Fire Damage

Text-only Preview

Cement & Concrete Composites 27 (2005) 255-259
Microstructure of fire-damaged concrete. A case study
B. Georgali, P.E. Tsakiridis *
Hellenic Cement Research Center Ltd., K. Pateli 15, 14123 Lykovrissi, Athens, Greece
Concrete is a poor conductor of heat, but can suffer considerable damage when exposed to fire. Unraveling the heating history of
concrete is important to forensic research or to determine whether a fire-exposed concrete structure and its components are still
structurally sound. Assessment of fire-damage concrete structures usually starts with visual observation of color change, cracking
and spalling. On heating, a change in color from normal to pink is often observed and this is useful since it coincides with the onset
of significant loss of concrete strength. This paper presents results of cores strength, as well as, optical microscopy investigations of
fire-damaged concrete. Samples were taken from concrete that had been exposed to fire. Optical microscopy has focused on
microstructure of cement paste, aggregates, microvoids and cracks, as well as, on quantification the crack patterns found in heated
concrete samples. The physical condition of concrete sample in combination with the microscopic examination, enable a petro-
grapher to make a reasonable estimation of the minimum exposure temperature and its relative impact to the depth of damage in
O 2004 Elsevier Ltd. All rights reserved.
Keywords: Concrete; Fire-damage; Microstructure
1. Introduction
shown in Fig. 1. On heating above 300 C the color of
concrete can change from normal to pink (300-600 C)
When reinforced concrete is subjected to high tem-
to whitish gray (600-900 C) and buff (900-1000 C).
perature as in fire, there is deterioration in its properties.
The pink discoloration results from the presence of iron
Of particular importance are loss in compressive strength,
compound in the fine or coarse aggregates [1-3].
cracking and spalling of concrete, destruction of the bond
The first effects of a slow temperature rise in concrete
between the cement paste and the aggregates and the
will occur between 100 and 200 C when evaporation of
gradual deterioration of the hardened cement paste.
the free moisture, contained in the concrete mass, oc-
Assessment of fire-damaged concrete usually starts
curs. Instant exposure can result in spalling through
with visual observation of color change, cracking and
generation of high internal steam pressures. As the
spalling of the surface. Concrete color provides a broad,
temperature approaches 250 C dehydration or loss of
general guide of temperatures, whether the color repre-
the non-evaporable water or water of hydration, begins
sents the original surface or one resulting from spalling.
to take place. The first sizable degradation in compres-
Crazing, cracking popouts caused by quartz or chert
sive strength is usually experienced between 200 and 250
aggregate particles, spalling and dehydration (crumbling
C. At 300 C strength reduction would be in the range
and powdering of paste) are general indications of
of 15-40%. At 550 C reduction in compressive strength
temperatures to which concrete has been exposed as
would typically range from 55% to 70% of its original
value [3-5].
Temperatures in the 550 C range are critical because
calcium hydroxide dehydration takes place. Calcium
hydroxide is a hydration product of most Portland ce-
* Address: Department of Mining Engineering and Metallurgy,
ment, the amount being dependent upon the particular
Laboratory of Physical Metallurgy, National Technical University of
cement being used. Aggregates also begin to deteriorate
Athens, 9, Iroon Politehniou Str., Athens 15780, Greece. Tel.: +30-10-
at about 550
2818453/210-7722179; fax: +30-10-2842140/6944-981306.
C. For example quartz expands at a higher
E-mail address: [email protected] (P.E. Tsakiridis).
rate around 300 C [6,7].
0958-9465/$ - see front matter O 2004 Elsevier Ltd. All rights reserved.

B. Georgali, P.E. Tsakiridis / Cement & Concrete Composites 27 (2005) 255-259
Concrete Color
Other Possible Physical Effects
950 C
900 C
Powdered, light colored, dehytdrated paste
Black Through
800 C
Spalling, exposing not more than 25 percent
Gray to Buff
of reinforcing bar surface
600 C
575 C
Popouts over chert or quartz aggregate particles
550 C
Deep Cracking
Pink to Red
300 C
300 C
Surface crazing
40 C
Fig. 1. Visual evidence of temperature to which concrete has been heated.
Two main types of spalling occur during fire.
2. Experimental procedure
Explosive spalling and sloughing off of concrete surface
layers. Explosive spalling looks like a series of popouts
The ten test specimens examined were concrete cores
and usually occurs within the first 30 min of fire-expo-
from the first floor of a reinforced concrete building
sure. Sloughing off is a gradual non-violent separation
(age: 15 years old) which had been exposed to fire on 5th
of the concrete that occurs primarily at the edges of
February 2000.
columns and beams. When concrete spalls, deeper layers
Five cylinder cores (cyl. 1-5) were subjected to com-
of concrete are exposed to the maximum fire-tempera-
pressive strength test according to method ASTM C-39,
ture, speeding the transmission of heat to the rein-
and three cylinder cores (cyl. 6-8) were subjected to
forcement. As the temperature within a member rises,
``tensile splitting test'' according to method ISO 4108
steel reinforcement expands more than concrete. This
(Brazilian test). The other two cores (cyl. 9-10) were
can lead to further spalling and cracking around the
used for the petrographic analysis based on ASTM
steel. Such cracks often develop where incipient cracks
C-856: ``petrographic examination of hardened con-
(due to drying shrinkage, flexural loading or other fac-
crete'' was carried out on two thin sections from each
tors) were present. Also, differing thermal expansion
core. Their construction was made in such a way, that
between aggregates and cement paste can create surface
the observation of the phenomena escalation due to fire-
crazing, which can lead to deeper cracking [4,7].
exposed would be feasible (Fig. 2).
Except the visual observation and the tests on con-
The thin section were manufactured by vacuum
crete cores, optical microscopy applied to petrographic
impregnation, of the selected sample (cut from the core,
thin section has been used in investigations of concrete
microstructure and has recently been applied to fire-
damaged concrete. Results suggest that the nature and
extent of cracking may be correlated with the actual
temperatures attained in the concrete and a form of
quantification of the crack patterns found was at-
tempted [3,4].
Thin sections for
The aim of the present research is to take advantage
petrographic analysis
of the optical microscopy results on thin sections, and
carry out a study to determine the microstructures of
concrete-pastes/aggregates, microvoids/cracks, and sep-
aration of cement paste from aggregates in fire-exposed
Fire damaged side
concrete samples, as well as the quantification of the
crack patterns.
Fig. 2. Specimens for (thin section) microscopic observation.

B. Georgali, P.E. Tsakiridis / Cement & Concrete Composites 27 (2005) 255-259
size 50 mm * 50 mm * 10 mm) with epoxy resin, followed
oxides present in the aggregate. Taking into account
by cutting, grinding, and polishing until a final thickness
that the reinforcement had been set in the depth of
of 20 lm is reached. During the process the thin section
3 cm and that the surface, exposed to fire, color has
is mounted upon an object glass.
been changed to gray, temperature is believed to exceed
The thin sections were analyzed by polarization
800 C [1,6].
microscopy. The concrete constituents, and its micro-
The absence of macroscopically visible glassy layer, at
the location close to fire origin, indicates that there was
separation of the cement paste from aggregates--in fire-
not any concrete melting and, as a result, temperature is
exposed concrete samples, were examined and signs of
believed not to exceed the limit of 1000 C [1-3].
the deterioration process were noted. The quantification
of the crack patterns was carried out by measuring the
3.2. Effect on compressive strength
crack density of concrete, in units of ``mm'' of crack
length per ``cm2'', correlating the extent of cracking,
Although a compendium of indirect information
because of temperature increase, with the depth from the
about the structural capability of individual elements
surface exposed to fire. The examination was carried out
can be obtained, it may still become necessary, and is
in filtered transmitted light, at a magnification of 32* so
indeed frequently desirable, to determine the actual
that the cracks present in an area of 3.2 mm * 3.2 mm
loading capacity of an integrated system of elements.
could be observed.
Furthermore, the most important effect of fire on con-
crete structure is the reduction of its compressive
strengths due to heating. As a result, to estimate the
3. Results and discussion
strength of the fire-damaged concrete, in the affected
building, five cylinder cores were subjected to compres-
3.1. Macroscopic observation
sive load test. The results are given in Table 1. More-
over, three of the extracted cylinder cores were subjected
All of the ten concrete cylinders examined, sustained
to ``Brazilian test'' according to method ISO 4108:
significant damage caused by exposure to heat. The
``tensile splitting test'' and the results are given in
damage in all cases confined, especially, to the surface
Table 2.
side near to the fire origin. The surface crazing was due
From the results of Table 1, it can be seen that the
to dehydration, which starts at 100 C and ends at 540
residual compressive strengths of fire-exposed concrete
C, resulting in the removal of free water. Cracking and
were very low, whose reduction has reached 70% of
softening, following by decrepitation of the concrete
initial compressive strength. This indicates that tem-
surface is caused first by expansion and then by
perature exposure exceeded 700 C [1,6,7].
shrinking of the cement paste due to transformation of
According to the results of ``Brazilian test'' (Table 2)
Ca(OH)2 to CaO in the temperature of 450 and 500 C.
the tensile strengths of the examined cores specimens
Though the remainder cylinder cores appeared with
were, also, too low, a fact that was predictable given
fewer damages, significant parallel cracking was ob-
that, random and radial microcracks were observed in
served toward the core axis. The existence of cracking
the load axis.
perpendicular to face (parallel to the axis concrete core)
in significant depth (bigger than 3 cm) indicates large
scale internal cracking because of internal shrinkage,
Table 1
which is caused by overheating following by rapid
Compressive strengths of fire-damaged concrete
cooling (due to fire-extinguishing). Obviously, this
Core no.
Diameter (mm)
Height (mm)
Strength (MPa)
cracking does not have the ability of self-restoration.
Spalling and popouts of concrete are confined to surface
Cyl. 1
Cyl. 2
close to areas exposed to fire, an indicator of heating
Cyl. 3
at 573 C. Deeper cracking beyond the depth of the
Cyl. 4
reinforcement indicates that temperature has exceeded
Cyl. 5
700 C [3,4].
Because the aggregates used in concrete, was mostly
limestone, heat transmission in the internal part of the
reinforced concrete was greater. The presence of red
Table 2
Tensile splitting test of fire-damaged concrete
aggregates around reinforcement indicates that temper-
atures up to 590 C must have been attained. Further-
Core no.
Diameter (mm)
Height (mm)
Strength (MPa)
more, in bigger depth pink aggregates can be observed
Cyl. 6
(heating at 300 C) which is a result of transformations
Cyl. 7
in the composition and/or structure of hydrated iron
Cyl. 8

B. Georgali, P.E. Tsakiridis / Cement & Concrete Composites 27 (2005) 255-259
3.3. Petrographic examination
3.4. Crack density
All samples received from the fire-damaged concrete
Information on the density and the distribution of
cores were studied microscopically giving the following
cracks is useful in determining not only the minimum
exposure temperature, but the thickness of concrete
(from the spalled surface) that may eventually be re-
1. At the surface near to the fire origin carbonate aggre-
moved in the case of repair work. It is also important in
gate in concrete have been transformed to CaO, a fact
determining whether fire-attacked elements and its
that indicates the temperature must have reached 900
components, such as steel reinforcement, are still struc-
C (Fig. 3a, 32*). The rest of the aggregates--in
turally sound and that the local loading conditions, in
greater depth--are seemed to have been preserved
the long term would not adversely affect the mechanical
relatively harmless.
properties and the durability of the elements.
2. The carbonation reaction has been completely devel-
Fig. 4 shows the extent of cracking, due to the tem-
oped in the cement paste, at the surface near to the
perature increase, in connection with the depth from the
fire-exposed side, contrary to the inner side of the
surface exposed to fire. Crack density is given in units of
specimen, where crystals of Ca(OH)2 has been
detected (Fig. 3b, 200*).
3. Large amount of gaps has been observed in the whole
area of examined specimens, which are related with,
either because of the pulverization of the aggregates,
)2 9
or because of the cement paste fragmentation, by rea-
son of its structure collapse (Fig. 3c, 32*).
4. Large amount of heavy cracking has been detected
in the cement paste-aggregates interface (Fig. 3d,
2*) in the whole area of examined specimens. More-
over, in the main area of the cement paste micro-
Crack Density (mm/cm 6
cracking, of various orientation, has been observed
which obviously, are not placed among these of
self-restored. The absence of a microscopically visible
0.25 0.5 0.75 1
1.25 1.5 1.75 2
2.25 2.5 2.75 3
3.25 3.5
Depth from the Surface (cm)
glassy layer, underlain by thin layers of altered paste
and aggregate, indicated that temperature exposure
Fig. 4. Crack density vs the depth from the surface of the examined
did not exceed 1000 C.
Fig. 3. Representative thin section photos of the examined fire-damaged concrete samples.

B. Georgali, P.E. Tsakiridis / Cement & Concrete Composites 27 (2005) 255-259
mm of crack length per cm2. Each point is an average of
Microscopic observation at the surface near to the
the crack length measured in 15 different squares of 3.2
fire origin showed that carbonated aggregates have been
mm * 3.2 mm covering the wide of the 50 mm * 50 mm
transformed to CaO, a fact that indicates the tempera-
thin section.
ture reached 900 C. The carbonation reaction has been
According to Fig. 4, the temperature in the inner side
completely developed in the cement paste, at the surface
of the core (at 3 cm depth) near to the reinforcement
near to the fire-exposed side, contrary to the inner side
bar, did not exceed 800 C, whereas in the outer side
of the specimen, where crystals of Ca(OH)2 has been
reached 950-1000 C [3].
detected. The large amount of gaps observed is related
with, either because of the pulverization of the aggre-
gates, or because of the cement paste fragmentation at
4. Conclusions
the temperature around 900 C. Finally, the absence of a
microscopically visible glassy layer, underlain by thin
Measurements, such as microscopically-petrographic
layers of altered paste and aggregate, indicated that
examination and loading tests, together with the mac-
temperature at the surface exposed to fire did not exceed
roscopically observation, were used in order to deter-
1000 C.
mine the thermal history of a fire-damaged concrete and
to provide information regarding the maximum tem-
perature at the surface exposed to fire.
Macroscopically observation showed significant par-
allel cracking toward to the core axis, a fact that indi-
cated large scale internal cracking because of internal
[1] Erlin B et al. Evaluating fire damage to concrete structures.
shrinkage, which is caused by overheating following by
Concrete Constr 1972;(2):76-82.
rapid cooling (due to fire-extinguishing). Spalling and
[2] Green JK. Reinstatement of concrete structures after fire. The
Architects' J 1971;(1):93-9.
popouts are confined near surface exposed to fire, an
[3] Guise SE. Petrographic and color analysis for assessment of fire
indicator of heating at 575 C. Deeper cracking beyond
damaged concrete. In: Jany L, et al., editor. Proceedings of the
the depth of the reinforcement indicates that tempera-
19th International Conference on Cement Microscopy. 1999.
ture has exceeded 790 C. Taking into account that the
p. 365-72.
surface, exposed to fire, color has been changed to gray,
[4] Powers-Couche L. Fire damaged concrete-up close. Concrete
Repair Digest 1992;1:241-8.
temperature is believed to exceed 800 C.
[5] Gustafero AH. Experiences from evaluating fire-damaged concrete
The results of compressive test showed that the con-
structures--fire safety of concrete structures. American Concrete
crete's fire residual compressive strengths were very low,
Institute SP-80, 1983.
whose reduction has reached the 70%. This fact indi-
[6] Chu TY. Radiant heat evaluation of concrete--a study of the
cates that the temperature exposure exceed 700 C.
erosion of concrete due to surface heating. Research paper SAND
77-0922. Sandia Laboratories. Albuquerque NM, 1978.
According to crack density measurements, the tem-
[7] Powers-Couche L. Observations of concrete exposed to very high
perature in the inner side of the core near to the rein-
temperature. In: Gouda GR. et al, editor. Proceedings of the
forcement bar, did not exceed 800 C, whereas in the
16th International Conference on Cement Microscopy. 1994.
outer side reached 950-1000 C.
p. 369-76.

Document Outline

  • Microstructure of fire-damaged concrete. A case study
    • Introduction
    • Experimental procedure
    • Results and discussion
      • Macroscopic observation
      • Effect on compressive strength
      • Petrographic examination
      • Crack density
    • Conclusions
    • References