Real scale fire test on an industrial hall in prefabricated concrete members
Prof. Dr. Ir. L. Taerwe, Dr. Ir. A.
Magnel Laboratory for Concrete Research, Department of Structural Engineering, Ghent University
In this paper attention is paid to:
the fire resistance of prefabricated concrete members
the temperature development during a real fire with heavy fire load
the overall structural behaviour of
the assessment o
f the structural concrete members during and after the fire
Summary of the contents of the paper
This paper deals with a real scale fire test
carried out in 1974,
on an industrial hall in prefabricated
concrete members, raised especially for this
test. A team of researchers and specialists in this field had
done extensive preliminary investigations concerning the temperature curves to which this building
would be submitted. Fifteen producers of prefabricated concrete elements cooperated to the pro
and produced the different components of the building.
This experiment was performed in order to gain knowledge concerning the behaviour of buildings
composed of prefabricated concrete elements during an actual fire. In this paper results are gathered
concerning the temperature development during the fire and the overall behaviour of the building
during and after the fire.
im of the experiment
Under normal circumstances, a fire is an accident
, an unforeseen event
y for reasons we can only examine afterwards.
In this paper, a real scale fire test on an
industrial hall in prefabricated concrete elements, raised especially for this test, is described. To the
authors’ knowledge, this is the first time that the test res
ults of this experiment will be presented to an
full scale experiment was performed in order to gain knowledge concerning the behaviour of
buildings composed of prefabricated concrete elements during an actual fire.
In those day
he regulations required 2 hours fire resistance for structural elements in high rise buildings. The
requirement was the same for prefabricated and for cast
place concrete. The difference was that
often only for prefabricated elements test e
vidence was required
place concrete this was
not required and the fire resistance could be assessed by analytical methods or by means of appropriate
for the relevant parameters. These latter methods appeared to yield more favo
results than fire tests, thus overestimating the fire resistance. Hence, the prefabricated concrete
producers appeared to be in a disadvantageous position.
This scientific test on the
prefabricated structure, with commercially available elements and
actual connections, with realistic loads, submitted to a completely controlled fire, had to yield
information concerning :
the general assumption that
structures offer a very good fire resistance, which appears to
be substantially higher
than the fire resistance of the separate elements during laboratory tests,
the influence of a series of parameters on the behaviour of the elements during fire,
the influence of certain factors that can’t be examined during a fire test in a laboratory,
but can be
important for the resistance of the structure, for example the thermal deformations,
the comparison of the fire resistance of a series of elements submitted to a real fire and the results
of laboratory tests on similar elements submitted to a
standardized temperature evolution,
the fire itself : maximum temperature, the spread of the temperature in the different parts of the
building and the influence of the air supply.
escription of the building
The building was 12 m x 18 m in plan, had a
free height of 6 m under the roof girders and consisted
completely of prefabricated concrete elements. The skeleton existed of three portal frames, each
consisting of two columns (cross section 400 mm x 400 mm) with a girder that supported the roof. The
ere anchored in the foundations
The type of concrete, the reinforcement and the concrete cover were the most important parameters.
These parameters were combined in such a way that their influence could be examined separately.
prestressed concrete roof girders had a span of 18 m, while the girders that had to support the
intermediate floor had a span of 6 m. The roof was composed of elements of different concrete types :
normal density concrete (NC), light
weight concrete (LC)
and cellular concrete (CC). To consider as
many parameters as possible, the walls were also composed of different types of prefabricated
elements : two masonry walls were made of concrete blocks (types NC, LC and CC); two were made
of wall elements in cell
ular concrete and light
Inside the building, at half height, a few floor elements were loaded. These elements were supported
by different types of girders or a masonry wall. This wall was part of a small closed cell inside the
t was exposed to the fire as well on the inside as on the outside.
All elements were standard products. The windows
and doors (with a total surface of about 8 % of the
surface of the exterior walls)
(with a surface of about 6 % of the roof
to give the fire the opportunity to develop in a realistic way.
Wood was selected as fuel. Two preliminary tests were carried out to determine the amount of wood
that would give
chance to reach the high temper
ISO Standard 834, during a
sufficiently long time. The total amount of wood used was 27 tons, meaning 125 kg/m², sawn in small
beams, loosely stacked and dried.
From tests it appeared that the experimental value of the
was 14421 kJ/kg resulting in a theoretical
kJ/m² or 3
easurements during the fire
A total of 71 thermocouples (Ni
ere installed at
the surface of
the prefabricated concrete
and inside the building along 5
vertical lines at levels of 0.1 m, 1.5 m, 3.0 m, 4.5 m and 6.0
Deformation and displacement measurements
Four cameras were placed at a height
of 8 m, to get an overview of the complete building. At
points displacements were measured by means of a series of levelling instruments.
Position of the thermocouples inside the building
de the building
The ignition of the wood
took about 2 minutes.
After 20 minutes of fire ignition, enormous
smoke clouds could be observed. Opening of the doors and stirring up the fire by means of four fans
was not sufficient to obta
n a fully devel
oped fire. It was only after 34 minutes, when the light
failed, that the fire
(flames escaping through the roof)
and the inside temperature
reached the level of the ISO
From that moment
he fire rate was roughly controlled
closing and opening of the doors.
It was found that
amount of fuel
that was used
and the ventilation of the building, the heating curve of the ISO stand
ard 834 was followed during
1 h 15 min (period between about 2
0 and 9
utes in figure 5)
After 70 minutes, flames no
longer escaped from the building, and i
t can be reasonably assumed that after 9
temperature decreased relatively fast. Reliable measurements were
t available anymore at that
because thermocouples had fallen down in the layer of burning charcoal, while others were
interrupted because of mechanical
The fire extinguished after 120 minutes by lack of fuel.
In the elements
Several of the prefabricated
were provided with thermocouples on the inside as
well as on the surface.
In this paper the results of the measurements at two of these elements are
Average air temperature at different levels inside the building
0, no. 20b in figure 1
One of the registered elements is th
e central column in one of the short façades, supporting the central
roof beam. Two side faces
of this column
were fully exposed to the fire and one side face only partly,
because it is in con
tact with a light
weight concrete wall. Figure
shows the position of the
thermocouples and the results of the temperature measurements.
From the measurements it follows that :
The surface temperature increases very rapidly after a time span of about 3
0 minutes. This
temperature follows the air temperature inside the building with a delay of 15 to 20 minutes. This
delay can be caused by the high thermal capacity of the building element. The rapid temperature rise
was attributed to the sudden increase of
heat transfer caused by, among other things, the breaking of
the glass of the windows, 27 minutes after the start of the fire.
The average temperature of the reinforcement increased slowly untill 30 minutes after the start of
the fire and then increase
d sharply. In  this sudden temperature rise is attributed to an unknown
phenomenon. However, it is most probable that spalling of the concrete cover occured.
Temperature of column
Central roof girder IV 120
shows the posi
tion of the thermocouples and the results of the temperature measurements
the central roof girder of the building
From the measurements it follows that:
The surface temperature increases very rapidly after a time span of about 30 minutes. This
rature follows the air temperature inside the building but the delay is smaller than for the other
elements. This is probably related to the position of the element inside the building.
The average temperature of the reinforcement remains almost the sam
e during 35 minutes. The
sudden temperature rise after these 35 minutes was, at that time, attributed to an unknown
phenomenon and the measurements were considered to be unreliable. However, it is most probable
that spalling of the concrete cover occured.
Average temperature of the reinforcement
Temperature of central roof girder
Observations and measurements after the fire
After the fire, the building was carefully examined to see how each element had behaved during the
experiment. It appeared that, only a limited number of roo
f elements with a smaller thickness behaved
not as favourably as the other elements. The
different types of
all stayed intact during the fire
without substantial reduction of the mechanical characteristics.
Concerning the colum
s, most damage
erved at the colums in light
weight concrete, because of spalling. Th
had an age of
8 to 9 months at the time of testing, and the moisture content was about 7 % by mass.
After the fire test the building was dismantled. A few elements (the most s
everely damaged) were
selected and subjected to a series of tests.
These tests can be classified in three series :
Series A aimed at a fairly complete analysis of the elements after the fire : static tests untill failure,
residual prestressing, material
characteristics of concrete and reinforcement, chemical analysis of the
Series B was meant to get more information about the possibilities to repair elements that were
damaged during the fire, and about their structural behaviour after the re
Series C included only the determination of material characteristics of the different types of concrete
From the static loading tests on a few members taken from the building, it appears that they show a
reasonable safety margin
. By repairing the damaged concrete, the safety level could be raised up to
Average surface temperature
Temperature in the upper flenge
Temperature of the reinforcement
mperature of the girder web
This means that, although the fire induced damage, the structural members could
generally be reused after proper repair.
The prestressing steel showed a decrease of
tensile strength by 15 %. Measurements indicate a
considerable loss of prestressing.
An extended description of the tests carried out on different members after the fire will be the subject
of further publications.
The structural behaviour of
the building was judged satisfactory as no major collapse occured.
skeleton fulfilled its loadbearing function during and after the fire. The fact that the building did
not show significant damage, is
attributed to the fact that the loadbearing
thermal displacements at the level of the roof girders were about 40 mm. However
t jeopardize neither the connections between the elements, nor the stability of the
fire has given, without any doubt, results that were much more favourable than
the tests carried out in laboratory conditions on separate elements. Other than the observation that
the thermal attack was not as severe as prescribed, the fact that the diffe
rent elements were not
ly to the fire but integrated into the whole of the
connections, has certainly contributed to this favourable result.
The very heavy fire load resulted in a fully developped fire during about
and 15 minutes
after an initial heating period of about 20 minutes. Hence, it can be concluded that in reality it is
almost impossible that a fire will last for two hours in the same compartment. However, during a
in laboratory conditions
, the fire is maintained during two hours under the most
unfavourable loading conditions.
Indeed, in the 1970's the regulations required to apply the full
service load during a fire test.
The authors gratefully acknowledge the financial s
the former “Institute for
Encouragement of Scientific Research in Industry and Agriculture (IWONL/IRSIA)” and the Belgian
Federation of the Concrete Industry (FeBe). The project was co
ordinated by a working group
consisting of representatives
from the Laboratory for Fuel Technology and Heat Transfer (Ghent
University), the Magnel Laboratory for Concrete Research (Ghent University), the Ministry of Public
Works (Building Division), the Belgian Research Centre of the Cement Industry (CRIC),
Building Research Institute (WTCB/CSTC), the National Society for Fire Safety, the Fire Brigade
Department of the city of Ghent and the Federation of the Concrete Industry.
F. Almey, R. Minne, A. Van Acker, “La cellule d’essai de l’U.A.C
.B. en proie aux flammes.
proefcel werd de prooi der vlammen”, Beton, n
F. Almey, R. Minne, A. Van Acker, “Les résultats d’un ‘incendie pas comme les autres’.
resultaten van een ‘niet alledaagse brand’ ”, Beton, n
o. 34, F
F. Almey, R. Minne, A. Van Acker, “Incendie expérimental d’un batiment industriel préfabriqué :
rapport de synthèse.
Experimentele brand van een geprefabriceerd industrieel gebouw :
syntheseverslag”, Beton, n
o. 40, A