Content uploaded by Bjarne Paulsen Husted
Author content
All content in this area was uploaded by Bjarne Paulsen Husted on Sep 23, 2016
Content may be subject to copyright.
PROTECTION AGAINST EXTERNAL FIRE SPREAD -
HORIZONTAL PROJECTIONS OR SPANDRELS?
1 Markus Nilsson, 1 Axel Mossberg, 2 Bjarne Husted & 3 Johan Anderson
1 Brandskyddslaget AB, Sweden
2 Division of Fire Safety Engineering, Lund University, Sweden
3 SP Technical Research Institute of Sweden, Fire Technology
ABSTRACT
The impact of horizontal projections on external fire spread was investigated using the
numerical tool Fire Dynamics Simulator (FDS). The numerical study was divided into a validation
study and a comparative analysis. The validation study was performed to evaluate FDS as a
calculation tool for modelling external fire spread and was conducted using experimental results from
a large-scale fire test done on a SP FIRE 105 test rig in Borås, Sweden. A previous validation study
on FDS version 5.5.3 showed promising results, which was further developed in order to validate FDS
6.2.0. It was concluded that FDS 6.2.0 could reproduce the experimental results. In the comparative
analysis the impact on the external fire from a smaller apartment was studied in FDS with different
configurations of horizontal projections and spandrels in the building exterior. The analysis showed
that at least a 60 cm deep horizontal projection results in lesser consequences at the facade compared
with scenarios built up by different spandrel heights.
INTRODUCTION
External fire spread between fire compartments is a risk that has been observed in both
experimental and numerical research 1. A common way that different countries have dealt with this
problem is by setting prescriptive spandrel (i.e. a vertical safety distance) and/or horizontal projection
configurations in their respective building regulation 2. E.g. in Sweden, a prescriptive spandrel
configuration of at least 1.2 meters between windows in the facade is applied 3. However, the spandrel
and/or horizontal projection configurations differ significantly between countries and a conclusion
drawn from a review of different building codes is that the level of protection differs and that more
research on the subject of external fire spread between fire compartments is needed 2.
In the case with the Swedish building regulation (BBR) no horizontal projection alternative is given.
However, the regulation is performance-based and the spandrel configuration is given as general
recommendation. This means that the designer can deviate from this recommendation if she/he can
show that (1) the functional requirement is reached, i.e. fire will not spread between different
compartments along the building facade, or (2) that an equal level of safety as stated in the general
recommendation is given by the solution. Since no quantitative target level is given by the functional
requirement it is implied that the accepted safety level is to be derived from the spandrel height of 1.2
meters mentioned above.
The fact that no horizontal projection alternative is given in the regulation poses a problem when
designing buildings with French, and regular, balconies. Because of the lack of guidance it is not clear
how a balcony needs to be designed to fulfil the requirement, without the given spandrel height.
Because of these “design uncertainties”, a study was initiated by Brandskyddslaget AB in cooperation
with Lund University. The results of this study, and some further elaborated work on the subject, are
presented in the paper below.
METHODOLOGY
The background to this work lies mainly in the results of an MSc thesis 2 conducted at Lund
University, investigating the impact of horizontal projections on external fire spread using the
numerical tool FDS 6.2.0 4. The numerical study was built up by a literature review followed by a
validation study and a comparative analysis. The literature review was about gaining understanding of
previous research and how different countries manage the protection against external fire spread
between openings in the building exterior. This enabled one to highlight important aspects from the
theory and results from previous studies of spandrels and horizontal projections as well as the
protection requirements in different building regulations.
In order to obtain valid results in the comparative analysis, it was first necessary to evaluate FDS as a
calculation tool for modelling external fire spread. This was done by performing a validation study of
FDS against a large-scale fire test with a geometry as closely linked as possible to the setup used in
the comparative analysis. The validation study was performed using experimental data from a large-
scale fire test on a SP FIRE 105 test rig in Borås 5, Sweden. A previous validation study on FDS 5.5.3
has been carried out on this particular test 6. This was further developed in the thesis 2 in order to
validate FDS 6.2.0. A description of the setup and key results from this part as well as key results
from the previous validation study on this test rig are presented in this paper.
The conclusions drawn from the validation study were taken into consideration when performing the
simulations in the comparative analysis. In the comparative analysis an apartment was built in FDS
with two opening configurations: a door and a window. Two types of fire sizes were modelled namely
3.1 MW and 4 MW. The impact of the external flames on the building facade where studied at
different heights along the exterior using mainly adiabatic surface temperature output data,
. This
basic case is referred to as the Spandrel-case with accompanying descriptive suffix depending on
opening type. Given the geometry used in the Spandrel-case, additional simulations were conducted
including balconies with various depths at different heights at the building exterior as the only
difference geometry wise. This case is referred to as the Balcony-case, divided into several sub-
scenarios depending on the opening configuration, the fire size, the depth of the balcony and the
position of the balcony.
By comparing the output data from the scenarios within the Spandrel-case and their associated
scenarios within the Balcony-case, the impact of horizontal projections on external fire spread was
shown at different heights above the underlying opening by observing the difference in the output data
in different graphs. Balcony-scenarios resulting in consistently lower values than the Spandrel-case at
each height means that the existence of these balconies results in lesser consequences at the facade on
all heights compared with the Spandrel-case. Furthermore, if these values are below the grey
horizontal line (BBR Limit), these balconies are considered to result in lesser consequences on the
facade at all heights compared with the accepted level in the prescriptive part of the Swedish building
regulations. The latter since the BBR Limit highlights the consequence at 1.2 m above the opening.
In the previous study 2 the approach was generally to compare the result of horizontal projections on
various positions and of various dimensions with scenarios built up by different spandrel heights
between openings in the building exterior. In this paper the aim is to investigate and compare the
results of a fix balcony depth while varying additional parameters using the same methodology as
described above. Additional parameters compared with the thesis 2 are additional fire configurations
and ventilation conditions as well as various balcony widths and the presence of an overlying balcony.
COMPARISON OF PROTECTION REQUIREMENTS BETWEEN COUNTRIES
A comparison of the protection requirements for external fire spread through windows stated
in different countries´ building regulations is summarized in Table 1.
Table 1: Summary of the protection requirements for external fire spread through openings for different countries 2
Country/Region
Spandrel
(m)
Horizontal
projection (m)
Note
Australia
0.9
1.1
-Non-combustible spandrel with an FRL of 60/60/60.
-Non-combustible horizontal projection with an FRL of 60/60/60, having
an extension at least 450 mm along the wall.
Denmark
-
-
There are no prescriptive requirements presented in the building
regulations in regards to the external vertical fire spread.
Finland
1.0*
-
There are no prescriptive requirements presented in the building
regulations in regards to the external vertical fire spread.
France
0.6 - 1 .3**
0.6 - 1.3**
-The dimensions of horizontal projection and spandrel are depending on
the C + D rule, which in turn is dependent on several variables.
Hong Kong
0.9
0.5
-Spandrel and horizontal projection are to be of a Fire resistance rating
not less than of the intervening storey.
New Zealand
1.5
0.6
-The New Zealand building codes also provides a table with further
combinations of different spandrel heights and projection lengths.
Norway
1.2
1.2
These requirements can be omitted if the building contains an automatic
sprinkler system. Other alternatives are also given.
Portugal
1.1
1.1 - the depth of
the horizontal
projection
-The horizontal projection need to have an extension of at least 1000
mm along the wall and to be of fire resistance EI60.
Spain
1.0
1.0 - the depth of
the horizontal
projection
-Spandrel of fire class EI60.
Sweden
1.2
-
-Vertical openings within 1200 mm need to be fire-rated, either one
window in fire class E 30 or both windows in fire class E 15 (BBR).
USA
The
International
Codes: 0.914
The International
Codes: 0.762
The International Codes: the openings shall be separately vertically
when the openings are within 1524 mm of each other horizontally and
the opening in the lower storey is not a protected opening with fire
protection rating of less than ¾ hour. The spandrel and horizontal
projection to have a fire-resistance rating of not less than 1 hour.
*This value is a local interpretation in Finland for the vertical distance between openings in the facade
**Values depending on several variables, the reader is referred to the text in Chapter 3.5 2 for a more complete description
A conclusion based on the information in Table 1 is that the level of protection of the building
regulations differs between the countries and there is very little consensus on the protective measures
between the countries presented. This is also supported by a recent study on facade regulation in the
Nordic countries 7 and implies that further research on the subject is needed.
VALIDATION STUDY
The fire spread on combustible external walls has been studied for a long time and several test
methods have been proposed through the years 8. One of these is the SP FIRE 105 9 that was
developed from a small scale method introduced in 1958 utilizing correlations with large scale tests 10.
In the eighties more intense fires were anticipated due to introduction of new materials and a
significant revision of the method was done. At this time also a criterion for vertical fire spread
through openings in the facade was introduced. The amended method for testing reaction to fire
properties of facade systems was released in 1985 and also included an increase in size of the sample
to a large scale test. The method is currently recognized as standardized test in Sweden, Denmark, and
Norway and has been proposed to be a certifiable method internationally.
The SP FIRE 105 test method specifies a procedure to determine reaction to fire properties of
different assemblies of materials, insulation, and claddings when exposed to fire from a simulated
apartment fire where flames emerge through a large window opening. The test rig consists of a 150
mm thick lightweight concrete wall, 4 000 mm (W) by 6 000 mm (H), above a fire compartment with
a 3 000 mm (W) by 710 mm (H) front opening. The compartment also has a horizontal opening in the
floor, close to its back wall, for air intake that measures 3 140 mm (W) by 300 mm (D). The fire
source consists of two trays positioned next to each other, each measuring 1 000 mm (W) by 100 mm
(H) by 500 mm (D). A flame suppressing lattice is installed on top of the tray edges and after each
tray has been filled with 30 liters of heptane, water is added such that the fuel level is touching the
underside of the lattice. Figure 1 shows the dimensions of the fire rig and the position of the fire trays
inside the fire compartment as well as pictures of the external flames from the test and in FDS.
Figure 1: Left - Experimental setup, dimensions and instrument positions 5. Right - External flames during the test and in FDS 2.
As a calibration of the numerical model one reference test was used where a non-combustible calcium
silica boards (Promatect®) with 10 mm thickness of the test specimen was used on the facade 5. In
this test the method was slightly altered to evaluate fire growth on external combustible ship surfaces
with FRP instead of a regular building facade system. The boards were attached on the lightweight
concrete wall, leaving an air gap of about 50 mm, and covered the two fictitious windows, as
illustrated in Figure 1. Each board measured 1 250 mm (W) by 3 000 mm (H), giving a total area of 3
750 mm (W) by 6 000 mm (H).
In order to assess the panel surface, the thermal exposure was measured by using nominally 0.7 mm
thick, 150 mm by 150 mm Inconel steel plates with wire thermocouples spot-welded on their
backside 11. The steel plates were fastened flush with the surface and covered a hole (Ø100 mm)
drilled through the panels for the thermocouple wires. The gas temperature was measured along the
centerline of the panels using Ø0.50 mm sheathed type K thermocouples. The bead of each
thermocouple was positioned 50 mm from the front surface of the panels and 50 mm offset the surface
temperature measurement steel plates (thermometers). In total, six thermometers and six gas
measurement thermocouples were installed, as illustrated in Figure 1. The tests were conducted under
an Industrial Calorimeter where measurements of gas temperature, velocity, and generation of
gaseous species such as CO2 and CO and depletion of O2 were made. Based on these measurements,
both the convective and the total heat release rate can be calculated.
On this setup numerical work has been performed by SP using FDS version 5.5.3 6,12,13. The Navier-
Stokes equations in the limit of low-speed, thermally-driven flow with an emphasis on smoke and
heat transport from fires are solved by the FDS software. Comparing the experimental data and the
corresponding results from the simulations, a good correlation could be found between the shielded
thermocouples. In the test the actual HRR is measured, which have been used in the simulations.
Discrepancies between measured and computed temperatures close to the fire source were observed.
Although, a rather rapid fire growth took place in the early stages of the test, a following shorter
colder period were observed. The full force of the fire commenced after 8 and lasted until 14minutes.
Since the difficulties of the geometry, i.e. the flame suppressing lattice consisting of a perforated steel
sheet with pipes of 25mm diameter placed in the fire tray, the numerical model could not resolve the
flows close to the fire tray. Good information on the actual HRR in the test is invaluable information
in reproducing the right dynamics of the test rig, in particular pulsation in the fire intensity was
observed during this test 6,12,13.
The previous validation studies by SP were further developed in the thesis 2 in order to validate FDS
6.2.0. In FDS 5.5.3 the algorithm used is an explicit predictor-corrector scheme that is second order
accurate in space and time where turbulence is treated by means of Large Eddy Simulation (LES) in
the Smagorinsky form. This algorithm has later been changed to Deardorff model in FDS 6. Note that
updates in FDS software and structure does not necessarily mean improvements in all application
areas, which was investigated in this validation study regarding the area of external fire spread.
The key results are presented in this paper, for all results the reader is referred to the thesis 2.
In Figure 2 the gas temperature data from the thermocouples in the reference test (EXP) and FDS
6.2.0 are compared at different positions along the height of the facade as per the left drawing in
Figure 1. Note that the following FDS results were produced using a 5 cm grid, which in this case
corresponds to a mesh resolution of approximately 30 for the 3.1 MW heptane fire.
In Figure 3 the retrieved surface temperatures from the thermometers are compared with the
output in FDS for the same setup.
Figure 2: Left – FDS, 5 cm cells, comparison of gas temperature during the reference test and those calculated in FDS for the
sheathed type K thermocouples close to the opening. Right – FDS, 5 cm cells, comparison of gas temperature further up along the
facade during the reference test. 2
0
100
200
300
400
500
600
700
800
900
1000
0200 400 600 800 1000 1200
Temperature [°C]
Time [s]
C27 EXP C28 EXP C29 EXP
C27 FDS C28 FDS C29 FDS
0
100
200
300
400
500
0200 400 600 800 1000 1200
Temperature [°C]
Time [s]
C30 EXP C31 EXP C32 EXP
C30 FDS C31 FDS C32 FDS
Figure 3: Left - Comparison of surface temperatures retrieved from the Inconel thermometers and AST from FDS close to the
opening during the reference test. Right - Comparison of surface temperatures further up along the facade during the reference
test. 2
0
100
200
300
400
500
600
700
800
900
0200 400 600 800 1000 1200
Surface temperature [°C]
Tiime [s]
C21 EXP C22 EXP C23 EXP
C21 FDS 5 cm C22 FDS 5 cm C23 FDS 5 cm
0
100
200
300
400
0200 400 600 800 1000 1200
Surface temperature [°C]
Time [s]
C24 EXP C25 EXP C26 EXP
C24 FDS 5 cm C25 FDS 5 cm C26 FDS 5 cm
Generally using this setup, a good correspondence is found close to the opening however FDS slightly
overestimates the temperatures further up, especially on position C29 and C30. The output data in
FDS 6.2.0 also seem to be slightly higher on all positions compared to the output data in FDS 5.5.3,
especially for the surface temperature data modelled in FDS.
In total, three grid sizes were used in the validation study, namely 20 cm, 10 cm and 5 cm large cells.
The higher mesh resolution from the 5 cm grid were considered important in order to resolve more of
the turbulence of the fire plume, which influenced more on the results close to the opening at position
C21/C27. Obtaining good results close to the opening were considered important since the main area
of concern in the comparative analysis is close to the fire source. The smaller grid size also
contributed to a more realistic appearance of the plume with better air entrainment, as seen in the right
pictures in Figure 1. Hence, one main conclusion from the validation study is the need of a high mesh
resolution of at least 30 to obtain credible results in the comparative analysis.
The results presented above were performed on a single mesh, which is particularly time consuming
for the finer grid cell setup. However, additional simulations were performed in the sensitivity
analysis where the computational domain was divided into different mesh alignments. Little or no
difference was seen in the output therefore the simulations in the comparative analysis were divided
into several meshes to save computational time.
Altogether, FDS 6.2.0 was deemed well suited as a calculation tool for modelling external fire spread.
The conclusions drawn from the validation study were taken into consideration when performing the
simulations in the comparative analysis.
COMPARATIVE ANALYSIS
One general conclusion in the thesis 2 was that the use of at least a 60 cm deep horizontal
projection resulted in less severe consequences above the projection compared with scenarios built up
by different spandrel heights. In addition, this balcony depth was considered to result in lesser
consequences at the facade on all heights compared with the accepted level in the prescriptive part of
the Swedish building regulations.
In this paper the balcony depth of 60 cm is held fix while varying the height and the width of the
projection. Also, possible influences of an overlying projection are studied as well as the change in
ventilation conditions and heat release rate per unit area (HRRPUA). The general setup, design fire
and process of generating the results in the thesis 2 are explained briefly below, from which also the
elaborated work presented further down are based on. The reader is referred to the thesis 2 for a more
thorough explanation of the setup and process.
Geometry, Measurements, Design Fire and Process of Generating the Results
The fire compartment was arranged as an apartment room with the dimensions 4.5 m (W) by 4.5 (D)
by 2.5 m (H), which corresponds to a 20 m2 room. A 1 m2 burner was placed in the middle of the
room. In the lower part of the rear wall a narrow 1.6 m2 opening was added for ventilation purpose,
simulating an open door into the room with the same size as the door in the front. The low
configuration prevents the room to leak out too much smoke through the ventilation intake. At the
front wall either a 0.8 m (W) by 2 m (H) door or a 1.2 m (W) by 1.2 m (H) window were placed
depending on the scenario. In Figure 4 the geometry of the general setups is seen.
The computational domain was divided into two vertically aligned meshes as seen in the left picture
in Figure 4. This configuration was chosen since in the validation study it was seen to result in lower
computational times compared with the other mesh configurations and at the same time have
negligible effect on the results. Because of the ventilation conditions into the room, the fire size of 3.1
MW used in the validation study were reused in the comparative analysis and a higher heat release
rate of 4 MW was used for sensitivity analysis purposes.
To estimate the consequence on the facade, the device quantity
in FDS 6.2.0 were used in each
cell at the facade surface, starting just above the opening and ending 3.2 m above. In total, 16-24
devices were included in each row depending on the opening type below, with a total of 64 rows
along the height.
The reason for using
is because it is an artificial temperature both taking into account the incident
radiation temperature
and the gas temperature
14. For example, if an exposed body is far away
from the fire source, radiation is the dominating part and hence the
will be closer to the radiation
temperature
. If however the exposed body is exposed by mainly convection from the hot gases,
will be closer to the gas temperature
. The quantity
can then be used as an alternative
means of expressing the thermal exposure to a surface in FDS without taking into account the energy
gain or loss through conduction from the exposed body. The adiabatic surface temperature is the
theoretical highest achievable temperature of an exposed body during a fire scenario. Hence, the
parameter is important in regards to the risk of ignition during longer fire exposure times.
Since FDS is transient and the
data values from each device at the facade are fluctuating over
time, it is difficult to describe the damage at the facade in a comparative sense. Also, the design of the
resulting graphs only allows for one value at each height describing the consequence. Hence, the
simulations continued until an equilibrium state was seen. In order to do this the specific heat of the
building materials (only concrete) were divided by a factor of 100 to speed up the process. Since this
procedure is done for all the simulations and the aim was to compare the data, this was not seen to
affect the credibility of the results. Since the interrelationship between the data values from each
device over time on every row was seen to be fairly constant, the maximum values observed from
each of the devices over time were used to describe the damage at the facade for that specific row.
Lastly, to express the consequence at the facade at a specific height a mean value was calculated of
the maximum values respectively. These values build up the graphs below for each height from which
the results are read out.
Window Spandrel
Door Higher Balcony
Window Higher Balcony
0.2 m
2.35 m
0.2 m
0.35 m
2 m
0.8 m
0.4 m
0.2 m
2.35 m
1.2 m
0.2 m
0.35 m
0.4 m
1.2 m
Door Lower Balcony
Window Lower Balcony
Figure 4: Geometry of the Window Spandrel-scenario and the Door/Window Higher and Lower Balcony-scenarios.
The Importance of Projection Width and Presence of an Overlying Projection
In Figure 5 the importance of projection width and presence of an overlying projection are studied for
the door configuration at 3.1 MW. The dotted lines highlight additional simulations for this paper.
Figure 5: A comparison of AST at the facade above the door at various heights between the Door Spandrel-scenario and the different
Door Balcony 60 cm-scenarios.
The similar setup is seen for the window configuration in Figure 6.
Figure 6: A comparison of AST at the facade above the window at various heights between the Window Spandrel-scenario and the
different Window Balcony 60 cm-scenarios.
The suffix - less wide corresponds to a horizontal projection as wide as the underlying opening, seen
in Figure 4 as the setups with rectangles filled in black. Descriptions without suffixes correspond to
the setups with horizontal projections extending 0.4 m on each side of the underlying opening, marked
in Figure 4 as outer rectangles. Lastly, the suffix - Additional Balcony corresponds to a similar sized
projection positioned above to an identical apartment, which is also seen in Figure 4 for two setups.
As seen in Figure 6, all Balcony-scenarios result in lower
at the facade compared with the
Spandrel-case and the BBR Limit. The similar is seen for the Door configuration in Figure 5, however
the Door Lower Balcony 60 cm - less wide-scenario results in higher
than the BBR Limit from
0.25 m to 1 m above the underlying door. Similar runs were conducted for both opening
configurations on the higher HRR of 4 MW, however not implemented in this paper due to space
limitations. In these simulations all the Balcony-scenarios resulted in lower
at the facade
compared with the Spandrel-case and the BBR Limit.
50
150
250
350
450
550
650
0.25 0.45 0.65 0.85 1.05 1.25 1.45 1.65 1.85 2.05 2.25 2.45 2.65 2.85 3.05 3.25
Adiabatic surface temperature [ºC]
Height above door [m]
AST Door Spandrel AST BBR Limit
AST Door Lower Balcony 60 cm AST Door Balcony less wide 60 cm - Additional balcony
AST Door Higher Balcony 60 cm - less wide AST Door Lower Balcony 60 cm - less wide
AST Door Higher Balcony 60 cm
50
150
250
350
450
550
650
0.25 0.45 0.65 0.85 1.05 1.25 1.45 1.65 1.85 2.05 2.25 2.45 2.65 2.85 3.05 3.25
Adiabatic surface temperature [ºC]
Height above window [m]
AST Window Spandrel AST BBR Limit
AST Window Lower Balcony 60 cm AST Window Balcony less wide 60 cm - Additional Balcony
AST Window Higher Balcony 60 cm - less wide AST Window Lower Balcony 60 cm - less wide
AST Window Higher Balcony 60 cm
The Influence of Ventilation Conditions and HRRPUA
The simulations presented in the thesis 2 were performed at a HRRPUA of 3100 kW/m2, which on the
1 m2 burner resulted in the 3.1 MW fire. To investigate the sensitivity in the HRRPUA parameter the
input value was halved and compared with the original simulation, suffix - 1500 kW/m2. Also, the
influence of varying the ventilation conditions were studied by removing the 1.6 m2 ventilation hole
in the rear wall for the original 3.1 MW simulation, suffix - no ventilation. The results for the Door
Spandrel-scenario and Window Spandrel-scenario are presented in Figure 7-8, where dotted lines
highlight the additional simulations for this paper.
Figure 7: A comparison of AST at the facade at various heights between the different Door Spandrel-scenarios.
Figure 8: A comparison of AST at the facade at various heights between the different Window Spandrel-scenarios.
As seen in both figures, the change in HRRPUA resulted in no visible deviations in the results for
both opening configurations. However, blocking the rear ventilation hole is seen to drastically
increase the
at the facade. This is especially seen for the window configuration, now reaching
higher values than for the 4 MW simulation, the latter which contain the ventilation opening in the
rear wall.
The drastic change in output for the no ventilation case were then investigated further by studying the
impact of the less wide 60 cm deep horizontal projections for this setup, as seen in Figure 9-10. The
external flames for these individual scenarios are visualized in Figure 11.
0
100
200
300
400
500
600
700
800
900
1000
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2
Adiabatic surface temperature [ºC]
Height above door [m]
AST Door Spandrel AST Door Spandrel - 4 MW
AST Door Spandrel - no ventilation AST Door Spandrel - 1500 kW/m2
0
100
200
300
400
500
600
700
800
900
1000
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2
Adiabatic surface temperature [ºC]
Height above window [m]
AST Window Spandrel AST Window Spandrel - 4 MW
AST Window Spandrel - no ventilation AST Window Spandrel - 1500 kW/m2
Figure 9: A comparison of AST at the facade above the door at various heights between the Door Spandrel - no ventilation-scenario and
the different Door Balcony 60 cm - no ventilation-scenarios.
Figure 10: A comparison of AST at the facade above the window at various heights between the Window Spandrel - no ventilation-
scenario and the different Window Balcony 60 cm - no ventilation-scenarios.
Similar to the Door Lower Balcony 60 cm - less wide-scenario in Figure 5 the corresponding scenario
for the no ventilation case in Figure 9 are extending the BBR Limit, however in this case not as
convincing and still lower than the Spandrel-case. The rest of the scenarios are resulting in lower
values compared with the Spandrel-case and the BBR Limit. A distinct difference in the impact on the
external fire spread is seen in Figure 11 depending on the opening configuration. For the door
configuration the horizontal projections mainly protect the facade by shielding the radiative and
convective heat from the fire. For the window configuration the horizontal projections are also seen to
project the fire plume further away from the facade.
50
150
250
350
450
550
650
750
0.25 0.45 0.65 0.85 1.05 1.25 1.45 1.65 1.85 2.05 2.25 2.45 2.65 2.85 3.05 3.25
Adiabatic surface temperature [ºC]
Height above door [m]
AST Door Spandrel - no ventilation AST BBR Limit
AST Door Higher Balcony 60 cm - less wide - no ventilation AST Door Lower Balcony 60 cm - less wide - no ventilation
50
150
250
350
450
550
650
750
0.25 0.45 0.65 0.85 1.05 1.25 1.45 1.65 1.85 2.05 2.25 2.45 2.65 2.85 3.05 3.25
Adiabatic surface temperature [ºC]
Height above window [m]
AST Window Spandrel - no ventilation AST BBR Limit
AST Window Higher Balcony 60 cm - less wide - no ventilation AST Window Lower Balcony 60 cm - less wide - no ventilation
Figure 11: A comparison of the external flames between the Door/Window - no ventilation-scenarios.
Door Spandrel
- no ventilation
Door Lower Balcony
60 cm - less wide
- no ventilation
Door Higher Balcony
60 cm - less wide
- no ventilation
Window Spandrel
- no ventilation
Window Lower Balcony
60 cm - less wide
- no ventilation
Window Higher Balcony
60 cm - less wide
- no ventilation
DISCUSSION AND CONCLUSIONS
Similar to the simulations performed in the thesis 2, the balconies used in this paper are
rectangular non-combustible balconies with completely open sides and no separation walls. One
reason for choosing this balcony type is because the design is geometrical simple and therefore better
simulated in FDS, which would bring less uncertainty into the results compared to simulations of
other balcony types. The simple design has also shown to give the best protection compared to other
balcony designs 1. The latter observations were concluded after it emerged that more enclosed
balconies were shown to trap more of the hot gases at floor levels above the fire floor and thus
increase the rate of vertical fire spread. However, even if the results from the thesis 2 were built on
simulations with the simpler balcony design, only one balcony was included between the apartments.
Given the fact that the exterior of multi-story buildings in reality often contain multiple balconies one
above the other for both esthetical and practical reasons, the effects of an overlying balcony were
investigated in this paper to study the above-mentioned effects. The inclusion of an overlying balcony
did not show any major changes in the results compared to the original setup.
One of the difficulties in using FDS as calculation tool for this problem area is the computational
capacity and time required to obtain reliable results, which in turn is based on the outcomes of the
validation study. This restricts the user from varying a large number of input variables and study the
outcomes since the amount of simulations to perform are limited. The fixed balcony depth of 60 cm
enabled the investigation of additional input parameters compared to the thesis 2. New input variables
that were varied in this paper were the change in ventilation conditions and HRRPUA. The HRRPUA
was treated in the validation study of the thesis 2 by doubling the value of 3100 kW/m2 to 6200
kW/m2 resulting in no major variations in the output. The simulations with HRRPUA of 1500 kW/m2
were used to examine if the value of 3100 kW/m2 in the comparative analysis were set too high.
Recommended values of HRRPUA found in the literature are between 500-2500 kW/m2 15 and if set
too high, there is a risk that the specified burner is injecting with a too high flow of fuel gas for the
given surface area, which may lead to a jet flame associated with large Froude numbers. This flame
type is less affected by the surrounding air and controlled by the momentum of the fuel gas rather than
the gravity. However, as already been presented, no visible deviations in the results were seen.
The change in ventilation conditions made a great impact on the results compared with the original
setup. By comparing the external flames in Figure 11 with the pictures of the external flames during
the original setups found in the thesis 2, a distinct difference is noticed in regards to the distance
between the external flames and the wall. In this case the external flames are seen to rise closer to the
wall compared to the case with ventilation. The reason for this is a result of a fire not being able to
vent in fresh air from the rear opening anymore. The fire plume is now ejecting out from the same
opening as the intake of fresh cold air. The cold air is then pushing the plume more upwards as a
result of two oppositely directed flows through a limited area in regards to the height. This is seen
especially for the less high window opening, where the fire plume ejecting out from the room is
covering a larger area of the opening, so that the impact of the inflowing air on the plume is greater.
On the contrary, for the door configuration the ejecting plume is not reaching as far down, which
allows the air to pass easier along the floor level resulting in less impact on the ejecting plume.
Studying the importance of projection width in this paper is an extension of the sensitivity analysis in
the previous thesis 2. This was performed since previous research 16 have shown that a wider balcony
than the underlying opening results in lower temperatures at the facade, because the external flames
are kept further away from the facade on the sides of the balcony. However, a wider balcony type is
not always the preferable solution in the exterior because of both esthetical and practical reasons. The
results presented here are partly in line with previous research 16, mainly seen for the Door/Window
Lower Balcony-scenarios in Figure 5-6. The wider balcony type results in lower
-values, which is
especially prominent for the door configuration in Figure 5 comparing the output from both opening
configurations. However, regarding the scenarios exceeding the BBR Limit in Figure 5 and 9, further
studies on the flow around the balcony are needed to conclude why a narrower opening type result in
the need of a wider balcony when the balcony is positioned just above the opening.
Similar to the findings in the thesis 2 there is a distinct difference in the consequences at the facade
depending on the opening configuration. A fire plume ejecting through a narrower door type is
causing higher gas velocities through the opening, resulting in a fire plume ejecting further away from
the facade. The outcome of a wider window type on the other hand is lower velocities and hence a fire
plume ejecting closer to the building exterior. Consequently, the impact of the latter scenario on the
facade is greater, which is also seen in the results. At the same time the results from the Window-
scenarios in this study and the thesis 2 generally show a larger difference in output to the Spandrel-
case compared to the Door-scenarios. This further suggests that a horizontal projection offers higher
protection compared to spandrels when the underlying opening is wide and short.
The results in this paper further shows that the use of a 60 cm deep horizontal projection results in less
severe consequences above the projection compared with scenarios built up by different spandrel
heights. The horizontal projections in this study were mainly 20 cm thick rectangular non-combustible
balconies with open sides and no separation walls positioned at two different heights, having the same
width as the underlying opening. The results in this paper suggest that the use of these balconies
generally result in lesser consequences at the facade compared with the accepted level in the
prescriptive part of the Swedish building regulations (BBR). This means that in general, given the
conditions in this study and the thesis 2, a spandrel height of at least 1.2 m as stated by the BBR can
be replaced by a 60 cm deep horizontal projection positioned at any height above the underlying
opening. This is similar to the protection requirements for France and New Zealand, seen in Table 1.
However, this is only seen for the window opening type in the building facade. A narrower opening
results in the need of a wider balcony type when the balcony is positioned just above the opening.
Hence, the general conclusion stated above is not applicable for this specific configuration.
REFERENCES
1 Mammoser III, J.H., & Battaglia, F., A computational study on the use of balconies to reduce flame spread
in high-rise apartment fires, Fire Safety Journal, Vol. 39, pp. 277-296, 2004.
2 Nilsson, M., The impact of horizontal projections on external fire spread - a numerical comparative study,
Report nr. 5510, Lund University, Division of Fire Safety Engineering, Lund, Sweden, 2016.
3 Boverket, Boverkets byggregler (Swedish building regulations), BBR 22, Karlskrona: Boverket, 2015.
4 McGrattan, K., et al., Fire Dynamics Simulator User’s Guide, FDS Version 6.2.0, SVN Repository
Revision: 22352, NIST Special Publication 1019, National Institute of Standards and Technology,
Gaithersburg, MD USA, April 2015.
5 Evegren, F., Rahm, M., Arvidsson, M., Hertzberg, T., Fire testing of external combustible ship surfaces,
11th International Symposium on Fire Safety Science, Christchurch, NZ: IAFSS, 2014.
6 Anderson, J., Jansson, R., Fire Dynamics in Facade Fire Tests: Measurement and Modeling, Proceedings of
Interflam 2013, pp. 93, Royal Holloway College, University of London UK, 2013.
7 Strömgren, M., Albrektsson, J., Johansson, A., Almgren, E., Comparative analysis of façade regulations in
the Nordic countries, 1st International Seminar for Fire Safety of Facades, MATEC Web of Conferences 9,
DOI: 10.1051/matecconf/20130901003, 2013.
8 Hildebrand, C., Facade Fire Testing, Workshop report CIB 14, Institut für Baustoffe, Leipzig, 1988.
9 SP Fire Technology, SP Fire 105, External wall assemblies and facade claddings: Reaction to fire, Issue
No: 5, SP Swedish National Testing and Research Institute: Fire Technology, Borås, 1994.
10 Johannesson, P. and Larsson, G., Fire Tests with light, non-bearing external walls, Meddelande 124,
Statens Provningsanstalt (SP Technical Research Institute of Sweden), Borås, 1958.
11 Wickström, U., Jansson, R., Tuovinen, H., Validation fire tests on using the adiabatic surface temperature
for predicting heat transfer, Fire Technology, SP Report 2009:19, Borås 2009.
12 Jansson, R., Anderson, J., Experimental and numerical investigation of fire dynamics in a facade test rig,
Proceedings of Fire Computer Modeling, Santander, p. 247, Spain, 18-19th October 2012.
13 Anderson, J. and Jansson, R., Facade fire tests - measurements and modeling, MATEC Web of
Conferences 9, 02003, 2013.
14 Wickström, U., Temperature calculation in fire safety engineering, New York: Springer International
Publishing, ISBN: 978-3-319-30170-9, 2015.
15 Best Practice gruppen, CFD Best Practice, Copenhagen, Denmark, November 2009.
16 Morgado, H.J.L., Rodrigues, J.P.C., & Laim, L.M.S., Experimental and numerical study of balcony effect
in external fire spread into upper floors, Journal of Application of Structural Fire Engineering, Czech
Republic, 19-20 April 2013.