Professional Documents
Culture Documents
BAILEY TOH (2007) - Behaviour of Concrete Floor Slabs at Ambient and Elevated Temperatures
BAILEY TOH (2007) - Behaviour of Concrete Floor Slabs at Ambient and Elevated Temperatures
Abstract
This paper presents the latest developments of a simple design method used to predict the membrane action of unrestrained concrete,
or concrete and steel composite, floors under fire conditions. The developments include the refinement of the assumed in-plane stress
pattern and a prediction of when concrete crushing occurs in the proximity of the corners of the slab. The design method is compared
against 44 small-scale tests carried out at ambient and elevated temperatures on horizontally unrestrained slabs with an aspect ratio of
1.0 or 1.55. The slabs were reinforced with either mild steel or stainless-steel welded mesh of different grades, ductility, sizes and bar
spacing. Both the ambient and fire tests highlighted the occurrence of membrane action, either supporting loads higher than the
theoretical yield-line load in the case of the ambient tests, or reaching higher failure temperatures compared to those calculated based on
yield-line theory. Comparison between the developed simple design approach and test results showed good correlation both at ambient
and elevated temperatures.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Ambient tests; Concrete slabs; Critical temperature; Fire tests; Membrane action; Simple design method
0379-7112/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.firesaf.2006.11.009
ARTICLE IN PRESS
426 C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436
2. Ambient tests
Fig. 1. Use of membrane action to remove fire protection from secondary A series of 22 small-scale reinforced concrete slabs were
beams. tested at ambient temperature (Fig. 2) at the University
ARTICLE IN PRESS
C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436 427
of Manchester from November 2004 to July 2005. between the clamps and slabs, allowing free horizontal
The slabs had a size of 1.2 m 1.2 m or 1.8 m 1.2 m movement. There was no horizontal restraint provided to
with a target thickness of 20 mm and a concrete cover of the slab’s perimeter. Due to the large displacements
5 mm. The slabs were supported vertically on 50 mm wide witnessed in the tests, it was observed that the slab was
angles, resulting in a clear span of 1.1 m 1.1 m or supported off the edge of the angles leading to the
1.7 m 1.1 m, giving an aspect ratio of 1.0 or 1.55. The assumption that the span of the slab could be taken as
corners of the slabs were lightly clamped, with rollers the clear span.
The proportions of the concrete mix comprised 44%
Granno aggregate (maximum size 6 mm), 30% sand, 20%
Reaction frame Support frame Concrete slab cement and 6% water, by weight. The reinforcement
consisted of either mild steel or stainless-steel welded wire
mesh of different grades, ductility, diameters and bar
spacing. Tables 1 and 2 summarise the details of the tests
together with the calculated theoretical yield-line loads P
and recorded maximum test loads Ptest.
Air bag
All the slabs supported a load well above the theoretical
yield-line load showing that membrane action occurred.
Table 1 shows that the value of Ptest (which includes the
Load cells self-weight of the slabs) ranged from 1.45 to 2.44 above the
yield-line load P for the slabs with mild steel mesh. For the
slabs with stainless-steel mesh, the value of Ptest/P, referred
to hereinafter as the enhancement factor e, ranged from
Fig. 2. Experimental set-up for ambient tests. 1.44 to 3.66, as shown in Table 2.
Table 1
Details of slab tests with mild steel mesh at ambient temperature
Test Dimensions Wires in long span Wires in short span Wire fcu P Ptest Ptest =P D20 Plimit Plimit =Ptest
L l H (mm) spacing (MPa) (kN/m2) (kN/m2) (mm) (kN/m2)
+ fy/fu Ductility + fy/fu Ductility (mm)
(mm) (MPa) (%) (mm) (MPa) (%)
M1 1700 1100 18.2 2.42 732/756 1.49 2.42 757/777 3.26 50.8 41.3 8.52 20.7 2.44 44.0 15.5 0.74
M2 1100 1100 19.1 2.42 732/756 1.49 2.42 757/777 3.26 50.8 38 13.8 27.0 1.96 28.5 20.3 0.75
M3 1700 1100 22.0 1.53 451/487 6.39 1.49 454/495 8.98 25.4 35.3 6.35 12.3 1.93 34.5 9.13 0.74
M4 1100 1100 20.1 1.53 451/487 6.39 1.49 454/495 8.98 25.4 35.3 8.17 18.3 2.24 22.3 11.9 0.65
M5 1700 1100 18.9 1.47 406/500 9.89 1.48 435/473 9.87 12.7 37.9 8.69 17.9 2.06 32.8 12.7 0.71
M6 1100 1100 21.6 1.47 406/500 9.89 1.48 435/473 9.87 12.7 38.6 15.7 27.03 1.72 21.2 21.2 0.78
M7 1700 1100 20.4 0.84 599/653 3.30 0.85 604/679 2.91 12.7 41.6 5.11 8.65 1.69 39.8 7.68 0.89
M8 1100 1100 19.0 0.84 599/653 3.30 0.85 604/679 2.91 12.7 42.9 6.68 10.7 1.60 25.8 10.1 0.94
M9 1700 1100 22.0 0.66 450/470 1.30 0.68 402/454 3.93 6.35 37.6 5.07 7.35 1.45 34.5 7.16 0.97
M10 1100 1100 19.7 0.66 450/470 1.30 0.68 402/454 3.93 6.35 37.3 6.36 9.89 1.56 22.3 9.13 0.92
Table 2
Details of slab tests with stainless steel at ambient temperature
Test Dimensions Wires in long span Wires in short span Wire fcu P Ptest Ptest =P D20 Plimit Plimit =Ptest
L l H (mm) spacing (MPa) (kN/m2) (kN/m2) (mm) (kN/m2)
(mm)
+ fy/fu Ductility + fy/fu Ductility
(mm) (MPa) (%) (mm) (MPa) (%)
S1 1700 1100 19.0 2.99 639/888 31 3 614/859 33 50.8 40.6 10.6 17.1 1.61 41.1 16.5 0.96
S2 1100 1100 20.4 2.99 639/888 31 3 614/859 33 50.8 41.2 18.7 26.8 1.44 26.6 25.5 0.95
S3 1700 1100 21.0 2.51 569/854 38 2.51 555/848 38 50.8 50 9.20 17.3 1.88 38.8 15.7 0.91
S4 1100 1100 19.0 2.51 569/854 38 2.51 555/848 38 50.8 50.7 11.6 22.8 1.96 25.1 20.7 0.91
S5 1700 1100 17.6 1.55 344/732 37 1.53 447/739 34 25.4 49.8 4.23 13.4 3.15 30.2 7.2 0.54
S6 1100 1100 20.6 1.53 344/732 37 1.55 447/739 34 25.4 49.8 7.76 26.8 3.45 22.2 12.7 0.48
S7 1700 1100 20.5 1.58 265/578 41 1.58 271/586 41 12.7 41.9 7.37 19.5 2.64 26.5 11.3 0.58
S8 1100 1100 19.3 1.58 265/578 41 1.58 271/586 41 12.7 43 9.84 36.0 3.66 17.1 15.7 0.44
S9 1700 1100 19.7 0.98 280/678 54 0.98 301/707 59 8 37.1 4.94 14.3 2.89 27.2 7.84 0.55
S10 1100 1100 18.8 0.98 280/678 54 0.98 301/707 59 8 37.2 6.66 20.2 3.03 17.6 10.8 0.54
S11 1700 1100 18.9 0.78 349/744 56 0.75 580/815 39 6.35 45.7 5.97 16.2 2.72 30.3 8.74 0.54
S12 1100 1100 22.3 0.78 349/744 56 0.75 580/815 39 6.35 48.2 10.5 31.5 3.02 25.3 17.2 0.55
ARTICLE IN PRESS
428 C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436
Observations from the slab tests with mild steel mesh load divided by the yield-line load at ambient temperature
showed two distinct modes of failure, which were found to (Ptest/P) that could be achieved by using the dead weights
be dependent on the reinforcement ratio, aspect ratio and was quite restrictive, due to the physical size of the slab’s
ductility of the reinforcement. Considering a constant footprint and the need to ensure enough space between the
aspect ratio, it was found that the first mode of failure, for weights so that they did not clash with each other as the
lightly reinforced slabs, comprised fracture of the rein- slab deflected downwards. As a result, the actual values of
forcement across the shorter span of the slabs (Fig. 3a). the load level (Ptest/P) varied significantly in the tests, from
The other mode of failure, for heavily reinforced slabs, 0.29 to 1.33, as shown in Tables 3 and 4. The strategy was
comprised compressive failure at the corners of the to try and apply, within the physical constraints, the
slabs (Fig. 3b). For the high ductile stainless-steel slabs, maximum achievable load level, but less than the unity, in
fracture of the reinforcement did not occur and failure each test. However, in the case of Test MF7, the load level
was by crushing of the concrete at the corners of the was accidentally applied beyond unity due to human error.
slab. An electric kiln, located beneath the slabs, provided a
heating rate of 300 1C/h, up to a maximum temperature of
3. Fire tests 1000 1C, which was maintained for the duration of the test.
The temperature inside the kiln, the temperature of bottom
A similar experimental programme at elevated tempera- and top surfaces of the slab, and the temperature of
tures, consisting of 22 slabs, with the same geometry, rein- reinforcement at three locations at the centre of the slab,
forcement and support conditions, adopted in the ambient were recorded during each test. Fig. 5 shows the typical
tests, was carried out from September 2005 to June 2006. temperature recording for a test slab.
The details of the tests are given in Tables 3 and 4. All the tests failed in a similar manner (Fig. 6) showing
All tests were carried out under a transient heating state, the classic membrane action [12] behaviour of horizontally
with a predefined working load (Ptest) uniformly applied on unrestrained slabs. Failure occurred by fracture of
the top surface of the slab by using dead weights, as shown reinforcement across the shorter span of the rectangular
in Fig. 4. Tables 3 and 4 show the values of the theoretical slabs or across one of the spans of the square slabs. Fig. 6
yield-line load at ambient temperature (P) and the applied shows clearly the mode of failure with the longer edges of
test load (Ptest) for the mild steel and stainless-steel slabs, the slab being pulled inwards as the slab deflects down-
respectively. The load ratio, defined as the applied test wards, which relives the reinforcement strains in the shorter
Fig. 3. Two typical modes of failure for test slabs at ambient temperature: (a) fracture of reinforcement along short span and (b) compressive concrete
failure at corners.
Table 3
Details of slab tests with mild steel at elevated temperatures
Test Dimensions Wires in long span Wires in short span Wire fcu P Ptest Ptest =P Tmesh DT Tpred T pred =T mesh
L l H (mm) spacing (MPa) (kN/m2) (kN/m2) (1C) based
+ fy/fu Ductility + fy/fu Ductility (mm) on DT
(mm) (MPa) (%) (mm) (MPa) (%) (1C)
MF1 1700 1100 19.7 2.43 695/727 2.86 2.41 722/742 3.46 50.8 43.2 9.52 5.280 0.55 764.9 54.38 641 0.84
MF2 1100 1100 23.1 2.41 684/736 3.19 2.43 780/795 1.07 50.8 43.3 19.00 5.519 0.29 747.1 37.33 679 0.91
MF3 1700 1100 19.0 1.54 336/404 18.76 1.54 349/420 12.57 25.4 39.1 4.07 3.655 0.90 727 41.74 543 0.75
MF4 1100 1100 19.8 1.54 336/404 18.76 1.54 349/420 12.57 25.4 39 6.36 5.429 0.85 700.1 30.74 537 0.77
MF5 1700 1100 20.1 1.51 402/463 12.84 1.52 467/498 6.53 12.7 37.1 13.47 5.280 0.39 721.9 43.88 647 0.90
MF6 1100 1100 19.5 1.51 402/463 12.84 1.52 467/498 6.53 12.7 38.5 14.35 7.900 0.55 782.4 32.73 620 0.79
MF7 1700 1100 18.8 0.85 405/444 5.40 0.83 470/486 3.65 12.7 43.8 3.36 4.460 1.33 556.5 44.95 439 0.79
MF8 1100 1100 20.9 0.85 405/444 5.40 0.83 470/486 3.65 12.7 43.5 5.65 4.646 0.82 653.9 32.13 545 0.83
MF9 1700 1100 21.6 0.71 371/382 3.40 0.68 449/455 1.44 6.35 47.1 5.23 3.697 0.71 652.1 41.82 568 0.87
MF10 1100 1100 21.0 0.71 371/382 3.40 0.68 449/455 1.44 6.35 40.4 7.18 5.494 0.77 686.0 31.07 555 0.81
ARTICLE IN PRESS
C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436 429
Table 4
Details of slab tests with stainless steel at elevated temperatures
Test Dimensions Wires in long span Wires in short span Wire fcu (MPa P Ptest Ptest =P Tmesh DT Tpred T pred =T mesh
L l H (mm) spacing (kN/m2) (kN/m2) (1C) based
+ fy/fu Ductility + fy/fu Ductility (mm) on DT
(mm) (MPa) (%) (mm) (MPa) (%) (1C)
SF1 1700 1100 21.1 2.99 589/891 65.0 2.99 570/885 65.4 50.8 42.7 12.32 5.31 0.43 892.3 50.21 849 0.95
SF2 1100 1100 20.5 2.99 639/888 31.1 3 614/859 32.6 50.8 38.5 18.47 7.92 0.43 879.0 37.66 857 0.97
SF3 1700 1100 18.6 2.48 608/874 47.8 2.5 583/834 45.5 50.8 40.6 7.56 5.25 0.69 870.6 52.29 772 0.89
SF4 1100 1100 18.7 2.51 569/854 38.2 2.51 555/848 37.6 50.8 45.7 11.17 5.41 0.48 4825a 37.23 845 –
SF5 1700 1100 17.9 1.58 343/725 69.0 1.55 460/766 47.8 25.4 40.0 4.50 3.64 0.81 886.6 42.79 699 0.79
SF6 1100 1100 18.9 1.53 344/732 37.0 1.53 447/739 34.5 25.4 46.5 6.77 5.41 0.80 897.6 31.52 698 0.78
SF7 1700 1100 18.4 1.56 256/585 78.1 1.57 316/621 40.0 12.7 35.2 6.50 5.24 0.81 887.6 38.34 681 0.77
SF8 1100 1100 22.2 1.58 265/578 41.3 1.58 271/586 41.1 12.7 41.2 12.19 7.97 0.65 878.9 27.34 731 0.83
SF9 1700 1100 19.8 0.98 280/678 53.9 0.98 301/707 59.2 8 40.4 5.00 3.68 0.74 881.6 38.66 715 0.81
SF10 1100 1100 22.6 0.98 280/678 53.9 0.98 301/707 59.2 8 39.7 8.74 7.98 0.91 874.9 27.64 602 0.69
SF11 1700 1100 19.8 0.78 349/744 56.0 0.75 580/815 38.6 6.35 40.8 6.38 5.29 0.83 826.4 41.83 606 0.73
SF12 1100 1100 23.3 0.78 349/744 56.0 0.75 580/815 38.6 6.35 41.7 11.07 8.00 0.72 839.9 29.39 629 0.75
a
Test stopped before failure.
1000
900 Test SF10 It was observed that the mild steel slabs normally
800 collapsed in a more sudden manner, whereas the collapse
Temperature (°C)
span until fracture of the reinforcement in the longer span 4. Simplified method at ambient temperature
occurs. Previous work [12,13] has shown that this failure
mode is characteristic of tensile membrane action occurring Based on the ambient and elevated test results, the
in the central region of the slab surrounded by a balancing original design method [12] was extended to include a
compressive membrane ring. No compression failure, as more precise in-plane stress pattern and to include
identified in the ambient tests (Fig. 3b), was observed in possible crushing of the concrete at the corners of the
any of the fire tests. slabs. The method calculates the enhancement (e) above
ARTICLE IN PRESS
430 C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436
the theoretical yield-line load due to membrane action. together with affinity rules, and is given by
As a slab deflects, it will be subject to membrane action, 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
provided all the perimeter edges are vertically supported n ¼ pffiffiffi 2 3ma2 þ 1 1 , (1)
2 ma
and fracture of the reinforcement does not occur. Since
membrane action is dependent on the in-plane forces, where a is the aspect ratio of the slab (L/l) and m is the
an increase in vertical displacement will result in an coefficient of orthotropy, which defines the ratio of the
increase in the load carrying capacity. In the case of yield moment capacity of the slab in the orthogonal
horizontally unrestrained slabs, membrane action com- directions (refer Fig. 7).
prises tensile membrane action in the central plan area of For a square slab, with isotropic or orthotropic
the slab surrounded by a balancing ring of compressive reinforcement, the ‘shorter’ span of the slab is defined by
membrane action. As the method is based on an enhance- the span with the smaller moment capacity so that the
ment due to membrane action (which is governed by the coefficient of orthotropy (m) is always less than, or equal to,
change of geometry) above the yield-line load (which is unity. Provided the coefficient of orthotropy (m) is less
based on rigid-perfectly plastic slab behaviour), the overall than, or equal to, unity, Eq. (1) will result in a value of n
method can be classed as plastic design with change of less than, or equal to, 0.5 leading to a valid yield-line
geometry. pattern.
The assumptions adopted in the design method are
summarised as follows:
4.2. First stress pattern
1. Based on a predefined yield-line pattern, the in-plane
For the failure mode of reinforcement fracture across the
forces increase with increase in vertical displacement.
shorter span, the in-plane stress distribution is defined by
2. The self-equilibrium in-plane forces comprise compres-
the constants k and b (Fig. 7a), which are calculated using
sive membrane action around the perimeter of the slab
and tensile membrane action in the central area of the 4na2 ð1 2nÞ
k¼ þ 1, (2)
slab. 4n2 a2 þ 1
3. Based on the in-plane stress distributions (as shown
in Fig. 7), the in-plane forces along the rigid plates 1 1 þ 2f 2 þ 2m
b¼ ð1 2mÞ pf 1 , (3)
are defined by the constants ‘k’ and ‘b’ or ‘v’ depen- KðA þ BÞ 3
ding on the stress pattern. The 1st stress pattern
in which
in Fig. 7 governs provided the value of ‘b’ is less
than f1 (f1 ¼ fu/fy). If ‘b’ is greater than f1 then the A ¼ ð1 4n2 Þa2 ,
2nd stress pattern must be used with the value of ‘b’ set 2k
to f1. B¼ ½ð2naÞ2 þ 1.
3
4. When the applied load on the slab exceeds a certain
limit, the fracture of the reinforcement across the central Considering a triangular compressive stress block (ml) at
crack will initiate at the slab’s centre and gradually both ends of the assumed fracture line, with the full slab
propagate towards its edges. For simplicity, the tensile depth H being in compression and the contribution of the
stress of reinforcement across the crack is assumed to reinforcement in compression ignored, the width factor m
vary linearly from its yield stress (fy) near the edge of the can be determined from equilibrium of the in-plane
slab to its ultimate stress (fu ¼ f2 fy) at the slab centre, horizontal forces across the full central fracture line as
as shown in Fig. 8. follows:
5. Two modes of failure are considered, comprising
1 0:8f cu H 1
fracture of reinforcement across the shorter span of m¼ 1þ . (4)
2 ð1 þ f 2 ÞT 0
the slab (as explained in assumption 4 above) and
compression failure of concrete in the corners of the By substituting the value of m from Eq. (4) into Eq. (3),
slab. The compression failure considers both membrane the constant b can be calculated. If the constant b is less
and bending action at the corner of the slabs. than or equal to f1, the 1st stress pattern (Fig. 7a) is valid
and the load carrying capacity for a given displacement can
The basic equations for the simplified design method are be calculated. However, if the constant b is greater than f1,
presented here. then the 1st stress pattern becomes invalid and the 2nd
stress pattern (Fig. 7b), which is discussed in Section 4.3,
4.1. Yield-line pattern must be used.
Assuming that the 1st stress pattern is valid, the load
Considering a rectangular slab of L l (length width), carrying capacity can be calculated with reference to
the in-plane stress distributions are based on an assumed Fig. 7a and Eqs. (2) and (3). The load carrying contri-
yield-line pattern determined by the factor n (Fig. 7), which bution from the membrane forces, and the increase in
can be calculated using general yield-line theory [14] bending resistance due to the in-plane compressive forces,
ARTICLE IN PRESS
C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436 431
L
kbKT0 nL
C C
4 1
Fracture
forming at
kLd S centre of slab S 1
1+k 3 2
Ld T2 T2
where 1+k
bKT0 T1
Ld = (nL) 2 + l 2 / 4
In-plane forces: Reinforcement: 2 l
T1 = bKT0 (1 − 2n) L Yield force = T0
bKT0 Ultimate force = f2T0
T2 = Ld Yield moment = M0
2 (1 + k)
Yield force = KT0
k2 bKT0 Ultimate force = f1KT0
C= Ld
2 (1 + k) Yield moment = M0 5
L
kbKT0 nL
C C
4 1
S Fracture
forming at
k(1 − v) Ld centre of slab S
1+k 1
3 2
(1 − v) Ld T2 T2
Fig. 7. In-plane stress distribution patterns shown along assumed yield lines: (a) 1st stress pattern for bof1 and (b) 2nd stress pattern for b ¼ f1.
are represented separately by enhancement factors above 2. Enhancement due to the effect of in-plane forces on
the theoretical yield-line load as follows: the bending capacity for Elements 1 and 2 (Fig. 7a) is
given by
1. Enhancement due to membrane forces for Elements 1
and 2 (Fig. 7a) is given by
a1 b b b2
e1b ¼ 2n 1 þ ðk 1Þ 1 ðk2 k þ 1Þ
4b w nð2 kÞ 2 3
e1m ¼ 1 2n þ ,
3 þ g1 d 1 3 þ ð1 2nÞð1 a1 b b1 b2 Þ,
2bK w 2 k a2 bK b b2 K 2
e2m ¼ . ð5Þ e2b ¼ 1 þ ðk 1Þ 2 ðk k þ 1Þ, ð6Þ
3 þ g2 d 2 3 2 3
ARTICLE IN PRESS
432 C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436
0.5L 0.5L
nL n
kbKT0 kbKT0
ml C
T0 C 1 T0 1
0.5T3 0.5T3
l
-ml 0.5T3 0.5T3 S
2 S
z z
T2
f2T0
f2T0 T2'
T2
0.5T bKT0 0.5T bKT0
1st Stress Pattern 2nd Stress Pattern
2 + f2 l
The in-plane lever arm z is given by z = ( − ml)
3 (1 + f2) 2
20 based on
b¼ 0:67f cu 0:45 T0 . w = 160 mm
kKT 0 2 2 120
80
(14) 15 40
0
4.5. Maximum allowable vertical displacement 0 100 200 300 400 500 600 700 800 900
Reinforcement temperature (°C)
The simple method presented so far in this paper cannot 160
predict the maximum allowable vertical displacement at 140
which actual fracture of the reinforcement or crushing of
Displacement (mm)
factors over the range of temperatures specified in various values of w. The calculated average reinforce-
Step 1. ment temperature–displacement response can then be
4. Compute the load carrying capacity by multiplying the compared with the test results, as illustrated in the lower
yield-line loads from Step 2 by the enhancement factors graph in Fig. 10.
from Step 3. Plotting the obtained load carrying capacities
versus the temperatures for each value of w will generate When considering the temperature–displacement re-
the four dash lines in the upper graph in Fig. 10. sponse in Fig. 10, it is worth noting that the simple
5. For a given constant working load, extract the method is based on rigid plastic behaviour with change of
corresponding temperatures for the yield-line load and geometry. Therefore, the method does not predict any
80 80
40 40
0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800
Average reinforcement temperature (°C) Average reinforcement temperature (°C)
160
Test MF6 120 Test MF8
Simple MF6 Simple MF8
120
80
80
40
40
0 0
0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700
Average reinforcement temperature (°C) Average reinforcement temperature (°C)
Tests MF9 and MF10
160
Test MF9
Simple MF9
Central deflection (mm)
80
40
0
0 100 200 300 400 500 600 700
Average reinforcement temperature (°C)
Fig. 11. Predicted deflection–temperature relationships for slabs with mild steel mesh at elevated temperatures.
ARTICLE IN PRESS
C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436 435
displacement until the theoretical yield-line load is reached The above design procedure can easily be conducted using a
(at 180 1C for Test SF12). The simple method then forms a standard spreadsheet. However, using Eqs. (2)–(14) will only
displacement–temperature failure envelop based on the provide an average reinforcement temperature–displacement
reduction in strength of the material and an increase in response as shown in Fig. 10. To calculate the failure point at
vertical displacement, which is required to mobilise the which reinforcement fracture occurs a maximum allowable
membrane effects. For a good comparison the test curve displacement needs to be specified. The original design
and predicted failure envelop should gradually merge approach [10] extended the maximum allowable displacement
together. (D20) at ambient temperature (Eq. (14)) to included the effects
160 160
Test SF2 Test SF4
Simple SF2 Simple SF4
120 120
80 80
40 40
0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Average reinforcement temperature (°C) Average reinforcement temperature (°C)
Tests SF5 and SF6 Tests SF7 and SF8
200 200
Test SF6 Test SF7
Simple SF6 Simple SF7
Central deflection (mm)
Central deflection (mm)
160 160
Test SF5 Test SF8
Simple SF5 Simple SF8
120 120
80 80
40 40
0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Average reinforcement temperature (°C) Average reinforcement temperature (°C)
80 80
60 60
40 40
20 20
0 0
0 100 200 300 400 500 600 700 800 900 0 100 200 300 400 500 600 700 800 900
Average reinforcement temperature (°C) Average reinforcement temperature (°C)
Fig. 12. Predicted deflection–temperature relationships for slabs with stainless-steel mesh at elevated temperatures.
ARTICLE IN PRESS
436 C.G. Bailey, W.S. Toh / Fire Safety Journal 42 (2007) 425–436
of thermal curvature to estimate the maximum allowable a simple design method for unrestrained concrete slabs at
displacement in fire conditions (DT) as follows [10]: ambient and elevated temperatures has been presented. In
s
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
addition, details of 44 small-scale tests on horizontally
aðT Bot T Top Þl 2 0:5f y 3L2 unrestrained concrete slabs at ambient and elevated
DT ¼ þ , (16)
19:2H E 8 temperatures are briefly discussed. The tests comprised of
mild steel or stainless-steel mesh reinforcement with
where a is the coefficient of thermal expansion of concrete
varying bar diameter, spacing and ductility. Observations
which can be taken as 18 106 for normal weight
from the tests showed that two modes of failure were
concrete. The difference between the temperature at the
observed at ambient temperature, comprising fracture of
bottom and top of the slab (TBotTTop) can be taken as
the reinforcement or crushing of the concrete at the corners
250 1C for the small-scale slabs considered, in accordance
of the slab. At elevated temperatures all the tests failed by
with the test results. The maximum allowable displacement
fracture of the reinforcement.
(DT) and corresponding critical temperature Tpred is shown
When compared against the test results, the developed
in Tables 3 and 4 for all slabs.
simple design method produced good predications both at
ambient and elevated temperatures. In some cases, where
6. Comparison between test results and simple analyses
the load ratio was low, the prediction for the response at
elevated temperatures was not so good. However, based on
When compared against the 22 ambient tests, the
the maximum allowable displacement, the simplified
simplified design method tends to give conservative
method always produced conservative predictions of the
predictions when estimating the load capacity (Plimit) based
maximum load capacity at ambient temperature and
on the maximum allowable displacement D20 given by Eq.
failure temperatures under fire conditions.
(15). As shown in Tables 1 and 2, it was found that the
ratio of the predicted allowable load to the maximum test
References
load (Plimit/Ptest) ranged from 0.97 to 0.65 for the mild steel
slabs and 0.96–0.44 for the stainless-steel slabs. [1] Bailey CG, Lennon T, Moore DB. The behaviour of full-scale steel
Comparison of the predicted displacement–temperature framed buildings subjected to compartment fires. Struct Eng 1999;
response (based on the principle of rigid plastic behaviour 77(8):15–21.
with change of geometry) and the test results, of the 10 [2] O’Conner MA, Kirby BR, Martin DM. Behaviour of a multi-storey
mild steel reinforced slabs at elevated temperatures, is composite steel framed building in fire. Struct Eng 2003;81(2):27–36.
[3] Fire resistance of concrete structures. Report of a Joint Committee of
shown in Fig. 11. The predictions for MF2, MF3 and MF6 the Institution of Structural Engineers and the Concrete Society, The
do not correlate very well with the test results. It was found Institution of Structural Engineers, 1975.
that slabs MF2 and MF6 had a low load ratio (Ptest/P), as [4] BSEN1992-1-2, Eurocode 2, Design of concrete structures. Part 1.2.
shown in Table 3, and further investigation is required to General rules. Structural fire design. London: British Standards
understand the poor correlation. The predicted displace- Institution; 2004.
[5] BSEN1994-1-2, Eurocode 4, Design of composite steel and concrete
ment–temperature response and the test results, for the structures. Part 1.2. General rules. Structural fire design. London:
stainless-steel reinforced slabs, are shown in Fig. 12. In British Standards Institution; 2006.
general, the predictions are very good except for slabs [6] Huang Z, Burgess IW, Plank RJ. Modelling of six full-scale fire tests
SF1–SF4, which again had a low load ratio (refer Table 4). on a composite building. Struct Engr 2002;80(19):30–7.
Based on the maximum allowable displacement (DT) at [7] Gillie M, Usmani AS, Rotter JM. A structural analysis of the
Cardington British Steel Corner Test. J Construct Steel Res
elevated temperature, given by Eq. (16), the corresponding 2002;58:427–42.
average mesh temperature can be calculated (Tpred). It was [8] Elghazouli AY, Izzuddin BA, Richardson AJ. Numerical modelling
found that the ratio of the predicted average mesh critical of the structural fire behaviour of composite buildings. Fire Saf J
temperature to the test average mesh critical temperature 2000;35:279–97.
Tpred/Tmesh ranged from 0.91 to 0.75 for the mild steel slabs [9] G.M. Newman, J.T. Robinson, C.G. Bailey, Fire safe design: a new
approach to multi-storey steel-framed buildings. SCI Publication
and 0.97–0.69 for the stainless-steel slabs, as shown in P288. Ascot: The Steel Construction Institute; 2000.
Tables 3 and 4, respectively. This comparison shows that [10] C.G. Bailey, Steel structures supporting composite floor slabs: design
the simple design method, presented in this paper, provides for fire, BRE digest 462. Watford: The Building Research Establish-
conservative predictions for the mesh reinforcement failure ment; 2001.
temperature. However, it is clear that further work is [11] G.M. Newman, J.T. Robinson, C.G. Bailey, Fire safe design: a new
approach to multi-storey steel-framed buildings. 2nd ed. SCI
required to improve the accuracy of predicting the failure Publication P288. Ascot: The Steel Construction Institute; 2006.
point corresponding to fracture of the reinforcement to [12] Bailey CG. Membrane action of unrestrained lightly reinforced
achieve a more efficient design. concrete slabs at large displacements. Eng Struct 2001;23(5):470–83.
[13] Bailey CG, White DS, Moore DB. The tensile membrane action of
7. Conclusions unrestrained composite slabs simulated under fire conditions. Eng
Struct 2000;22(12):1583–95.
[14] Wood RH. Plastic and elastic design of slabs and plates, with
The recent developments of refining the in-plane stress particular reference to reinforced concrete floor slabs. London:
distribution, and prediction of compression failure, within Thames and Hudson; 1961.