properties were used beyond that. This can affect the uncertainty calculation. Overall, the model predictions for the wall systems showed reasonable accuracy. The model can therefore be used to predict the fire resistance of different external LSF wall systems under flame zone conditions.

5 Parametric study

Since the focus of this study is to simulate the heat transfer of external wall systems exposed to bushfires, the internal wall systems T1–T4 were then modified to be external wall systems PS1 to PS4. A 0.42 mm thick steel cladding was introduced on the FS using a 40 mm batten fixed to the FS plasterboard. This modification was also done for the external wall system T5 due to the early failure during the experiment (refer to Subsection 4.2).

During bushfires, the external walls play an important role in protecting the external envelope. The introduction of non-combustible steel cladding is expected to act as a protection layer for the wall during bushfires, by eliminating the chance of firebrand collection inside the wall system (these firebrands can result in bushfire-induced building fires) and act as a heat barrier to the internal wall cladding. The external wall systems considered in the parametric study are shown in Tab.5. The PS6 wall system is the same as the wall system T6 and the PS7 wall system is similar to the PS6 wall system except for cavity insulation where no cavity insulation was used in PS7. All these wall systems (PS1–PS7) were exposed to two different fire curves, where Fire Curve 1 is based on AS 1530.8.2 [2] and Fire Curve 2 is based on Ref. [35] which simulated a realistic bushfire time–temperature curve.

5.1 Fire curves

Two temperature-dependent bushfire curves were used in the parametric study and are shown in Fig.14. Fire Curve 1 is based on the Australian Standard, AS 1530.8.2 [2] which specifies that the building elements should be exposed to the standard fire curve for 30 min and the temperatures should be monitored for another 60 min after the flame exposure (Fire Curve 1 in Fig.14). Fire Curve 2 shows a bushfire time–temperature curve derived based on the bushfire flame characteristics (obtained from the ABCB bushfire verification method) [35], which includes the specific characteristics of a realistic bushfire curve such as a rapid increment in temperature and sustained higher temperature for 2 min. The rate of temperature increment and the maximum temperature in Fire Curve 2 are higher than that of Fire Curve 1. However, the wall specimen is exposed to a higher heat load during the exposure to Fire Curve 1. The influence of these two characteristics on different external wall systems was evaluated by comparing the surface temperatures across the wall thicknesses.

5.2 Model predictions

Fig.15–Fig.19 present the predicted temperatures in the seven external wall systems when exposed to the two different bushfire flame zone curves. A comparison of the maximum AS surface temperatures till 90 min into the simulation is given in Tab.5. Out of the seven wall systems, the maximum AS wall temperature was recorded in PS5 for both fire curves (218 and 150 °C for Fire Curves 1 and 2, respectively). Steel stud temperatures exceeded 400 °C when exposed to Fire Curve 1 and showed a negligible lag time in reaching the peak temperature when compared to the FS of the wall (Fig.17 and Fig.18). PS6 showed the best performance recording the least temperature increment on the AS followed by PS7 and PS4. Both PS4 and PS7 wall systems showed approximately similar AS wall temperatures, but the steel stud temperatures were significantly low in PS7 (Fig.18 and Fig.20).

6 Discussions

6.1 Wall lining material

In this study, LSF wall systems with different wall lining materials including steel cladding, plasterboards, AAC panels and fiber cement boards were investigated. These lining materials showed different influences on the heat transfer through the wall systems.

6.1.1 Steel sheets

The comparison of temperature predictions between the wall systems T1–T4 and PS1–PS4 showed that the introduction of thin steel cladding on the FS of the wall system resulted in a notable reduction in the FS temperatures. Furthermore, the steel sheets delayed the temperature rise and reduced the maximum temperatures on the FS plasterboard as shown in Fig.15 and Fig.17. Both sides of the steel cladding reached temperatures beyond 200 °C at which the external coating starts to degrade, and the steel batten fixed to the FS plasterboard reached temperatures beyond 600 °C. Furthermore, experimental studies on wall systems with external steel cladding exposed to flame conditions have demonstrated that even though the cladding was subjected to out-of-plane deformations, no joint opening was observed [17]. This is favorable in bushfire conditions as one of the main challenges of construction for bushfire conditions includes maintaining the integrity of external building envelope to eliminate ember penetration. However, steel loses its capacity at elevated temperatures therefore, further studies are required on the integrity of connections to ensure that the steel sheeting will remain in place for the duration of fires. The model predictions show that given adequate measures have been taken to retain the steel cladding on the walls, the heat transfer across the wall systems is significantly improved.

6.1.2 Gypsum plasterboards

Gypsum plasterboards undergo two stages of dehydration when exposed to fire as shown in Eqs. (8) and (9) at temperatures 80–250 °C [49]. These chemical reactions are endothermic and their effect can be observed in the time–temperature curves with lower temperature increment rates, and in temperature-dependent specific heat profile (Fig.8(b)).

(8){\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_4\cdot 2{\mathrm{H}}_2\mathrm{O}\left(\mathrm{s}\right)\to {\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_4\cdot \frac{1}{2}{\mathrm{H}}_2\mathrm{O}\left(\mathrm{s}\right)+\frac{3}{2}{\mathrm{H}}_2\mathrm{O}\left(\mathrm{g}\right)

(9){\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_4\cdot \frac{1}{2}{\mathrm{H}}_2\mathrm{O}\left(\mathrm{s}\right)\to {\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_4\left(\mathrm{s}\right)+\frac{1}{2}{\mathrm{H}}_2\mathrm{O}\left(\mathrm{g}\right)

In Fig.21, the temperature range where the gypsum plasterboards undergo dehydration reactions is colored in gray for better visualization. It was observed that the FC temperature of wall systems PS1–PS4 fell within the colored region which depicts that the first and second gypsum dehydration reactions had taken place throughout the board thickness in all those tests. The presence of chemically bound water acts favorably toward the fire resistance of the wall systems by absorbing the heat energy during the dehydration process.

6.1.3 Autoclaved aerated concrete panels

When the wall systems externally lined with AAC panels (PS6 and PS7) are compared, the maximum temperature recorded on the FC surface was 115 °C in PS6 with cavity insulation. Previous studies have highlighted that a long plateau region is observed in the AAC temperatures between 50 and 350 °C due to the endothermic reactions that release chemically bound water [16,50]. Since the wall system PS6 has only reached 115 °C at the end of the test (Fig.22(b)), the chemically bound water might not have been fully removed from the AAC panel (i.e., dehydration is not completed across the thickness of the panel), thus the wall may be reused under repeated bushfire exposures as well. However, when the AAC panel was exposed to repeated specific heat tests, a constant specific heat value of 1 kJ/kg·K (approximately) was observed until 575 °C [16]. This is mainly due to the complete removal of chemically bound moisture during the first exposure, and this will change the heat transfer across the panel. Therefore, separate studies must be undertaken if the wall systems are designed for repeated bushfire exposure. Furthermore, the wall systems lined with AAC panels showed the best performance when the FC temperatures are considered. FC temperatures can be used to ensure the safety of service lines and ducts inside the wall system.

6.1.4 Fiber cement boards

Wall systems lined with fiber cement boards demonstrated the highest heat transfer across the fiber cement board and recorded the maximum cavity temperatures of 470 and 634 °C (at 30 min) with and without the steel cladding on the FS. Furthermore, it showed an approximately similar temperature increment in the FC when exposed to both Fire Curves. This is a significant deviation from the other wall systems considered in the study, where the FC temperatures did not experience a rapid increment when exposed to Fire Curve 2.

6.2 Light gauge steel framed wall stud temperatures

The reduction of the yield strength and elastic modulus of steel is considered to reduce the capacity of load-bearing LSF walls when exposed to fire. During fires, the stud hot flange temperatures can go beyond 600 °C, at which the yield strength is less than 40% of its original value [51]. This can lead to buckling of the load-bearing studs and the wall systems will fail early. Based on a previous study conducted on the steel temperatures and the yield strength and elastic modulus [52], for the wall systems considered in this study, a less than 20% reduction of the yield strength and elastic modulus will be observed when exposed to bushfire flame zone conditions. The analysis of stud temperatures showed that the stud hot flange reached a maximum temperature of 456 °C (in PS5), which is less than the critical stud hot flange temperature of 550 °C used for the commonly used load ratio of 0.4, i.e., the applied load in fire is 40% of the ambient temperature stud capacity. The maximum hot flange temperatures recorded in all other external wall systems were less than 350 °C. Therefore, the LSF wall studs can be considered to retain their original load-carrying capacity [52,53]. These predicted temperature profiles can be used to conduct structural analysis using FEM to predict the failure behaviors. However, the focus of this study is to predict heat transfer across the wall systems to help improve large-scale heat transfer models of buildings exposed to bushfire conditions and thus, FEM analysis is not within the scope.

Furthermore, the results showed that the use of cavity insulation resulted in reduced AS temperatures but increased the FC temperatures in wall systems PS3 and PS6. This creates a higher temperature difference between the hot and cold flanges of the wall studs. Even though this is favorable for the non-load bearing conditions by delaying the heat transfer to the AS, the load-bearing walls have shown reduced fire resistance due to structural failure when cavity insulation is used because of increased temperature differences between hot and cold flanges and increased hot flange temperatures [21]. Since external walls are constructed as load-bearing walls, it is necessary to consider the effect of cavity insulation on the structural performance of stud walls under fire conditions. Other experimental studies [54] showed that using insulation outside the cavity (external insulation) acted favorably for these conditions as it does not create a large temperature difference between the flanges of the stud. Furthermore, external insulation can be recommended for bushfire conditions as it can be replaced along with the steel cladding and the batten in the above-discussed wall systems after a bushfire event.

The effect of cavity insulation on heat transfer was evaluated using two different wall systems ((PS1 versus PS3) and (PS6 versus PS7)). Fig.22 compares the predicted FC and AC surface temperatures with and without cavity insulation. Higher FC temperatures and lower AC temperatures were observed in both settings when cavity insulation was used. This implies that the use of cavity insulation results in reduced AS wall temperatures. Even though the glass fiber insulation melts when its temperature exceeds 600 °C, it did not reach the melting temperature during both bushfire exposure simulations. Therefore, it is concluded that cavity insulation will not be affected during the bushfire exposures for the given wall systems, provided the gaps in the walls are properly sealed.

Overall, this study investigated the heat transfer of external wall systems lined with non-combustible wall lining materials using FDS. The model-predicted temperatures showed a good agreement with the experimental results of six different wall configurations measured across the wall thickness. Furthermore, the parametric study consisted of seven external wall systems exposed to two bushfire flame zone curves and provided a comparison of the performance of different cladding materials and their contribution to the performance of wall systems under bushfire flame zone conditions. These models help in improving large-scale heat transfer models of buildings exposed to bushfires (attacks come from the external side of the buildings). The heat transfer across the external envelope is important in these studies as opposed to internal fires. Furthermore, full-scale fire tests of buildings exposed to bushfires are expensive and small-scale components can be tested at a significantly lower cost which can then be used to validate the building component models. However, the authors acknowledge the differences in the behavior of individual elements in comparison to the buildings as a whole and further studies are required to investigate this effect.

Furthermore, the factors considered in designing buildings for bushfire conditions are different for the building fires. The main focus in bushfires is to ensure the integrity of the external building envelope. Therefore, the integrity of external wall lining materials is important to avoid fires igniting the combustibles in wall cavities as well as the inside of the buildings. The model predictions in this study can be further used to determine the maximum temperatures across these external walls and use them in designing the service lines.

6.3 Model limitations

In this numerical study, the model showed limitations in simulating the effect of plasterboard fall-off when exposed to fires of a longer duration. However, a bushfire is expected to last for a shorter duration than a building fire and thus the model predictions are considered reasonably accurate. If the model is used to simulate prolonged fire conditions, the effect of plasterboard fall-off needs to be considered. Plasterboard fall-off can be manually induced when the surface reaches a specific temperature. However, it is challenging to determine the exact locations of the fall-off and the size of the area affected, and therefore, further studies are recommended. Furthermore, the thermal properties used in this study are based on a constant rate of temperature rise, and further studies are needed on the effects of different rates of temperature rise on thermal properties.

7 Conclusions

This study has presented a numerical investigation of the bushfire resistance of external LSF walls under flame zone conditions using FDS heat transfer models. It provides the details of developing a heat transfer model with verification and a mesh sensitivity analysis. The model was then used to predict the temperatures of six external and internal wall systems exposed to bushfire conditions and compared them with experimental results, which showed a good agreement. The study was then extended to predict the temperatures across seven external LSF wall systems exposed to two different bushfire flame zone exposure curves. The results were used to evaluate the heat transfer characteristics of external wall systems lined with four materials (steel cladding, gypsum plasterboards, fiber cement boards and AAC panels).

The parametric study results showed that the use of thin external steel cladding on the external wall systems reduced and delayed the FS board surface temperatures and is therefore suitable for applications in bushfire-prone areas. Evaluation of wall systems lined with gypsum plasterboards (13 and 16 mm thick) along with steel cladding showed adequate bushfire performance. However, the plasterboards had been through both dehydration processes by the end of the simulated bushfire exposures. Furthermore, 75 mm thick AAC panels in LSF walls demonstrated superior performance under bushfire conditions. The AS temperatures of the walls were less than 50 °C in both cases (with and without cavity insulation), while the steel stud hot flange temperatures were below 100 °C. The AAC panels can withstand repeated bushfire exposures and thus provide a good solution for the external wall systems used in bushfire-prone areas. However, further studies are required to confirm this. The results showed that the load-bearing capacity of thin-walled steel studs was adequate during the bushfire exposures of wall assemblies with gypsum plasterboards and AAC panels on the FS.

Overall, this study has improved the knowledge of numerical modeling of heat transfer of external wall systems exposed to bushfire flame zone conditions. The modeling methodology can be used in other applications of wall systems exposed to other fire conditions. Furthermore, the outcomes of this study are important in improving the heat transfer models of full-scale buildings exposed to bushfires.

References

[1] Buechi H, Weber P, Heard S, Cameron D, Plantinga A J. Long-term trends in wildfire damages in California. International Journal of Wildland Fire, 2021, 30(10): 757–762

[2] StandardsAustralia Limited. Methods for Fire Tests on Building Materials, Components and Structures, Part 8.2: Tests on Elements of Construction for Buildings Exposed to Simulated Bushfire Attack––Large Flaming Sources. Sydney: Standards Australia Ltd., 2018

[3] StandardsAustralia Limited. Construction of Buildings in Bushfire-Prone Areas. Sydney: Standards Australia Ltd., 2018

[4] NFPA1144. Standard for Reducing Structure Ignition Hazards from Wildland Fire. Quincy, MA: NFPA, 2013

[5] IWUIC 2018. International Wildland Urban Interface Code. Washington, D.C.: International Code Council, 2018

[6] Jessop D, Abu A, Wade C, Spearpoint M, Gerlich H. Performance of a light timber-framed compartment in natural fire subjected to lateral load. Fire and Materials, 2019, 43(2): 175–188

[7] Jiang J, Lu Y, Dai X, Li G Q, Chen W, Ye J. Disproportionate collapse of steel-framed gravity buildings under travelling fires. Engineering Structures, 2021, 245: 112799

[8] Zhan Q, Xiao Y, Musso F, Zhang L. Assessing the hygrothermal performance of typical lightweight steel-framed wall assemblies in hot-humid climate regions by monitoring and numerical analysis. Building and Environment, 2021, 188: 107512

[9] Ariyanayagam A D, Mahendran M. Numerical modelling of load bearing light gauge steel frame wall systems exposed to realistic design fires. Thin-walled Structures, 2014, 78: 148–170

[10] Ariyanayagam A D, Mahendran M. Experimental study of non-load bearing light gauge steel framed walls in fire. Journal of Constructional Steel Research, 2018, 145: 529–551

[11] Kesawan S, Mahendran M. A review of parameters influencing the fire performance of light gauge steel frame walls. Fire Technology, 2018, 54(1): 3–35

[12] Gnanachelvam S, Ariyanayagam A, Mahendran M. Fire resistance of LSF wall systems lined with different wallboards including bio-pcm mat. Journal of Building Engineering, 2020, 32: 101628

[13] Tao Y, Mahendran M, Ariyanayagam A. Fire tests of cold-formed steel walls made of hollow section studs. Journal of Constructional Steel Research, 2021, 178: 106495

[14] XuQHofmeyerHMaljaarsJvan HerpenR A. Thermomechanical modelling of sandwich panels with connections in fire resistance tests. In: Proceedings of the SiF 2022––The 12th International Conference on Structures in Fire. Hong Kong, China: PolyU, 2022, 703–714

[15] de Boer J, Hofmeyer H, Maljaars J, Van Herpen R. Two-way coupled CFD fire and thermomechanical FE analyses of a self-supporting sandwich panel facade system. Fire Safety Journal, 2019, 105: 154–168

[16] Pancheti J, Mahendran M. Fire resistance of external light gauge steel framed walls clad with autoclaved aerated concrete panels. Thin-walled Structures, 2021, 167: 108201

[17] Hendawitharana S, Ariyanayagam A, Mahendran M, Steau E. Bushfire resistance of external light steel wall systems lined with fibre cement boards. Fire Safety Journal, 2023, 139: 103806

[18] Pancheti J, Mahendran M, Steau E. Fire resistance of external LSF walls with corrugated steel cladding. Journal of Constructional Steel Research, 2022, 188: 107008

[19] Pancheti J, Mahendran M. Fire resistance of external light gauge steel framed walls with brick veneer cladding. Thin-walled Structures, 2023, 182: 110162

[20] Hendawitharana S, Ariyanayagam A, Mahendran M, Steau E. Evaluating the bushfire resistance of a safe room using full-scale experiments. Structures, 2023, 49: 995–1015

[21] Ariyanayagam A D, Mahendran M. Influence of cavity insulation on the fire resistance of light gauge steel framed walls. Construction & Building Materials, 2019, 203: 687–710

[22] Rusthi M, Keerthan P, Mahendran M, Ariyanayagam A. Investigating the fire performance of LSF wall systems using finite element analyses. Journal of Structural Fire Engineering, 2017, 8(4): 354–376

[23] Peiris M, Mahendran M. Advanced numerical modelling of light-gauge steel framed walls subject to eccentric compression. Engineering Structures, 2022, 256: 114063

[24] MaCLuHLiRQuM. One-dimensional finite difference model and numerical simulation for heat transfer of wall in chinese solar greenhouse. Transactions of the Chinese Society of Agricultural Engineering, 2010, 26(6): 231–237 (in Chinese)

[25] MagarabooshanamHAriyanayagamA DMahendranM. Numerical study of double stud LSF walls exposed to fire conditions. In: Proceedings of Cold-Formed Steel Research Consortium Colloquium 2020. Baltimore, MD: JScholarship, 2020

[26] Hadjisophocleous G, Jia Q. Comparison of FDS prediction of smoke movement in a 10-storey building with experimental data. Fire Technology, 2009, 45(2): 163–177

[27] Lázaro D, Puente E, Lázaro M, Lázaro P G, Peña J. Thermal modelling of gypsum plasterboard assemblies exposed to standard fire tests. Fire and Materials, 2016, 40(4): 568–585

[28] Nguyen Q, Ngo T, Tran P, Mendis P, Aye L, Baduge S K. Fire resistance of a prefabricated bushfire bunker using aerated concrete panels. Construction & Building Materials, 2018, 174: 410–420

[29] Zhou J, Zhou X, Cong B, Wang W. Comparison of different CFD-FEM coupling methods in advanced structural fire analysis. International Journal of Thermal Sciences, 2023, 193: 108465

[30] Zhou J, Zhou X, Cong B, Wang W. Numerical study of the convective heat transfer coefficient for steel column surrounded by localized fires. Fire Safety Journal, 2023, 141: 103987

[31] Zhou J, Zhou X, Cong B, Wang W, Gu M. Simulation of steel beam under ceiling jet based on a wind−fire−structure coupling model. Frontiers of Structural and Civil Engineering, 2023, 17(1): 78–98

[32] Glasa J, Valasek L, Weisenpacher P, Halada L. Cinema fire modelling by FDS. Journal of Physics: Conference Series, 2013, 410: 012013

[33] Cicione A, Beshir M, Walls R, Rush D. Full-scale informal settlement dwelling fire experiments and development of numerical models. Fire Technology, 2020, 56(2): 639–672

[34] Hendawitharana S, Ariyanayagam A, Mahendran M, Gonzalez F. Lidar-based computational fluid dynamics heat transfer models for bushfire conditions. International Journal of Disaster Risk Reduction, 2021, 66: 102587

[35] AustralianBuilding Codes Board (ABCB). Bushfire Verification Method Handbook. Canberra: ABCB, 2019

[36] McGrattanKHostikkaSFloydJMcDermottRVanellaMMuellerE. Fire Dynamics Simulator-Technical Reference Guide. Gaithersburg, MD: NIST Special Publication, 2023, 1018–1

[37] Kuzyk A, Skorobagatko T, Yemelyanenko S, Borys O, Dobrostan O. Computer simulation of fire test parameters façade heat insulating system for fire spread in fire dynamics simulator (FDS). Series of Geology and Technical Sciences, 2020, 4(442): 35–44

[38] BlakePPhylaktouHAndrewsG. Validating FDS against a full-scale fire test. In: Proceedings of 2018 Fire and Evacuation Modeling Technical Conference. Gaithersburg, MD: FEMTC, 2018

[39] Li Q, Zhang C, Li G Q. Symmetric modeling of the thermal actions in a structural fire experiment on a long-span composite floor beam in a compartment. Fire Safety Journal, 2021, 120: 103079

[40] HowellJ RMengücM PDaunKSiegelR. Thermal Radiation Heat Transfer. Boca Raton, FL: CRC Press, 2020

[41] McGrattanKMcDermottRVanellaMHostikkaSFloydJ. Fire Dynamics Simulator Technical Reference Guide Volume: 2 Verification. Gaithersburg, MD: NIST Special Publication, 2021, 1018–2

[42] Zehfuß J, Sander L. Gypsum plasterboards under natural fire—Experimental investigations of thermal properties. Civil Engineering Design, 2021, 3(3): 62–72

[43] AriyanayagamAMahendranM. Thermal Finite Element Analysis of Peer Stud Walls Lined with Gypsum Plasterboards––Part 1. Brisbane: Queensland University of Technology, 2016

[44] Jatheeshan V, Mahendran M. Experimental study of cold-formed steel floors made of hollow flange channel section joists under fire conditions. Journal of Structural Engineering, 2016, 142(2): 1409

[45] TaoYMahendranMAriyanayagamA. Load-bearing walls made of cold-formed steel hollow section studs exposed to fire. In: Proceedings of 9th International Conference on Steel and Aluminium Structures: ICSAS19. London: Independent Publishing Network, 2019, 1533–1544

[46] PanchetiJ. Fire Resistance of external light gauge steel framed wall systems. Dissertation for the Doctoral Degree. Brisbane: Queensland University of Technology, 2022

[47] GnanachelvamS. Fire and energy performance of cold-formed steel frame wall systems. Dissertation for the Doctoral Degree. Brisbane: Queensland University of Technology, 2020

[48] McGrattanKMcDermottRVanellaMHostikkaSFloydJ. Fire Dynamics Simulator Technical Reference Guide Volume: 3 Validation. Gaithersburg, MD: NIST Special Publication, 2021, 1018–3

[49] Kolaitis D I, Founti M A. Development of a solid reaction kinetics gypsum dehydration model appropriate for CFD simulation of gypsum plasterboard wall assemblies exposed to fire. Fire Safety Journal, 2013, 58: 151–159

[50] Wakili K G, Hugi E, Karvonen L, Schnewlin P, Winnefeld F. Thermal behaviour of autoclaved aerated concrete exposed to fire. Cement and Concrete Composites, 2015, 62: 52–58

[51] Ariyanayagam A, Mahendran M. Fire performance of load bearing LSF wall systems made of low strength steel studs. Thin-walled Structures, 2018, 130: 487–504

[52] Ariyanayagam A, Mahendran M. Residual capacity of fire exposed light gauge steel frame walls. Thin-walled Structures, 2018, 124: 107–120

[53] Kankanamge N, Mahendran M. Mechanical properties of cold-formed steels at elevated temperatures. Thin-walled Structures, 2011, 49(1): 26–44

[54] Gnanachelvam S, Ariyanayagam A, Mahendran M. Effects of insulation materials and their location on the fire resistance of LSF walls. Journal of Building Engineering, 2021, 44: 103323

RIGHTS & PERMISSIONS

Higher Education Press