Beruflich Dokumente
Kultur Dokumente
Lecture notes
BYGDTU
2007
ISSN 1396-4046
CONTENTS
Contents
Symbols
Preface
Deterioration curves
Reinforcement
10
11
Cold-worked reinforcement
13
15
17
19
Stiffness of steel
23
Properties of concrete
24
Parameters
27
Heating rate
28
Cooling
29
Strength
30
Siliceous concrete
31
32
Lightweight aggregate
33
34
Alumina cement
35
Stiffness of concrete
36
40
Stress distribution
43
References
51
SYMBOLS
A
Ac
As
At
a
a
ao
a1
au
B
b
bo
C
Ceff
C'
Ci
c
D'
d
dc
dN
dP
dP+N
ds
ds,ave
ds'
E
E'
Eco
Eco20
EP
Es
Es20
F
FcE
Fc
Fcr
Fcrc
Fcu
Fs
FsE
Fsu
f
fcc
fcc,ave
fcc20
fct
fct20
fs
fsu
fs20
f0.2
h
I
Ic
Ic20
Is
i
k
k
l
M
Mo
M1
M+
Mm
mw
N
P
T
TM
Ts
Ts,ave
T1- T64
t
u
V
Vc
Vct
Vs
x
y
z
strength
compressive strength of concrete
average of fcc
fcc at 20C
tensile strength of concrete
fct at 20C
tensile strength of idealized steel
ultimate tensile stress of steel
fs at 20C
0.2 pct. proof stress
opening height
moment of inertia
I of a reduced concrete cross-section
Ic at 20C
I of a steel cross-section
number
parameter defined as N/Fcu
minimum fraction of a property in general
length of a beam or a column
moment load
mid-span moment
constant moment
positive moment capacity
negative moment capacity
total moment load
iterated moment
normal load
prestressing force
temperature
temperature at the centre line
steel temperature
average Ts
parameters describing a damage curve
time
mid-span deflection of a column
shear capacity
V width respect to concrete in compression
V width respect to concrete in tension
V width respect to the steel links
depth of the neutral axis
depth of a compression zone
co-ordinate
o
c
creep
cu
cup
cu20
load
p
s
smin
th
th,ave
tr
creep
load
th
tr
c
cM
s
s,ave
o
c
p
s
increment
deflection
strain
concrete strain at min. compressed edge
concrete strain at max. compressed edge
creep strain of concrete
ult. concrete strain at the stress fcc
ult. plastic strain of concrete (>cu)
cu at 20C
instantaneous stress related concrete strain
steel strain at prestress
steel strain
minimum s to obtain the stress fs
thermal strain
average th
transient strain of concrete
stress distribution factor
inclination of compression stresses
curvature
caused by creep
caused by load
thermal curvature
caused by tr
Poisson's ratio
reduction of fcc due to heating
c at centre line of a cross-section
reduction of fs
average s
stress
concrete stress at minimum compressed edge
concrete stress at maximum compressed edge
prestress of steel
stress of steel
Indices:
,ave
B
c
cc
cr
ct
cu
eff
fict.
i
M
N
P
s
su
sy
T
th
tr
u
w
20
0.2
average
bottom of a slab
concrete
concrete compression
critical
concrete tension
concrete ultimate
effective
fictive
index number
middle
normal force
prestressing
steel
steel ultimate
steel yield
top of a slab
thermal
transient
ultimate
web
at 20 C
0.2 pct. proof (stress)
PREFACE
As a consulting engineer, the author has felt a need for a rational basis for fire safety design of concrete structures. This subject has therefore been a main area of his research
for 25 years leading to a system of methods for design of beams, slabs, columns, walls,
prestressed beams and frames and analyses of anchorage and spalling based on a continuous material research program and supplemented with full-scale tests for verification of the design methods. The aim is to be able to calculate the load bearing capacity
of any structure of any concrete at any time of any fire exposure. The resulting methods
and data are published in a number of papers and lecture notes and parts of them are
adopted by the Eurocode for Concrete Structures EN1992 and the Danish code for Concrete Structures DS411 and the Danish Action code DS410.
These lecture notes are made exclusively for the course Structural Fire Safety Design.
They are developed over a period of 20 years for bringing research results quickly into
teaching and thereby into practise. They are not published or released for public access,
and it is forbidden to distribute them. Publication of the content is in progress as a series
of journal papers.
Kristian Hertz
1.2
s ( T)
1.2
1
s02 ( T)
c2(T)
0.8
s(T
)
c1 ( T)
c1test( T)
c2 ( T)
0.6
c1(T)
0.4
c2test( T)
0.2
0
200
400
600
800
1000
1200
C
1200
DETERIORATION CURVES
Almost any material is damaged by heat so that the damage can be described by an
S-shaped curve, which in some cases tends to be a straight line and in other cases show a
sudden damage at a certain temperature.
The author has found that these curve shapes for the purpose of design can be expressed by
the very same mathematical expression, given as
(T) = k +
1 k
2
T T T T
1+ + + +
T1 T2 T8 T64
64
Where the Greek letter means a ratio between a material property such as the tensile
strength at a given temperature TC and the same property at 20C. T1 to T64 are parameters with the unit C describing the curve and k the ratio between the minimum and the
maximum value of the property, which is often 0 except for residual properties after a fire
exposure.
An S-shaped curve is obtained by using ratios between the variable T and fixed parameters
in the denominator. A fixed value determines at which temperature the ratio reduces the entire expression, and the power of a ratio determines how steep the reduction will be. Compared to other possible S-shaped curves given by, for example logarithmic or tangential expressions, this is found to give the best fit for the extreme curves which can be expected describing reductions of material properties.
In figure 1, the flexibility of the expression is demonstrated.
T
s02 (T) = 1 +
for 0C < T 600C
T
767 ln
1750
s02 (T) =
0.108 (1000 T)
T 440
10
1.8
20 ( )
1.8
1.6
1.4
300 ( )
1.2
1
500 ( ) 0.8
0.6
600 ( )
0.4
0.2
0
0.01
0.02
0.03
0.04
0
0.040
Figure 2. Stress-strain curves for a mild steel with a 0.2% proof strength of 253 MPa
REINFORCEMENT
Some of the lattice lines of a steel crystal are incomplete. When the crystal is subjected to a
shear stress a loose end of a lattice line may combine to a neighbour line, leaving a new
loose end and the line has moved one step. The material yields. Movable irregularities like
these are called dislocations, and movement of dislocations can give rise to formation of
more dislocations when a crystal is stretched. When sufficiently many are produced, they
may hinder the movement of each other, increasing the stress necessary for deformation.
This is a short explanation of cold-working. If atoms of, for example, carbon and nitrogen
are present, or vanadium or niobium are added to the steel, they may serve as anchors for
the dislocations, increasing the energy necessary for movement and therefore increasing the
yield strength. Nitrogen atoms can move to the dislocations in the time after the steel has
been produced, increasing the strength of the material, and the nitrogen content is often reduced to ensure that the material does not become too brittle in time. The carbon content is
often reduced to make welding possible.
If the temperature is increased, the energy necessary for moving the dislocations is reduced
because temperature represents oscillations of the atoms, and the yield stress or the 0.2%
stress will be reduced. Within the first 200-300C an increase in temperature also means
that more new dislocations can be produced when a stress is applied, and this effect of coldworking at strains of about 2.0% or more can lead to an increase of the ultimate strength.
This effect is shown in Fig. 2 for a mild steel with a 0.2% yield strength at 20C of 253
MPa tested by Skinner [11].
Figure 2 also illustrates that the value of the strain, at which the steel becomes more plastic that is, the strain of the 0.2% stress - seems to be almost constant for all temperature levels.
The same observation can be made for stress-strain curves of most other reinforcing bars.
11
1.2
1.2
1
s20 ( T)
0.8
CEN 20 ( T)
0.6
s ( T)
CEN 02 ( T) 0.4
0.2
0
200
400
600
800
1000
1200
0
T
1200
Figure 3. Reduction of 0.2% strength and 2.0% ultimate strength of mild steel and hot-rolled bars
(solid curves) and the same properties according to CEN (dashed curves)
12
1.2
s100 ( )
s200 ( )
1.2
s300 ( ) 0.8
s400 ( )
s500 ( )
s600 ( )
s700 ( )
0.6
0.4
s800 ( ) 0.2
0
0
0
0.01
0.02
0.03
0.04
0.04
Figure 4 Idealized stress-strain curves for hot rolled bars according to CEN [3]
The curve for 2.0% strain should only be applied if it can be shown that these large strains
can be obtained. As mentioned this is seldom possible for steel constructions, but it is possible for some concrete constructions, such as T-shaped cross-sections, where small depths
of the compression zones may lead to large strains in the tension reinforcement.
Idealised stress-strain curves from the CEN code are shown in figure 4, and calculating the
corresponding 0.2% stresses by solving the complicated equations for the stress strain
curves, the resulting curve for the 0.2% stress is shown as a dotted line CEN02 in figure 3. It
can be seen that this curve fits reasonably well with the well-documented curve s, which
must be recommended to use for nearly all steel constructions and for most concrete constructions.
This means that for hot rolled bars the curve found previously can be applied given by
(k,T1,T2,T8,T64) = (0,6000,620,565,1100) for the 0.2% strength s(T). And if it can be
proved that a strain of 2.0% or more can be obtained, a curve can be applied for the 2.0%
strength s20(T) given by (k,T1,T2,T8,T64) = (0,100000,100000,593,100000) and shown in
figure 3.
Experience from tests and fires show that, after a fire, mild steel and hot-rolled bars can be
considered to regain their strength, which is obtained through the chemical composition. (If
the general formula is used to express this in a program the parameters (k,T1,T2,T8,T64) =
(1,100000,100000,100000,100000) are obtained).
13
1.2
scw20( T)
1.2
CEN 20 ( T) 0.8
DTU 20 ( T)
0.6
scw( T)
CEN 02 ( T) 0.4
DTU 02 ( T)
0.2
0
0
0
200
400
600
T
800
1000
1200
1200
Figure 5. Reduction of 0.2% strength and 2.0% ultimate strength of cold-worked steel
of 560 MPa
COLD-WORKED REINFORCEMENT
Cold worked reinforcing bars may have an increased 0.2% strength of about 560 MPa obtained by twisting, stretching, or both, or by cold rolling. Twisting creates a chaotic system
of dislocations and many sources for formation of new dislocations, which can be utilised
by stretching. If a bar is first twisted and then stretched, the steel often becomes so brittle
that it is not recommended for structural use. Therefore, steels with the combined effect can
be produced by first stretching and then twisting or by a simultaneous process (Caprani and
Buchwald [13]. Steel like this has been made for many years in Denmark called Danish
Tentor, which was stretched 5% and twisted one turn over a length of 15 times the diameter.
Its properties at high temperatures have been measured in the author's laboratory by Petersen and Hansen [14].
In figure 5 the results for the 0.2% stress and the 2.0% stress from these tests, supplemented
with data from Caprani and Buchwald[13] re called DTU02 and DTU20 respectively. Except
for an increase of both strengths between 100C and 300C the test data for these bars are
seen to follow the step curves that can be derived from the CEN code for cold-worked bars
shown as CEN02 and CEN20 and new idealised curves can be described by the parameters
(k,T1,T2,T8,T64) = (0,100000,900,555,100000) for the 0.2% stress scw and
(k,T1,T2,T8,T64) = (0,100000,5000,560,100000) for the 2.0% stress scw20.
It seems that these curves can be recommended for cold-worked slack reinforcement of
about 550 MPa no matter how it is cold-worked.
14
1.2
1.2
1
scwC ( T)
0.8
DTU 02C ( T)
scw20C ( T)
0.6
0
0
200
400
600
T
800
1000
1200
1200
One explanation of the small reduction of the 0.2% stress between 100C and 300C, and
the close position of the two curves, may be that the steel has an initial cold-working strain
and dislocation structure, and that it does not take the same strain to reach an improved
stress level at these temperatures as it does for hot rolled bars.
Above 300C the crystal structure becomes looser, and the borders between the steel grains
disappear. From approximately 600C, the 0.2% and 2.0% stresses of the cold worked bars
are reduced to the level of the same properties for hot rolled bars.
In a cold condition after a fire exposure the effect of cold-working is permanently reduced if
the temperature has been above 400C. And after heating to approximately 800C there is
no effect left. This means that the 0.2% stress, scwC, of a cold worked bar is reduced to the
original strength of the steel that it had before it was cold-worked - for example 325 MPa in
stead of 560 MPa - and the parameter k is 0.58 (Fig. 6).
The residual reduction scw20C of the 2,0% stress seems to be larger in percent mainly because the 2.0% stress before the fire was larger, and therefore k is assessed as 0.52.
The idealized curves can be given by
(k,T1,T2,T8,T64) = (0.58,100000,5000,590,730) for the 0.2% stress, scwC, and
(k,T1,T2,T8,T64) = (0.52,100000,1500,580,650) for the 2.0% stress, scw20C, in a cold condition after a fire.
15
/fs201.2 1.2
.
204C
93C
21C
( )
21
93 ( )
204 ( )
315C
0.8
429 ( )
429C
540 ( ) 0.4
20 C
100 C
1400
200 C
1200
300 C
800
400 C
600
654 ( )
540C
654C
0.2
1600
1000
315 ( ) 0.6
MPa 1800
0
0
0.02
0.04
0.06
0.08
400
500 C
200
0
0.02
0.04
0.06
0.080
0.08
600 C
700 C
800 C
900 C
16
1.2
1.2
sp20 ( T)
CEN 20 ( T)
sp ( T)
0.8
CEN 02 ( T)
0.6
Ruge 20 ( T)
Ruge 02 ( T)
0.4
Cahill02 ( T)
Harma 02 ( T) 0.2
0
200
400
600
800
1000
1200
1200
Figure 9. Reduction of 0.2% strength and 2.0% ultimate strength of coldworked prestressing steels of 1600 MPa
Figure 9 shows test data derived from Ruge and Linnemann [8], Cahill [16] and Harmathy
and Stanzack [15] and compared to the 2.0% stress and the 0.2% stress derived by calculation from the CEN codes (dotted) and new idealized curves (solid) given by
(k,T1,T2,T8,T64) = (0,2000,360,430,100000) for the 0.2% stress, sp, and
(k,T1,T2,T8,T64) = (0,100000,490,450,100000) for the 2.0% stress, sp20.
1.2
1.2
1
sp20C ( T)
Kord 20C ( T)
0.8
0.6
spC ( T)
0
0
200
400
600
T
800
1000
1200
1200
Figure 10 Residual reductions of 0.2% strength and 2.0% ultimate strength of a coldworked prestressing steel of 1600 MPa
Figure 10 shows residual reductions of 0.2% and 2.0% stresses for a 1600 MPa cold-drawn
prestressing steel (dotted), and idealised curves (solid) can be proposed as
(k,T1,T2,T8,T64) = (0.20,100000,750,550,650) for the 0.2% stress, spC, and
(k,T1,T2,T8,T64) = (0.20,100000,950,550,650) for the 2.0% stress, sp20C, in a cold condition
after a fire.
17
1.2
1.2
0.2%
1
spq ( T)
0.8
RL02 ( T)
DDM02 ( T)
0.8
RL20 ( T)
0.6
DDM20 ( T)
0.6
CEN02 ( T) 0.4
CEN 20 ( T) 0.4
0.2
0.2
2.0%
1
spq20 ( T)
200
400
600
T
800
1000
1200
200
400
600
800
1000
1200
Figure 11. Reduction of the 0.2% strength and the 2.0% ultimate strength of quenched
and tempered prestressing steel of 1500 MPa in a HOT condition and the same properties according to
Ruge and Linnemann [8] (RL Dashed), Dannenberg et al. [25] (DDM Dash-dot) and CEN [3] (CEN Dotted).
18
1.2
1.2
0.2%
1
spqC ( T)
0.8
spq20C ( T) 0.8
RL02C ( T)
0.6
RL20C ( T)
0.6
DDM02C ( T)
0.4
DDM20C ( T)
0.4
0.2
0.2
200
400
600
T
800
1000
1200
2.0%
200
400
600
800
1000
1200
Figure 12. Reduction of the 0.2% strength and the 2.0% ultimate strength of quenched and tempered
prestressing steel of 1500 MPa in a COLD residual condition and the same properties according to Ruge
and Linnemann10 (RL Dashed), Dannenberg et al. [25] (DDM Dash-dot).
Stress-strain curves are measured for quenched and tempered prestressing steels of approximately 1500 MPa by Dannenberg et al. [25] and treated by Kordina [19] and
Leonhardt [26] and later a comprehensive study is made by Ruge and Linnemann [8].
The idealised curve is shown as a full line in figure 1 and it is given by the parameters
(k,T1,T2,T8,T64) = (0,1100,100000,430,100000) for the 0.2% stress spq(T).
If it can be proved that a strain of 2.0% or more can be obtained, a curve can be applied for
the 2.0% stress spq20(T) given by (k,T1,T2,T8,T64) = (0,3000,1400,450,100000).
The idealised curve for the residual 0.2% stress spqC(T) of quenched and tempered
prestressing steels is shown as a full line in figure 12 and it is given by the parameters
(k,T1,T2,T8,T64) = (0.213,100000,10000,590,660).
And for the residual 2.0% stress spq20C(T) the same curve is a good idealisation
(k,T1,T2,T8,T64) = (0.213,100000,10000,590,660).
Although the curves of the relative reductions of the two strength parameters are equal in
this particular case, the initial values are not the same and therefore the final values are not
equal.
19
1.2
1.2
0.2%
sq20 ( T)
sq ( T)
D1020 ( T)
D1002 ( T) 0.8
D1420 ( T)
D1402 ( T) 0.6
Ds0212
Ds0225
Ds2012t
0.8
0.6
Ds2025t
0.4
Ds2012
0.2
0
2.0%
Ds2025
200
400
600
800
T , T , T , T12 , T25 , T
1000
1200
0.4
0.2
200
400
T T T T
600
T
800
T
1000
T
1200
Figure 13. Idealised curves for the reduction of the 0.2% strength and the 2.0% ultimate strength of
quenched and self-tempered reinforcing steel of 550 MPa in a HOT condition and the same properties
found by Dotreppe11 (+ 12 mm transient, x 25 mm transient, o 12 mm steady, 25 mm steady) and
according to the test series reported in this paper (Dotted 10 mm transient and Dashed 14 mm transient).
20
Figure 14
Transportable ring
oven in test
machine.
Figure 15
Principle for
deformation
measurement.
This steel is therefore completely different from other reinforcing steels on the market,
and none of the curves presented so far in this paper and in Hertz [21] nor in the codes
can be expected to represent the strength variation of it at high temperatures. The author
has therefore decided to test the properties in order to establish a complete design basis.
A previous study has been made by Prof. Dotreppe [27] where properties are found in a
HOT condition for single temperatures between 480-680C for steels of quality S500
and diameters of 12 and 25 mm. These tests are made in a gas oven, where the steels are
subjected to a standard fire exposure, i.e. a rapid heating rate compared with the heating
rate in a steel which has a cover of concrete. These results are included in figure 3 as
single points.
The authors test series comprises transient tests of the HOT curves during heating and
steady state tests of the residual stress-strain curves in a COLD condition after heating
of a quenched and self-tempered steel (of quality K550TS produced by Fundia Ltd. according to the Danish Standard DS 13080) with a 0.2% stress of 550 MPa.
With financial support from the Danish light aggregate concrete industry, an electrical
ring oven was built for the HOT tests (figure 14). The transportable oven has an insulated cylindrical chamber of diameter 200 mm and height 250 mm and it can be placed
in a test machine. A measurement device was constructed in order to measure the deformation of the reinforcing bar in the oven (figure 15). Two pairs of heat resistant
Kanthal steel rods are fixed to the reinforcing bar at a distance of 150 mm. These two
pairs of measuring rods are penetrating the bottom of the oven together with the reinforcing bar. The one pair is welded to a plate to which two potentiometers are fixed, and
the other pair is welded to a plate to which the potentiometers can measure the elonga-
21
tion. The elongation measured is the sum of the thermal and mechanical elongation of
the reinforcing bar between the two fix points minus a difference of the thermal elongation of the two pairs of Kanthal rods caused by their different lengths. By measuring this
total elongation for a loaded reinforcing bar and subtracting the same measurement for
an unloaded reinforcing bar heated by the same heating curve, the result is the mechanical elongation by means of which the mechanical strain can be found.
1.2
1.2
0.2%
1
0.8
sq20C ( T) 0.8
D1002C ( T) 0.6
D1020C ( T) 0.6
sqC ( T)
D1420C ( T)
D1402C ( T)
0.4
0.4
0.2
0.2
0
2.0%
200
400
600
T
800
1000
1200
200
400
600
800
1000
1200
Figure 16. Reduction of the 0.2% strength and the 2.0% ultimate strength of quenched and selftempered reinforcing steel of 550 MPa in a COLD residual condition. sqC and sq20C are proposed idealized curves and D1002C and D1020C are curves measured for 10 mm steel at steady state conditions
and D1402C and D1420C are curves measured for 14 mm steel at steady state conditions.
A number of tests have been made by the author and his students Ramolla [28] and
Blichert [29] heating reinforcing bars of 10 mm and 14 mm at a heating rate of 5C per
minute until they brake while a constant load is applied. This was repeated for stress
levels of 0, 20, 40, 50, 60, 70, 80, 90, 95, 100, 104 and 108 percent of the 0.2% stress at
20C, and for each test the strain was measured during heating for each 100C.
Strains obtained without load (0%) are subtracted from the strains measured with load,
and from these results transient stress-strain curves can be constructed and the
0.2% stress and the 2.0% stress derived.
The result is shown in figure 13, where the Dotted curves refer to 10 mm bars and the
Dashed curves to 14 mm bars. I addition single points are shown found by
Dotreppe [27].
The Full lines represent the idealised curves given by the parameters
(k,T1,T2,T8,T64) = (0,6000,1150,540,700) for the 0.2% stress sq(T), and
(k,T1,T2,T8,T64) = (0,100000,100000,590,700) for the 2.0% stress sq20(T).
In order to assess the residual COLD properties, reinforcing bars of 10 mm and 14 mm
were heated 5C per minute in an electrical oven to maximum temperatures varying
from 100C to 1000C with 100C interval. The maximum temperatures were kept for
60 minutes and the oven was cooled max 5C per minute. This heat treatment answers
roughly the heating curve for reinforcement with a concrete cover in a structure. After
22
cooling, the steel was drawn in a test machine and the stress-strain curves were recorded
by two extensometers. The results are shown in figure 6 as Dotted curves for 10 mm and
Dashed curves for 14 mm. The Full lines represent the idealised curves given by the parameters (k,T1,T2,T8,T64) = (0.418,100000,100000,700,900) for the 0.2% stress sq(T),
and (k,T1,T2,T8,T64) = (0.437,100000,100000,700,900) for the 2.0% stress sq20(T).
The relation between the deterioration curves for slack reinforcement steels is shown in
figure 17 and the same is shown for prestressing steels in figure 18.
It is seen that the quenched steels keep their strength better than the hot rolled and cold
worked steels until the tempering temperature 450-500C where the microstructure responsible for the increased strength has been formed. It is also seen that the quenched
steels have a lower strength at higher temperatures, which means that the effect is rapidly lost at 500C.
s ( T)
1.2
1.2
0.8
0.8
sp20 ( T)
scw( T) 0.6
spq20 ( T)
sq ( T)
0.6
0.4
0.4
0.2
0.2
200
400
600
800
1000
1200
200
400
600
800
1000
1200
23
T1
6000
100000
100000
100000
100000
100000
100000
100000
2000
100000
100000
T2
T8
k
0.000
0.000
1.000
1.000
0.000
0.000
0.580
0.520
0.000
0.000
0.200
620
100000
100000
100000
900
5000
5000
1500
360
490
750
565
593
100000
100000
555
560
590
580
430
450
550
1100
100000
100000
100000
100000
100000
730
650
100000
100000
650
0.200
100000
950
550
650
0.000
0.000
0.213
0.213
0.000
0.000
0.418
0.437
1100
3000
100000
100000
6000
100000
100000
100000
100000
1400
10000
10000
1150
100000
100000
100000
430
450
590
590
540
590
700
700
100000
100000
660
660
700
700
900
900
T64
Table 1. Recommended parameters for design reductions of 0.2 % and 2.0 % stresses of hotrolled, cold-worked bars, cold-worked prestressing steels, quenched and tempered, and
quenched and self-tempered reinforcement during and after fire exposure.
STIFFNESS OF STEEL
24
PROPERTIES OF CONCRETE
When an ordinary not fire proof concrete is heated, free water evaporates, and above
approximately 150C, chemically bound water is released from the hydrated calcium
silicate Hertz [32]. This process has a local peak at 270C. The dehydration of the matrix and the thermal expansion of the aggregate give rise to internal stresses
(figure 20a), and from 300C micro cracks will pierce through the material (figure 20b).
The compressive- and the tensile strength, the thermal conductivity and the modulus of
elasticity are reduced and an unloaded specimen will be subject to an irreversible expansion.
Concrete heated to less than 300C can recover after a fire by soaking moisture from the
air, but when the micro cracks are formed the strength loss is permanent, and the material is susceptible to environmental impact such as frost etc. The author therefore prescribes that material heated at more than 300C should be removed when a repair work
is done.
At 400-600C, crystals of calcium hydroxide decompose into calcium oxide and water,
a process reaching its highest intensity at about 535C (figure 20c). This weakens the
concrete, but during the cooling phase and within the first days after a heat exposure,
the calcium oxide absorbs water from the ambient air and expands, which opens the
cracks already formed. You may say that the new calcium hydroxide act as jackets widening the cracks as shown on Figure 1d. The compressive strength is reduced further 20
pct. during the cooling phase and the minimum strength is found up to a week after the
fire depending on the geometry of the structure. This is why the author seldom investigates a fire damaged concrete structure before a week has passed and the minimum residual load bearing capacity is reached. This often requires that the site should be bared
until the damages are assessed.
Above 600C, the hydrated calcium silicate decomposes further and this second phase
of the process has a peak at 710C. At 800C, concrete can often be crumbled to gravel
by the fingers, and above 1150C feldspar melts and the remaining minerals of the cement paste turns into a glass phase.
25
H2O
H2O
H2O
H2O
26
If the concrete is loaded in compression, the compressive stresses in the matrix should
first be unloaded before tensile stresses can be established and micro cracks formed
when the aggregates expand (figure 21). The author has found this simple explanation
for the so-called transient strain, which is a strain contribution that only occurs if the
specimen is loaded during heating Hertz [32].
Since the micro cracking is reduced the drop of the compressive strength in a hot condition and after a fire is also reduced. The concrete can be up to 25 percent stronger than
an unloaded fire exposed concrete if compressive stresses of 25 to 30 percent of the
original strength are applied.
However, the engineer should only consider this effect if the compressive load is known
through the entire fire course and it is in general safer to use values obtained from tests
of unloaded specimens instead of the transient values of the compressive strength.
The loss of compressive strength in the cooling phase is one reason why a design for a
fully developed fire course must include the time where the concrete is the weakest.
Since reinforcing bars usually are the weakest in a hot condition during the fire, it is
necessary to investigate the load bearing capacity at two different times. The author
calls the conditions at these times for HOT and COLD and finds the capacity during a
full fire course as a minimum of the capacities at these two conditions.
Steam pressure from water released during heating combined with compressive thermal
and static stresses at the fire-exposed surface can give rise to explosive spalling of dense
or wet structures. This is no problem for indoor constructions of traditional concrete.
Traditional concrete means a concrete not containing particles smaller than the cement
grains such as micro silica or small filler as for example used for high strength or selfcompacting concrete Hertz [30]
If a concrete is wet or dense, a test can be made in order to document that explosive
spalling is avoided perhaps by adding polypropylene fibres or to reject the use of an actual material according to Hertz and Sorensen [31].
Sound
Damage cc
C
27
Concrete
Hard Burned Brick
Chinaware
Crystal Glass
Glass
Stoneware
Cracked Tiles
Wood
Papier Mach
1.00
0.97
0.92
0.84
0.67
0.38
0.15
0.05
0.02
20
100
200
300
400
500
600
700
800
As mentioned the hydrated cement paste shrinks when heated, and this causes tensile
stresses in the matrix and compressive stresses in the aggregates. The author has found
that these stresses give rise to changes in the sound picture of the material, which has
proven to be a valuable tool for determining the maximum temperature and degree of
fire damage of a concrete. Sound is a complex phenomenon, which depends on the
shape of the structure, and therefore the author has experienced from teaching that it can
be described most precisely by comparing with the sound of a material, for which the
sound picture is familiar. This comparison is given in Table 1. The damage is given by
cc, which is the ratio between the compressive strength of the heated concrete and the
corresponding strength at 20C. (cc = fcc/fcc20)
The relation between damage and sound can be used for most concretes. The temperatures shown are valid for the main group of concretes as defined later.
By means of this method, a trained person can estimate the damage and the maximum
temperature in the actual depth within 50C immediately and without leaving further
damage to the structure. The alternative would be to drill a specimen, make a section
and count the number of cracks in an electron microscope. This leaves a hole, takes time
and is less precise.
PARAMETERS
A variation of the water cement ratio has no effect on the loss of compressive strength
during a fire down to a value of approximately 0.40 because excessive water evaporates.
For lower values, unhydrated cement may act as fire resistant filler, which can reduce
the strength loss Hertz [32]
The age of the concrete has no influence except for fresh concrete where heating will
give a forced hydration within the first 150C before the concrete is damaged. This may
course a slight increase of the compressive strength, and the following damages must be
assessed related to this increased final strength level.
Super plasticizing additives have no known effect on the fire resistance, but they can be
used for adding small particles increasing the denseness and the risk of spalling. Some
super plasticizers produce ammonia when heated, which may require full breath protection of a fire brigade entering a building made from these materials Hertz [38].
28
1.2
cc ( T)
ccrapid( T)
0.8
0.6
ccrapidtest( T)
0.4
0.2
200
400
600
800
1000
1200
Figure 22. Residual compressive strength of rapid heated main group concrete.
HEATING RATE
The values of the compressive strength given in this paper are valid for slow heated
concrete, where the heating rate is 2C per minute or less. These values can be illustrated as S-shaped curves of cc as a function of the temperature T.
More rapid heating will occur for areas near the surface of a structure and this can be of
importance especially for slender cross sections. The problem of measuring the strength
of rapid heated concrete is that a specimen should have a certain size if its cross section
should be representative for the distribution of aggregate and cement paste. If the specimen is sufficiently big such as a 150 mm diameter cylinder, a rapid fire-exposure of 10or 20C per minute in an oven will cause a considerable temperature difference of approximately 400C between surface and core giving rise to thermal stresses, which will
damage the specimen before it is tested.
In order to investigate the effect of rapid heating the author has developed a test method
using microwave power to heat up the specimen combined with a thermal insulation to
regulate the heat loss from the hot specimen to the cold oven. By means of this method,
it is possible to heat concrete cylinders at rates of 20C per minute with a limited thermal difference Hertz [35].
A series of 300 specimens of a main group concrete (defined later) have been tested for
a heating rate of 10C per minute, and the resulting curve for cc(T) is a straight line in
stead of the S-shaped curve intersecting the slow heated curve for the same concrete at
approximately 500C (figure 22). The reason why the residual strength is smaller for
small temperatures is that the matrix does not have the same time to creep when the aggregate expands and the matrix shrinks. The matrix therefore appears to be more brittle
giving more micro cracks, when the specimen is heated rapidly. For higher temperatures, the rapid heated concrete is stronger because it takes time for the calcium hydroxide to decompose and damage the concrete. The idealised curve for the rapid heated
main group concrete is described by the parameters
(k,T1,T2,T8,T64) = (0,1500,580,520,690).
29
H2O
H2O
COOLING
If a hot specimen is taken from an oven and quenched in water the micro cracks can be
water filled and the calcium hydroxide can be established in the cracks and voids filling
them out instead of opening them as shown at figure 23b. The result may be a stronger
material in the COLD condition after the fire than in the HOT condition during the fire.
Although this mechanism is not in general relevant for building fires, it is mentioned in
order to give an understanding of the influence of the cooling process.
If the specimen is cooled down slowly, i.e. at cooling rates of 1C per minute or less, in
an oven or in a hot massive building or tunnel the reformation of calcium hydroxide has
sufficient time to take place, and the concrete has its maximum damage after the cooling
(figure 23a).
If the cooling is faster, the reformation of calcium hydroxide will take place at a slow
rate in the days after the cooling, and the concrete reaches its minimum strength several
days after the fire. The speed of this regeneration process depends also on the moisture
content of the ambient air and the size of the structure.
A concrete, which has been damaged by a fire and is subjected to a new fire exposure or
is exposed to a few thermal cycles from an industrial process will deteriorate further, if
the temperature exceeds the level from the first exposure. It will be damaged approximately as if it reaches the new temperature in the first exposure. This is logical from the
nature of the deterioration, and it is observed in the authors lab and reported in literature on structures for nuclear power plants and military purposes Hertz [32].
If many thermal cycles are made the crack movements in the microstructure may cause
an increasing damage to the concrete in time. The author has observed this in buildings
for fire training in the Danish Civil Defence, where violent fire exposures are repeated
daily for years and it is known to be a problem for military installations such as ramps
for missiles and runways for jet planes.
30
STRENGTH
Idealised values of the reduction cc of the compressive strength are made for various
concretes because it is seldom possible to make a test series of an actual concrete when
a structure is designed. As seen from the description of the deterioration processes
above the aggregate is of a major importance for the result. Therefore, idealised curves
are given on the safe side related to test results for the reduction as a function of temperature for a number of concretes characterised by their aggregates. Since the curves
need to be on the safe side, a test series for an actual concrete may give improved values, and as mentioned later it is even possible to design a concrete with a fire resistance
as wanted for almost any purpose.
It is necessary to give the concept safe a special consideration in this context before
creating recommended curves. In CEN [3], data are presented based on transient tests,
where the concrete is loaded during heating. These values of cc are up to 0.25 larger
than the values from unloaded tests with the largest difference between 500C and
700C and the effect is found for tests in a HOT as well as in a COLD condition after
cooling. As mentioned previously the compressive stresses should be between 25 and
30 percent of the ultimate strength at 20C in order to obtain this effect, and if the problem is not considered in detail it might seem reasonable to presume that these stresses
are applied in most compression zones.
However, the safety factors for strength and for load used for calculations of the ultimate limit resistance and the safety from application of a characteristic value of the
compressive strength as a 5% quantile of the real strength means, that the physical stress
level usually is near the limit for obtaining the increase of the strength from transient
tests. In addition engineers usually only calculate a few hard loaded members for the ultimate load and assume the others less loaded to be safe. Furthermore, during a fire,
beams, slabs, columns and walls often have large deflections, which can give considerable changes of the stresses in time. Changes of stresses also occur due to the deterioration of the cross sections giving rise to changes of the geometry of the compression
zones. In addition, the considerable reduction of the E-modulus above 500C means that
parts of the compression zones with high temperatures near the surface where the transient effect is predominant are unloaded and central parts with smaller temperatures are
loaded. Moreover, engineers often assume structural members as simply supported although they are not in the real structure, and they do so especially for buildings, where
fire load is important. This is a safe consideration for the ultimate load, because it means
that the moment forces at mid span are less than presumed in the calculations. Nevertheless, what is safe for the ultimate load is unsafe for the assessment of the reduction of
the compressive strength at fire. In addition, during a fire the real hogging moments at
the supports unload the sagging moments at mid span further.
The author therefore recommends average unloaded values for the compressive strength
of concrete when curves are made as a general reference for the property. For the applications, this can very roughly be estimated to represent a quantile of 5-10% of a normal
distribution of cases including variations of materials data for unloaded and transient
tests and corresponding variations of applied and changed stresses in compression zones
in building design.
31
1.2
1.2
0.8
ccHOT( T) 0.8
ccCOLD( T)
cchigh( T)
cclow( T)
ccCEN ( T)
cchigh( T)
0.6
0.6
cclow( T)
0.4
0.4
CEN
0.2
0.2
200
400
600
T
800
1000
1200
200
400
600
800
1000
1200
SILICEOUS CONCRETE
can be considered to form a group of aggregates, which expands the most, and to give
the largest damage. It is therefore often used as a safe reference for design values if the
aggregate is not known.
Several investigations like Abrams [36] and Malhotra [37] have clearly shown that also
siliceous concrete is further damaged in the cooling phase giving a reduction of cc of up
to 0.20.
In Hertz [32], 15 test series are compared for the compressive strength of siliceous concrete in a HOT condition. The property cc is given as the ratio between the HOT average strength at the test and the original average strength at 20C. Excluding two series
due to their test procedures, the others show a relatively close main band of results. In
figure 24a, the limits of the band are shown as dotted lines, and a full drawn design
curve is made as an average of the curves. The curves are valid for unloaded concrete. If
they were made by means of transient tests with a load of 30% of the compressive
strength at 20C, the values of the high limit curve could be increased up to 0.25 at temperatures above 500C as discussed previously.
In CEN [3], a recommended transient curve is given shown as CEN at figure 24a, and it
has larger values especially above 500C, where the cracks caused by the calcium hydroxide are significant. The full drawn curve recommended by the author for a HOT
condition is given by the parameters (k,T1,T2,T8,T64) = (0,15000,800,570,100000).
In figure 24b, curves are given for a residual COLD condition after a fire based on 12
test series Hertz [32]. As seen the relative residual strength cc is about 0.20 smaller than
the HOT strength for temperatures above 400C. The CEN code does not give any curve
for this property, because this code only contains data for standard fires.
The recommended curve for a COLD condition is given by the parameters
(k,T1,T2,T8,T64) = (0,3500,600,480,680).
32
1.2
1.2
Sea Gravel
cchot( T)
0.8
ccCOLD( T) 0.8
cchigh( T)
cclow( T)
ccCEN ( T)
cchigh( T)
0.6
cclow( T)
ccSGtest( T)
0.4
0.6
0.4
CEN
0.2
0.2
200
400
600
T
800
1000
1200
200
400
600
800
1000
1200
33
1.2
1.2
cchot( T)
cchot( T)
0.8
cchigh( T)
cchigh( T)
cclow( T)
ccPumice( T)
0.8
0.6
cclow( T)
ccPumice( T)
0.4
0.4
0.2
0.2
0.6
200
400
600
T
800
1000
1200
200
400
600
800
1000
1200
LIGHTWEIGHT AGGREGATE
like expanded clay causes only a small thermal expansion and therefore only a small
damage to the concrete, and at the same time, it ensures a small thermal conductivity.
However, due to the small density the temperature of central parts of cross sections
made from these materials are seldom smaller than temperatures of similar cross sections based on main group aggregates.
Lightweight aggregate concrete may have densities of 600 kg/m3 for the lightest over
900- and 1200 kg/m3 to 1800 kg/m3 for heavy and strong qualities. The lightest qualities
have typically a characteristic compressive strength of 3-4 MPa, where it may be about
20 MPa for the heavy. They are usually not included in the structural codes for heavy
concrete, but this seems to be a result of arbitrary traditional limitations.
The author has investigated the properties of light aggregate concretes as a part of a series of projects continuing for more than 10 years where design methods and parameters
are established for the Danish lightweight aggregate concrete industry, and has proved
that the same design methods apply for fire safety design of lightweight aggregate concrete as for heavy concrete Hertz [41]. Figure 26a is based on data from this project and
10 test series on expanded clay aggregates in a HOT condition reported in Hertz [32]. In
addition, a curve on Pumice is shown obtained by Jensen [43].
Figure 7b is also based on data from the design project, 8 test series from Hertz [32] for
the residual COLD strength of expanded clay concrete, data from later projects like
Mailund and Larsen [42] and a curve on pumice from Harada et al. [44]. The deviations
of the pumice curves from the average of the expanded clay curves can be regarded as
random, and therefore the recommended curves can be applied for natural lightweight
aggregates as well. The full drawn curve recommended for a HOT condition is given by
the parameters (k,T1,T2,T8,T64) = (0,100000,1100,800,940).
The curve recommended for a COLD condition is given by the parameters
(k,T1,T2,T8,T64) = (0,40000,650,830,930).
34
1.2
Chamotte
1
ccProof ( T)
0.8
ccTestProof( T)
ccDTU( T)
ccTestDTU( T)
Mo-clay
0.6
0.4
0.2
200
400
600
800
1000
1200
Figure 27. Examples of fire resistant concrete and design curves for them.
The author and his students have made a humble attempt to design a fire resistant concrete based on Scandinavian raw materials. Here the pulverised chamotte is exchanged
by a natural powder of a special Danish clay called mo-clay, and the aggregates are
gravel of burned mo-clay Muff [45]. An example of a curve of a residual compressive
strength is shown in figure 27 and a design curve is made given by (k,T1,T2,T8,T64) =
(0,10000,4000,710,1100).
Since free calcium hydroxide is avoided by means of the pozzolanas, these concretes
have the same strength in a HOT condition during a fire as in a COLD condition after
one or more heat exposures, and only one curve is needed for each material.
The examples illustrate how concrete can be designed to meet almost any requirements
of fire resistance. However, design curves must be created for each fire resistant material based on tests, and the curves presented here can only be examples illustrating the
possibilities of improving the fire resistance of concrete.
35
ALUMINA CEMENT
are available as heat resistant materials for industrial applications. Nekrasov and Tarasova
[5] write that they are used in the West, but they have disadvantages in a moist climate. This
has also been realised by experience for structural use in buildings.
As an illustration, the author has made a small test series of 12 specimens with one of these
products called Hasle BST1200A with a compressive strength at 20C of 58 MPa. After
heating to 800C, the compressive strength was reduced to 50 percent. However, after only
48 hours in a moist climate at 38C and 75 % RH the compressive strength was reduced to
42 percent.
These materials are well suited for ovens and other industrial applications to operate in a
dry climate, but they are not seen as an alternative for building structures.
k
0.00
0.00
0.00
0.00
0.00
0.00
T1
15000
3500
100000
10000
100000
4000
T2
800
600
1080
780
1100
650
T8
570
480
690
490
800
830
T64
100000
680
1000
100000
940
930
Table 3 Recommended parameters for design reductions of compressive strength of concrete in a HOT
condition during a fire and in a COLD condition after a fire.
The curves shown in the this chapter are valid for a description of the reduction of strength
of slowly heated concrete, which is predominant in large cross-sections exposed to ordinary
fire courses.
However, by means of microwave power heating the author has shown that the curve
for residual compressive strength becomes much straighter, if the concrete is heated rapidly, i.e. by 10C per minute.
36
Figure 28. Residual modulus of elasticity of concrete with Danish sea gravel,
w/c = 0.87, Ec20= 27.9 GPa. Hertz [12].
STIFFNESS OF CONCRETE
Comparing curves for the development of the modulus of elasticity to curves for the development of the compressive strength with temperature for the same concretes exposed
to the same temperature-time courses and subjected to the same loading, it appears to be
a universal truth, that the reduction of the modulus of elasticity is the square of the reduction of the compressive strength.
(See for example the curves in Hertz [32]).
2
E co = c E co 20
The relation appears to be valid for a large variety of concretes in a hot condition as well
as after the cooling phase, and applicable for maximum temperatures above 200C,
where the effects of the initial moisture condition are negligible.
During the investigation previously mentioned on concrete based on Danish sea gravel, the
author has noticed a considerable increase of the ultimate strain cu with temperature
(Hertz [12]). The same observation can be made from the stress-strain curves for various
concretes with - or without application of load during heating and tested in a hot or in a cold
condition, as they are reported in the literature. (See for example Schneider [48], Harmathy
and Berndt [46], Harada et al. [44] and Fischer [47]).
37
Figure 29 Residual Poisson's ratio of concrete with Danish sea gravel, w/c = 0.87, c20= 0.16.
Hertz [32].
From the stress-strain curves it can also be seen that the increase in strain follows the
decrease in stress, and the simple model is suggested that the product of stress and strain
remains a constant for each point of the stress-strain curve while the material is weakened due to the heat.
38
Figure 31 Idealized stress-strain curves for concrete exposed to high temperatures in a hot condition and in
a cold condition after the heat exposure.
39
E co = c E co 20
The lack of knowledge about these relationships has been a main obstacle so far to the
development of simple and rationale methods for calculations of the stress-strain developments of fire exposed concrete structures.
Applying a certain variation of the single parameter c it is now possible to generate the
heat induced changes of the full, idealized stress-strain curves whether the idealisations
are elasto-plastic or curved lines are used.
40
The thermal expansion, the creep and the so called transient strain of concrete exposed
to high temperatures have been analysed by a number of authors as described in Hertz
[32], and a theoretical model for calculating the total strain has been developed by
Anderberg and Thelandersson [49].
The total strain is assumed to be a sum of 4 contributions: the thermal strain th, the instantaneous stress-related strain load, the creep strain creep and the so called transient
strain tr:
= th + load + creep + tr
The thermal strain of the concrete depends mainly on the aggregate. Yet for rough calculations up to approximately 500C, it might be a reasonable approximation to consider the thermal expansion of ordinary concrete based on quartz or limestone as linearly increasing with temperature by the coefficient 11x10-6 C-1.
-6
-1
th = 11 10 ( T - 20C ) C
41
= 530 10
f cc
6
t
3h
3.04(T 20 C)
1000 C
where /fcc is the ratio between the actual compressive strength of the concrete at the
temperature T during the time t.
It is seen that even at a stress ratio of 1, the duration should be at least 10-20 h, if the
creep strain should be comparable to the thermal strain at temperature levels below
500C.
Creep strains therefore are important mostly for applications of concrete in industry and
reactor technology, where sustained high temperatures may occur.
As regards concrete structures exposed to short-time heat-pulses from ordinary fire
courses creep strains may be important for calculation of the time dependent deflections
after the fire exposure.
42
Figure 33. Thermal strain of concrete and steel from Harada et al. [20].
As regards concrete structures during the time of the fire exposures, creep strains are
important mostly to structural members being close to collapse, and seldom of importance to structural members, which are designed to resist fully developed fire courses.
In Hertz [32] it is shown that the transient strain may be regarded as a hindered part of
the thermal expansion for loaded concrete specimens exposed to heating.
The transient strain is found to be proportional to the ratio between the compressive
stress and the compressive strength of the concrete at 20C, and furthermore it is proportional to the thermal expansion.
tr = -2.35
f cc20
th
43
STRESS DISTRIBUTION
One of the main hindrances for developing calculational procedures for determination
of the load-bearing capacities and other mechanical properties of concrete structures has
been the fact that the maximum temperature and he material properties vary considerably throughout a fire exposed cross-section.
The problems can be handled by application of finite element analyses using an appropriate computer and time for generating the in-data.
However, many of these calculations can be executed much more easily by introducing
a basic concept: the stress distribution factor.
Consider a cross-section exposed to fire at two parallel surfaces.
The isotherms will all be parallel to the surfaces at any time of the fire exposure, and the
reduction of the compressive strength of the concrete c is then a function of the depth
from the surface.
The maximum temperature occurring at the centre of the cross-section until the actual
time is denoted TM, and the corresponding reduction of the compressive strength of the
concrete is cM.
The average compressive strength of the concrete in a cross-section of the thickness C is
expressed as
f cc,ave = cM f cc 20
where fcc20 is the compressive strength at 20C, and is a factor called the STRESS
DISTRIBUTION FACTOR.
44
This factor represents a very useful concept by means of which the stress-strain conditions of a loaded and fire exposed concrete section can be calculated as a whole and almost just as easy as by the calculation of the same cross-section without fire exposure.
Considering that the alternative procedure is to split up the cross-section into a number
of finite elements each of uniform maximum temperature and to solve the problems for
the entire system of elements, the simple procedures using stress distribution factors are
less laborious and just as precise as the finite element analysis.
In addition, the procedures based on stress distribution factors are identical to the procedures used for cross-sections without a fire exposure if the fire exposure is reduced to
nil.
The stress distribution factor is determined by
=
1
C
0 c ( T ( z ) ) dz
C cM
and is the ratio of the average compressive strength of the cross-section to the compressive strength at the centre of the cross-section, which is
fccM = cMfcc20
Furthermore is utilised that if the compressive strength at any point of the cross-section
is reduced by the factor c to
fcc = cfcc20 ,
and that the initial modulus of elasticity of the concrete at the same point is
Eco = c2 Eco20
and the ultimate strain is
cu = cu20/c
where cu20 is often considered to be 0.35 pct.
From the elasto-plastic stress-strain relations it is seen that the ultimate stress is reached
at the strain cu20 for most of the temperature levels.
It is therefore a reasonable approximation to assume the cross-section being able to act
at its ultimate stresses at every point, when compressed to a uniform strain of cu20 in a
plastic analysis.
If cM is less than unity the approximation is even more valid when the entire crosssection is compressed to cuM = cu20/cM > cu20.
45
This means that the cross-section can be loaded to an ultimate resistance equal to the
average compressive strength multiplied by the thickness of the section, before the ultimate strain is reached at the centre-line.
For applications, where the strain may vary along the centre-line, but has a constant
value across the section, the material could be considered to be uniformly damaged
through the section.
The stress-strain curve of the material is assessed to be the one of an impaired concrete
with a compressive strength equal to the mean value through the cross-section, but with
an ultimate strain not exceeding cuM.
The ultimate resistance per unit length of the cross-section thus is
cMfcc20C
and the initial stiffness per unit length is
C
0 c dzE co 20
46
4 / 3
10
1
10 cM
Figure 35. Relation between the summation of c2 and the approximation 4/3.
However, calculations of the integral for a large number of different fire exposures and
thicknesses of the concrete cross-section show that a better approximation would be
C
o c dz 3 cM C
The average deviation was less than 5 pct., and no single deviation was above 10 pct.
The values of c and may be found for a HOT condition, where the temperature is a
maximum at 30 mm from the surface, and for a COLD condition after the fire exposure.
Thus, the initial stiffness per unit length is
4
3 cM E co20C
47
Figure 36. Comparison between the model using a reduced cross-section and the actual distribution of
stiffness for a typical fire development. (C = 0.40 m, a = 348x10-9m2/s, Ah/At = 0.04 m,
If the elastic parts of the elasto-plastic stress-strain relations are used, or a more detailed
analysis is made based on a stress-strain curve, the cross-section could be considered to
have a fictive thickness of
4
3 C
and to consist of concrete with the mechanical properties of the concrete at the centreline, i.e.
2
f cc = cM f cc20 , E co = cM E co20 , cu = cu20 / cM
48
4cM2
EI
EI 20
Figure 37. Relation between the relative flexural stiffness and the approximation 4cM2.
This rough model, where the Eco-modulus is zero at both sides and of full central value
over a thickness of 4/3C at the middle, is very close to the distribution of c2 over the
cross-section.
The agreement is so close, that even the moment of inertia weighted by the Eco-modulus
is almost identical for the two models.
This means that
C
2 o2 z 2 c2 E co 20 dz
1 4 3 2
C cM E co20
12
Also the validity of this approximation has been tested by calculation of the integral for
the previously mentioned fires and cross-sections as presented in the figure.
The average deviation was less than 1 pct. and the maximum deviation less than 4 pct.
of the initial value of the moment of inertia times the Eco-modulus before the fire exposure.
The model using a reduced cross-section of thickness 4/3C is therefore valuable to the
calculation of the load-bearing capacity of a fire exposed wall or column as well as for
the elastic- or curved-line analysis of a cross-section.
However, in case a plastic analysis is carried out, all parts of the cross-section will be
able to act by their ultimate stresses, when they are compressed at a large uniform strain.
49
For this analysis the cross-section could be reduced to one of a thickness C having the
uniform mechanical properties of the concrete at the centre-line
f cc = cM f cc20 , cu = cu20 / cM
In the early phases of a fire, development of large temperature gradients may occur near
the surface of the cross-section.
If the cross-section is not loaded, large thermal stresses may arise. However, if the
cross-section is loaded the thermal strains and the transient strains should be added to
the compression strain, which is assumed to be uniform through the cross-section.
If the cross-section is loaded by 30 pct. of the compression strength before the fire, a
temperature difference of say 500C between a surface layer and the core would give
rise to a difference in strain of
3
th + tr = 1.1 ( 1 - 2.35 0.3 ) 500 / 10 = 0.16 pct.
And the ultimate strain of the concrete in a hot condition at 500C is increased by
50
These values are calculated for an unloaded concrete with Danish sea gravel aggregates,
which will represent conservative estimations for most other Danish concretes and
loads, or they are calculated for German siliceous aggregates, representing conservative
estimations for most other concretes and loads.
The engineer may of course use data for the actual concrete, if these can be documented,
in order to utilise a possible benefit of the actual data compared to the worst case data
found in codes or textbooks.
Finally, the engineer may decide to improve the fire resistance of the concrete, which is
possible by means of well-known techniques, and alter the parameters accordingly.
For example the author has developed a fire proof concrete based on Danish materials,
which is undamaged and has c = 1.0 up to a temperature level of 800C, and can sustain a temperature of 1100C.
And it is obvious that any concrete between the siliceous and the fire resistant materials
can be applied for construction.
The calculation methods presented in this book has the big advantage, that they take a
variation of the properties of the concrete into account, whether it is a natural variation
caused by different aggregates or a deliberate variation caused by the development of
fire resistant concretes.
51
REFERENCES
1.
2.
3.
4.
5.
. . . . .
, 1969.
(NEKRASOV K. D. and TARASOVA A. P. Heat resistant concrete of Portland cement. Structural Literature Publisher, Moscow 1969.)
6.
7.
8.
9.
10.
52
11.
12.
13.
14.
15.
16.
CAHILL T. The Behaviour of Prestressing Wire at Elevated Temperatures. Proceedings of a Symposium on Fire Resistance of Prestressed Concrete, Braunschweig, 1965.
17.
18.
19.
KORDINA K. Das Verhalten von Stahlbeton- und Spannbetonbauteilen unter Feuerangriff. Heft 2. Institut fr Baustoffkunde und Stahlbetonbau der Technischen
Hochschule Braunschweig, February 1963. 19p.
20.
21
53
22
23
DS: Danish Code of Practice for Concrete Constructions. DS411, version 4.1.
Danish Standardisation Organisation, 1999.
24
JNICHE, W.U. PUZICHA, W: Technische Mitteilungen des Httenwerkes Reinhausen. Heft 2, Dezember 1953.
25
26
27
DOTREPPE, J.C.: Mechanical Properties of Quenched and Self-Tempered Reinforcing Steel at Elevated Temperatures Compared with Recommendations of Eurocode 2Part 1-2.
Materials and Structures Vol.30, August September 1997. pp.430-438.
28
29
30
31
HERTZ, K.D. SRENSEN, L.S.: Test Method for Spalling of Fire Exposed Concrete.
Fire Safety Journal. Vol.40, No.5, pp.466-476.
Elsevier Science Ltd. July, 2005.
32
54
33
34
35
36
37
38
HERTZ, K.D.: Danish Investigations on Silica Fume Concretes at Elevated Temperatures. ACI Materials Journal Vol. 89, No. 4. pp. 345-347.
Detroit July-August 1992.
39
KHOURY, G.A., MAJORANA, C.E. PESAVENTO, F. SCHREFLER, B.A.: Modelling of Heated Concrete.
Magazine of Concrete Research. Vol.54, No.2,
April 2002, pp.77-101.
40
41
55
42
43
44
45
46
HARMATHY, T.Z. BERNDT, J.E.: Hydrated Portland Cement and Light weight
Concrete at Elevated Temperatures.
Journal of the ACI Vol.63, No.1, pp.93-112.
Research Paper No. 280.
Division of Building Research. Ottawa 1966.
47
FISCHER, R.: ber das Verhalten von Zementmrtel und Beton bei hheren Temperaturen. Deutcher Ausschuss fr Stahlbeton. Heft 214, pp.61-128.
Berlin 1970.
48
49