Professional Documents
Culture Documents
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Design model for the verification of the separating function of light timber frame
assemblies
Andrea Frangi ∗ , Vanessa Schleifer, Mario Fontana
ETH Zurich, Institute of Structural Engineering, 8093 Zurich, Switzerland
permitted to define some rules about detailing of wall and floor Table 1
assemblies that have been included for example in EN 1995-1-2 [5]. Thickness and mean density of the gypsum boards tested under ISO fire exposure.
Thus, the criterion E (integrity) may be assumed to be satisfied if Manufacturer Type Thickness Density
the criterion I (insulation) has been satisfied and panels remain tested (mm) (kg/m3 )
fixed to the timber structure on the unexposed side. Manufacturer 1 Gypsum fibreboard (GF) 10, 12.5, 15, 18 1186
In timber buildings, walls and floors are mostly built up by Gypsum plasterboard 15 908
adding different layers to form an assembly. For the verification of (GP) of type A
the separating function of timber assemblies, component additive Manufacturer 2 Gypsum plasterboard 15 853
methods are common. These models are called component (GP) of type F
Gypsum fibreboard (GF) 12.5 1504
additive models, since the fire resistance of a layered construction
Gypsum plasterboard 10, 12.5, 15, 25 810
is obtained by adding the contribution to the fire resistance of
(GP) of type A
the different layers. In [6] calculation models for the verification Manufacturer 3 Gypsum plasterboard 15 889
of the separating function of light timber frame wall and floor (GP) of type F
assemblies used in the UK [7], Canada [8] and Sweden [9] as Gypsum fibreboard (GF) 12.5 1313
well as according to ENV 1995-1-2 [10] have been presented and
reviewed. The current design method according to EN 1995-1-2 • Gypsum plasterboards:
(Annex E) is based on the Swedish component additive method. As Type A, H and F according to EN 520 [14].
an enhancement of the method of ENV 1995-1-2 and the Canadian Type X according to ASTM C1396 [15] or CAN/CSA-82.27-
method, the Swedish component additive method takes into M91 [16].
account the influence of adjacent materials on the fire performance • Gypsum fibreboards according to EN 15283-2 [17].
of each layer and therefore describes the real fire performance • Rock fibre insulation (density ≥ 26 kg/m3 ).
more appropriately. However, the design method is based on input • Glass fibre insulation (density ≥ 15 kg/m3 ).
data that was deduced from a limited number of fire tests on wall The developed design method is based on the additive compo-
assemblies and therefore only covers a limited range of timber nent method given in EN 1995-1-2. Thus, the fire resistance tins of
structures. the timber assembly is taken as the sum of the contributions from
A comprehensive research project on the separating function the different layers (claddings, void or insulated cavities) according
of light timber frame wall and floor assemblies with cladding to their function and interaction as follows (Fig. 1):
made of gypsum plasterboards and wood-based panels has
i=n−1
been carried out at ETH Zurich in collaboration with the Swiss X
tins = tprot,i + tins,n (1)
Laboratories for Materials Testing and Research (Empa). The
i=1
objective of the research project was the development of an
improved design model for the verification of the separating with
Pi=n−1
function (insulation and integrity criteria) of light timber frame i =1 tprot,i Sum of the protection values tprot,i of the layers (in
wall and floor assemblies. A large number of small-scale fire tests direction of the heat flux) preceding the last layer of
permitted the analysis of different parameters (material, thickness, the assembly on the fire-unexposed side (min).
position and number of the layers) on the thermal behaviour of tins,n Insulation value tins,n of the last layer of the assembly
protective cladding made of gypsum plasterboards and wood- on the fire-unexposed side (min).
based panels [11,12]. The results of the fire tests allowed the Protection and insulation values of the layers can be calculated
verification and calibration of thermal properties used for thermal according to the following general equations taking into account
finite element (FE) analysis. Based on an extensive FE parametric the basic values of the layers, the coefficients for the position
study, the coefficients of the design model for the verification of the of the layers in the assembly and the coefficients for the joint
separating function of light timber frame wall and floor assemblies configurations:
were calculated [13]. The design model was verified by means of tprot,i = (tprot,0,i · kpos,exp,i · kpos,unexp,i + 1ti ) · kj,i (2)
full-scale fire tests.
The paper first describes the basic structure of the design model tins,n = (tins,0,n · kpos,exp,n + 1tn ) · kj,n (3)
for the verification of the separating function of light timber frame with
wall and floor assemblies. Then, the main results of experimental tprot,0,i Basic protection value (min) of layer i (Fig. 1
and numerical analyses are presented. Their results permitted
and Table 3).
the calculation of the coefficients of the design model for the
tins,0,n Basic insulation value (min) of the last layer
verification of the separating function of light timber frame wall
n of the assembly on the fire-unexposed
and floor assemblies.
side (Fig. 1 and Table 3).
2. Design method for separating function of timber construc-
1ti , 1tn Correction time (min) for layers protected
by gypsum plasterboards of type F or type X
tions
as well as gypsum fibreboards (Table 5).
A comprehensive design method for the verification of the kpos,exp,i , kpos,exp,n Position coefficient that takes into account
separating function of timber constructions has been developed the influence of layers preceding the layer
based on an extensive experimental as well as finite element considered (Table 4).
thermal analysis [13]. The design method is capable of considering kpos,unexp,i Position coefficient that takes into account
timber assemblies with an unlimited number of layers made of the influence of layers backing the layer
gypsum plasterboards, wood panels or combinations thereof. The considered (Table 6).
cavity may be void or filled with insulation made of rock or glass kj,i , kj,n Joint coefficient (Table 7).
fibre. The design method is valid for following materials:
• Solid timber panels (density ≥ 400 kg/m3 ). 3. Fire tests
• Oriented Strand Board (OSB) (density ≥ 550 kg/m3 ).
• Particleboards (density ≥ 500 kg/m3 ). A series of 17 small-scale fire tests (in the following mentioned
• Plywood (density ≥ 400 kg/m3 ). as V1 to V17) was performed with non-loaded specimens consist-
1186 A. Frangi et al. / Engineering Structures 32 (2010) 1184–1195
Fig. 1. Timber frame wall and floor assemblies: Numbering and function of different layers.
Table 2
Measured fire resistance of the small-scale tests V4 and V9 with regard to the insulation criterion (temperature rise criteria: 140 ◦ C/180 ◦ C at fire-unexposed side of
assembly).
Fire test Test specimen Insulation time (min) Figure
V9 10 + 10 + 10 mm gypsum fibreboards 57
V4 15 + 15 mm gypsum fibreboards with cavity filled with 80 mm thick wood fibre batts (ρ = 178 kg/m3 ) 72
V4 15 + 15 mm gypsum fibreboards 83
V9 15 + 15 mm gypsum fibreboards with cavity filled with 80 mm thick glass fibre batts (ρ = 27 kg/m3 ) 85
V9 15 + 15 mm gypsum fibreboards with cavity filled with 40 mm thick rock fibre batts (ρ = 31 kg/m3 ) >85 (test stopped)
V9 15 + 15 mm gypsum fibreboards with cavity filled with 40 mm thick rock fibre batts (ρ = 110 kg/m3 ) >85 (test stopped)
V4 15 + 15 mm gypsum fibreboards with cavity filled with 80 mm thick rock fibre batts (ρ = 42 kg/m3 ) >105 (test stopped)
T1 RF
T11
WF
Fig. 3. Left: Temperature development for the small-scale fire tests V4 and V2; Right: Position of the thermocouples (RF = rock fibre, WF = wood fibre).
same thickness (15 mm). Even though gypsum boards of different boards backed with rock fibre or wood fibre. Thus, insulating
type and manufacturer (with different density, fibres and fillers) batts caused the fire-exposed gypsum boards to heat up more
were tested, the duration of the plateau at about 100 ◦ C was more rapidly. The temperature development measured behind the fire-
or less the same and the overall temperature development mea- exposed gypsum fibreboard for the assembly with void cavity
sured up to about 200 ◦ C was quite similar (the differences in (see T15) followed up to 250 ◦ C roughly the same development
time for the same temperature measured were less than 5 min). as for gypsum boards backed with rock fibre or wood fibre (see
Thus, it can be assumed that the water content and therefore of T25 and T35). Subsequently, the temperature increase was slower.
gypsum for the different boards was about the same. Further, the For the assembly with a double layer of 15 mm thick gypsum
addition of fibres and fillers to the gypsum core did not signifi- fibreboards, the temperature development measured behind the
cantly change the thermal behaviour of the boards. However, the first gypsum fibreboard (see T5) followed up to 250 ◦ C roughly
reinforcement of the gypsum core with fibres and fillers generally the same development as for gypsum fibreboards tested as single
improves the stability and the mechanical properties (shrinkage, board (see T10). But then the temperature increase was slower.
cracking, ablation, falling-off) of the boards after complete dehy- A possible reason may be the effect of some moisture migration
dration. Noticeable is that the density of the gypsum fibreboards from the second gypsum board towards the first gypsum board.
was much higher than for the gypsum plasterboards (Table 1), but Noticeable is that the temperature measured behind the fire-
no significant difference was observed with regard to the thermal unexposed gypsum fibreboard for the assemblies with a void cavity
behaviour. Thus, the density does not seem to be a relevant pa- and double layer (see T1 and T11) were quite similar. Thus, the void
rameter to describe the thermal behaviour of gypsum boards. The cavity did not significantly influence the temperature development
main parameter is the water content and therefore the thickness on the fire-unexposed side of the assembly.
of the board (Fig. 2). Cone calorimeter tests conducted with reg- Temperature measurements between gypsum board and tim-
ular gypsum plasterboards as well as gypsum plasterboards with ber frame (see for example thermocouple T20 in Fig. 3) as well
improved core cohesion from different countries (Sweden, Canada, as additional results of fire tests with gypsum boards backed by
USA, New Zealand and Japan) and tested as fire protective cladding timber boards (OSB, particleboards) showed that the timber had
for a wooden member showed similar results [19]. The density of a small influence on the thermal behaviour of the gypsum boards
the boards tested varied between 651 and 864 kg/m3 . in comparison to gypsum boards tested as a single board (see also
Since all results shown in Fig. 2 were obtained by testing single thermocouples T10 and T20 in Fig. 3). Fig. 4 summarizes the mea-
gypsum boards directly exposed to fire, the thermal behaviour of sured time taken to reach the temperature rise of 250 ◦ C (average)
the gypsum boards was studied without the influence of additional and 270 ◦ C (at any point) on the fire-unexposed side of 15 mm thick
layers. However, gypsum boards used as cladding for light timber gypsum fibreboards tested as a single board or backed with differ-
frame assemblies are usually backed by other boards (gypsum, ent materials.
timber), insulation batts (rock, glass or wood fibre) or void cavities. Table 2 summarizes the measured fire resistance of the small-
Therefore, the effect of the backing material, or void, was studied scale fire tests V4 and V9 with regard to the insulation criterion.
in a separate series of tests. Fig. 3 left shows the temperatures According to most standards, for example EN 13501-2 [20], the
measured for the small-scale fire test V4 with 15 mm thick insulation failure is the time taken for the average temperature on
gypsum fibreboards (Fig. 3 right), which permitted the analysis the fire-unexposed side of the whole construction to increase by
of the influence of the void cavity or cavities filled with different 140 ◦ C and at any point by 180 ◦ C. It can be seen that assemblies
insulation (RF = rock fibre, WF = wood fibre) as well as the with cavities filled with rock fibre batts showed the highest fire
influence of multiple layers. Further, Fig. 3 left also shows the resistance although the thermal behaviour of gypsum boards
temperature development of 15 mm thick gypsum fibreboards backed by insulating batts is influenced unfavourable (Figs. 3 and
tested as single board (fire test V2, Fig. 3 right). It can be seen 4). The shortest fire resistance was measured for the assembly
that the temperatures measured behind the fire-exposed gypsum consisting of 3 layers of gypsum fibreboards, although the total
fibreboards for the assemblies with the cavities filled with rock thickness of the gypsum fibreboards was the same as for the other
fibre or wood fibre (see thermocouples T25 and T35) were similar assemblies given in Table 2. The main reason is that the 10 mm
and increased faster than for gypsum fibreboards tested as a single thick gypsum fibreboards fell off sooner than the 15 mm thick
board (see T10). Additional fire tests with gypsum boards backed gypsum fibreboards. Noticeable is that the insulation time of the
with glass fibre showed the same thermal behaviour as for gypsum assembly with the cavity filled with 80 mm thick wood fibre batt
1188 A. Frangi et al. / Engineering Structures 32 (2010) 1184–1195
Fig. 4. Time taken to reach the temperature rise of 250 ◦ C (average) and 270 ◦ C (at any point) on the fire-unexposed side of 15 mm thick gypsum fibreboards tested as a
single board or backed with different materials.
was lower than for the assembly with a void cavity. The reason was Table 3
that first the wood fibre batt caused the fire-exposed gypsum board Basic insulation values tins,0,n as well as basic protection values tprot,0,i for different
materials. For rock and glass fibre insulation only the basic protection value tprot,0,i
to heat more rapidly, and secondly the wood fibre batt started to is given as wall and floor assemblies with the insulation as last layer of the assembly
shrink strongly when directly exposed to fire and fell off. Unlike are rarely used in buildings.
the wood fibre batt, the rock fibre batt remained in place during Material Basic insulation Basic protection value
the whole time of fire exposure. value tins,0,n (min) tprot,0,i (min)
1.4 1.2
hi hi
Gypsum 24 · 30 ·
4. FE thermal analysis 15 15
plasterboard,
gypsum fibreboard
For the calculation of the coefficients of the design method
hi
1.4
hi
1.1
Solid timber panel 19 · 30 ·
(basic values tprot,0 and tins,0 , correction time 1t as well as position 20 20
1.4 1.1
coefficients kpos,exp and kpos,unexp ) an extensive finite element
hi hi
Particleboard 22 · 20
33 · 20
thermal analysis was conducted using ANSYS. The heat transfer
1.4 1.1
to the surface of the member was calculated using temperature- OSB, plywood 16 ·
hi
23 ·
hi
20 20
independent constant values according to EN 1991-1-2 [21] for the
(0.75·log(ρi )−ρi /400)
resultant emissivity by radiation εres = 0.8 and the coefficient of Rock fibre – 0.3 · hi
heat transfer by convection αc ,exp = 25 W/m2 K and αc ,unexp = Glass fibre – for hi < 40 mm: 0
for hi ≥ 40 mm: (0.0007 · ρi
4 W/m2 K. Density, thermal conductivity and specific heat capacity
+ 0.046) · hi + 13 ≤ 30
of gypsum, timber and insulation vary as a function of temperature.
With hi : Thickness of the layer considered (mm), ρi : Density of the layer considered
Unfortunately, the thermal properties of these materials at high
(kg/m3 ).
temperatures are difficult to measure. Transient effects and the
method of measurement can have a significant effect on the test
After calibration with the fire tests on single boards, the material
results [22,23]. Further, the thermal properties are influenced by
properties were verified with additional 15 small-scale and large-
mass transfer of moisture into or out of the gypsum and timber
scale fire tests [27–29]. For example, Fig. 7 right shows the compar-
board as well as cracking and ablation. In order to consider these
ison between FE thermal analysis and fire tests on gypsum boards
effects the thermal properties are usually calibrated with results of
backed by timber boards or particleboards. It can be seen that the
fire tests. Thus, thermal properties of gypsum and timber used in
FE thermal analysis also predicted the temperature development
FE thermal analyses are often ‘‘apparent or effective’’ values rather
of gypsum boards satisfactorily. Additional detailed comparisons
than ‘‘real’’ physically-correct material properties and explain the
between FE thermal analysis and fire tests can be found in [13].
large scatter of values used by different authors [23].
Density, thermal conductivity and specific heat of gypsum, tim-
ber and insulation were assumed based on measurements found 5. Coefficients of the design method
in the literature [24–26] and calibrated to results of fire tests with
single gypsum or timber boards as well as insulation batts. Figs. 5 5.1. Basic values
and 6 show the temperature-dependent thermal conductivity and
specific heat of gypsum and timber used for the FE thermal analy- The basic insulation value tins,0 corresponds to the fire resis-
sis. In the figures values measured and assumed by other authors tance of a single layer without the influence of adjacent materials
are also given. The value of the thermal conductivity of gypsum and joints, i.e. the average temperature rise over the whole of the
assumed between 20 and 70 ◦ C, which is higher in comparison to non-exposed surface is limited to 140 K, and the maximum tem-
values assumed by other authors, takes into account the increased perature rise at any point of that surface does not exceed 180 K. The
heat transfer due to mass transfer of moisture. Shrinkage, cracks basic insulation value can be assessed by tests or FE thermal analy-
and ablation of gypsum boards as well as cracks in charred layer sis. For FE thermal analysis only the temperature criterion 140 ◦ C is
of timber panels increase the heat flux due to radiation and con- used. Further, the temperature of the layer at the beginning of the
vection. For simplicity, these effects were modelled by increasing analysis on the fire-exposed side as well as on the fire-unexposed
the thermal conductivity starting from 600 ◦ C. This approach has side is assumed as 20 ◦ C. The definition of the basic insulation value
already been used by other authors [2,22]. tins,0 is illustrated in Fig. 8.
The results of small-scale fire tests on single gypsum and timber Wall and floor assemblies with only one single layer are rarely
boards were used to calibrate the material properties assumed for used in buildings. Most wall and floor assemblies consist of
the FE thermal analysis (FEA). In this way, the thermal behaviour of assemblies having two or more layers (Fig. 1). The contribution to
the boards was not influenced by other layers. For example Fig. 7 the separating function of the construction of each layer (except
left compares the results of the FE thermal analysis with the results the last layer of the assembly on the fire-unexposed side) is
of the small-scale fire tests on single gypsum boards. The FE ther- mainly the protection of the following layers (Fig. 1). Therefore
mal analysis predicted the temperature development of the gyp- it seems more appropriate to introduce a basic protection value
sum boards with a thickness of 10, 12.5 and 15 mm satisfactorily. tprot,0 defined as the time until failure of the protective function.
For the gypsum board with a thickness of 18 mm the increase of the In analogy to the calculation for fire protective claddings of
temperature after complete dehydration based on the FE thermal load-bearing timber constructions according to EN 13501-2 the
analysis occurred about 5 min earlier than in comparison to the definition of the basic protection value tprot,0 is illustrated in Fig. 9.
fire test, i.e. the result of the FE thermal analysis was conservative. The testing method for fire protective claddings according to EN
A. Frangi et al. / Engineering Structures 32 (2010) 1184–1195 1189
Fig. 5. Temperature-dependent thermal conductivity (left) and specific heat (right) of gypsum used for the FE thermal analysis in this study as well as in previous studies.
The points given in the figure indicate measured values of the thermal conductivity and the specific heat.
Fig. 6. Temperature-dependent thermal conductivity (left) and specific heat (right) of timber used for the FE thermal analysis in this study as well as in previous studies.
The points given in the figure indicate measured values of the thermal conductivity and the specific heat.
13501-2 is performed with a particleboard with a thickness of Fig. 10 right shows the comparison of the basic insulation
19 mm backing the layer studied. The contribution of the cladding value tins,0 for gypsum boards measured in the small-scale fire
to the fire protection of the particleboard may be assumed to be tests and calculated according to the equation given in Table 3 as
satisfied if the average temperature rise over the whole exposed well as according to EN 1995-1-2 for all fire tests with gypsum
surface of the particleboard is limited to 250 K, and the maximum boards tested as a single board. It can be seen that the basic
temperature rise at any point of that surface does not exceed 270 K. insulation values calculated according to EN 1995-1-2 are in a
For FE thermal analysis only the temperature criterion 250 ◦ C is good agreement only with the results of fire tests with 10 mm
used. Further, the temperature of the layer at the beginning of the thick gypsum boards. By increasing the thickness of the board the
analysis on the fire-exposed side as well as on the fire-unexposed basic insulation values are underestimated, i.e. EN 1995-1-2 leads
side is assumed as 20 ◦ C (Fig. 9). to conservative results. Vice versa, the results based on the FE
Table 3 gives the equations for the calculation of the basic parametric study (new equation in Table 3) agree better with the
insulation value tins,0 as well as the basic protection value tprot,0 test results. Unlike EN 1995-1-2 the new equation assumes a non-
for different materials, that were systematically calculated using linear relationship between thickness of the board and the basic
numerical finite elements simulations and verified with fire insulation value.
tests [12,13]. For example Fig. 10 left shows the comparison of
the basic protection value tprot,0 for gypsum boards calculated 5.2. Position coefficients
according to the equation given in Table 3 as well as according to
EN 1995-1-2 with all results of the small-scale fire tests performed The position coefficient considers the position of the layer
with gypsum boards backed by timber board or timber frame. It within the assembly (in direction of the heat flux), because the
can be seen that the difference between the new equation and layers preceding and backing the layer considered have an influ-
EN 1995-1-2 is quite small and both agree well with the test ence on its fire behaviour. The physical meaning of the position
results. Unlike EN 1995-1-2 the new equation assumes a non- coefficient can be explained looking at the fire behaviour of a tim-
linear relationship between thickness of the board and the basic ber assembly with 3 layers as shown in Fig. 11. For simplicity, it
protection value. is assumed that each layer is of the same material, has the same
1190 A. Frangi et al. / Engineering Structures 32 (2010) 1184–1195
Fig. 7. Comparison between measured and calculated temperatures for different fire tests [13].
Fig. 8. Definition of the basic insulation value tins,0 using FE thermal analysis.
Fig. 9. Definition of the basic protection value tprot,0 using FE thermal analysis.
Fig. 10. Left: Comparison of the basic protection value tprot,0 for gypsum boards calculated according to the equation given in Table 3 as well as according to EN 1995-1-2
with all results of the small-scale fire tests performed with gypsum boards backed by timber board or timber frame. Right: Comparison of the basic insulation value tins,0 for
gypsum boards measured in the small-scale fire tests and calculated according to the equation given in Table 3 as well as according to EN 1995-1-2 for all gypsum boards
tested as single board.
A. Frangi et al. / Engineering Structures 32 (2010) 1184–1195 1191
(a) Start of fire: t = 0. (b) Second layer exposed to fire: t = t prot,1. (c) Third layer exposed to fire: t = t prot,1 + tprot,2.
Fig. 11. Temperature distribution at different times of timber assembly with 3 layers.
lower than the contribution of the first layer, i.e. tprot,2 < tprot,1 . The
position coefficient kpos,2 of the second layer can be described as the time when the layer considered is exposed directly to fire as
the ratio tprot,2 to tprot,0,2 and is <1.0. For the same physical reasons well as the material and thickness of the layer considered, while
EN 1995-1-2 assumes that after failure of a cladding, charring takes the influence of preheating is small. As the design model for the
place at an increased rate of initially unprotected surfaces [31]. verification of the separating function of timber assemblies is de-
The third layer is protected by the second layer. After failure veloped for ISO fire exposure the furnace temperature is a func-
of the protection by the second layer (temperature of 270 ◦ C tion
P of the sum of the protection values of the preceding layers (i.e.
at the interface between second and third layer) at time t = tprot,i−1 ). The thickness of the layer considered is expressed by
tprot,1 + tprot,2 , the third layer is directly exposed to fire. The its basic protection value tprot,0,i or basic insulation value tins,0,n .
third layer is the last layer in the assembly and takes care of the For example, Fig. 12 shows the position coefficient kpos,exp of gyp-
insulation function (Fig. 1). Therefore for this layer the temperature sum boards with different thickness as a function of the sum of the
criterion 140 ◦ C is applied and an insulation value (tins,3 ) has to protection
P values of the layers preceding the gypsum board (i.e.
be calculated. Because of the further increased temperature in the tprot,i−1 ). It can be seen
P that the position coefficient kpos,exp de-
fire compartment and the missing protective char layer as well creases with increasing tprot,i−1 , because the longer the gypsum
as the temperature rise criterion of 140 ◦ C instead of 250 ◦ C, the board is protected by the preceding layers, the higher the furnace
contribution of the third layer to the total fire resistance is lower temperature will be when exposed directly to the fire. Further, the
than the contribution of the first and second layer, i.e. tins,3 < position coefficient kpos,exp decreases with decreasing thickness of
tprot,2 < tprot,1 . The position coefficient kpos,3 of the third layer the gypsum board. For timber boards and insulation batts similar
can be described as the ratio tins,3 to tins,0,3 and is <1.0. The fire effects were observed. Thus, it was possible to determine the po-
resistance of the timber assembly can be calculated as the sum sition coefficient kpos,exp as a function of the sum ofP the protection
of the contributions from the different layers, i.e. tins = tprot,1 + values of the layers preceding the layer considered ( tprot,i−1 ) and
tprot,2 + tins,3 . the basic values of the layer considered (tprot,0,i and tins,0,n ), mak-
The influence of the layers preceding and backing the layer con- ing the calculation of the position coefficient kpos,exp easy for the
sidered was analysed separately. The position coefficient kpos,exp designer (Table 4).
considers the influence of the layers preceding the layer studied, The position coefficients kpos,exp given in Table 4 were
while the influence of the layer backing the layer studied is con- calculated assuming that the layers fall off when the temperature
sidered by kpos,unexp . The position coefficients were systematically of 270 ◦ C is reached on the fire-unexposed side of the layers.
calculated using numerical finite elements simulations. The results Fire tests showed that this assumption is conservative for gypsum
of the numerical FE simulations showed that the position coeffi- plasterboards of type F or type X (commonly used in North
cient kpos,exp is mainly influenced by the furnace temperature at America) and gypsum fibreboards [13]. Therefore the protection
1192 A. Frangi et al. / Engineering Structures 32 (2010) 1184–1195
Table 4
Position coefficients kpos,exp,i and kpos,exp,n . For rock and glass fibre insulation only the position coefficient kpos,exp,i for the protection value tprot,i is given as wall and floor
assemblies with the insulation as last layer of the assembly are rarely used in buildings.
Material Position coefficients kpos,exp,i and kpos,exp,n
q
tprot,0,i tprot,0,i
0.5 · tprot,i−1 >
P
P
tprot,i−1
for 2
q
tprot,0,i tprot,0,i
0.5 · tprot,i−1 >
P
P
tprot,i−1
for 2
h i(0.75−0.002·ρi )
t tprot,0,i
(0.001 · ρi + 0.27) · P prot,0,i tprot,i−1 >
P
tprot,i−1
for 4
Table 5
Correction times 1ti and 1tn of protection and insulation values tprot,i and tins,n of layers protected by gypsum plasterboards of type F or type X as well as gypsum fibreboards.
or insulation values of layers protected by gypsum plasterboards board has even a positive effect on the fire behaviour of the
of type F or type X as well as gypsum fibreboards can be increased solid timber panel). Table 6 gives the equations for the position
using a correction time 1t (Eqs. (2) and (3)). Table 5 gives coefficient kpos,unexp for different materials backed by insulating
the correction times 1t that were systemically calculated using batts, that were systematically calculated using numerical finite
numerical finite element simulations. For the calculation of 1t it elements simulations. For backing layers made of timber or
was assumed that for floor assemblies the gypsum plasterboards gypsum kpos,unexp,i = 1.0 is assumed as a simplification.
type F or type X and the gypsum fibreboards fall off when the
temperature of 400 ◦ C is reached on the fire-unexposed side of
5.3. Void cavities
the board, while for wall assemblies the falling-off of the boards
was assumed by a temperature of 600 ◦ C. The temperature criteria
were based on the evaluation of a large number of full-scale fire The influence of void cavities between two layers is considered
tests found in the literature [29,32–37]. in the design method by modifying the position coefficient kpos,exp
The influence of the layer backing the layer studied is for the layer on the fire-unexposed side of the cavity and the
considered by the position coefficient kpos,unexp . Results of fire position coefficient kpos,unexp for the layer on the fire-exposed
tests supported by finite element thermal simulations showed side of the cavity. The results of the fire tests supported by
that the influence of the layer backing the layer considered is finite element thermal simulations showed that the temperature
small if the backing layer is made of gypsum or timber, while development measured behind claddings made of gypsum or
insulating batts backing the layer considered caused the layer to timber backed with the void cavity followed up roughly the same
heat up more rapidly, reducing the protection time of the layer. development as for claddings backed with insulation (see for
Further, it was observed that the thickness hi+1 of the layer backing example T15, T25 and T35 in Fig. 3). Thus, for simplicity, for
the layer considered does not influence the position coefficient claddings made of gypsum or timber on the fire-exposed side of
kpos,unexp,i for hi+1 ≥ 30 mm. For example Fig. 13 shows the the cavity the same position coefficients kpos,unexp as for insulation
calculated position coefficient kpos,unexp,i for a 10 mm thick solid as baking layer are used (Table 7), while the influence of the void
timber panel backed by layers made of different material and cavity on insulations on the fire-exposed side of the cavity can be
thickness. There is a strong influence of insulating batts on the neglected (i.e. kpos,unexp = 1.0, Table 7).
fire behaviour of the solid timber panel (kpos,unexp,i < 0.6), while Results of finite element thermal simulations showed that the
the influence of backing layers made of timber or gypsum is positive influence of the void cavity on the thermal behaviour of
small (for gypsum board kpos,unexp is about 1.1, i.e. the gypsum claddings or insulations on the fire-unexposed side of the cavity
A. Frangi et al. / Engineering Structures 32 (2010) 1184–1195 1193
Fig. 13. Position coefficient kpos,unexp for a 10 mm thick solid timber panel (STP) backed by layers made of different materials and thicknesses (STP: Solid timber panel, PB:
Particleboard; GP: Gypsum plasterboard; RF30: Rock fibre insulation with a density of 30 kg/m3 ; RF120: Rock fibre insulation with a density of 120 kg/m3 ).
Table 6 Table 7
Position coefficient kpos,unexp,i . Modification of position coefficients kpos,exp and kpos,unexp in the case of void cavities.
Material of the layer kpos,unexp,i for kpos,unexp,i for layers backed by Material Layer on the Layer on the fire-unexposed side
considered layers backed by insulation fire-exposed side of of the cavity
cladding made of the cavity
gypsum or timber
Cladding kpos,unexp,i 3 × 1ti according to
Gypsum 1.0 0.5 · h0i .15 according to 1.6 × kpos,exp,i Table 5
plasterboard, gypsum Table 6, column 3 according to
fibreboard Insulation kpos,unexp,i = 1.0 Table 4 1ti according to Table 5
Solid timber panel 1.0 0.35 · h0i .21
Particleboard 1.0 0.41 · h0i .18
OSB, plywood 1.0 0.5 · h0i .15
(0.001·ρi +0.08)
Rock fibre 1.0 0.18 · hi
h2i
Glass fibre 1.0 0.01 · hi − 30 000
+ ρi0.09 − 1.3
With hi : Thickness of the layer considered (mm), ρi : Density of the layer considered
(kg/m3 ).
Table 8
Joint coefficients kj,i and kj,n .
[22] Thomas G. Thermal properties of gypsum plasterboards at high temperatures. [31] König J. Structural fire design according to Eurocode 5—Design rules and their
Fire Mater 2002;26:37–45. background. Fire Mater 2005;29:147–63.
[23] Källsner B, König J. Thermal and mechanical properties of timber and some [32] Kodur VKR, Sultan MA, Denham EMA. Temperature measurement in full-scale
other materials used in light timber frame construction. In: Proc. Meeting 33 of wood stud shear walls. Internal report no. 729. Ottawa: Institute for Research
the working commission W18-timber structures. CIB. Int. Council for Research in Construction, National Research Council of Canada; 1996.
and Innovation. Paper no. CIB-W18/33-16-3. 2000. [33] Kodur VKR, Sultan MA, Latour JC, Leroux P, Monette RC. Experimental studies
[24] Mehaffey JR, Cuerrier P, Carisse G. A model for predicting heat transfer on the fire resistance of load-bearing steel stud walls. Internal report no. 833.
through gypsum-boards/woos-stud walls exposed to fire. Fire Mater 1994;18: Ottawa: Institute for Research in Construction, National Research Council of
297–305. Canada; 2003.
[25] Bénichou N, Sultan MA, MacCallum C, Hum J. Thermal properties of wood, [34] Sultan MA, Séguin YP, Leroux P. Results of fire resistance tests on full-scale
gypsum and insulation at elevated temperatures. Internal report no. 710. floor assemblies. Internal report no. 764. Ottawa: Institute for Research in
Ottawa: Institute for Research in Construction, National Research Council of Construction, National Research Council of Canada; 1998.
Canada; 2001. [35] Sultan MA, Lougheed GD. Results of fire resistance tests on fullscale gypsum
[26] Harmathy TZ. Properties of building materials. In: The SFPE handbook board wall assemblies. Internal report no. 833. Ottawa: Institute for Research
of fire protection engineering. 2nd ed. Boston: Society of Fire Protection in Construction, National Research Council of Canada; 2002.
Engineers/National Fire Protection Association; 1988. [36] Sultan MA. Fire resistance of wood joist floor assemblies. Fire Technol 2008;
[27] König J, Norén J. Timber frame assemblies exposed to standard and parametric 44:383–417.
fires—Part 1: Fire tests. Trätek. Rapport I 9702015. Stockholm; 1997. [37] Sultan MA. Fall-off of gypsum plasterboard in fire. In: Proceedings of the fifth
[28] König J, Walleij L. One-dimensional charring of timber exposed to standard and international conference structures in fire. 2008.
parametric fires in initially unprotected and post-protection situation. Trätek. [38] Richardson LR, Batista M. Fire resistance of timber decking for heavy timber
Rapport I 9908029. Stockholm; 1999. construction. Fire Mater 2001;25:21–9.
[29] König J, Rydholm D. Small-scale fire tests of heavy timber components. Trätek. [39] König J. Fire resistance of timber joists and load bearing wall frames. Trätek.
Rapport P 0310036. Stockholm; 2003. Rapport I 99412071. Stockholm; 1995.
[30] Frangi A, Erchinger C, Fontana M. Charring model for timber frame floor [40] Collier PCR, Buchanan AH. Fire resistance of lightweight timber framed walls.
assemblies with void cavities. Fire Safety J 2008;43:551–64. Fire Technol 2002;38:125–45.