ArticlePDF Available

Effect of wind speed and direction on facade fire spread in an isolated rectangular building

Authors:

Abstract and Figures

This paper investigates the influence of wind speed and direction on external fire spread in an isolated rectangular building using computational fluid dynamics models validated with wind tunnel data and facade fire tests. Two wind speeds (2 m/s, 4 m/s) are considered for each of four wind directions (0°, 45°, 90°, 180°) and compared to a reference case of no wind. Results indicate that facade fire spread is heavily influenced by the near-wall flow fields generated by the building geometry. These flow fields explain counterintuitive findings such as the upstream tilting of flames under the influence of reverse flow near the side walls. The presence of external wind was found to inhibit the initial development of facade fires, but can greatly exacerbate fire spread once the fire has fully developed. The largest fire occurred for the case of no wind (7.5 GJ in 15 minutes) while the smallest fire occurred for the 4m/s diagonal wind case (2.2 GJ). An additional case with temporally varying wind conditions demonstrated a 50% increase in fire spread area compared to no wind. The study provides valuable insight into wind and fire interaction in building facades that can help improve the fire safety of buildings.
Content may be subject to copyright.
Accepted for publication in Fire Safety Journal (March 17, 2022)
https://doi.org/10.1016/j.firesaf.2022.103570
1
Effect of wind speed and direction on facade fire spread
in an isolated rectangular building
Yousef Abu-Zidana,b*, Shyanaka Rathnayakab, Priyan Mendisb, Kate Nguyena
a Civil and Infrastructure Engineering, RMIT University, VIC 3001, Australia
b Department of Infrastructure Engineering, University of Melbourne, VIC 3010, Australia
* Corresponding author. Email address: yousef.abu-zidan@rmit.edu.au
Abstract
This paper investigates the influence of wind speed and direction on external fire spread in an isolated
rectangular building using computational fluid dynamics models validated with wind tunnel data and
facade fire tests. Two wind speeds (2 m/s, 4 m/s) are considered for each of four wind directions (0°,
45°, 90°, 180°) and compared to a reference case of no wind. Results indicate that facade fire spread
is heavily influenced by the near-wall flow fields generated by the building geometry. These flow
fields explain counterintuitive findings such as the upstream tilting of flames under the influence of
reverse flow near the side walls. The presence of external wind was found to inhibit the initial
development of facade fires, but can greatly exacerbate fire spread once the fire has fully developed.
The largest fire occurred for the case of no wind (7.5 GJ in 15 min) while the smallest fire occurred
for the 4m/s diagonal wind case (2.2 GJ). An additional case with temporally varying wind conditions
demonstrated a 50% increase in fire spread area compared to no wind. The study provides valuable
insight into wind and fire interaction in building facades that can help improve the fire safety of
buildings.
Keywords
External wind; facade fire; combustible cladding; CFD simulation; near-wall reverse flow; synergistic
effect; external flame spread
2
1 Introduction
Recent incidents around the world have demonstrated the potentially devastating impact that
combustible cladding can have on the loss of life and property [1]. Modern trends towards
environmental sustainability and energy efficiency have resulted in the use of combustible facade
materials that are highly vulnerable to fire. Every year, an average of five tall buildings experience
fire spread along the external facade [2], with recent events such as the Grenfell Tower fire in
London, UK [3, 4] (Figure 1) and the Lacrosse fire in Melbourne, Australia [5] demonstrating the
significant influence that combustible facade materials can have on the speed of fire propagation
along the external walls of the building.
Figure 1: Recent example of combustible facade fires in Grenfell Tower, London, UK on 14 June 2017.
Although it is known that wind has a considerable effect on the spread of urban fires [6-9], the nature
of wind and fire interaction in facade fire propagation is yet to be investigated in detail. Current
standards for testing the fire performance of external facade systems, such as the ISO 13785 [10], BS
8414-1 [11], and AS 5113 [12], acknowledge the influence of external wind on large scale fire tests
and hence require wind speed at the beginning of testing to not exceed a maximum limit in order to
ensure consistency of results. The standards however do not require testing to be performed for
different wind speeds and directions. This results in uncertainty regarding the real-life performance of
facade systems under external wind.
Recent studies on ejected facade flames indicate that wind can influence facade fires in two main
ways: it can affect the flame height and trajectory, and it can affect the heat release rate of the fire.
Wind acting perpendicular to the facade was seen to reduce the height of ejected flames [13, 14],
while wind acting parallel to the facade was seen to tilt the ejected flame towards the downwind
direction [15-19]. The tilting angle increases with wind speed, and for very strong winds, the fire may
even spread horizontally along the same floor. This has been shown to reduce the effectiveness of fire
protection features such as overhangs and balconies that are designed to block vertical overflow of
fire plume spread [20]. The tilted trajectory of the flame allows the fire to penetrate the space between
balconies and overhangs, leading to vertical fire spread to upper floors.
The presence of external wind is also reported to influence the temperature and heat release rate of
facade fires. Hu et al. [13] mention two competing effects when subjecting a window-ejected flame to
external wind: the “air supply effect” and the “venting effect”. Below a certain threshold of wind
speed, the presence of external wind increases the air supply to the fire and consequently increases the
heat release rate. In such a case, the air supply effect is said to be dominant. As the wind speed
exceeds the threshold, the venting effect becomes dominant where the presence of external wind
causes unburnt fuel vapours to be vented away from the heat source before being ignited. Under such
3
conditions, an increase in wind speed will result in a reduction of heat release rate and flame
temperature.
The competing interaction between the air supply and venting effects help explain seemingly
conflicting findings in the literature. Wind acting perpendicular [21, 22] and parallel [17] to the facade
was reported to reduce the temperature of window-ejected flames, while an outdoor experiment by
Anderson et al. [22] reports an increase in fuel consumption on days when wind speed was higher.
The interaction of wind and fire is further complicated by various parameters including the size and
geometric configuration of facade openings and the presence of cross ventilation [23-26].
Previous studies on wind and fire interaction have mostly focused on window-ejected flames. Studies
on fire propagation along the external building walls are scarce and mostly consider fire spread on a
localised facade section [27] without accounting for the effect of the wind flow structures generated
by the full building geometry. The presence of a building in a wind flow stream generates complex
flow patterns that vary considerably from those experienced by a thin facade section. These flow
structures dictate the wind conditions at various locations on the building surface and can therefore
influence the spread of facade fires. The main flow structures generated by a rectangular building are
shown in Figure 2 and include the following features:
Stagnation region. This region forms in front of windward surfaces where wind flow is blocked
by the presence of the building. The region is characterised by low wind speed and large positive
pressures. For tall buildings, a downdraft may occur along the windward surface as high wind
speed at the top of the building is diverted downwards.
Wake region. Flow separation at the leading edges of the building results in the formation of a
wake region that encompasses the top, side, and leeward surfaces. The wake region is
characterised by high turbulence, low wind speed, and negative pressure. Vortex shedding may
also occur in the wake region, particularly for tall and slender buildings.
Wind shear layers. Flow separation at the leading edges results in the formation of wind shear
layers that define the boundary between the undisturbed streamwise flow and the wake regions.
Shear layers are characterised by high wind speeds and a large velocity gradient. The trajectory
of these layers is highly unstable, and very large negative pressures occur immediately
downstream of the point of separation where shear layers originate.
Figure 2: Top view of wind flow past a rectangular building. Wind flow direction from left to right [28].
4
A fire travelling along the building facade may experience varying wind conditions as it moves
through the different wind flow structures generated by the building. These conditions also vary
temporally due to flow instability caused by atmospheric and building-generated turbulence as well as
changes in weather conditions. Such effects need to be considered when predicting fire propagation
along building facades.
This paper presents a systematic investigation of the influence of wind flow patterns generated by an
isolated rectangular building on the propagation of fire along the external facade. The study is
performed using computational fluid dynamics (CFD) simulations that are validated with wind tunnel
data and facade fire tests. Fire Dynamics Simulator (FDS) 6.7.5 is used to simulate both the wind flow
field and facade fire propagation. The study involves the following steps:
1. Validating the wind flow field in FDS with results from the wind tunnel.
2. Validating the fire properties of the combustible facade material in FDS with results from
facade fire tests.
3. Performing a parametric study that involves varying the wind speed and wind direction
relative to the fire source on the building facade.
The findings of this study provide valuable insight into wind and fire interaction in building facades
that can help improve fire safety of buildings.
2 Methodology and numerical setup
The use of computational wind engineering (CWE) principles and tools for fire safety applications has
gained recent interest [6], and best practice guidelines for developing wind and fire coupled modelling
have been developed [29]. Developing a CWE model involves defining a building geometry within a
computational domain and specifying appropriate boundary conditions and meshing parameters.
These details are specified as follows.
2.1 Building geometry and wind profile
An isolated rectangular building with height = 20 m and width = 10 m is adopted in this study.
The building is subjected to an atmospheric boundary layer profile with reference wind speeds
corresponding to highly recurrent winds. Wind speed in FDS is specified in terms of the logarithmic
law [Eq. (1)]. The highest wind speed in this study is  = 4 m/s at  of 10 m. Open terrain
roughness is selected ( = 0.01 m) and von Karman constant is equal to 0.4. Inflow turbulence is
ignored in this study to reduce complexity and computational cost.
=
ln
(1)
For each simulation case, the selected wind speed and direction is held constant for the entire duration
of the simulation. Later in section 6.6, additional cases are presented where wind speed and direction
are varied during the simulation to investigate the effect of variable wind conditions on facade fire
propagation.
5
2.2 Mesh configurations and domain size
Fire analysis requires mesh refinement in the region of the fire plume, which in this study is located
near the building surfaces. The recommended grid size is estimated using the following equation [30]:
=
/
(2)
where is the characteristic fire diameter, is the total heat release rate of the fire, is the air
density, is the specific heat of air, is the ambient temperature, and is gravity. A grid with
/ between 4 and 16 is desirable, where is the nominal size of the mesh cell. Using Eq. (2), an
acceptable grid size in the current study is 0.25 m for a fire heat release rate of 3.4 MW. The resulting
/ value is around 6 which falls within the acceptable range.
A high level of mesh resolution near the building is also needed for better resolution of wind flow
features. To reduce the total number of computational elements, selective refinement is specified near
the building surface while coarser elements are specified elsewhere in the domain. Using a blocking
technique, a structured cartesian grid is generated with 0.25 m elements near the building surfaces, a
maximum element size of 2m away from the building (Figure 3). The total cell count is 1.7 million
elements.
The size of the computational domain is minimised to limit the total number of cells in the model. It is
important to note that wind flow simulations generally require large computational domains to ensure
appropriate development of the flow field around the building with minimal influence from domain
boundaries [31, 32]. However, adopting a large domain in the current study was found to be
computationally prohibitive. FDS requires the use of highly structured grids with stringent
requirements for grid alignment in all directions, and this low level of mesh control makes it very
difficult to limit the number of computational elements in large domains through selective refinement.
The computational domain in this study has an upstream and downstream length = 2, side
clearances = 2 and domain height = 2 (Figure 3).
6
Figure 3: Mesh configuration with increasing refinement near the building. Mesh count is 1.7 million elements.
2.3 Turbulence modelling and near-wall treatment
Flow turbulence is resolved using large-eddy simulations (LES) with the Smagorinsky sub-grid scale
model [33] and the default value of model constant Cs = 0.2. The Smagorinsky model has been shown
to perform reasonably well for predicting wind flow around a rectangular building of similar aspect
ratio to the current study [34]. A wall boundary condition is specified at the ground and building
surfaces, while a velocity inlet condition is specified at the lateral boundary corresponding to the wind
approach direction. All other lateral boundaries, as well as the top boundary, are specified open
boundary conditions. At the building surface, near-wall flow is treated using wall functions based on
the logarithmic law of the wall [35]. The wall functions are used for determining the wall stresses
(), while the eddy viscosity () in the wall-adjacent cells are determined using the wall-adapting
local eddy-viscosity (WALE) model [36].
3 Validating wind flow field in FDS
The wind flow field is validated by comparing results from FDS to experimental results from the wind
tunnel. The experimental results of Meng and Hibi [37] are used which provide wind speed
measurements at various locations around a rectangular building (height to width ratio of 2:1)
subjected to an atmospheric wind profile. The inflow wind speed profile in FDS is selected to closely
reflect the wind profile from in the wind tunnel experiment. The FDS profile is defined in terms of the
logarithmic law [Eq. (1)], with = 0.24 m/s, = 0.4, and an aerodynamic roughness length of =
0.01 m. This corresponds to a reference windspeed  = 4 m/s at the reference height  = 10 m.
A good agreement between the numerical and experimental profile is observed as shown in Figure 4.
Domain
Height
2
7
Figure 4: Comparison of wind speed profile in FDS simulation and wind tunnel experiment [37].
To validate the numerical model, wind speed is measured in the FDS model at 186 locations around
the building. The locations of these measurements are specified in Meng and Hibi [37]. Computations
are performed for 100 s to initialise the flow field and a further 300 s over which time histories of
wind speeds are recorded. Mean wind speeds at each measurement location is calculated by averaging
the time history signal and the results are compared with experiments in Figure 5, where the vertical
axis corresponds to numerical results from FDS, the horizontal axis corresponds to experimental
results, and the shaded areas show the error margins. The dashed diagonal line represents an exact
correlation between experimental and numerical results, and the data is fitted with a linear regression
curve (red line) with intercepts set at 0.
Figure 5: Comparison of FDS wind speed predictions with experimental results [37] with shaded areas showing error
margin.
The results in Figure 5 indicate an acceptable level of agreement between the FDS model and
experiment where 70% of all predictions fell within an error margin of ±30%. The R-squared value
(coefficient of determination) is 0.9 and the linear regression slope is almost equal to 1.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 0.2 0.4 0.6 0.8 1 1.2 1.4
z/H
U/U
ref
FDS Inlet
Experiment
y = 1.034x
R² = 0.9
0
1
2
3
4
5
6
0123456
Experiment wind speed (m/s)
8
Nonetheless, some noticeable discrepancies occur at a limited number of locations in Figure 5 where
the wind speed was overpredicted by the FDS model. These problematic points occur mostly at the
boundary of the wake region and near the edge-induced shear layer, which suggests that the
discrepancies are due to the FDS model underpredicting the size of the wake region. The FDS model
seems to predict a narrower wake with the shear layers located closer to the sides of the building
compared to the experiment. The exact cause of this error is not immediately evident and could be due
to several factors including insufficient mesh refinement and lack of inflow turbulence. The relatively
small domain size may have also contributed to the narrower wake and acceleration of flow near the
sides of the building [32]. Aside from these discrepancies, the FDS model performed reasonably well
at other locations near the building surface with a mean absolute error of 27%.
To further validate the FDS model, a qualitative investigation of the wind flow field around the
building is performed. Figure 6 presents contour plots of instantaneous wind speeds on the central
vertical plane (Figure 6a) and a horizontal plane at the mid-height of the building (Figure 6b). These
plots confirm the formation of wind flow structures that include a stagnation region at the windward
surface, a wake region on the sides and leeward faces, and fluctuating shear layers at the leading
edges of the building. These structures are consistent with those observed around an isolated
rectangular building in previous wind tunnel experiments and numerical studies. Hence, the FDS
model is deemed adequate for the purposes of investigating wind and fire interaction in building
facades, despite some inaccuracies in the quantitative prediction of wind speeds at a limited number
of locations in the model.
Figure 6: Side view (a) and top view (b) of the instantaneous wind flow field around the building from the FDS model.
Building
Shear layer
Stagnation
region
(a) Side view
Wake
WIND
Building
(b) Top view (z/H = 0.625)
Stagnation
region
WIND
Shear
layer
Wake
0 1 2 3 4 5 6 7
Wind speed (m/s)
9
4 Validating combustion properties of facade material
Before investigating wind and fire interaction, a numerical model for the cladding material is
developed in FDS and validated with experimental data. Aluminium composite panels (ACP) with a
polyethene (PE) core was selected for this study due to their high combustibility and readily available
experimental data on fire performance. Large-scale experimental testing of various ACPs was
commissioned by the Department for Communities and Local Government (DCLG), London in
response to the Grenfell cladding fire incident. Testing was performed per the BS 8414-1 standard for
assessing the behaviour of non-load bearing external cladding systems exposed to an external fire
[11]. The material from DCLG fire test 1 [38] is selected in this study (Figure 7), which consists of a
front-facing aluminium composite panel (ACP) constructed from two 0.5 mm aluminium sheets that
surround a 3 mm PE core on either side. A 55 mm air gap is provided behind the ACP, followed by a
100 mm-thick rigid polyisocyanurate foam (PIR) insulation board.
Figure 7: Cladding cross-section showing material composition and thicknesses.
Figure 8a shows the DCLG testing rig. The cladding extends 8.5 m high (6.5 m above fire source),
and the widths of the main wall and wing wall are 2.6 m and 1.3 m, respectively. Thermocouples are
installed at two levels with heights of 2.5 m and 5 m above the fire source.
Polyethylene (PE) core
Polyisocyanurate (PIR)
55 mm
Air gap
3
mm
0.5
mm
0.5
mm
100 mm
Aluminum
composite panel
(ACP)
Insulation
outdoor
indoor
10
Figure 8: (a) Full-scale DCLG test [38], (b) equivalent FDS model, and (c) model dimensions in mm (not to scale).
To validate the material properties in the numerical model, an equivalent testing rig is constructed in
FDS (Figure 8b) with equivalent locations of thermocouples and an equivalent fire source.
Combustion properties of the ACP material were selected by calibrating the FDS model to
experimental measurement from the cladding fire test. A computational grid of 0.25m is used which is
equivalent to that adopted in the full building model. Although this grid is quite coarse considering the
scale of the testing rig, maintaining a consistent grid between the fire rig model (Figure 8b) and the
full building model (Figure 3) was necessary as previous studies have demonstrated that the
computational grid can influence combustion behaviour of materials [39]. Having a consistent grid
will ensure that calibrated material is appropriate for use in the full building model.
4.1 Fire source properties
The fire source in the DCLG test is based on BS 8414-1 specification represents an external fire
source, or a fully-developed fire in a room, that exposes the cladding to external flame [38]. The fire
source in the cladding experiment involves a wooded crib with approximate dimensions of 1.5 m x 1
m x 1 m. Figure 9 shows the heat release rate function of the wooden crib which was calibrated by
Dréan et al. [39]. The fire source has a maximum heat release rate of approximately 3.4 MW.
(a) DCLG test
(b) FDS model
Fire
source
Wing
wall
Level 1
Main wall
Level 2
Thermo-
couples
(c) Model dimensions
Level 1
2600
1340
2000
220
2000
Combustion
chamber
2500
2500
1500
500
500
500
500
Level 2
M1 - M5
Main wall
Wing
wall
M6 - M10
11
Figure 9: Heat release rate of fire source [39].
4.2 Thermal properties of facade material
Simulating ACP material in FDS is challenging as this material consists of multiple layers with
varying combustion and thermal properties. The thermal properties of the ACP components are listed
in Table 1.
Table 1: Thermal properties of ACP materials [40-42]
Material
Aluminium
Polyethylene (PE)
Polyisocyanurate
(PIR)
Air
gap
Density (kg/m3)
2700
1360
36
1.2
Specific heat (J/g K)
0.9
2
1.1
0.85
Thermal conductivity (W/m K)
237
0.43
0.048
-
Emissivity
0.7
0.9
1
0.8
A challenging aspect of modelling ventilated cladding is accounting for the insulating air layer
between the exterior facade panel and internal insulation. Earlier studies have modelled the air gap
explicitly by providing computational cells across the air gap and specifying two distinct surfaces to
model the external panel and internal insulation [39, 40]. This approach is however computationally
restrictive in the current study because the entire building is simulated and not just a portion of the
facade. Explicitly modelling flow in the narrow air gap between the exterior panel and insulation
requires a very fine grid which considerably increases the number of computational cells. Also, a
smaller timestep is needed to ensure solver stability which greatly increases computational cost.
Hence, an alternative approach is adopted where the air gap is modelled with a solid material with
thermal properties equivalent to those expected in an air gap. This approach was proposed by Drean et
al. [43] who derived an expression for equivalent thermal conductivity that accounts for the total heat
transfer across the air gap due to the combined effect of convection, conduction, and radiation. Figure
10 presents the thermal conductivity function for a 55 mm air gap adopted in this study.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0 5 10 15 20 25 30 35
Heat release rate (MW)
Time (min)
12
Figure 10: Equivalent thermal conductivity of 55 mm air gap as a function of the average temperature in the air gap
It should be noted that the treatment of the air gap as a solid layer, rather than explicit modelling of
airflow in the gap, will result in considerable limitations such as the exclusion of the stack (chimney)
effect from the analysis, and also the exclusion of the insulation combustion effect and the effect of
cavity barriers or other means of fire stop. In tall buildings, the stack effect can be significant because
of the large difference in wind pressure between upper and lower levels of tall buildings due to the
shape of the atmospheric boundary layer [44]. This effect can contribute to the acceleration of fire
spread through the ventilated gap and should therefore be considered in the analysis. However, for the
relatively short building considered in this study, the stack effect is likely minimal and so the
simplified approach of modelling the air gap as a solid layer is deemed acceptable.
A single surface is defined in FDS for modelling the external panel, air gap, and PIR insulation
simultaneously. The surface is assigned a layered material with thicknesses described in Figure 7 and
thermal properties described in Table 1. The use of a single surface to represent a multilayer cladding
panel was necessary to reduce the computational cost of the model given the available computing
resources. The limitations of this approach should however be taken into account when considering
the results of this study.
4.3 Calibrating combustion properties of facade material
There are two methods for modelling the combustion of materials in FDS. This can be done either
using pyrolysis models that simulate the chemical decomposition of solid material into combustible
gases or by using a simplified approach where combustion parameters are imposed explicitly. The
simplified approach is adopted in the current study where combustion properties of ACP are
calibrated with data from the DCLG test 1. The calibrated properties include the maximum heat
release rate per unit area (HRRPUA) and ignition temperature. A t-squared ramp function is also
calibrated to control temporal development of the heat release rate, which was needed to improve fit
to experimental data.
Calibration is performed by varying combustion properties until agreement is achieved between
temperature profiles from the numerical model and corresponding experimental data. A reasonable
agreement was achieved by selecting a maximum heat release rate per unit area of 0.51 MW/m2 and
an ignition temperature of 370 °C. Figure 11 compares temperature plots for thermocouples on the
main wall shown in Figure 8c. Aside from locations close to the wing wall, the plots in Figure 11
show reasonable agreement between the FDS model and experiment on the main wall of the facade
fire test.
0
2
4
6
8
10
12
0200 400 600 800 1000
Equivalent thermal conductivity
(W/m K)
Temperature (°C)
13
Figure 11: Thermocouple temperatures on the main wall from experiment and FDS model. Thermocouple error is within
0.75% of temperature reading.
There are some limitations of the calibrated material that should be noted. Firstly, the wing wall
experienced accelerated fire development compared to the experiment, and it was difficult to calibrate
temperature profiles on the main walls and wing wall simultaneously for the simplified material
model and coarse grid used. The localised ignition points and the convective heat transfer in the
proximity of the joint between the wing wall and main wall may have contributed to this issue.
Calibration based on the main wall was nonetheless deemed adequate for the current study
considering that the final model involves an external facade that is flush with the fire source. The
calibrated material would however not be suitable for cases with complex geometries where the
cladding is not flush with the fire source.
An important limitation to consider is the uncertainty associated with the properties of the fire source.
The combustion of the wooden crib is not consistent throughout the volume of the crib due to the
complex interaction between the combustion of timber and the geometry-dependent ventilation of
cold air entering the combustion chamber and hot air leaving the chamber. FDS does not capture well
the impact of the geometry when there is a thermal response from the secondary wing, and this
contributes to less accurate prediction of temperatures at M1-M2 and M6-M7. Further research is
needed to resolve this issue which would lead to improved material calibration of combustible
cladding in FDS.
Another limitation of note is the lack of validation data beyond the duration of the experiment.
Because the DCLG fire test was performed for testing fire performance, the test was halted once the
temperature exceeded performance limits specified by the BS 8414-1 standard. The lack of
experimental data beyond this point introduces uncertainty in the behaviour of the fully-developed
fire, such as the maximum temperature at which the material burns. This limitation can be addressed
0
200
400
600
800
1000
1200
1400
0200 400 600
Temperature (
°
C)
M6
0200 400 600
M7
0200 400 600
Level 2
M8
0200 400 600
M9
0200 400 600
M10
0
200
400
600
800
1000
1200
1400
0200 400 600
Temperature (
°
C)
M1
0200 400 600
M2
0200 400 600
Time (s)
Level 1
M3
FDS
Exp
0200 400 600
M4
0200 400 600
M5
14
in future studies by performing validation testing where the fire test is continued beyond the point
where the facade specimen fails the performance criteria specified in fire testing standards. For the
current study, however, the model is deemed adequate for replicating the initial propagation of facade
fires as demonstrated by the comparison in Figure 11. The validation of the numerical model could be
further improved by including an estimation of experimental uncertainty in the comparison, but this
could not be performed in the current study because details of experimental uncertainty are not
provided in the DCLG fire test report [38].
5 Parametric study on wind and facade fire interaction
Having validated the wind flow around the building and the combustion properties of the cladding
material, the final component of this study involves investigating the effect of wind speed and
direction on the propagation of facade fire for an isolated rectangular building described in section
2.1. The calibrated facade material from section 4 is applied to the exterior walls of the building on all
sides, while a non-combustible surface is specified at the roof.
To simulate a window-ejected flame, a simple burner surface with dimensions of 2.0 m by 1.0 m is
applied to the centre of the building surface at a height of 3.5 m from the ground. The fire source is
specified a constant heat release rate of 3.4 MW which is representative of an external flame venting
from a fully-developed post-flashover room fire [11]. From preliminary simulations with a non-
combustible facade, the heat release rate of the fire source was verified to not be affected by external
wind conditions. Hence, it should be noted that this study only addresses wind effects on the
combustion of the external facade, and does not account for wind effects on the development of the
fire source inside the room.
The atmospheric wind profile described in section 3 is applied to the inlet boundary of the model.
Simulation cases are generated by varying the reference wind speed  and the direction of the
wind. All other parameters of the model are kept constant, including the fire source, material
properties, and mesh configuration. The study involves 9 simulation cases as listed in Table 2.
15
Table 2: List of simulation cases in the parametric study
Number of cases
Wind direction
Wind speeds
1
No wind
2
Windward
2 m/s, 4 m/s
2
Leeward
2 m/s, 4 m/s
2
Side wind
2 m/s, 4 m/s
2
Diagonal wind
2 m/s, 4 m/s
Figure 12: Wind directions relative to fire source (top view)
Four wind directions are considered in the study. These are labelled according to the location of the
fire source relative to the approach wind direction (Figure 12). For the windward cases, the fire source
is located on the windward surface of the building. Similarly, the fire source is located on the leeward
surface for the leeward case. The side wind is only considered from one direction due to the symmetry
of the model. The diagonal wind approaches the windward surface of the building at a 45° angle. For
each of the wind directions, two highly recurring wind speeds are considered:  = 2 m/s and 4 m/s
at reference height  = 10 m. The 2m/s wind speed aligns closely with the maximum limit in BS
8414-1 standard for facade fire testing. An additional case with no wind is included to serve as a
benchmark for investigating the influences of various wind conditions.
Turbulence modelling and near-wall treatment are performed as per the approach described in section
2.3. The simulations are initially solved for 30 s with the fire source deactivated in order to initialise
the wind flow field around the building, and further 900 s (15 min) once the fire source was activated.
The models are solved on a single computing node with two 6-core Intel (R) Xeon (R) E5-2620 v3
processors at 2.4 GHz and a total RAM of 64 GB. The average CPU time for each case was 34.6 h.
Figure 13 shows simulation results of the instantaneous temperature contours of the facade fire under
the different wind conditions considered in this study. For quantitative analysis of fire propagation, a
post-processing script was developed using Python 3.8 and the Open-source Computer Vision
(OpenCV) library to calculate the time histories of fire spread from contours of wall temperature.
Two metrics were used to quantify fire spread:
Vertical spread (meters), calculated as the vertical distance between the fire source (3.5 m
above ground) and the highest point on the building surface with a wall temperature greater
than 400°C (approx. the ignition temperature of ACP).
Fire spread area (m2), calculated as the total surface area on the building with a wall
temperature greater than 400°C.
Windward
θ = 0°
Diagonal wind
θ = 45°
Side wind
θ = 90°
Leeward
θ = 180°
16
The time history of vertical fire spread is presented in the plots on the left in Figure 14, while the fire
spread area is presented in the plots on the right. The results in Figure 14 are time-averaged with an
averaging period of 30 s.
Figure 15 presents the heat release rate (HRR) from the combustion of the facade material under the
various wind conditions considered in this study. The plots are calculated by subtracting the heat
release rate of the fire source (3.4 MW) from the total heat release rate in the simulation. Because the
HRR of the fire source is assigned using a fixed curve in FDS, the HRR is not expected to be affected
by external wind conditions. Preliminary simulations were performed using non-combustible
cladding material to verify that the heat release rate of the fire source remained unchanged for all
wind conditions considered in this study.
The total energy released by the combustion of the facade is determined by calculating the area under
the plots in Figure 15. This provides a convenient metric for quantifying the overall magnitude of the
fire using a single value. The results are presented in Figure 16 in units of Gigajoules (GJ).
17
t = 1 min
t = 5 min
t = 10 min
t = 15 min
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
(j)
(k)
(l)
(m)
(n)
(o)
(p)
(q)
(r)
(s)
(t)
Figure 13: Effect of wind direction on facade fire. Plots show instantaneous temperature contours of external flames at
reference wind speed = 4 m/s.
No wind
Windward
Leeward
Side wind
Diagonal wind
200 400 600 800 1000
Temperature (°C)
18
Figure 14: Time histories of facade fire spread under various wind speeds and directions. Plots show vertical fire spread
(left) and overall fire spread area (right) on the building surface. Results are time-averaged with an averaging period of 30
s.
0
5
10
15
20
0 5 10 15
Vertical spread (m)
Time (min)
Windward
No wind
2 m/s
4 m/s
(a)
0
20
40
60
80
0 5 10 15
Fire spread area (
m2)
Time (min)
Windward
No wind
2 m/s
4 m/s
(b)
0
5
10
15
20
0 5 10 15
Vertical spread (m)
Time (min)
Leeward
No wind
2 m/s
4 m/s
(c)
0
20
40
60
80
0 5 10 15
Fire spread area (
m2)
Time (min)
Leeward
No wind
2 m/s
4 m/s
(d)
0
5
10
15
20
0 5 10 15
Vertical spread (m)
Time (min)
Side wind
No wind
2 m/s
4 m/s
(e)
0
20
40
60
80
0 5 10 15
Fire spread area (
m2)
Time (min)
Side wind
No wind
2 m/s
4 m/s
(f)
0
5
10
15
20
0 5 10 15
Vertical spread (m)
Time (min)
Diagonal wind
No wind
2 m/s
4 m/s
(g)
0
20
40
60
80
0 5 10 15
Fire spread area (
m2)
Time (min)
Diagonal wind
No wind
2 m/s
4 m/s
(h)
19
Figure 15: Heat release rate from the combustion of facade panels under various wind speeds and directions
Figure 16: Effect of wind speed and direction on total energy released from the combustion of facade material
0
5
10
15
20
25
0 5 10 15
Heat release rate (
MW)
Time (min)
Windward
No wind
2 m/s
4 m/s
(a)
0
5
10
15
20
25
0 5 10 15
Heat relase rate (MW)
Time (min)
Leeward
No wind
2 m/s
4 m/s
(b)
0
5
10
15
20
25
0 5 10 15
Heat release rate (MW)
Time (min)
Side wind
No wind
2 m/s
4 m/s
(c)
0
5
10
15
20
25
0 5 10 15
Heat release rate (MW)
Time (min)
Diagonal wind
No wind
2 m/s
4 m/s
(d)
7.42
4.50 4.50
3.02
7.49
4.79 4.61
3.56
2.23
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
No wind Windward Leeward Side wind Diagonal wind
Total energy released (GJ)
2 m/s
4 m/s
Wind speed
20
6 Results and discussion
The effect of wind conditions on facade fire propagation is investigated by comparing simulation
results in section 5 for cases of different wind speeds and directions. The plots of heat release rate
(Figure 15) and total energy released (Figure 16) provide an overview of wind effects on the facade
fire. Overall, these plots suggest that the presence of external wind causes a reduction in the intensity
of the facade fire, with larger wind speeds causing a greater reduction. This indicates that the venting
effect [13] is dominant for most wind conditions considered in this study. The largest fire occurred in
the case of no wind where the total energy released during the 15-min simulation was 7.49 GJ (Figure
16). This is considerably larger than cases with a wind speed of 4 m/s where the total energy released
ranged from 2.23 to 4.79 GJ depending on wind direction.
Although the plots of heat release rate and total energy released provide valuable insight into the
overall fire behaviour, a detailed analysis of facade fire propagation under wind effects can be
performed by examining results of temperature contours of the external flame (Figure 13) and plots of
vertical spread and fire spread area (Figure 14). These results are discussed in detail in the following
subsections.
6.1 Benchmark case with no wind
Of all cases considered in this study, the benchmark case with no wind resulted in the fastest vertical
spread of the fire (Figure 14a), although the fire spread was mostly limited to the centre of the
building surface (Figure 13a-d). It took approximately 9 min for the fire to reach the top of the
building (average vertical spread rate = 1.9 m/min), and by the end of the simulation (15 min),
approximately 32% of the frontal surface area (63.1 m2) was experiencing temperatures above 400°C.
The lack of external wind seems to provide highly favourable conditions for vertical buoyant forces to
drive the flame upwards along the building facade.
The vertical spread rate for the no wind case (1.9 m/min) is lower than that observed in the initial
stages of the Grenfell Tower fire (3.6 m/min) where the same cladding material was used [45]. This
discrepancy could be explained by differences in the heat release rate of the initial fire source. The
initial fire in the Grenfell incident was estimated to be as high as 4 MW [46] which is greater than the
3.4 MW fire used in this study. This may have contributed to the greater vertical spread rate in the
Grenfell fire. Despite this discrepancy, the vertical spread rate achieved in this study is still a
reasonable reflection of an ACP facade fire given that it falls within the uncertainty range given for
the quantitative analysis of the Grenfell fire by Guillaume et al. [45].
6.2 Effect of perpendicular wind
Results from the windward cases indicate that external wind acting perpendicular to the fire source
causes an initial delay in the vertical spread of the fire (Figure 13e-f). This delay is influenced by
wind speed. As shown in Figure 14a, a wind speed of 2 m/s resulted in a minor delay compared to the
case of no wind (10 min to reach the top of the building), while a wind speed of 4 m/s resulted in a
considerable delay and reduction in the total vertical spread. At t = 15 min, the total vertical spread
was 16.6 m for both the no wind case and for 2 m/s wind, while the 4 m/s case had a total vertical
spread of 12.2 m. This corresponds to a 26% reduction in total vertical propagation for the 4 m/s case.
The reduction in vertical fire propagation under perpendicular wind may be attributed to the formation
of a high-pressure stagnation region on the windward surface that could inhibit the initial spread of
21
the flame as seen in Figure 13e-f. Moreover, the presence of perpendicular wind may have generated
a downdraft on the windward surface that counteracts the vertical buoyant forces generated by the fire
plume, thereby reducing the speed of vertical fire propagation. This finding aligns with previous
research that reports a reduction in flame height under perpendicular wind [13, 14].
Despite a reduction in vertical propagation, the windward case of 4 m/s experienced only a minor
reduction in the total area of fire spread. Figure 14b shows that at 15 min, the total area on the facade
with temperatures greater than 400°C was 65.2 m2 for the cases of no wind and 2 m/s, while the case
of 4 m/s showed only a marginal reduction (< 5%) with a fire spread area of 58.4 m2. The windward
case of 4 m/s experienced greater lateral fire spread towards the side edges of the building compared
to the case with no wind (Figure 13h). The lateral spread is likely due to the behaviour of the
approaching wind that, once blocked by the presence of the building, is forced to travel around the
side edges and above the top edge at accelerated speeds carrying the flames with it.
Hence, perpendicular wind tends to inhibit the vertical spread of fire while increasing the lateral
spread along the windward surface; however, the extent of this influence is largely dictated by the
wind speed.
6.3 Effect of leeward wake
In the leeward cases, the wind is approaching from the opposite direction of the fire source, so the
facade fire occurs on the leeward surface that is encompassed by the wake of the building. Figure 14c
indicates that the leeward wake resulted in a considerable reduction in vertical fire propagation. For
wind speeds of 2 m/s and 4 m/s, the maximum vertical spread at 15 min was 9.4 m and 8.4 m,
respectively. This represents a 43-48% reduction in total vertical spread compared to the case of no
wind. This is likely due to large negative pressures in the wake region causing the fire to be drawn
away from the building surface, thereby resulting in a reduction of vertical fire spread.
Despite a considerable reduction in vertical spread, the leeward cases experienced a large amount of
lateral spread which resulted in an overall fire spread area that is comparable to the no wind case at 15
min (Figure 14d). The lateral spread of fire in the leeward wake is likely driven by the vortex
shedding phenomenon that causes the fire plume to fluctuate back and forth as shown in Figure 17.
The frequency of vortex shedding is known to be influenced by many factors including wind speed
and building shape and size. However, it is not clear how changes in vortex shedding behaviour will
affect facade fire propagation. This is an important issue to consider in taller buildings where stronger
vortex shedding behaviour is more likely to occur.
22
t = 4min 56s 4min 57s 4min 58s 4min 59s 5min 00s
Figure 17: Side-to-side fluctuation of facade fire in leeward wake during a 5 s period. Reference wind speed = 4 m/s.
Although the shedding of vortices is known to be influenced by wind speed, this seems to have a
minimal effect on fire spread behaviour for the building considered in this study. Figure 14c-d shows
similar plots for wind speeds of 2 m/s and 4 m/s. At t = 15 min, the higher wind speed of 4 m/s only
resulted in a marginal reduction of vertical spread (< 10%) and total spread area (< 5%) compared to
the 2 m/s case.
Overall, the results indicate that the leeward wake is likely to reduce the vertical propagation of
facade fire considerably, while only marginally reducing the overall fire spread area due to increased
lateral spread. Moreover, fire spread in the leeward wake is not significantly affected by wind speed.
6.4 Effect of side wind
The presence of side wind caused a reduction in the vertical spread and total fire spread area
compared to the case of no wind (Figure 14e-f). Interestingly, the presence of side wind caused the
flame to tilt in the opposite direction to the approaching wind (Figure 13m-p). This contradicts
findings in the literature that report tilting of the flame in the direction of the side wind in facade test
sections [15-19]. Nonetheless, this behaviour can be explained by the flow structures generated at the
sides of the rectangular building (Figure 2). Flow separation that occurs at the leading edges creates
large wake regions on the sides of the building with large negative pressures occurring immediate
downstream the leading corners. The large negative pressures produce a recirculating flow that draws
the flame towards the leading edge of the building (Figure 18a). This behaviour of reverse flow in the
wake region is a well-established phenomenon in wind-engineering literature [47-50]. Notably, the
fire is unable to spread beyond the leading edge (left edge in Figure 13m-p) due to the presence of a
strong streamwise wind shear layer at that location. The flames spread upwards once they reach the
leading edge.
The discrepancy between the current findings and previous studies can be explained by differences in
the geometries being considered and resulting differences in the size of the wake regions. Previous
studies on side wind were performed for thin facade sections which are expected to generate minimal
flow separation and smaller side wakes. In such cases, the fire is expected to tilt towards the
downwind direction under the influence of side wind rather than experiencing large negative pressures
that draw the flame toward the upwind direction as seen in the rectangular building of this study.
These results highlight the importance of including the full building geometry and resulting flow
fields when assessing wind effects on facade fires.
23
It should be noted that the results of fire spread under side wind may have been affected to some
extent by limitations of the FDS model in replicating the flow field near the side walls as described
previously. Despite the FDS model underpredicting the size of the wake region, the generated wake
was still sufficient to fully encompass the side walls of the rectangular building and to produce the
distinctive reverse-flow behaviour seen in Figure 18a. This suggests that the general behaviour of fire
spread under side wind was not affected by limitations of the FDS model, although further
improvement to flow field predictions in FDS is needed to improve confidence in the quantitative
results.
Figure 18: Top view of flow past building at z/H = 0.625 for cases of (a) side wind and (b) diagonal wind. Close-up view of
flow near the fire source is shown on right.
6.5 Effect of diagonal wind
The case of diagonal wind experiences very different behaviour to side wind. The lack of flow
separation for the diagonal case results in an undisturbed and streamlined parallel flow near the
building surface (Figure 18b). As a result, the fire plume experiences very strong tilting in the
direction of the wind (Figure 13q-t). This behaviour is consistent with studies on side wind in facade
test sections [15-19].
For the diagonal cases, the wind speed seems to have a considerable influence on the rate of vertical
and overall fire spread (Figure 14g-h). The case of 4 m/s experienced the lowest fire spread of all
cases considered in this study with a maximum vertical spread of 3.2 m and a total spread area of 13.5
m2 at 15 min. The case of 2 m/s experienced considerably larger values of fire spread compared to 4
Building
WIND
(a) Side wind (θ = 90°)
Fire source
Reverse flow
near wall
Building
(b) Diagonal wind (θ = 45°)
Fire source
Flow direction
near wall
Shear layer
Recirculation region
24
m/s with a vertical spread of 10.3 m and a total spread area of 30.1 m2 at 15 min. However, these
values are still considerably lower than the benchmark case of no wind. Hence, it may be concluded
that larger wind speeds for the diagonal cases reduce the risk of fire propagation. However, the strong
tilting of the flame towards the downwind direction as seen in Figure 13t could potentially lead to
further spread of fire to the side surfaces of the building.
6.6 Increased risk of fire propagation under variable wind conditions
The results presented so far indicate that the presence of external wind is likely to inhibit fire spread
during the initial stages of facade fire development. These cases were performed with constant wind
conditions that are present from the initial ignition of the facade and remain unchanged for the entire
duration of the simulation. Under such conditions, the presence of any combination of wind speed and
direction was seen to reduce the risk of fire propagation compared to the case of no wind. However,
real wind conditions can vary with time and this can potentially increase the risk of fire propagation.
To demonstrate this point, an additional case is performed with variable wind conditions that are
likely to increase the risk of facade fire propagation. These conditions were selected based on the
findings of this study with the aim of maximising fire spread. The simulation is performed for a total
duration of 30 min divided into three segments as follows:
t = 0 - 10 min: no wind to allow for initial fire development (Figure 19a)
t = 10 20 min: diagonal wind with = 45° and  = 4 m/s (Figure 19b)
t = 20 30 min: wind direction is changed to = -45° while wind speed is unchanged at 
= 4 m/s (Figure 19c)
The results of the variable wind case are presented in Figure 19, where it can be seen that the lack of
external wind from t = 0 - 10 min causes the fire to spread vertically (Figure 19a) at a rate equivalent
to that of the no wind case (Figure 19d). As a diagonal wind is introduced at t = 10 min, the vertical
plume that has already developed experiences lateral spread to the downwind direction. This causes
the fire to reach the trailing edge of the building and spread to the side wall (Figure 16b). At t = 20
min, the total fire spread area for the variable wind case has increased by 17 % compared to the case
of no wind (Figure 19d).
As wind conditions are varied again at t = 20 min, the fire undergoes further lateral spread on the
main wall but in the opposite direction, reaching the left edge of the building (Figure 19c). The wind
also causes further lateral spread of fire on the side wall that has now fully ignited. This increases the
total fire spread area on the building facade from 113.1 m2 for the case of no wind to 170.1 m2 for the
case of variable wind, which represents a 50% increase in total fire spread area at t = 30 min.
An animated comparison of fire spread for variable wind and no wind cases is presented in Appendix
A.
The variable wind conditions in this study have been specifically selected to maximise fire spread, but
these conditions are not unrealistic. The results clearly demonstrate the potential of a synergistic effect
and a resulting increased risk of facade fire spread under variable wind conditions. These effects
should be considered when assessing the fire safety of buildings.
25
Figure 19: Fire spread on building facade under variable wind conditions at (a) t = 10 min, (b) t = 20 min, and (c) t = 30
min. Plot (d) compares the fire spread area on the building surface with temperatures greater than 400°C.
7 Conclusions
This study investigated the influence of wind speed and direction on the propagation of facade fires in
an isolated rectangular building. A computational fluid dynamics (CFD) model was constructed in
Fire Dynamics Simulator (FDS) and validated with experimental data from wind tunnel and facade
fire tests. A series of parametric studies with four wind directions and two wind speeds were
performed with the validated model. The main conclusions from this study are summarized as
follows:
Wind flow patterns generated by the building geometry were found to dictate the behaviour of
facade fire propagation. The full building geometry and resulting wind flow patterns need to
be considered when assessing wind effects on facade fires.
Both wind speed and direction have a considerable influence on facade fire propagation. The
case of no wind experienced the quickest vertical propagation but with minimal lateral spread.
Perpendicular wind inhibits the vertical spread of fire while increasing the lateral spread along
the windward surface. The leeward wake region generates highly unstable conditions that
increase the lateral fire spread while reducing vertical propagation.
For the rectangular building considered in this study, side wind caused the fire to spread
laterally in a direction opposite to the wind flow, while diagonal wind caused a very strong
0
50
100
150
200
0 5 10 15 20 25 30
Fire spread area (m2)
Time (min)
No wind
Variable wind
(+50%)
170.1 m2
113.1 m2
(d)
Diagonal wind (θ= 45°)No wind Diagonal wind (θ= -45°)
(+17%)
96.2 m2
82.0 m2
θ
Lateral
spread
θ
WIND
(b)
Vertical
spread
(a)
Spread
to side
wall
WIND
(c)
26
lateral spread in the downwind direction. These behaviours are explained by differences in
near-wall flow structures generated under the influences of side and diagonal winds.
There is a synergistic effect on the spread of facade fires with variable wind conditions. Once
the fire has fully developed; the presence of wind can considerably increase the risk of fire
spread along multiple facade surfaces on the building. Variable wind conditions need to be
considered when assessing the fire safety of buildings.
The study reveals interesting aspects of wind and fire interaction that could contribute to reduced fire
safety performance in building facades. Most notably, the increase in lateral fire spread due to side
wind and diagonal wind can potentially reduce the effectiveness of fire protection features such as
overhangs and balconies as lateral spread can cause the fire to penetrate the space between these
protective features. Further research is needed to investigate the performance of these safety features
under temporally varying wind conditions and for complex building geometries.
Due to the large number of parameters included in this study and limitations in available computing
resources, the study only considered a limited range of wind speeds and directions. Although these
were sufficient to provide an insight into the general behaviour of wind and fire interaction under
highly recurrent wind speeds, further research is needed to consider intermediate values of wind
speeds and direction not considered in this study, as well as considering the effect of extreme wind
events.
A notable limitation of the study is the use of a simplified approach for modelling the ACP material,
where a single surface was used to model the entire cladding system. This approach has considerable
limitations as it does not account for the combustion of individual cladding components and excludes
the modelling of fire spread in the air gap between the ACP panel and the insulation layer. The
simplified material was calibrated to replicate the general fire spread behaviour of ACP which was
deemed adequate for the purpose of investigating wind and fire interaction at the full building scale.
This limitation should however be addressed in a future study with more intricate models of ACP and
other cladding systems.
The study assumed a fully-developed, post-flashover flame as the fire source for the initial ignition of
the facade. The study does not account for the effect of external wind on the initial development of the
localised fire inside the room, and further research is needed to include this effect in the model. The
study also does not account for the development of secondary fires caused by the re-entry of the
external flame through openings. Modelling these behaviours requires accounting for the interactions
between external and internal parameters which considerably increases the complexity of the model.
The influence of secondary fires on facade fire spread under wind effects is an important area for
future research.
Further research is also needed to investigate other parameters not considered in this study including
the effect of inflow turbulence and turbulence generated by nearby structures, as well as the influence
of the stack effect within the ventilation gap between the exterior facade panel and internal insulation.
These will improve confidence in the fire safety performance of building facades under various wind
conditions.
27
Funding sources
This work was supported by the Australian Research Council [grant numbers: DE190100217 - Facade
fire failures in buildings: a robust nanocomposite solution, IH200100010 - Transformation of
Reclaimed Waste Resources to Engineered Materials and Solutions for a Circular Economy TREMS,
and CRCPEIGHT00084 - Upcycling solutions for hazardous cladding and co-mingled waste].
Appendix A. Supplementary data
[INSERT LINK TO VIDEO HERE]
Video caption: Side-by-side comparison of fire spread under no wind and variable wind conditions.
References
1. Nguyen, K.T., et al., Performance of modern building façades in fire: a comprehensive review.
Electronic Journal of Structural Engineering, 2016. 16(1): p. 69-86.
2. Bonner, M. and G. Rein, Flammability and multi-objective performance of building façades: towards
optimum design. International Journal of High-Rise Buildings, 2018. 7(4): p. 363-374.
https://doi.org/10.21022/IJHRB.2018.7.4.363.
3. Spinardi, G. and A. Law, Beyond the stable door: Hackitt and the future of fire safety regulation in the
UK. Fire Safety Journal, 2019. 109: p. 102856. https://doi.org/10.1016/j.firesaf.2019.102856.
4. Bisby, L., Grenfell Tower Inquiry report. Phase 1: Expert report. 2018.
5. Genco, G., Lacrosse Building Fire: 673 La Trobe Street, Docklands on 25 November 2014. 2015, City
of Melbourne: Melbourne, Australia.
6. Węgrzyński, W. and T. Lipecki, Wind and Fire Coupled Modelling—Part I: Literature Review. Fire
Technology, 2018. 54(5): p. 1405-1442. https://doi.org/10.1007/s10694-018-0748-5.
7. Pitts, W.M., Wind effects on fires. Progress in Energy and Combustion Science, 1991. 17(2): p. 83-134.
https://doi.org/10.1016/0360-1285(91)90017-H.
8. Canton, L.G., San Francisco 1906 and 2006: An Emergency Management Perspective. Earthquake
Spectra, 2006. 22(2_suppl): p. 159-182. https://doi.org/10.1193/1.2181467.
9. Suzuki, S. and S.L. Manzello, Characteristics of Firebrands Collected from Actual Urban Fires. Fire
Technology, 2018. 54(6): p. 1533-1546. https://doi.org/10.1007/s10694-018-0751-x.
10. International Organization for Standardization, ISO 13785-2:2002 : Reaction-to-fire tests for facades -
Part 2: Large-scale test. 2002.
11. British Standards Institution, BS 8414-1:2015 + A1:2017 - Fire performance of external cladding
systems. Test methods for non-loadbearing external cladding systems applied to masonry face of a
building. 2017.
12. Standards Australia, AS 5113:2016 - Fire propagation testing and classification of external walls of
buildings. 2016, SAI Global Limited: Sydney.
13. Hu, L., et al., Facade flame height ejected from an opening of fire compartment under external wind.
Fire Safety Journal, 2017. 92: p. 151-158. https://doi.org/10.1016/j.firesaf.2017.06.008.
14. Fang, X., et al., Facade flame height ejected from opening of a compartment under the coupling effect
of side walls and ambient wind. Fire Safety Journal, 2020. 112: p. 102966.
https://doi.org/10.1016/j.firesaf.2020.102966.
15. Bai, Z.P., Y.F. Li, and Y.H. Zhao, Study on characteristics of fire plume in building facade window
under lateral blow. PLoS ONE, 2019. 14(11): p. e0225120.
https://doi.org/10.1371/journal.pone.0225120.
28
16. Hu, L., et al., Facade flame height and horizontal extending distance from opening of compartment fire
with external sideward wind. Proceedings of the Combustion Institute, 2019. 37(3): p. 3859-3867.
https://doi.org/10.1016/j.proci.2018.06.201.
17. Li, J., et al. Side wind effect on the flow behavior of the window plume. in Asia-Oceania Symposium on
Fire Science and Technology. 2018. Springer. https://doi.org/10.1007/978-981-32-9139-3_9.
18. Li, J.M., et al. Computational study of wind effect on window flame spread across the exterior wall of
high-rise building. in Applied Mechanics and Materials. 2013. Trans Tech Publications.
https://doi.org/10.4028/www.scientific.net/AMM.438-439.1898.
19. Sugawa, O., D. Momita, and W. Takahashi, Flow behavior of ejected fire flamelplume from an opening
effected by external side wind. Fire Safety Science, 1997. 5: p. 249-260.
https://doi.org/10.3801/IAFSS.FSS.5-249.
20. Wang, T.-h., H. Zhao, and Y. Zhou, Analysis on the influence of fire overhangs on the window to
exterior wall vertical fire plume spreading under external wind. Procedia Engineering, 2016. 135: p.
384-392. https://doi.org/10.1016/j.proeng.2016.01.146.
21. Ren, F., L. Hu, and X. Sun, Experimental investigation on lateral temperature profile of window-
ejected facade fire plume with ambient wind. Fire Technology, 2019. 55(3): p. 903-913.
https://doi.org/10.1007/s10694-018-0809-9.
22. Ren, F., et al., An experimental study on vertical temperature profile of facade fire plume ejected from
compartment with an opening subjected to external wind normal to facade. International Journal of
Thermal Sciences, 2018. 130: p. 94-99. https://doi.org/10.1016/j.ijthermalsci.2018.04.008.
23. Gao, W., et al., Fire spill plume from a compartment with dual symmetric openings under cross wind.
Combustion and Flame, 2016. 167: p. 409-421. https://doi.org/10.1016/j.combustflame.2016.01.011.
24. Cao, L. and Y. Guo, Large eddy simulation of external fire spread through openings. International
Journal on Engineering Performance-based Fire Codes, 2003. 5(4): p. 176-180.
25. Li, M., et al., Wind effects on transition and extension characteristics of the external flame projecting
from a compartment with opposing openings. Fire Safety Journal, 2020: p. 103102.
https://doi.org/10.1016/j.firesaf.2020.103102.
26. Li, M., et al., Wind effects on flame projection probability from a compartment with opposing
openings. Fire Safety Journal, 2017. 91: p. 414-421. https://doi.org/10.1016/j.firesaf.2017.04.037.
27. Zheng, L., et al. Influence of Wind Direction on Fire Spread on the Exposed XPS insulation Wall. in
IOP Conference Series: Earth and Environmental Science. 2020. IOP Publishing.
https://doi.org/10.1088/1755-1315/526/1/012013.
28. Abu-Zidan, Y.F., Verification and validation framework for computational fluid dynamics simulation
of wind loads on tall buildings, in Department of Infrastructure Engineering. 2019, The University of
Melbourne. https://doi.org/11343/228846.
29. Węgrzyński, W., T. Lipecki, and G. Krajewski, Wind and Fire Coupled Modelling—Part II: Good
Practice Guidelines. Fire Technology, 2018. 54(5): p. 1443-1485. https://doi.org/10.1007/s10694-018-
0749-4.
30. McGrattan, K., et al., Fire dynamics simulator user’s guide. NIST special publication, 2019. 1019(6).
31. Abu-Zidan, Y., P. Mendis, and T. Gunawardena, Impact of atmospheric boundary layer inhomogeneity
in CFD simulations of tall buildings. Heliyon, 2020. 6(7): p. e04274.
https://doi.org/10.1016/j.heliyon.2020.e04274.
32. Abu-Zidan, Y., P. Mendis, and T. Gunawardena, Optimising the computational domain size in CFD
simulations of tall buildings. Heliyon, 2021. 7(4): p. e06723.
https://doi.org/10.1016/j.heliyon.2021.e06723.
33. Smagorinsky, J., General circulation experiments with the primitive equations: I. The basic
experiment. Monthly Weather Review, 1963. 91(3): p. 99-164. https://doi.org/10.1175/1520-
0493(1963)091<0099:GCEWTP>2.3.CO;2.
34. Okaze, T., et al., Benchmark test of flow field around a 1:1:2 shaped building model using LES:
Influences of various calculation conditions on simulation result. AIJ Journal of Technology and
Design, 2020. 26(62): p. 179-184. https://doi.org/10.3130/aijt.26.179.
29
35. Pope, S.B., Turbulent Flows. 2000, Cambridge: Cambridge University Press.
36. Nicoud, F. and F. Ducros, Subgrid-scale stress modelling based on the square of the velocity gradient
tensor. Flow, Turbulence and Combustion, 1999. 62(3): p. 183-200.
https://doi.org/10.1023/A:1009995426001.
37. Meng, Y. and K. Hibi, Turbulent measurements of the flow field around a high-rise building. Wind
Engineers, JAWE, 1998(76): p. 55-64. https://doi.org/10.5359/jawe.1998.76_55.
38. BRE Global Ltd, Report No B 137611-1037 (DCLG test 1). 2017.
39. Dréan, V., et al., Numerical simulation of the fire behaviour of facade equipped with aluminium
composite material-based claddings - Model validation at large scale. Fire and Materials, 2019. 43(8):
p. 981-1002. https://doi.org/10.1002/fam.2759.
40. Dréan, V., et al., Numerical simulation of the fire behaviour of façade equipped with aluminium
composite material-based claddings - Model validation at intermediate scale. Fire and Materials, 2019.
43(7): p. 839-856. https://doi.org/10.1002/fam.2745.
41. Periodic Table. Aluminium Periodic Table. 2020; Available from: https://www.periodic-
table.org/Aluminium-periodic-table/.
42. Xometry, Datasheet: HDPE (High Density Polyethylene). 2021.
43. Drean, V., R. Schillinger, and G. Auguin, Assessment of an insulating air layer model of façade
external system: contribution to fire simulation of facade performance fire test. Journal of Physics:
Conference Series, 2018. 1107: p. 042004. https://doi.org/10.1088/1742-6596/1107/4/042004.
44. Mijorski, S. and S. Cammelli, Stack effect in high-rise buildings: a review. International Journal of
High-Rise Buildings, 2016. 5(4): p. 327-338. https://doi.org/10.21022/IJHRB.2016.5.4.327.
45. Guillaume, E., et al., Reconstruction of Grenfell Tower fire. Part 1: Lessons from observations and
determination of work hypotheses. Fire and Materials, 2020. 44(1): p. 3-14.
https://doi.org/10.1002/fam.2766.
46. Guillaume, E., et al., Reconstruction of Grenfell Tower fire. Part 2: A numerical investigation of the
fire propagation and behaviour from the initial apartment to the façade. Fire and Materials, 2020.
44(1): p. 15-34. https://doi.org/10.1002/fam.2765.
47. Castro, I.P. and A.G. Robins, The flow around a surface-mounted cube in uniform and turbulent
streams. Journal of Fluid Mechanics, 1977. 79(2): p. 307-335. 10.1017/S0022112077000172.
48. Martinuzzi, R. and C. Tropea, The Flow Around Surface-Mounted, Prismatic Obstacles Placed in a
Fully Developed Channel Flow (Data Bank Contribution). Journal of Fluids Engineering, 1993.
115(1): p. 85-92. 10.1115/1.2910118.
49. Ogawa, Y., S. Oikawa, and K. Uehara, Field and wind tunnel study of the flow and diffusion around a
model cubeII. Nearfield and cube surface flow and concentration patterns. Atmospheric
Environment (1967), 1983. 17(6): p. 1161-1171. https://doi.org/10.1016/0004-6981(83)90339-6.
50. Paterson, D.A. and C.J. Apelt, Simulation of flow past a cube in a turbulent boundary layer. Journal of
Wind Engineering and Industrial Aerodynamics, 1990. 35: p. 149-176. https://doi.org/10.1016/0167-
6105(90)90214-W.
... Numerical modelling of façade fire tests has emerged as a cost-efficient alternative that has allowed researchers to perform sensitivity analysis on a wide range of empirical factors [4]. While there have been several attempts to develop numerical models for façade fire tests [5][6][7][8][9][10][11], the accuracy of such models is often limited by uncertainty in the fire source properties used in the simulations. Despite the availability of experimental data, calibrating the fire source in numerical models of façade fire tests has proved challenging. ...
Article
Full-text available
Calibration is an important step in the development of predictive numerical models that involves adjusting input parameters not easily measured in experiments to improve the predictive accuracy of the numerical model compared to the real system. For complex models of façade fires, model calibration can be difficult due to the large number of input parameters that need to be calibrated simultaneously. This paper proposes a machine-learning-based surrogate modelling technique to help with calibrating the fire source in simulations of façade fire tests. Two case studies are presented to assess the feasibility of the proposed method: a simple fire source with a single burner surface based on the JIS A 1310:2015 test, and a complex fire source of a wooden crib based on the BS 8414-2:2015 test. The properties of the fire sources are calibrated based on thermocouple temperatures measured near the cladding surface. In both case studies, the ML-based surrogate model successfully calibrated the fire source properties, resulting in a high level of agreement between the calibrated model and results for experiments (average error = 2.8% and 14.3% for case studies 1 and 2). The proposed method can be applied for various optimisation problems in fire engineering research and design.
Conference Paper
Full-text available
Design of tall buildings is a multidisciplinary process and requires a wide range of measures to take to withstand in various emergencies. Fire is one of them and can cause of chaotic situations considering hundreds of users in multistorey. The measures are divided into two categories, active and passive fire protection. Architectural design and detailing are part of the passive systems. Along with the fundamental solutions in fire protection, tall buildings require more specific and often design-specific solutions. Detection and extinguishing systems are part of the active fire protection systems. The failure of the World Trade Centre pointed the vulnerability of steel structures; the fire at the Grenfell Tower pointed the significance of the enveloping materials for facades, and the fire at the One Meridian Tower pointed the significance of the active systems for every floor. The paper focuses on the fire case studies of tall buildings and follows with an assessment of the approaches in architectural design and detailing applicable from the structural design to the finishing interior detailing.
Article
Full-text available
The construction of tall and slender buildings has seen recent growth in many cities around the world. Tall buildings are susceptible to dynamic excitation under wind effects which typically govern the structural design for strength, stability, and serviceability. This paper presents the state of the art in the analysis and design of tall buildings against wind effects. Structural design criteria are discussed in detail, with serviceability criteria relating to occupant comfort noted as being of particular importance. The latest in wind analysis tools and techniques is also presented. Wind tunnel testing remains the gold standard for determining wind loads on tall buildings, while the emerging use of computational fluid dynamics (CFD) is noted as being particularly useful for concept design stages. The paper aims to provide a valuable reference for engineers, architects, and designers involved in wind analysis and design of tall buildings.
Article
Full-text available
Recently, there has been a growing interest in utilizing computational fluid dynamics (CFD) for wind resistant design of tall buildings. A key factor that influences the accuracy and computational expense of CFD simulations is the size of the computational domain. In this paper, the effect of the computational domain on CFD predictions of wind loads on tall buildings is investigated with a series of sensitivity studies. Four distinct sources of domain error are identified which include wind-blocking effects caused by short upstream length, flow recirculation due to insufficient downstream length, global venturi effects due to large blockage ratios, and local venturi effects caused by insufficient clearance between the building and top and lateral domain boundaries. Domains based on computational wind engineering guidelines are found to be overly conservative when applied to tall buildings, resulting in uneconomic grids with a large cell count. A framework for optimizing the computational domain is proposed which is based on monitoring sensitivity of key output metrics to variations in domain dimensions. The findings of this paper help inform modellers of potential issues when optimizing the computational domain size for tall building simulations.
Article
Full-text available
Recently, there has been a growing interest in utilizing computational fluid dynamics (CFD) for wind analysis of tall buildings. A key factor that influences the accuracy of CFD simulations in urban environments is the homogeneity of the atmospheric boundary layer (ABL). This paper aims to investigate solution inaccuracies in CFD simulations of tall buildings that are due to ABL inhomogeneity. The investigation involves two steps. In the first step, homogenous and inhomogeneous ABL conditions are generated in an empty computational domain by employing two different modelling approaches. In the second step, the homogenous and inhomogeneous conditions are each applied to an isolated tall building, and simulation results are compared to investigate impact of ABL inhomogeneity on wind load predictions. The study finds that ABL inhomogeneity can be a significant source of error and may compromise reliability of wind load predictions. The largest magnitude of inhomogeneity error occurred for pressure predictions on the windward building surface. Shortening the upstream domain length reduced inhomogeneity errors but increased errors due to wind-blocking effects. The study proposes a practical approach for detecting ABL inhomogeneity that is based on monitoring sensitivity of key output metrics to variations in upstream domain length.
Article
Full-text available
Medium-sized wall fire experiments were carried out using extruded polystyrene (XPS) panels to build an insulation layer to simulate the flame spread over the exterior thermal insulation system during the exposed stage. During the experiments, the influences of the wind direction on the lateral and vertical flame spread speed without fire barrier designs under the wind speed of 2.5m/s were studied quantitatively and qualitatively. According to the experimental results, it is found out the averaged peak lateral flame spread rate increases and the averaged peak vertical flame spread rate decreases when the angle between the wind direction and the wall increases from 0° to 60°. It can be seen from the analysis that the largest average lateral flame spread rate and the largest average vertical flame spread rate also show regular changes at same wind direction. This study results can provide valuable references for optimizing the fire safety design of exterior insulation system.
Thesis
Full-text available
Wind loading is significant in tall buildings and typically governs their structural design. Controlling wind effects in tall buildings often requires major design modification to the building geometry and dynamic properties. Hence, there is a need for an accurate and computationally efficient wind analysis tool that would facilitate assessment of aerodynamic building performance during different stages of design. Computational fluid dynamics (CFD) is a numerical wind analysis tool that overcomes limitations of traditional wind tunnel testing including higher cost and time requirements. However, CFD is scarcely utilised in the building industry as a quantitative wind analysis tool due to a lack of confidence in the reliability of its predictions. Therefore, this thesis aims to develop a methodology for evaluating the performance of CFD models in predicting wind loads on tall buildings, so that confidence in CFD as a wind design tool can be established. To achieve this aim, the thesis first presents a review and synthesis of the theoretical knowledge required to adopt the generic verification and validation (V&V) framework from the field of Computational Science and Engineering (CS&E) to the specific context of CFD for tall building design. To the author's knowledge, no comprehensive work was dedicated to holistically implementing the V&V framework to CFD simulations of wind loading on tall buildings. A critical distinction between verification and validation activities is noted, where verification assesses whether the numerical model accurately reflects the modeller's conceptualisation of the problem, while validation evaluates the performance of the numerical model in predicting physical reality. Verification and validation activities relating to CFD simulations of tall buildings are described. Next, sensitivity studies are performed to investigate the nature of various numerical errors in CFD simulations of tall buildings, including ABL inhomogeneity, domain-size, and spatial discretisation errors. Notably, the study finds that errors due to ABL inhomogeneity and short upstream domain lengths can be especially significant and should therefore be controlled. Moreover, peak negative pressures that occurred in the first computational elements directly following flow separation edges are found to be highly mesh dependent and therefore unreliable. Next, the thesis implements the proposed methodology to Large-eddy simulations (LES) of a simplified tall building case. A validation experiment is conducted in a boundary layer wind tunnel and the results are used to validate LES predictions. The study reports on the accuracy of LES predictions of surface pressures, base reactions, and floor acceleration. The study discusses merits and limitations of the proposed methodology, as well as practical challenges to its implementation. These challenges include encouraging tall building industry projects to invest in validation-grade experiments and working with researchers to develop efficient strategies for the sharing of validation results. URI: http://hdl.handle.net/11343/228846
Article
Full-text available
The overflow of the flame plume from the window is the main cause of the vertical spread of the fire on the facade of the building. This paper considers the geometry of the window by taking measures to prevent the flame from propagating along the vertical wall. In this paper, a residential building is taken as an example to evaluate the flame plume characteristics through experimental tests and numerical simulations. The objective of the present study is to study the flame plume characteristics under the air blow on the outer window side of the building. The theoretical equations of the flame tilt angle, non-dimensional temperature and non-dimensional velocity are derived. A series of experimental tests were carried out in a reduced-scale building model corresponding to the changes of lateral blow ventilation velocity. Reduced scale numerical simulations were conducted to verify the experiments. Results showed that the flame tilt angle increases with ventilation velocity increases. Meanwhile, the experimental results were compared with the reduced-scale tests and numerical simulations. These showed a good agreement between experimental results and numerical simulations. All these findings provide theoretical basis for building fire prevention outside window.
Article
Full-text available
The Grenfell Tower fire occurred on 14 June 2017, killing 72 people. The pattern and speed of vertical and horizontal fire spread characterize this catastrophic event. Plentiful video and photographic data of the fire spread available has been carefully verified and concatenated into a database. The verified data have been superimposed on a projection of the Grenfell Tower in order to track the development of the fire. The surface that is unburnt, burning, or extinguished, as well as the presence of internal fire at any given location, is thus recorded for the duration of the fire. An analysis of the results showed that the initial vertical propagation can be divided into three phases. After the façade ignited at the fourth floor, vertical propagation over time is linear, with a vertical fire spread rate of around 3.5 m/min until the fire reached the sixth floor. Then fire propagation decelerated. Finally, fire spread accelerated with a power four dependence. The maximum vertical fire spread rate was around 8 m/min as the fire reached the crown at the top of the building. Horizontal spread proved to be greatest at the level of the crown (0.293 ± 0.005 m/min). There is a linear relationship between speed of horizontal fire spread and height. These correlations and observations yield important conclusions, and eight different hypotheses capable of explaining the global behaviour of the fire are suggested.
Article
Full-text available
The dramatic event of the Grenfell Tower in 2017 reminds the importance of addressing fire issues as a whole and clearly highlighted one of the major roles played by the façade as fire propagation vector. To understand and analyse this disaster, numerical simulation allows particular phenomena to be evaluated more easily. The numerical model addressed for the fire behaviour of the façade system was developed using a multiscale approach and validated at different scales. In this paper, the fire behaviour of the façade and of its window frames is addressed. A computational fluid dynamics (CFD) model is used to investigate the fire spread from the initial apartment to the overall façade with different scenarios for the fire source and ventilation. Fire propagation through windows to the façade and to upper apartments is addressed. General curves representing the re‐entry of flames into upper apartment are extracted from simulations. The numerical results are validated by comparison with observations from videos and pictures of the real fire. Numerical results show that whatever the initial fire location and ventilation conditions, even if the fire source is of hundreds kilowatts, it is enough to ignite the adjacent element early and to the appearance of external flames shortly after.
Article
The assisting wind force effect on the transition and extension characteristics of the flame projecting outside the window of a compartment was experimentally investigated. The transition criteria, length, and maximum temperature profile of the external flame were studied. As the wind velocity increases, three states: namely a spill-like plume, horizontally ejected flame and no flame outside the window were identified in the experiments. The Froude number can be used to determine the three states of the external flames, with the critical values separating these stages being 2.3 and 6.4. Two length scales, l1 and l2 are proposed to modify the developed theories without a wind effect. The length scales are used to analyze the length and maximum temperature profile of the external flame. The dimensionless flame lengths are proportional to the 1.5 and 1.2 power of the dimensionless convective heat flow rate for the spill-like plume and horizontally ejected flame respectively. Correlations for the maximum temperature rises of the external flame are established between the dimensionless convective heat flow rate and the length (height) of the plume by taking into account the thermodynamic and dynamic effects of the external wind.
Article
Appropriate large-eddy simulation (LES) guidelines for pedestrian wind environments are being established by the working group of the Architectural Institute of Japan. We conducted LESs for a flow field around an isolated building to clarify the influence of computational conditions on turbulent statistics. We performed a cross-comparison study by examining both experimental and LES results for various computational conditions such as grid arrangements, subgrid scale models, spatial derivation schemes for convection term, and convergence criterion for a coupling algorithm of flow and pressure fields. Results can be used to provide LES guidelines for predicting the pedestrian wind environment.
Article
This paper presents an experimental investigation and analysis on the flame height ejected from an opening of fire compartment over the facade under the coupling effect of two side walls and ambient wind normal to the facade. Experiments were carried out in a reduced-scale model consisting of a cubic fire compartment with an opening, a facade wall and two side walls. Side walls were axi-symmetrically placed at the two sides of opening along the facade, while the opening and the facade were set normal to ambient wind generated by a wind tunnel. Facade flame heights were measured for various heat release rates, side wall distances and ambient wind speeds. Results showed that facade flame height increased as side wall distance decreased, meanwhile decreased as wind speed increased. Theoretical analysis was performed to interpret the behavior based on the air entrainment of ejected flame under their coupling effect. A global non-dimensional factor K as a function of side wall distance, ambient wind speed, the two characteristic length scales of opening as well as the dimensionless excess heat release rate was proposed to predict the facade flame height of various conditions. The experimental results were shown to be well represented.