Numerical simulation of the fire behaviour of façade equipped with aluminium composite material‐based claddings—Model validation at intermediate scale

Increasing the energy performance of buildings is a crucial sustainable development objective. However, building features, products, mounting, and fixing of façade components have a large impact on fire safety. Authors in previous study performed façade fire propagation tests according to ISO13785‐1 on different combinations of ACM claddings and insulants.


Summary
Increasing the energy performance of buildings is a crucial sustainable development objective. However, building features, products, mounting, and fixing of façade com- The external thermal insulation (ETI) of residential building façades is exhibiting strong growth in Europe. 1 Many existing buildings are renovated, aiming for better energy efficiency and aesthetic improvements. However, building features and products (on their own or in combination) may have an impact on fire safety, because fire spread through façades can be one of the fastest ways for a fire to spread insulation system. Façade fire spread can be increased by the combustible load of the façade. As the fire safety design of buildings, in general, aims to slow fire spread by deliberate compartmentation, propagation via the façade may lead to quicker fire spread than any compartmentation requirement assumes, leading to difficulties in evacuation and firefighting. composite systems," comprising insulation fixed to the wall and a finish of thick or semi-thick render, between which are successively inserted a reinforcement and intermediate coats. In the context of fire safety, it should be noticed that for systems such as ETICS or ventilated façades, the materials used (cladding or insulation) may be combustible. 2 Moreover, in ventilated facades, the air gap may be a vector of fire propagation through chimney effect. [3][4][5][6] In recent years, several research studies have investigated the fire performance of ventilated façade systems and shown that flame spread through the air cavity of a ventilated façade can be several times higher than that along the outside the facade. 4,5 Furthermore, the presence of an air cavity tends to increase the energy released by the facade system. 6 Thus, both the materials taken independently and the whole system (combination of materials and assembly) are potentially a source of propagation of a fire. At present times, the use of fire barriers or compartment systems, as requested by national regulations, can hinder these problems, but they too constitute additional variables in the system.
The fire behaviour of an external façade insulation system is dependent on the overall system's performance, rather than the performance of the individual components. A façade system includes not only the cladding and the insulant's characteristics but also those of cavities, cavity barriers, mounting and fixing, substrate, and any singularities, such as window frames. All these elements interact strongly when involved in a fire; thus, as discussed in Bonner and Rein, 3  The appropriate scale for fire tests of such façade systems should also be discussed. For a scale to be appropriate, it must make it possible to evaluate the fire performance of the whole façade system by including all of the system's characteristics and fixing methods.
Assessment of a specific façade system's fire performance can be undertaken using large-scale testing in accordance with local regulations, and large assembly tests often include the methods described in BS 8414-1 and BS 8414-2, FM 4880-1, or NFPA 285. Many combustible façade systems, incorporating aluminium composite materials (ACMs), have recently been tested using these large-scale test methods. 10, 11 However, these large-scale tests are pass/fail orientated; they give very little quantitative information for further interpretation of the fire behaviour of the tested systems and are expensive and time consuming to prepare.
Recently, a series of façade fire propagation tests have been performed according to the ISO 13785-1 standard, 12  Façade fires have been studied numerically for different test facilities or methods and using different simulation codes. [14][15][16][17][18][19][20][21][22] Published results have shown the feasibility of modelling such test methods using large eddy simulation (LES), especially when incombustible claddings were considered. However, great attention must be paid to the numerical model sensitivity, in particular to correctly representing the behaviour of the flames near the façade system and thus the thermal stresses received by the facades. In the case of a combustible façade system, the thermal degradation suffered by the façade materials leads to the release of pyrolysis gases. The unburnt fuel gases are then transported by convection where they react with the ambient air, itself driven in the thermal plume. Moreover, the hot gases are cooled when they ascend as air is naturally convected by the external thermal plume. These aeraulic and combustion phenomena need to be correctly reproduced by the numerical approach.
In this paper, numerical simulations were performed to reproduce these intermediate fire tests. 13 The aim of this study is to validate a numerical model that could predict the behaviour of the whole systems at this scale, for later upscaling studies. Special attention was given to flow and thermal conditions at all locations in the tested system. This allows for additional investigation and understanding of the relative contribution of insulant and ACM cladding. In particular, the relative contribution of each component to the fire behaviour of the system can be numerically assessed.
The numerical simulations as well as the tests performed show that the ACM cladding is the most important element driving the global fire behaviour of the tested façade systems. In particular, ACM-PE-based cladding systems, whatever the insulant used in the system, show very marked fire propagation. Moreover, the integrity of the cavity is affected by the dripping and the destruction of the burning ACM-PE cladding.
Once the model for the fire behaviour of the façade system is validated at intermediate scale, larger façade systems will be investigated numerically to evaluate the influence of scaling.

| Test facility
The set-up of the experimental facility was developed according to ISO 13785-1 specifications. 12  Two types of external façade insulation system are generally encountered: (a) ventilated façades, consisting of an assembly of several layers (such as cladding, air gap and insulation) applied to the wall, and (b) rendered systems or ETICS "external thermal insulation (CalSil) boards as a support for the tested system and on which the system is installed ( Figure 1A).
A 100-kW sand propane burner with dimension 1200 × 100 × 150 mm (l × w × h) is installed with its upper surface placed 250 mm below the lower edge of the sample. The complete system is then placed under a large hood to collect the effluents. The instrumentation used during the test is fully detailed in Guillaume et al 13 ( Figure 1B).

| Tested systems in the experimental reference 13
In the reference paper, 13

| Main results of the experiments
The façade fire tests described above 13

| NUMERICAL SET-UP
The objective of the numerical study is to reproduce accurately the thermal loads imposed on the tested system, the thermal behaviour of the system, and the fire propagation through the system, in particular the observations of temperature, heat release rate, smoke, and chemical species (mainly CO and CO 2 ) released. The thermal characteristics of the system components are integrated in the model.  The default submodels of FDS were used for the gas phase radiation exchanges even if a sensibility analysis performed with 100 (default value), 500, and 800 solid angles was addressed. The combustion model with primitive and lumped gas species definition, to solve a transport equation for each species to be tracked, was also investigated, as well as the use of a single step reaction for CO production.  This technical choice is made to conserve reasonable calculation costs and regarding the later upscaling application of this numerical model. Compromises were needed to develop a robust numerical model able to be used for this application. The selected cell size is enough to capture the main features of local effects, not in details, but sufficiently to reproduce the fire behaviour in the present application. Furthermore, quickly after the beginning of the test, the fire propagation from the burner to the system leads to its combustion.
Thus, the cladding panels, and thereby the gaps between them, disappear in the first minutes of the test.

| Numerical model for thermal analysis
The thermal characteristics of the system components are integrated in the numerical model in terms of density, thermal conductivity, heat capacity, emissivity, heat of combustion, ignition temperature, mass loss rate, and species release rates, for every material involved. All thermal and combustion properties considered, for the material making up the systems, are given below.
In this study, the thermal and aeraulic phenomenon involved in the fire development from the initial fire source and in the fire spread through the tested system depends mostly of the fire behaviour of all the materials and of their arrangement in the system. FDS has several approaches for describing the pyrolysis of solids, and the selected approach depends on the availability of material properties and the appropriateness of the underlying pyrolysis model in simulation tools.
In FDS, the thermal degradation of materials can be simulated with either a simple pyrolysis model or solid fuels that burn at a specified rate. The second model is used in this study. The material heats up until its surface temperature reaches locally the given ignition temperature for every combustible materials. Once the surface of an individual solid cell has ignited locally, the heat transfer into the solid is still calculated, and the fire spread is modelled cell by cell. The appropriate mass loss rate derived from literature review for every combustible material is prescribed when ignition occurs.
The fuel burnout in each solid numerical cell is calculated from the combustible mass and the heat of combustion of the object through specification of its bulk density. The bulk density approach used in this study allows taking into account the appropriate heat of combustion of each fuel and is related to the mesh size. Thus, this parameter is calculated for a given cell size and will not be sensitive to the meshing. The density of PIR foam is taken at 36 kg/m 3 as presented in the product datasheet. 28 As no indication about the emissivity of PIR foam materials can be found in the literature, a value of 1 is assumed. This is justified by the surface charring of the foam. The heat capacity of PIR is around 1.10 J/g/K at ambient temperature. 25 Its thermal conductivity is around 0.021 W/m/K at ambient temperature 15,28,29 and can go up to 0.0531 at 90°C. 25 A temperature-independent value of 0.048 W/m/K is taken in this study. According to the literature, the heat of combustion of PIR can be taken at 26.3 MJ/kg with a CO and HCN yields of 0.038 and 0.01 g/g respectively. 26,27,30,31 The soot yield is 0.1 g/g. 27 The ignition temperature of PIR can be found around 370°C 30 and the maximum HRR is around 160 kW/m 2 . 25,29 The heat of decomposition of PIR is taken at 1750 J/g. 30 In Marquis et al 25,32 and Purser and Purser, 26 the effect of oxygen concentration depletion on the mass loss rate and on the residual mass fraction of PIR foam at 50 kW/m 2 is studied. The maximal value of mass loss rate of 0.006 kg/ m 2 /s used in this study is then representative of the maximum value achieved in well ventilated conditions and high thermal solicitations.
Thus, this value is always secure in the numerical models. In Marquis et al 25 Lyon and Janssens 30 and Purser, 33 the CO yield released by PIR foam is indicated as a function of the equivalence ratio Φ accounting for the recovery fraction due to ventilation conditions. The average value of 0.038 g/g used in this study is thus representative of well-ventilated regimes 30,33 to well-ventilated regimes with medium thermal loads (20% O 2 , 15 kW/m 2 ). 25 The thermal properties of PE used in this study are indicated in Trouvé 34 and Hietaniemi and Mikkola. 35 According to material datasheet, PE has a density of 1360 kg/m 3 . Its heat of decomposition is taken at 2300 J/g. 30,34 The ignition temperature of PE can be estimated around 380°C. 35 PE has a heat of combustion ranging between 40.3 and 46.3 MJ/kg. 27,30 The CO yield is of 0.024 g/g, 27,30 and the soot yield is 0.056 g/g. 27 The char yield is close to 0%, 27,30 and the asymptotic mass loss rate is 0.04 kg/m 2 /s. 27,30,35 In previous works, 27 The overall uncertainty of a numerical prediction is the combination of the uncertainties of both the numerical model and of the input parameters. 39 The numerical uncertainties are evaluated following 40,41 and are indicated in Table 2

| Fire behaviour observations
A comparison of the experimental fire behaviour of the system with numerical observations is presented in Figure 2    temperature at the side wall and in the air cavity, a "double maximum" is observed between 4 and 7 minutes (Figure 4). This double peak behaviour appears on the cavity temperature measurement before those of the side wall, with a delay in the range 30 seconds to 1 minute, due to the thermal inertia of the insulant and cladding materials.
The first peak is due to the insulant ignition, with a maximum mass loss rate observed around 4 minutes. Then, at side wall, the horizontal cavity barrier intumesces around 4 minutes 30 seconds. The second There is an excellent agreement between the numerical prediction and experimental data for the heat release rate during the test without the burner contribution (100 kW) ( Figure 5). The maximum values are close: 5.0 ± 0.5 MW during test (10% uncertainty) and 5.3 ± 0.9 MW in simulation (17% uncertainty) and reached at 5.5 minutes of test. This is correlated with the observations of the fire development (Figure 2) inside the system. The numerical model does not reproduce the small contribution early in the test (between 1 and 4 min) and predicts a longer contribution after the mean peak (between 6.5 and 9 min).
Regarding the energy released (THR), there is a good match between experimental and numerical results. The maximum values are close to (683 ± 68) MJ during the test and (710 ± 120) MJ in the simulation, corresponding to a difference of only 4%.
Moreover, the numerical approach allows the evaluation of the contribution of each part of the system (ACM-PE cladding or PIR insulant) during the test. The HRR evaluated for the cladding and for the insulant is presented in Figure 6. The insulant has a much lower contribution to the global HRR of the system than the ACM-PE cladding.
Moreover, for the [ACM-PE + PIR] configuration, the insulant shows a maximum heat release rate of 0.46 MW, while it is around 5 MW for the cladding. Additionally, the peak of HRR for the insulant happens the cavity performance. The insulant will then be exposed to the fire contribution of the cladding and the flames in the cavity. During the fire test, the insulant can then burn in well-ventilated conditions because it is quickly exposed to the external environment once the cladding has disappeared. This effect is well reproduced by the model, thus indicating that the fuels stoichiometry and the fuel mass released are correctly considered in the simulation.

| Sensibility analysis of the PIR ignition temperature
According to the literature, PIR ignition temperature can commonly range between 350 and 420°C. 30  • An easily flammable PIR with an ignition temperature of 350°C, • A PIR with a moderate flammability associated with an ignition of 370°C, • A PIR with a low flammability associated with an ignition temperature of 420°C.
The comparison between experimental data and the numerical prediction for the different values of peak HRR attributed to the PIR insulant to the ACM cladding and to the system is synthetized in Table 4. A relative deviation of 9% is found comparing with experimental measure.
It can be concluded that the ignition temperature of PIR does not play a significant role in the overall behaviour of the system as the predicted maximum HRR value for the PIR insulant is always in the margin of incertitude of the measurements of 10% (Table 2).
However, ignition temperatures ranging between 370°C and 420°C decrease the difference between experimental and numerical data for the HRR of the whole system. Furthermore, using values of between 350°C and 420°C is conservative with respect to the incertitude of measurement.

| Variance of the numerical model for the [ACM-PE + PIR] system
The method described in ISO 16730-1:2015 42 is used to further validate the numerical model against the experimental results. The relative difference (hybrid method) and the cosine associated with each quantity to be validated are presented in Table 5. The minimum cosine value evaluated, close to 0.72, is associated with the maximum relative difference value close to 57% and concerns the CO concentration.
However, this low cosine value and high relative difference value are explained by the CO concentration that is slightly overestimated in the numerical model. As justified previously, this difference can be      In Figure 15 The energy released (THR) is addressed in Figure 16   Additional information, such as the relative contribution of each part of the system (ACM-PE cladding or PIR insulant) during the test, which cannot be measured experimentally, was investigated with this numerical approach. The fire behaviour of each component of the overall system was thus validated numerically. The HRR evaluated for the cladding and for the insulant shows that the insulant has a much lower contribution to the global HRR of the system than the ACM-PE cladding. Additionally, the peak of HRR for the insulant happens earlier than the peak of HRR for the cladding contribution. Thus, the numerical modelling approach can help to understand the relative contribution of one material versus another in complex system. Hence, the ACM-PE represents more than 90% of the value of the peak of HRR and of the total energy released.
An additional sensitivity analysis was performed to evaluate the influence of the ignition temperature considered for the PIR insulant for the [ACM-PE + PIR] configuration. It can be concluded that the ignition temperature of PIR does not play a significant role in the overall behaviour of the system as the predicted maximum HRR value is always in the margin of incertitude of the measurements (10%).
Temperature profiles at different depths inside the PIR insulant and different location on the back wall were evaluated numerically for the [ACM-PE + PIR] configuration. The model predicted that an average thickness of 10 to 25 mm of PIR would achieve ignition temperature, depending on location. This is consistent with the experimentally observed char depth, of close to 10 mm. 13 Further validation of the numerical model was addressed through simulations of the system including a non-combustible mineral wool insulant and of the system including an inert non-combustible cladding.
This allows a deeper analysis of the material contribution to the fire spread in the façade systems and is then useful to further understand the role played by the individual components of the system, and to provide a reliable evaluation of façade systems in case of fire.
When compared with the heat flux measured above the system during the test of the [ACM-PE + PIR] configuration, higher heat fluxes are evaluated when PIR is used rather than MW. This is not related to the PIR combustion, but to the insulating properties of this material.
In all the simulated configurations and experimental results, the CO/CO 2 ratio is lower than 0.1 indicating that the combustion occurred in well-ventilated conditions. The strong combustion of the polyethylene in the ACM-PE leads to the consumption of the material quickly after its ignition. The external cladding burns in well-ventilated conditions because of its external location, and it burns away at early stage of the fire. The insulant is exposed to the fire contribution of the cladding and resulting flames in the cavity, even if the cladding has disappeared. During the fire test, the insulant can then burn in wellventilated conditions because it is quickly exposed to the external environment once the cladding has disappeared. This effect is well reproduced by the model, thus indicating that the fuels' stoichiometry and the fuel mass released are correctly taken into account in the simulation. For the [ACM-PE + MW] configuration, the CO 2 release is five times higher than for the inert cladding configuration, but comparable with the concentration evaluated for the [ACM-PE + PIR] system.
Thus, the high CO 2 release seems mainly due to the PE cladding combustion.
A comparison of the experimental HRR and that from the numerical models for the [ACM-PE + MW] and [ACM-PE + PIR] configurations shows that when PIR is used, a higher peak of HRR (of around 0.4 MW) is observed, but delayed of around 1 minute compared with that for the configuration with MW. This higher value is due to the small contribution of the PIR to the heat released. The delay is because ACM-PE combustion begins more slowly for the configuration with PIR. This is due to the energy absorbed by the PIR for charring, thermal cracking, and pyrolysis, leading to a competition between thermal and thermochemical effects. larger façade systems will be investigated numerically to evaluate the influence of scaling. The change in the ignition temperature or other thermal parameter will be investigated while changing the mesh resolution. The approaches for the use of ignition temperature model accompanied by area adjust for coarse mesh fire spread simulations following the work of Janardhan and Hostikka 43 will be investigated.
Additionally, further researches could address extended investigation in terms of numerical modelling with finest grid in cavities to observe the impact of more resolved mesh, or 3D thermal transfer for solid materials. Experimental developments could also be needed to evaluate the velocity in the cavity.