1

Kristian Hertz

Analyses of Concrete
Structures Exposed to
Fire
Part 2. Structures

Lecture notes
BYGDTU

2007
ISSN 1396-4046

CONTENTS
Contents

Symbols

Ultimate Limit State Analysis of Beams in Bending

Slabs and T-Sections

15

Shear

17

Analysis of Rectangular Cross-Sections

21

Deflections

26

Columns

32

Simplified Column Calculation Method

40

Derivation of Formulas

46

Full Scale Tests

51

Tabulated Data

53

References

55

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

area of an opening of a compartment

area of a concrete cross-section
area of a steel cross-section
tot. area of the surf. of a compartment
thermal diffusivity
parameter proportional to c
a for o in stead of c
calculation value
ultimate value of a
coefficient of total deflection
a function of a: b = (a - 1 + e-a) /a
b for a = ao
thickness of cross-section
effective width = 4/3 C
half period of a fire heat pulse
constant of number i
modified thickness
thermal amplitude
depth of a cross-section
depth of the force of the compr. zone
depth of a normal force N
depth of a prestressing force P
depth of the resultant of P and N
depth of steel reinforcement
average of ds
depth of steel opposite the steel at ds
E-modulus
constant temperature of a fire
initial tangent modulus of concrete
Eco at 20C
tangent modulus of a prestressed steel
E-modulus of a steel
Es at 20C
force
Euler force of a concrete column
force at a concrete cross-section
critical force
Fcr of a plain concrete column
ultimate force of a concrete cross-section
force of a steel cross-section
Euler force of a steel column

Fsu
f
fcc
fcc,ave
fcc20
fct
fct20
fs
fsu
fs20
f0.2
h
I
Ic
Ic20
Is
i
k
l
M
Mo
M1
M+
Mm
mw
N
P
T
TM
Ts
Ts,ave
t
u
V
Vc
Vct
Vs
x
y
z

ultimate force of a steel cross-section

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
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
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
increment
deflection

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

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)

ULTIMATE LIMIT STATE ANALYSIS OF BEAMS IN BENDING

The ultimate limit state analysis has become a common tool
for calculating the load-bearing capacities of structures.
The simple calculations of this state facilitates the design of
the structural elements, and introducing yield hinges an acceptable approximation is often obtained for the determination
of the distribution of moments and forces in hyperstatic structures.
Yet it is a precondition for genuine plastic considerations that
the deformations necessary can take place to ensure the postulated distributions of forces and stresses.
However, the increased plasticity of the concrete exposed to
high temperatures make the analysis even more allowable for
fire-exposed structures than for structures without fire exposure.
In the ultimate limit state prestressed structures are generally
treated like structures with slack reinforcement only, but the
strain of the prestressing steel corresponding to the prestress
must be taken into account calculating the strain of the reinforcement (Brndum-Nielsen [25]).
Furthermore, the following assumptions will be made for the
ultimate limit state analysis in this chapter:
Plane cross-sections are assumed to remain plane, which
means that the resulting changes in strain are assumed proportional to the distance from the neutral axis.
The strain at the ultimate stress of the concrete is

cu20 = 0.35 pct.

cu =
cM
cM
A rectangular stress block is assumed as an idealized distribution of the stresses in the compression zone of the concrete.
The depth of the stress block y is defined as
y=

4
x
5

where x is depth of the neutral axis.

The tensile strength of the concrete is considered to be zero.
All these assumptions are even more allowable for fireexposed structures than for structures without a fire-exposure.

Beam subjected to bending.

Consider a cross-section of a beam with a compression zone of
uniform width confined by a neutral axis, a compressed edge
and two parallel surfaces.
At first the two parallel surfaces are considered to be fire exposed, and the compressed edge is considered insulated, which
are the most common conditions for a rectangular crosssection subjected to a moment load.
Using these conditions the moment is defined as positive, and
if the more compressed edge is exposed to fire the moment is
defined as negative.
The isotherms of the compression zone of the cross-section
subjected to a positive moment are almost parallel to the two
fire exposed surfaces, and the ultimate compressive force per
unit length of the compression zone of width C is
C cM f cc 20
according to the definition of the stress distribution factor .
The compression zone could be considered to be of a constant
width C and loaded by the average ultimate compression stress
cMfcc20.
But just as well it could be considered to be of the reduced
thickness C and loaded by the ultimate compression stress
cMfcc20.

Using this assumption the compressive strength is neglected in

two surface layers of thickness

C
(1 - )
2
and the concrete in the core is considered to be of the same
strength as the concrete at the centre line.
The temperature T is calculated for each reinforcing bar and
the reduced strength s(T)fs20 is determined whether fs is defined as an yield stress, a 0.2 pct. proof stress or an ultimate
stress: the total ultimate force of the reinforcement is found by
a summation of contributions of each bar with cross-sectional
area Asi.
Fsu = i A si si f s 20i

The force is considered to act in a depth ds from the compressed edge of the cross-section, and the depth of the compression zone is calculated as
y=

Fsu
C cM f cc 20

The moment capacity is determined by

M + = Fsu ( d s -

y
)
2

It is a precondition for the determination of Fsu as the total ultimate force that a minimum strain smin is reached in each reinforcing bar.
If the steel has a well-defined yield point the minimum strain
would be
smin =

s f s20
= f s20
s E s20
E s20

i.e. smin is not influenced by the temperature, because the idealized stress-strain curve of the steel is changed by a linear affinity in the strain axis.

10

The strain of the reinforcement is calculated as

s =

5
y 0.35
4
pct .
5

cM
y
4

ds -

and if s is less than smin the cross-section is over-reinforced.

This means that a new force of the reinforcement Fs must be
determined in accordance with the strain distribution of the
cross-section.
Usually over-reinforced cross-sections are not allowed in design, because the structure may collapse without warning by
the formation of cracks in the tensile zone and the development of relatively large deflections.
The risk of emergence of over-reinforced cross-sections is often limited during the fire, as the reinforcing bars are hot and
have small ultimate stresses.
But when the structure is cooled down, the ultimate stresses of
the reinforcement will be regained partly or fully - depending
on whether the steel is cold-worked or not - while the compressive strength of the concrete is subject to a further decrease.
This may lead to over-reinforced cross-sections.
If the structure has been designed for a standard fire exposure
without a cooling phase (as for example proposed by FIP/CEB
[26] and [27] or the structural codes [2] and [4], the designer
will often compensate the small ultimate stresses of the hot reinforcement by application of larger bars. In this case the risk
of achieving a dangerous over-reinforced construction during
the cooling phase of a real fire course is especially high.
Although the author has participated in the development of
both structural codes and the simplified methods developed by
the author has been adopted in them, the codes still supports a
calculation without a cooling phase, because many national
legislations are only concerned with evacuation and not with
the conditions in the cooling phase.
The risk of achieving an over-reinforced construction is also
high if the cross-section is loaded by a negative moment.
The compression zone is then exposed to the fire at three surfaces, and the reinforcement is protected by a large part of the
concrete cross-section. In this case the strength reduction of
the concrete is of a special importance for the load-bearing ca-

11

pacity of the cross-section.

However, the problem in estimating the ultimate moment capacity is more complex, because the isotherms of the compression zone are curved.
It will be a reasonable assumption for practical calculations
that the variation in the reduction of the compressive strength
of the concrete is the same, whether the original strength is reduced by the heat conducted from the two parallel surfaces, or
the average strength due to this variation is reduced by the heat
conducted from the third surface.
This assumption is in accordance with the definition of the
stress distribution factor as independent of the strength of
the concrete at the centre line of the cross-section, cMfcc20.
Using the idealisation that the isotherms are parallel to the two
surfaces until the distance C/2 from the most compressed edge,
which often is the bottom, the average compressive strength at
this level is
cM f cc 20

as it is at all levels above.

12

Beam subjected to bending with reversed moment.

The average compressive strength of the zone within the depth
C/2 from the more compressed edge or the bottom is then
2

cM f cc 20

and the ultimate compressive force of this zone is

C
C cM f cc 20
2
This is achieved, and a proper consideration to the depth of the
resultant force is taken as well applying the average ultimate
compressive stress: cMfcc20 from a level at the distance
C
(1- )
2

from the most compressed edge.

13

Using a concentrated stress block of the compressive strength

at the centre line cMfcc20, the simple procedure is established
that the concrete strength is neglected in the surface layers of
the thickness
C
(1- )
2

from all fire exposed surfaces.

The total ultimate force of the reinforcement is found by a
summation of contributions of each bar with respect to the
strength reductions due to their individual temperatures.
Fsu = i A si si f s 20i
The force is considered to act in the depth ds' from the tensioned edge of the cross-section, the total depth of the crosssection is d, and the depth of the compression zone is calculated as
y=

Fsu
C cM f cc 20

Then, the negative moment capacity is

M - = Fsu ( d - d s -

C
y
(1- ) - )
2
2

Using the model of a concentrated stress block a conservative

estimate is obtained for the strain of the reinforcement

s =

d - d s -

C
5
( 1 - ) - y 0.35
2
4
pct.
5

cM
y
4

14

Positions of concentrated stress blocks.

which should be larger than smin, if the cross-section should
not be deemed to be over-reinforced.
It seems justified to use this rough and conservative estimation
of s, because the risk of achieving an over-reinforced crosssection is especially high in negative bending, and because the
uncertainties of using more laborious procedures would be
large anyway.
The reasons are not only uncertainties in estimating the thermal properties of the concrete, the temperature distribution of
the cross-section and the precise stress-strain curve of the concrete, but also the fact, that spalling and thermal stresses at the
convex corners causes a bevelling of the edges of most fire exposed concrete beams.

15

Slab with top-side insulation and subjected to bending

SLABS AND T-SECTIONS

The total ultimate force of the reinforcement of a slab is in

principle calculated by a summation like the one used for
beams, but often the contributions of all bars are equal, because they are placed in the same depth from the fire exposed
surface and hence will have the same temperature.
In estimating the stress distribution of the compression zone
the problem is different from the one for beams, because the
isotherms are now parallel to the neutral axis.
This means that the strain and the temperature vary simultaneously, and the stresses at different strain levels must be estimated using different stress-strain curves.
However, drawing up the stress distributions of various slabs
provided with various thermal insulation it appears that the
rectangular stress block of a depth equal to 4/5 of the depth x
to the neutral axis still represents a suitable idealisation of the
actual stress distribution with the ultimate stress at the top of
the slab.

16

The depth of the neutral axis becomes

x=

5
5
Fsu
y=
4
4 cM f cc 20

and cM is the reduction of the compressive strength of the

concrete at the top of the slab (one meter is considered).
However, especially if the slab acts as a compression zone of a
beam, i.e. the slab is the flange of a T-shaped cross-section, the
depth y of the compression zone might be so large, that the
concrete is heavily damaged in a part of the compression zone
near the fire exposed surface.
Therefore, the compression zone should not comprise a layer
of damaged concrete of thickness
C
(1- )
2
from the fire exposed surface of the slab.
The dimension C is a modified thickness of the slab, which is
the double of the real thickness if the top of the slab is totally
insulated, and somewhat more if it is not depending on the
ability of heat release from the top. (See the discussion on
modified thicknesses of columns).
Also for a cross-section of a slab loaded by a negative moment
the compression zone should be separated from the fire exposed edge by a surface layer of width
C
(1- )
2
in accordance with the theory for beams with negative moment
load.
Some wide and low beams approximate slabs, and for these
constructions the modified thickness C is determined as the
minimum of the modified slab thickness and the geometrical
thickness of the beam.
The analysis in order to avoid an over-reinforced cross-section
is made using the same procedure as derived for beams.

17

SHEAR

Consider an idealized model describing the statically function

of a simply supported high beam or wall as a compression arch
and a straight tensile reinforcement between the supports.
In each point the shear of the beam is represented by the vertical component of the compression force, and at the supports
the horizontal components are transferred by the main reinforcement.
Decreasing the height of the beam, the horizontal component
of the compression arch force increases, and if the compression zone and the tensile reinforcement are strong enough, the
horizontal component might reach the anchorage capacity of
the reinforcement, and a tensile shear fracture will occur.
The limited anchorage capacity requires a minimum inclination of the compression arch at the supports or a minimum
height of the arch.
In a case where a further reduction of the height of the beam is
wanted, the necessary track of the compression arch would intersect the reinforcement between the supports, and additional
supports may be introduced by the application of links, which
are able to transfer the vertical component of the compression
arch force to the top of the beam from where the arch can continue.
This simple idealized model shows that shear is not an isolated
phenomenon, but it is closely related to the distribution of
stresses in the compression and tension zone.

18

Stresses in a beam subjected to bending and shear.

If no external loads are applied to the part of the beam or wall
considered the compression arch will consist of straight sections, and the model is identical to the truss analogy proposed
by Ritter a hundred years ago.
In design, the anchorage capacity of the main reinforcement at
the support is often required to be at least equal to the reaction
transverse to it corresponding to a traditional assumption of an
45 inclination the compression arch.
Using a variable inclination the anchorage capacity may vary
accordingly.

19

Forces due to shear in a beam.

If the distance between the centre lines of the tension and the
compression zones is ds - dc the capacity due to compression
failure is
V c20 = C ( d s - d c ) cos sin f cc20
where C is the width of the cross-section and fcc20 is the compression strength, and is an empirical effectiveness factor
(Nielsen [29]). According to a theory of the author must be
dependent on the slenderness of the cross section, because it
compensates for the concentration of stresses where the compression arch acts upon the corners of the stirrups.
The shear capacity due to tension of the links of the total
cross-section area As per unit length of the beam or wall is

Vs20 = A s ( d s - d c ) ( cos / sin ) f s20

For a fire exposed construction the temperatures of the links
are easily calculated, because they are often placed in the same
depths from the surface along the entire beam or wall, and only
a few reduction factors s of Asfs20 have to be determined.
The temperature distribution of an inclined compression arch
in a web or a wall of thickness C is often identical to the temperature distribution of the compression zone, and the strength
reduction cM is already determined at the bending analysis.
Near the main reinforcement the average strength is further reduced, because heat penetrating from the bottom influences the
temperature distribution. In this zone the stress concentrations
causing the factor are found, so a further research on for
fire-exposed constructions may be relevant.

20

However, comparisons between calculated and actual shear

capacities indicate that the factors applied for cold calculation also are sufficient for fire-exposed constructions.
Finally the shear capacity is determined as the minimum of the
reduced capacity as to the tension of the links and the reduced
capacity as to the compression of the concrete.

V = min

s A s ( d s - d c ) ( c o s / sin ) f s 2 0
cM C ( d s - d c ) c o s sin f cc2 0

For walls, beams and especially slabs without a shear reinforcement the tensile strength fct of the concrete can be taken
into account, and assuming for example = 45 the shear capacity becomes
V ct20 =

1
C ( d s - d c ) f ct20
2

which should be reduced according to the reduction of fct, and

at least by cM. Reductions of the tensile strength of concrete
are discussed for example in Hertz [12].
The cohesion of the construction should also be estimated
from the temperature distribution.

21

ANALYSIS OF RECTANGULAR CROSS-SECTIONS

The rectangular cross-section is by far the most common

cross-section in reinforced concrete. It therefore appears to be
justified developing special methods for calculating the stressstrain distributions on cross-sections of this basic shape.
In addition such methods could also be applied in a great variety of other cross-sections such as slabs, T-shaped sections and
box sections with none or only few modifications.
In the following methods are developed for calculation of the
complete stress-strain distributions on cracked and uncracked
rectangular cross-sections loaded by eccentric normal forces
acting in the lines of symmetry or bending moments in main
axes. Curved stress-strain curves are applied and tensile
stresses are neglected for the concrete, and elasto-plastic
stress-strain curves are applied and compression stresses are
neglected for the reinforcement.
By means of the same procedure it will be possible to analyse
any cross-section with constant width of the compression zone
reinforced by a symmetrical slack, prestressed or partially
prestressed reinforcement.
The procedure could be a part of a calculation of the deflections of a beam or a slab or the instability of a wall or a column with eccentric load, and the theoretical basis will be the
same as used for calculation of instability of a centrally loaded
column.
Formulas have been developed for the stress-strain distribution
of a rectangular cross-section of concrete loaded by eccentric
normal force acting in the line of symmetry.
Using these formulas the following procedure is made for a
cross-section of constant width C and depth d loaded by the
normal force N acting in the line of symmetry in a depth dN
from the compressed edge.
In case the cross-section is loaded by a bending moment only
the section will be cracked.
Otherwise, the ultimate compression force of the concrete section Fcu and the ratio k between the normal load and Fcu are
found.
Fcu = cdf cc , k = N/ Fcu

22

The calculation parameter a1 and the ultimate value of the variable a called au are found

dk
, a u = E co cu
a1 = d
f cc
- dN k
2
a 1 ( 1 - k ) > 1 - e a1

If

then the cross-section will be uncracked; if not so it will be

cracked.
In case the cross-section is uncracked, the positive value y,
which is at maximum au is found by the equation
1
1
1
dN k
= k - + ( - y ) (1- k )
d
2
y e -1

Stress and strain distribution of

an uncracked cross-section.
and from y the parameter a and the depth of the neutral axis
d is calculated by
y

-1
e
a = ln

y ( 1- k )

x=

ad
y

and

23

The maximum concrete strain c and the curvature is then

f cc a , = c
c =
x
E co
and the stress as a function of the distance z from the neutral
axis is
z

( z ) = f cc ( 1 - e-a x )
In case the cross-section is cracked, it could be subjected to a
bending moment M only.
Otherwise the moment is defined by
M = N ( ds - d N )
where ds is the depth of the main reinforcement from the compressed edge.
The main reinforcement has a total area As and an average Emodulus Es.
The parameter a and the depth x of the neutral axis is found by
solving the equations
xbcf cc = N +

ds - x Es As a
f cc
x
E co

a - 1 + e- a
x x
b=
, dc =
a
2b a
xbcf cc ( d s - d c ) = M
The tensile force of the reinforcement then is
Fs = xbcf cc - N
and the maximum concrete strain c and the curvature can be
calculated as
f cc a , = c
c =
x
E co

24

and the stress is a function of the distance z from the neutral

axis
z

( z ) = f cc ( 1 - e-a x )

Stress and strain distribution of

a cracked cross-section.
For a prestressed cross-section the prestressing is considered to
be included in the normal force N, and the prestressing steel is
included in the total area of the main reinforcement using an
E-modulus and an ultimate stress found by the stress-strain
curve from the strain at the point of prestress.
It is to be checked that the tensile force Fs of the main reinforcement is less than the maximum, whether this is defined by
a yield stress, a 0.2 pct. proof stress, an ultimate stress or a
combination.
In case the parameters a and x are found by an iterative process, the check could be made for each step of the iteration.

25

26

DEFLECTIONS

Large deflections of fire exposed concrete structures do often

occur because of the decrease of the E-modulus and the yield
strength of the reinforcement, and the decrease of the Emodulus of the concrete at any stress level and the increased
plasticity.
The deflections often become so large that the structure may
be deemed to be insufficient, although it still may have a sufficient load carrying capacity.
The integrity of the fire-exposed structure could fail increasing
the risk of the spread of fire and blocking escape routes.
In hyperstatic structures the development of deflections may
cause changes in force distributions provoking collapse.
The deflections and the corresponding stress-strain distributions are important for estimations of possibilities of reusing
fire exposed concrete structures.
This is especially important for prestressed structures, where
the main reasons for using a prestress are to prevent deflections and in some cases to prevent the development of cracks.
And especially these properties of the prestressed structure are
most susceptible to the impact of fire, because the prestress is
usually decreased or lost.
Furthermore the application of a prestressing technique often
leads to relatively slender cross-sections, which are heated
throughout more easily than massive cross-sections.
The prestressing force is regarded as a normal force P in a certain depth dP of the cross-section, and the resulting depth of the
combined action of the prestressing force and the external normal load N in the depth dN is denoted dP+N.
Using inclined prestressing reinforcement the prestress in addition contributes to the shear force, and P is considered to be
the component perpendicular to the cross-section.
In case a normal load N is present, the structure may be regarded as a centrally loaded or cross-loaded column, and the
depth dN of the normal load may often depend on the deflection.
In this case the calculation of the deflections presented in the
following could be regarded as a basis of iteration, and in case
the iteration converge the actual deflection is determined; otherwise the column is not stable.
In this analysis plane cross-sections are assumed to remain

27

plane, and the deflection of the structural member considered

is assumed to depend on the distribution of the curvature only.
Previously it was found that the strain of the concrete could be
regarded as superimposed by strains caused by the instantaneous application of load, the transient conditions, which are the
influence of load and temperature simultaneously, the thermal
expansion and the creep.
Using this model the curvature of a cross-section is assessed
by superimposing curvatures due to the strain contributions
mentioned above.
Since the integration of the sum of the curvatures along a
structural member is equal to the sum of integrals for each partial curvature, the total deflection is most conveniently superimposed by contributions for each strain contribution. And it is
often possible to estimate the deflection of the whole member
due to one cause of curvature from calculation of the curvature
of a single cross-section.
As a basic member, a straight and simple supported beam of
the length 1 is considered, and the curvature of a significant
cross-section is calculated.
The cross-section chosen depends on the load distribution, but
is often the cross section at mid-span.
The temperatures Tsi are calculated for the reinforcing bars and
the corresponding strength reductions si are determined.
The average depth of the tensile reinforcement is calculated
weighted by the strength reductions, which are also assumed to
be the reductions of the modulus of elasticity according to the
idealized model for reinforcing steel.
d s,ave = i si E s20i A si d si / i si E s20i A si

and

E s = si si E s20i A si / A s where A s = i A si

and

s,ave = i si f su20i A si / i f su20i A si

The average steel temperature weighted by the E-modulus is

calculated as
T s,ave = i si E s20i A si T si / A s E s

28

Linear affinity of the stress-strain curve

of a prestressing steel.
In case a prestressing reinforcement is used with a prestress P
the ultimate stress increment fsu - P is reduced by s, and so is
the tangent modulus EP at the level of prestress.
From the previous discussion of stress distribution it is known
that the elastic reaction of a fire exposed concrete section is
nearly equal to the reaction of an idealized section reduced by
the thickness
C
( 1 - 4/3 )
2
from all fire exposed surfaces and made of a concrete of a
quality equal to the concrete at the centre line i.e. with
f cc = cM f cc20 ,

2
E co = cM E co20 ,

cu = cu20 / cM

where C is the width of the original section between two parallel fire exposed surfaces.

29

Analysis of the curvature of a cross-section.

If the cross-section is rectangular or if it fulfils some additional
preconditions the procedure developed in the chapter "Analysis of Rectangular Cross-sections" could be applied using the
data found for the fire exposed concrete and reinforcement fcc,
Eco, cu, fsu, Es, ds,ave and the width of the section
4/3
Ceff = C

By means of this procedure the depth is calculated of the neutral axis x, the maximum concrete stress c and the curvature
due to the instantaneous stress-related strains load.
In case the preconditions are not fulfilled an empirical procedure like the one proposed by Branson [29] could be applied
for the reduced cross-section.
The deflection of the beam is calculated due to the instantaneous stress-related strains, and if the beam is simply supported
with a uniform load distribution the mid-span deflection can be
estimated as
load =

5
2
load l
48

From the stress distribution chapter it is known that redistribution is possible of the thermal stresses of the compression zone
within the limits of the increased ultimate strains of the heated
parts of the loaded cross-section.

30

In order to obtain a reasonable estimation of the thermal expansion of the compression zone it is assumed that the free
thermal expansion of the cross-section initially is reduced to
nil by application of a compression force, then the stresses are
redistributed to the level of the elastic compression stress at
the centre line, and finally the cross-section is reloaded to an
average value of the thermal strain
th =

thM
4/3

Using the values of Ts,ave and ds,ave calculated for the reinforcement and using the same coefficient of thermal expansion
for concrete and steel the curvature due to thermal expansion
becomes

th =

11 10-6
d s,ave

M
( T s,ave - T4/3
)

However, a more detailed expression for the thermal expansion of the actual concrete should be applied for calculations
of instability of walls and columns with eccentric fire exposure.
This curvature is uniform along the beam and the corresponding mid-span deflection is
th =

1
2
th l
8

The curvature due to transient strain is estimated from the

transient strain at the edge of maximum compression of the
concrete section, where the compression stress is c, and the
average thermal strain has been determined.
It is further assumed that the transient strain is nil at depth of
the reinforcement of a deflected beam, and the curvature becomes
TM

2.35 c th
25.9 10-6
=
c 4/3 - 20C
tr =
d s,ave f cc20
d s,ave f fcc20

which varies along the beam as c and almost as the bending

moment. The mid-span deflection of a uniformly loaded and

31

simply supported beam is estimated as

tr =

5
2
tr l
48

If the concrete is compressed to the depth of the reinforcement

ds,ave, i.e. if ds,ave < x, the depth ds,ave could be substituted by the
depth of the neutral axis x, and
tr =

TM

25.9 10-6
- 20 if d s,ave < x
c 4/3
x f cc20

If the concrete is so damaged that the creep strain of the concrete compression zone is of any major importance, the deflections will be large, and the structure will usually be close to
collapse.
So, the deflection due to concrete creep could often be neglected, because the calculation of the load-bearing capacity
and the calculation of the other contributions to the deflection
would clearly show that the characteristics of the fire-exposed
structure are insufficient.
However, in case a conservative approximation of the curvature due to creep is preferred, it could be
creep =

530 10-6
d s,ave f cc

t 3.04 T M
e
3h

1000

where ds,ave is substituted by x if ds,ave < x, and t is the time of

heat- or fire exposure in hours, and the deflection for uniform
load is estimated as
creep =

5
2
creep l
48

The formulas cannot be applied to long time heat exposures

with t more than say 12 h, and after the cease of heat exposure,
a separate creep investigation could be made for the weakened
concrete.

32

COLUMNS

Because the reduction of the E-modulus of a heated concrete is

large compared to the reduction of the compressive strength,
the risk of buckling failure is increased compared to the risk of
compression failure, when a concrete column is exposed to
fire.
The fast heating of slender cross-sections and the large reduction of the E-modulus make slender concrete columns susceptible to special fire hazards although their load-bearing capacities might be independent of the use of reinforcing steel.
This matter clearly shows the importance of taking into account the weakening of the concrete in a precise manner and
rejecting any traditional opinion regarding concrete as a "fire
safe" material.
In the chapter on stress distribution the weakening of a concrete cross-section of width C has been discussed, and it was
found that
2

C/2

C/2

2
E co dz = 2 0 c dzE co20

4/3

cM E co20 C

and
2

C/2

2
z c E co20 dz

1
4
2
C3 cM E co20
12

i.e., the average elastic response as well as the moment of inertia could be calculated for a reduced cross-section of thickness
4/3C with the uniform modulus of elasticity cM2Eco20.
It is therefore most convenient to make an assumption for rectangular concrete columns similar to the one introduced for
beams: that the variation of the weakening through the depth
from one side caused by the heat conducted from this side will
be equal for all fire exposed sides of the cross-section.

33

Consequently, the reduced column or wall is obtained by neglecting the contributions of the concrete in surface layers of
the thickness
C
( 1 - 4/3 )
2
from all fire exposed surfaces, and by using the properties of
the concrete at the centre line of a cross-section of width C
equal to the smallest dimension of the column or equal to the
wall thickness.
In case a concrete column with a rectangular cross-section of
depth d and thickness C (where C < d) is exposed to fire on all
four sides, the product of the moment of inertia and the
modulus of elasticity becomes
Ic E co =

1
4/3
4
2
( d - C ( 1 - ) ) C3 cM E co20
12

and the area of the reduced cross-section becomes

Ac = ( d - C ( 1 -

4/3

) ) 3 C

If one or more surfaces of the column are insulated the thickness C should be modified subtracting a layer thickness of
C/2(1 - 4/3) from the fire exposed surfaces.
For a rectangular column the thickness C is determined as the
smaller of the two thicknesses of the column modified for the
ability of heat conduction.
A modified thickness is the double of the geometrical thickness if one surface is totally insulated. It will be infinite if both
surfaces are insulated totally, and if both surfaces are fire exposed it will be the geometrical thickness.

34

Modified thicknesses of columns.

35

It is the ambition to establish a total system for analysing concrete structures comprising methods for calculating deflections
of beams, deflections and instability of cross-loaded columns
and instability of centrally loaded columns, all methods based
on the same assumptions.
This system is advantageous, because it ensures homogeneity
in the calculational treatment of the structural members, which
is especially required for the eccentrically loaded or crossloaded columns.
Further it ensures continuity in the calculations from the load
case Fire to other load cases, and the system offers simplicity
to the every day work of the engineer, who is then able to use
the same procedures or subroutines on his computer, whether
he deals with a beam or a column, and whether the structure is
fire exposed or not.
Finally the theory will fit into the procedures used for the ultimate limit state analysis of beams, because the load causing
the ultimate stress-strain conditions will correspond to the
load-bearing capacity obtained from the ultimate limit state
procedures presented.
The calculation of the deflections of beams has been treated,
and the basic assumption was the application of Ritters expression for the modulus of elasticity of concrete loaded by a compression stress :
E c = E co ( 1 -

f cc

and consequently the stress-strain relation:

E co

= f cc ( 1 - e- f cc ) .

For the special case of a centrally loaded column simple design

procedures can be developed for the determination of the loadbearing capacity.

36

Consider a plane simply supported concrete column or wall

without reinforcement.
The cross-section is symmetrical as well as the fire exposure,
and therefore the initial curvature of the column will be zero,
when a central column force Fc is applied.
The stability of the column is assessed at a given time of the
fire exposure by applying a sinusoidal deflection curve with an
infinitesimal mid-span deflection u.
In instantaneous stress-related deflection due to the sinusoidal
moment distribution is
= load =

uFc 2 ,
l
Ic E c

using the moment of inertia of the reduced concrete section

and the E-modulus of the loaded concrete
Fc ) , with
E c = E co ( 1 Fcu = A c f cc
Fcu
If < u the deflection will be reduced until the column is
straight, and the column is stable.
If > u the column is unstable, and if = u the applied column
load is the critical load Fcrc
2

Fcrc

= Ic2 E co ( 1 - Fcrc ) = FcE ( 1 - Fcrc )

l
Fcu
Fcu

where FcE is the Euler force of the fire exposed column.

We obtain the well-known Rankine formula
1
Fcrc

1
Fcu

1
FcE

1
A c cM f cc20

l
2
Ic E co

and from the discussion it is seen that the formula is just as

valid for a fire exposed column as for a column without a fire
exposure. Usually the Rankine formula has a more complicated expression. The author has developed the simple expression presented in order to make it easier to understand.

37

Furthermore, this formulation divided in two contributions is

advantageous, because it even indicates the failure mode of the
fire-exposed column.
If Fcu > FcE the column will be subject to a buckling failure,
and if FcE > Fcu it is most likely that the failure mode will be
described as a compression failure.
This is in accordance with the observations noted in the test
reports for the fire-exposed columns.
In a case where the reinforcement of a column is taken into account, the ultimate compressive yield- or 0.2 pct. proof force is
summarised as
Fsu = i A si si f s20i

In this summation, only reinforcement sufficiently embedded

in the column should be taken into account.
The longitudinal compression reinforcement is often placed
near the surfaces of the column in order to improve the moment of inertia of the column cross-section as much as possible.
However, the positions near the surfaces will give rise to high
maximum temperatures of the reinforcing bars and of the concrete surrounding them and especially of the cross reinforcement transverse to them, being placed closer to the fire exposed surfaces.
The summation of Fsu shows the reduction of the ultimate
compressive force of the reinforcement; but it does not show
the risk of instability of the reinforcement.
For example it may be required that the reduced ultimate tensile force of the cross reinforcement should be at least 15 pct.
of the compression force taken into account for each longitudinal bar over any partial length of 15 times the diameter of
the longitudinal reinforcing bar.
Furthermore it should be required that the concrete surrounding the bar is not totally damaged, and that for example at least
10 pct. of the compressive strength is left at the temperature of
the longitudinal bar.
The yield point of mild steel and hot rolled bars disappears at
rather low temperatures, and cold-worked bars do not have any
yield point at all.
This means that the E-modulus of the reinforcement like the Emodulus of the concrete decreases, when a compression stress

38

is applied.
Therefore, it could be proposed to use a curved-lined relation
between the force and the strain of the reinforcement.
Using the same possible deflection curve as used for plain
concrete columns the critical column force will be
2
IE
Fcr =
2
l

and an approximate stiffness of the column cross-section is

proposed to be
IE = Ic E co ( 1 -

F
F
) + Is E so ( 1 )
Fcu + Fsu
Fcu + Fsu

being suitable for columns fulfilling the requirement

Is A s
Ic
Ac
which most columns do, because the reinforcement usually is
placed so that its moment of inertia is relatively large.
If the reinforcement has a yield point, the strain of the column
might be more than the yield strain; in this case, however, it
must be remembered that the strain will decrease for almost
half of the reinforcing bars if the loaded column deflects, and
these bars will act with their full E-modulus at the occurring
large column forces for which
Fc = F - Fsu >
( 1-

1
( Fcu - Fsu )
2

F
1
)<
2
Fcu + Fsu

thus the reduction of the stiffness of the reinforcement is safe.

For the concrete we obtain using F = Fc + Fsu :
Fc Fcu + Fsu Fcu > Fc Fcu + Fc Fsu

39

(1-

F
) < ( 1 - Fc )
Fcu + Fsu
Fcu

and the reduction of the concrete stiffness is safe as well.

If the column strain is less than the yield strain all bars are in
elastic state, and for small values of the critical column force
and relatively small values of Is the reduction of the concrete
stiffness might be too small, and the approximation of IE
might be slightly unsafe; but if Is/Ic As/Ac the deviation is
found to be within 5 pct.
So, the approximation is acceptable provided that Is/Ic As/Ac.
Defining the Euler forces of concrete and reinforcement as
2

FcE =

Ic E co
Is E so
and FsE =
2
2
l
l

the stiffness of the reinforcement is calculated by summation

for the same bars taken into account when calculating the ultimate compression force of the reinforcement Fsu
2
Is E so = i z i A si si E so20i

where zi is the lever arm of bar number i.

The critical column force becomes
Fcr = ( FcE + FsE ) ( 1 -

Fcr
)
Fcu + Fsu

and an extended Rankine formula can be derived as

1
Fcr

1
1
+
if Is A s
Fcu + Fsu
FcE + FsE
Ic
Ac

The formula appears to be the simplest possible extension of

the Rankine formula for plain concrete columns and is just as
advantageous in use.

40

SIMPLIFIED COLUMN CALCULATION METHOD

The simplified calculation methods are made in order to allow calculation by hand of the load bearing capacity of constructions made
of any concrete at any time of any fire exposure. This is, as it may
be seen, a much wider area of application than most simple procedures have. The tabulated data of the Eurocode CEN [2] and more
previously proposed calculation methods are restricted to cover
fixed times of a standard fire without a cooling phase and constructions made of a fixed material, which most often is chosen to be
German siliceous concrete. By means of such methods it is impossible to benefit from good properties of the actual concretes or to
improve the properties by designing concrete with an adequate fire
resistance. By means of such methods it is furthermore impossible
to meet reasonable requirements of the clients or apply a function
based fire safety design, as it will become more and more common
to do in the years to come.
Furthermore, the calculated load bearing capacity should vary continuously towards the capacity found by the cold design (except for
the influence of safety factors) when the fire varies towards no fire.
This is obtained by using a reduced cross-section, where the capacity of the concrete in a damaged zone is neglected and the concrete
in the reduced section is considered to be of a uniform quality.
The concrete quality is assessed as the one found in the middle of
the cross-section. This is defined as the centre line of a two-sided
exposed wall of the same thickness as the smallest dimension of the
fire exposed cross-section. The thickness of the damaged zone of
the cross-section is assessed as the damaged thickness for the same
wall. This is because a rectangular cross-section has isotherms
almost parallel with the long sides and is therefore in the central
part like a two-sided exposed wall of thickness equal to the length
of the short side.
For column calculations the reduced cross-section is expected to
model the reduction of the stiffness of the actual fire exposed crosssection.
By studying the relations between the temperature dependent
strength and E-modulus of more than 170 test series the author has
found that they all fit satisfactory to the following idealized expression:
Ec = c2Ec20
where

41

Ec is the E-modulus of the concrete at the temperature level in question,

Ec20 is the E-modulus of the concrete at 20C,
c is the reduction of the compressive strength,
fcc = cfcc20
where
fcc is the compressive strength of the concrete at the temperature
level in question,
fcc20 is the compressive strength of the concrete at 20C.
This finding is the necessary key for deriving calculation methods
for fire exposed columns.
At first the axial stiffness of a two-sided exposed wall is considered.
By means of the expression above this is found to be

E c,ave =

1 C 2
c E c20dx
C 0

where
Ec,ave is the average E-modulus across the cross-section,
C is the thickness of the wall,
x is the depth from the surface,
cM is the reduction of the compressive strength in the middle.
If the average compressive strength throughout the cross-section is
called cMfcc20 which means

cM f cc20 =

1 C
1
cf cc20dx or =

C 0
C cM

c dx

Then the author has found an empirical approximation for the average axial stiffness as

E c,ave = 4/3 cM 2 E c20

which means that

4/3

C
1
2dx
2 0 c
C cM

42

The cross-section is thereby reduced with the damaged zone

az =

C
(1-4/3 )
2

From all fire exposed sides, and the E-modulus in the reduced
cross-section is uniformly cM2Ec20
The flexural stiffness of this cross-section is then per unit length
EI

1 4/3 3 2
( C) cM E c20
12

EI

1 4 3 2
C cM E c20
12

The following figures validate these two approximations. They

show the relations between the axial and flexural stiffness found as
4/3

2
10 cM

10

2z and 4 cM 2
i=0 c

2
1 3
C
12

10
i=0

z i 2 c 2z

where zi is the distance to the z thick lamella no. i, and where the
reduced cross-section is calculated by dividing the half crosssection into 10 layers and summarising the contributions.
The calculations are made for wall cross-sections of width of 0.15
m, 0.20 m, 0.40 m and 0.60 m of Danish concrete with thermal diffusivity 34810-9m2/s. The approximations are shown at a hot and a
residual condition during four fully developed fires with (opening
factor, fire load) = (0.04,200), (0.04,400), (0.08,400), (0.08,800)
(m1/2,MJ/m2) which covers standard fires as well as a suitable variation of other fires.
The figures on the next page show that the reduced cross-section
leads to a good approximation for the axial stiffness and for the
flexural stiffness.
This reduced cross-section is therefore applied for calculation of
columns and walls and it may also be applied for calculation of deflection of beams.
The calculation methods may be the same as used for 20C calculations but taking into account the properties of the reinforcement
and the concrete at the high temperatures which occur in the crosssection. These properties may be the correct properties of the actual material.

43

1,20

Axial stiffness by lamellas

1,00
0,80
0,60
0,40
0,20
0,00
0,0

0,2

0,4

0,6

0,8

1,0

Axial stiffness calculated by reduced cross section

Comparation between axial stiffness from detailed calculations and

reduced cross-section.

Flexural stiffness by lamellas

0,8

0,6

0,4

0,2

0
0,0000

0,2000

0,4000

0,6000

0,8000

1,0000

Flexural stiffness calculated by reduced cross

section

Comparation between flexural stiffness from detailed calculations

and reduced cross-sections.

44

In practise some idealized models for the properties can be adopted

and values assessed as conservative values provided by codes or
textbooks.
For the concrete the loss of stiffness is modelled by the reduced
cross-section, where the E-modulus is considered to be uniformly
cM2Ec20 and the compressive strength cMfcc20.
For cross-loaded columns or columns not exposed to fire on all
sides, a thermal deflection must be added to the eccentricity. To
calculate this, the so-called transient strains must be considered.
This can be explained as the part of the thermal elongation that does
not take place, because the concrete is loaded during the process of
heating. When the concrete is heated, the aggregate expands, and
the cement paste shrinks, and at 300C micro cracks are formed in
the cement paste leading to a non-reversible thermal elongation. If
the concrete is loaded in compression, the compression stresses in
the matrix must be unloaded before tensile stresses are introduced,
and therefore the formation of micro cracks is delayed and reduced
and the thermal elongation is reduced. From this model of the
physical behaviour it seems obvious that the sum of the thermal expansion and the transient strain cannot become negative, since a
crack can hardly be more than closed. However, some results ([22]
and [30]) indicate that lager transient strains may occur, although it
is difficult to judge whether some of it is caused by altered constitutive conditions, errors of measurement or shrinkage. If such larger
transient strains are able to occur a modification is called for to
model the redistribution of stresses and strains, which the curved
temperature profile leads to. If the thermal deflection reduced due to
transient strains are calculated by means of a curvature assessed by
means of the strain at two points of a reduced cross-section, at
rough approximation seems to be valid not allowing the sum of
thermal expansion transient strain to be negative.
The assessment of the transient strain as well as any other material
property of the actual concrete is not a part of the simplified column calculation method, but input for it.

45

Regarding the reinforcement, the actual stress strain curve of the

steel at high temperatures may be used for the calculation. If the
steel has a yield point, this will disappear at about 300C and the
stress strain curve becomes a curved line. If a bilinear stress strain
curve is used as a simplified material model, this simplification is
the same as if a bilinear curve is used for steel without a yield point
at a calculation of a column at 20C. In some codes this is allowed.
The proposed method for calculation of fire-exposed columns therefore leaves this judgement to the designer, and the method works
well with curved lined as well as bilinear stress strain curves. This
means that the approximation if a bilinear curve is used corresponds
to the one at 20C, and this accords with the demand that the
method should vary towards the results at 20C if the fire varies towards no fire.

46

DERIVATION OF FORMULAS
Many empirical and semi-empirical formulas exist for concrete columns based on experience for constructions at 20C. However, in
order to take the influence of the changed material properties at
high temperatures into account in a reasonable way, it is preferable
to use derived expressions based on a more profound understanding
of the mechanical function.
Some formulas are therefore derived in the following in order to establish a suitable basis for calculation of reinforced concrete columns subjected to fire. Besides of this application, some of the formulas seem to be beneficial for calculation of the load bearing capacity at 20C, since they are easier to remember and understand
than most other formulas.

CENTRAL LOAD
F
A sinus deflection is presumed of a simply supported, axially loaded
column with a stress-independent E-modulus E0.

dy

d2y
2

y = sin( x),
= cos( x),
= - 2 sin( x),
2
L
dx L
L
dx
L
L
2
2
2
d y

E0I
:=
.
-M
=
F
=
-E
I
=
E
I,
F
=
0
0
E
dx 2
L2
L2

This force is known as the Euler force. For a system consisting of

concrete and reinforcement both systems can be considered to deflect equally, i.e. and are equal for reinforcement and concrete,
and the two contributions to the moment and the Euler force may be
added. FE = FsE + FcE.
As mentioned, it is a precondition for the Euler expression that the
stress strain curve is linear for concrete and reinforcement. Especially for concrete, but also for fire-exposed reinforcement it is
more reasonable to adopt a curved lined stress strain curve.

fcc

Ec0

Ec

Ritter proposed an expression for an idealized curved lined stress

strain curve as E = E 0 (1 ) which means that the E-modulus at

fu
the stress level is reduced by one minus the ratio between the
stress and the strength.

47

If this is inserted into the Euler expression a formula for equilibrium at a stress level is obtained as
2 IE 0
F
1
1
1
FR =
(1- ), where F = A and Fu = f u A.
=
+ .
2
L
Fu
FR
Fu
FE

This is a formulation of the well-known Rankine formula, which is

easy to remember.
A much more complicated formulation of exactly the same formula
i found in the Danish code DS [4] in clause 6.2.5.1(4).
If the column contains reinforcement with a curved lined stress
strain curve, the Rankine formula may be extended simply as:
1
1
1
=
+
FR
Fsu +Fcu
FsE +FcE
This states that the reverse of the Rankine force is the reverse capacity of the cross-section plus the reverse sum of the Euler forces
for reinforcement and concrete. Since the stress strain curves of all
reinforcement become curved lined when exposed to high temperatures, this formula may apply for fire calculation of columns no
matter which kind of reinforcement they contain.
This replaces the clause 6.2.5.1(5) in the Danish code DS [4] in the
calculations of full-scale tests. (In [4] the reinforcement can be
placed anywhere, which cannot be right).

48

ECCENTRIC LOAD
For an eccentric loaded column a condition of equilibrium is made
similar to the one for central load. This leads to
2
2E0I
F
-M = -EI (e+)F = 2 EI (e+)F =
(1- )
2
L
L
Fu

F
e

F
F F
= (1- - )
FE
Fu FE

F
F F
1
+
F
F
Fu
FR FE
e+=e
e
= (1- ) e+ = e
F
F
FE
FR
11
FR
FR
1-

This formula is equal to for example the one found in the Danish
code proposal [4] as 6.2.5.2(13)

COMPRESSED CROSS-SECTION
e+
F

2(e+)

Assuming a stress block to represent the ultimate capacity of the

cross section
F
1
F h 2(e + )
F
e
Fu
=

= 1 2
Fu
h
Fu
h 1 F
FR

F
e
e
=2
F = FR (1 2 )
FR
h
h

h
This corresponds to some extent to formula 6.2.5.2(17)
in the Danish code [4]

49

CROSS-SECTION WITH TENSION AND COMPRESSION

If the cross section cannot sustain its load as compressed, the deflection is assessed from the curvature calculated from the strain in concrete and reinforcement:

e+
F

Fs1

1 cu + sy 2
L , Ms = ds1Fs1 ds2Fs2, Fs = Fs1 Fs2,
10 hef

h
(F + F)
s
M = F( e ), M = Ms (F + F)
,
s
2
2Cfcc
ds1

ds2
Fs2

h
F2 + (2Fs ( e )2Cfcc )F + Fs2 Ms 2Cfcc = 0
2

where C is the width of the cross section.

b + b 2 4Fs 2 + 8Cf cc M s
h
If b = 2Fs ( e )2Cf cc then F =
2
2

ADDITIONAL ECCENTRICITY CAUSED BY THERMAL EXPOSURE

c1

CROSS SECTION WITH TENSION AND COMPRESSION

c1

1 1.1 105Ts k tr1 th (Tc )) 2

1 2.35
f cc20
e th =
L where k tr1 = max
8
h eff

heff

COMPRESSED CROSS SECTION

c1

c2

k tr1

1 k tr2 th (T2 ) k tr1 th (Tc )) 2

L where
8
h
c1

1 2.35

f cc20 and k tr2 = max

= max

e th =

c2

1 2.35

f cc20

th(T) is the thermal strain of the actual concrete at the temperature

T.

50

ktr1 and ktr2 are factors taking the transient strain into account. The
values correspond to those proposed by Anderberg and Thelandersson [22] and approximately in accordance with the first part of the
so called Master curves by Khoury, Grainger and Sullivan [30].
The condition > 0 provides a rough model taking into account the
redistribution of stresses and strains caused by the non-linear temperature distribution.
The expressions above have proven to model the thermal deflection
well for columns exposed to fire at all four sides.
For walls, and especially for walls exposed to fire on one side and
eccentrically loaded with the compression zone towards the fire, the
transient strain development in time has to be modelled, and it is
recommended to apply an iterative calculation of the thermal deflection and the stress distribution. In the last case the thermal bending counteracts the bending from the static moment applied, and
thereby reduces the eccentricity from this. The compression stresses
are therefore reduced, and the contribution to the thermal deflection
from the transient strains may be reduced.

e F
One primitive way of estimating when the reinforcement may be considered to be tensioned is when

e>

cs

h-cs

h + 2cs
6

51

FULL SCALE TESTS

Since the simplified calculation method for fire-exposed columns is
derived to model the reduced flexural and axial stiffness of a crosssection, this already represents some evidence for its applicability.
In order to extend this evidence, results of full-scale tests have been
compared to the results of the method. 50 tests have been recalculated representing variations in column length, cross-section dimensions, reinforcement, concrete quality, axial load and eccentricity of
load and of fire exposure.

The results are shown in the figure below.

45 of the tests are reported in Hass [31]. (Two column tests have
been neglected because the test results were very deviating from the
test results of columns with almost the same properties and load,
and obviously some test error must have occurred).
2 columns are from Lie et.al. [32] (Also shown in Hertz [1] and [3])
2 tests are from Seekamp et.al. [33] (Also shown in Hertz [1] and
[3])
1 tree sided exposed column is from Anderberg and Forsn [34].
The first 45 columns are all calculated by the same spreadsheet (in
Mathcad) and thus results of the same algorithm.
The other 5 tests have been calculated by hand using the same basics formulas.
Generally the calculation results for the 50 columns are on the safe
side compared to test results, and a reasonable agreement is observed. It must therefore be concluded that the calculation method is
suitable for design purpose.

52

Average fire resistance at test 80

Average deviation in minutes 6,9
Standard deviation in minutes 11,7

160

Test resistance in minutes

140
120
100
80
60
40
20
0
0

20

40

60

80

100

120

Calculated resistance in minutes

Comparation between calculations and 50 full scale
fire tests of concrete columns.

140

160

53

TABULATED DATA
The tabulated data from ENV1992-1-2 (reference [2]) Table 4.1 refer to siliceous concrete columns exposed to a standard fire without
a cooling phase. The German siliceous concrete must be considered
to be the concrete, which has the highest thermal conductivity and
the largest drop of compressive strength i Europe, and application
of the tabulated data should therefore be conservative for other concrete materials.
For the 45 tests from Hass [31] the siliceous concrete material may
be considered to be the one used for the tabulated data, and for
these columns the fire resistance time is assessed by means of the
table.
The assessments are included in the table of the column calculations
at line 2 from the bottom. It seems that these assessments are generally on the unsafe side compared to the tests results.
It seems that the method of the tabulated data does not consider the
influence of an eccentricity of the load and is not able to consider
eccentric fire exposure.
Furthermore, the tabulated data method does not consider the slenderness of the column. This is the most obvious reason for the deviations. It is confirmed by the fact that in general the short columns are assessed to have a fire resistance time closest to the time
at test, while the most slender columns are in general assessed very
much on the unsafe side.

54

160

Test resistance in minutes

140
120
100
80
60
40
20
0
0

20

40

60

80

100

120

140

Tabulated resistance in minutes

Comparison between tabulated data and 45 full scale fire tests of

concrete columns
Average fire resistance at test 74
Average deviation in minutes 36,3 (unsafe)
Standard deviation in minutes 26,2

55

REFERENCES

[1] Hertz, K.: Analyses of Prestressed Concrete Structures Exposed to Fire.

Report No. 174, Institute of Building Design,
Technical University of Denmark. 1985. CIB document W14/85/9 (DK).
[2] CEN: ENV 1992-1-2. Design of Concrete Structures; General Rules, Structural
Fire Design.
Secretariat of CEN/TC 250/SC 2. 1994.
[3] Hertz, K.: Simplified Calculation Method for Fire Exposed Concrete Structures.
Supporting Document for CEN PR-ENV 1992-1-2.
Report No. 203, Institute of Building Design,
Technical University of Denmark. 1993.
Published by CEN together with ENV 1992-1-2.
[4] Danish Standardisation Organisation: DS411ver. 4.1. Danish Code of Practice
for Concrete Constructions.
DS, 1999.
[5] Skinner, D.H: Measurement of High Temperature Properties of Steel.
Melbourne Research Laboraties. MRL 6/10. May 1972.
[6] Harmathy, T.Z. Stanzak, W.W.: Elevated Temperature Tensile and Creep Properties of some Structural and Prestressing Steels.
Nat. Resrearch Council of Canada.
Research paper No. 424, NCR 11163.
Division of Building Research.
Ottawa 1970.
[7] Danish Standardisation Organisation: DS412 ver.2. Danish Code of Practice for
Steel Constructions.
DS, 1983.
[8] Cahill, T.: The Behaviour of Prestressing Wire at Elevated Temperatures.
Proc. of Symposium on Fire Resistance of Presteressed Concrete.
Braunschweig 1965.
[9] Dannenberg, J.: Report on Methods of Assessment of the Fire Resistance of
Concrete Structural Members.
Cement and Concrete Association. 91p.
Wexham Springs 1978.

56

[10] Voves, B.: Behaviour of Prestressed Concrete Structures during Fire.

Czechoslovakian Technical University. Prauge 1978.
[11] Anderberg, Y.: Behaviour of Steel at High Temperatures.
RILEM-Comittee 44-PHT. Report 114p.
February 1983.
[12] Hertz, K.D.: Betonkonstruktioners brandtekniske egenskaber.
(Properties of Fire Exposed Concrete Structures) Part 1 and 2 of Ph.D. Thesis.
Report No. 140. Institute of Building Design, 210p in Dansih.
Technical University of Denmark. Lyngby July 1980.
[13] Hertz, K.D.: Heat-Induced Explosion of Dense Concretes.
Report No. 166. Institute of Building Design, 20p.
Technical University of Denmark. Lyngby February 1984.
CIB W14/84/33 (DK).
[14] Hertz, K.D.: Limits of Spalling of Fire Exposed Concrete.
Research Paper, Department of Buildings and Energy. 11p.
Technical University of Denmark. Lyngby November 1998.
[15] , .. , ..: .
(Fire Resistant Concrete on Portland Cement). In Russian 192p.
.
1969.
[16] Muff, M.: Brandbestandig beton baseret p mler.
(Fire Resistant Concrete Based on Mo-clay.)
M.Sc. thesis in Danish..
Institute of Building Design, 138p.
Technical University of Denmark. Lyngby 1983.
[17] Hertz, K.D.: Microwave Heating for Fire Material Testing of Concrete
- An Experimental Study.
Report No. 164. Institute of Building Design, 38p.
Technical University of Denmark. Lyngby August 1983.
CIB W14/83/19 (DK).
[18] Schneider, U.: Festigkeits- und Verformungsverhalten von Beton unter stationr
und instationrer Temperaturbeanspruchung.
Die Bautechnik heft 4. pp.123-132.
Wilhelm Ernst und Sohn. Berlin 1977.

57

[19] 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.
[20] Harada, T. Takeda, J. Yamane, S. Futumura, F.: Strength, Elasticity and Thermal
Properties of Concrete Subjected to Elevated Temperatures.
ACI, SP-34, pp.377-406.
Detroit 1972.
[21] Fischer, R.: ber das Verhalten von Zementmrtel und Beton bei hheren Temperaturen.
Deutcher Ausschuss fr Stahlbeton. Heft 214, pp.61-128.
Berlin 1970.
[22] Anderberg, Y. Thelandersson, S.: Stress and deformation characteristics of concrete at high temperatures.
2. Experimental Investigation and Material Behaviour Model.
Bulletin 54, Lund Institute of Technology. Lund 1976.
[23] Hertz, K.D.: Stress Distribution factors.
Report No. 158. Institute of Building Design, 60p.
Technical University of Denmark. Lyngby July 1981.
CIB W14/81/14 (DK).
[24] Ruge, J. Linnemann, R.: Festigkeits- und Verformungsverhalten von Bau-, Beton- und Spannsthlen bei hohen Temperaturen.
Sonderforschungsbereich 148, B4. p.159-221.
Technische Universitt Braunschweig. 1983.
[25] Brndum-Nielsen, T.: Structural Concrete.
Polyteknisk Forlag, 136p.
Copenhagen, January 1973.
[26] FIP/CEB: Report on Methods of Assessment of the Fire Resistance of Concrete
Structural Members.
Cement and Concrete Association, 91p.
Wexham Springs 1978.

58

[27] CEB: Design of Concrete Structures for Fire Resistance.

(Draft of an Appendix to the CEB/FIP Model Code on Concrete Structures).
Bulletin d'information No. 145.
Comit Euro-International du Beton. Paris January 1982.
[28] Nielsen, M.P.: Limit Analysis and Concrete Plasticity.
Prentice-Hall Inc. 420p.
New Yersey 1984.
[29] Branson, D.E.: Design Procedures for Computing Deflections.
ACI Journal, September 1968, pp.730-742.
[30] Khoury, G.A. Grainger, B.N. Sullivan, P.J.E.: Strain of Concrete during First
Heating to 600C under Load.
Magazine of Concrete Research, Vol.37, No.133, December 1985. pp.195-215.
CIB W14/86/45 (UK)
[31] Hass, R.: Zur Praxisgerechten Brandschutztechnischen Beurteilung von Sttzen
aus Stahl und Beton.Heft 69.
Institut fr Baustoffe, Massivbau und Brandschutz der Technishen Universitt
Braunschweig. Braunschweig 1986.
[32] Lie, T.T. Allen, D.E. Abrams, M.S.: Fire Resistance of Reinforced Concrete
Columns.
DBR Paper No. 1167.
National Research Council Canada. Ottawa 1984.
[33] Seekamp, H. Becker, W. Struck, W.: Brandversuche an Stahlbetonfertigsulen.
Deutscher Ausschuss fr Stahlbeton. Heft 162, Teil 1.
Wilhelm Ernst und Sohn. Berlin 1964.
[34] Anderberg, Y. Forsn, N.: Fire Resistance of Concrete Structures.
Nordic Concrete Research No.1. Oslo 1982.