You are on page 1of 15

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/240776671

New classification system for fire-damaged


concrete based on the strain energy dissipated
in a hysteresis loop

Article in Magazine of Concrete Research · January 2000


DOI: 10.1680/macr.2000.52.4.287

CITATIONS READS

2 28

1 author:

Ayman Nassif
University of Portsmouth
18 PUBLICATIONS 114 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ayman Nassif on 11 April 2014.

The user has requested enhancement of the downloaded file.


A new classification system for fire-damaged concrete based
on the strain energy dissipated in a hysteresis loop.

Dr A.Y. Nassif , BEng (civil), PhD (London)


Senior Lecturer
Department of Civil Engineering
Burnaby Building
Burnaby Road
Portsmouth PO1 3QL
Telephone +44 (0) 1705 84 2392
Fax. +44 (0) 1705 84 2521
E-mail ayman.nassif@port.ac.uk

University of Portsmouth 1 Ayman Nassif


Synopsis

This paper reports an experimental research into further developing the Stiffness Damage Test as a quantitative tool to
be used in assessing fire-damaged concrete. The research programme reported here aimed at establishing the
correlation between the strain energy dissipated in a hysteresis loop, performed on fire damaged concrete core, and the
postfiring residual stiffness characteristics. The dependency of such correlation on the level and mechanism of
damage was investigated by involving five test variables, namely: maximum temperature of exposure, method of
cooling, duration of thermal exposure, grade of concrete and type of aggregate used. The investigation involved
approximately 180 concrete cores (75mm Dia., 175mm long) which were extracted from river gravel and limestone
aggregate concrete blocks. The cores were subjected to various heating and cooling regimes in the temperature range
(217oC~470oC ).

The extent of fire-damage was assessed using the Stiffness Damage Test for determining the characteristics of
quasi-static low-stress load-unload cycles. The strain energy dissipated in a hysteresis loop was adopted as a
dimensionless Damage Index by dividing it by the stress range. Both the Damage Index and the residual Plastic Strain
were found to correlate well with the percentage residual initial and chord modulii. Such strong correlation is used to
introduce a new classification system for the level of fire-damage in concrete. A Damage Index of 5 is associated with
a 24% reduction of the chord modulus while a Damage Index of 20 is associated with a 66% loss of the chord modulus.
The coefficient of variation, which is used as a repeatability indicator, of the Stiffness Damage Test parameters was
found to be within acceptable limits.

The correlation between the maximum temperature of exposure and the residual stiffness characteristics was also
established. It is evident that the residual stiffness of fire-damaged concrete is not a sole function of the maximum
temperature of exposure. The method of cooling, duration of exposure, type and grade of the fire-affected concrete
materials are major determinants in the extent of damage.

University of Portsmouth 2 Ayman Nassif


Introduction

The rise in temperature of concrete, during a fire incident, leads to modification in its structural properties to various
degrees but rarely causes total structural collapse.

Attempting to predict the behaviour of concrete structures during a fire and designing it to withstand a probable fire
scenario is one section of the overall fire engineering practise. The structural response is traditionally predicted using
the modified material properties as a simplistic function of the maximum temperature of exposure(1). Such prediction
ignores the effect of other characteristics of fires, such as the thermal shock caused by spraying hot concrete with
water during fire-fighting. Previous research work does not seem to have sufficiently investigated this issue in spite of
its importance in establishing the structural behaviour of fire-damaged concrete.

After a fire incident, practitioners are concerned with the appraisal of the damaged structures and assessing the
residual load bearing capacity in an overall redesign exercise especially for important structures and installations.

The published literature contains examples of severe fires in various parts of the world which caused obvious structure
damage to concrete. The assessment of the extent of damage was carried out mainly by visual inspection. The
available methods of Ultrasonic Pulse Velocity (UPV) and Rebound Number were sometimes used but many
practitioners expressed their concern about the effectiveness and reliability of these methods. It is noteworthy that in
a previous publication(2), it was shown that UPV can be misleading and can significantly underestimate the extent of
damage when the moisture content of concrete is altered because of fire fighting activities. Indeed, the reliability of
using the UVP method in estimating sound concrete is questionable(3).

Dorsch(4) reported a fire which lasted for 3 hours and a possible temperature of 730°C was reached. The decision on
what to do with the damaged concrete areas was taken by carrying out sounding tests on all concrete elements. A
"dull" sound indicated damaged material which was cut before any repair started. Some cores were occasionally taken
and tested for compressive strength as a verification of the sounding tests.

Tomlinson(5) reported a very severe fire in Kellogg Company Factory in Great Britain which caused serious structural
damage to the concrete building, in particular to the ground floor where the fire originated. The fire was intense and
consumed around 2000 tonnes of paper and wax spreading very fast for a duration of 8 hours. It appears that the
method of assessing the structural damage was mainly carried out by visual inspection of every individual member for
discoloration and surface cracks, especially those along the principal planes. The reinstatement of the severely
damaged concrete elements in the ground floor was carried out by replacing or strengthening them.

In another example of real fire-damage to concrete, Fleischer and Chapman(6) reported on the work carried out to
assess fire and bomb damaged concrete buildings. The assessment was carried out visually and the condition of
concrete was evaluated by means of both UPV and Schmidt Hammer.

Gustaferro(7) wrote about his experience in assessing fire-damaged concrete structures. He investigated a destructive
fire in Chicago's McCormick Place, in 1967. The cast in place waffle slab suffered serious spalling exposing the
reinforcement in many areas. In spite of the severity of the damage, it was possible to repair the slab.

An apartment fire in a reinforced concrete high rise building consumed all combustible materials in the flat but no
serious structural damage was detected. Gustaferro used the Schmidt Hammer to outline the areas which were
severely damaged but he cautioned that "determining the strength of fire-damaged concrete with an impact rebound
hammer must be regarded as highly suspect".

University of Portsmouth 3 Ayman Nassif


In other cases, the damage inflicted by fire was so severe that demolition and rebuilding was cheaper than repair. Such
fires occurred in a four storey plant of beam-and-slab construction. The fire caused shear and flexural cracks in the
beams as a result of the redistribution of bending moments and reactions.

Gustaferro also reported a fire, in a petrochemical plant in Japan. The cast in place platform was subjected to high
intensity and short duration fire. The postfire assessment was carried out visually by scrutinising the platform for
cracks, especially in the negative moment areas which could have indicated redistribution of moment. The Schmidt
Hammer was also used in an attempt to establish the severely damaged areas.

In spite of the fact that the assessment of the platform established that the damage was superficial, the designer of the
platform still felt the need to strengthen it. Had the method of assessment been direct, reliable and conclusive, such
conservatism would not have been justified and obvious saving of capital and time would have been achieved.

The damage caused by fires to concrete structures involves other aspects in addition to the modification in the material
properties such as explosions and thermal expansion. Differential thermal expansion can be extremely damaging
which was the case in the fire reported by Kordina(8). It was calculated that the thermal expansion in some cases was as
much as 54mm causing an increase in the moment of 580 kNm.

Forrest(9) reported some incidents of fire in lightweight concrete structures in several countries. The extent of damage
in one incident in the UK was assessed by pushing a screwdriver into the concrete. As the screwdriver was pushed
only 6mm into the concrete, it was somehow concluded that the concrete was not seriously damaged. However, in
most cases the damage was assessed by observing the degree of spalling and the exposure of reinforcement. More
information on the traditional methods used in assessing fire damaged concrete can be found in references 1 & 10.

It is evident that the traditional methods of assessing the extent of fire-damage to concrete are not adequate and tend to
be qualitative. A technological development in this field was needed. Therefore, the need for an efficient, accurate
and reliable method for assessing the fractured state of concrete after a fire incident was paramount. The availability
of such reliable method should reduce the disruption period and save money. An initial investigations into adopting
the Stiffness Damage Test (SDT), as a quantitative tool in assessing the degree of fracture and loss of structural
properties of fire damaged concrete, was reported in a previous publication(11). This paper reports a subsequent
extensive experimental programme into the use of the strain energy dissipated in a quasi-static hysteresis loop as a
dimensionless numerical Damage Index (DI) after normalisation over the stress range.

University of Portsmouth 4 Ayman Nassif


Research programme

The main objectives of the research programme were:

1. to establish the correlation between the strain energy dissipated in a hysteresis loop, performed on fire damaged
concrete core, and the postfiring residual stiffness characteristics,

2. and to establish the correlation between the maximum temperature of exposure and the residual stiffness properties.

Five test variables were chosen to examine the dependency of such correlation on the degree and mechanism of
damage as well as the type and grade of fire-affected concrete materials. The variables included: maximum
temperature of exposure, method of cooling, duration of thermal exposure, grade of concrete and type of aggregate
used.

Six laboratory cast concrete blocks (700x500x200mm) together with standard cubes were involved in the
investigation. Details of concrete constituents and mix proportions can be found in Table (1).

The concrete was cured in water at 20°C for 28 days and the average cube compressive strength was determined
according to BS 1881 Part 116. The concrete blocks were conditioned by storing them in a controlled environment
(20°C and 65% RH) for a few months to allow the moisture content to stabilise. Subsequently the blocks were "cored"
to obtain 200mm long by 75mm diameter cores. A total of 180 cores were obtained from six blocks. The core ends
were then cut, using a diamond saw, to reduce the length to 175mm and to produce partially prepared ends. The cores
were then exposed to various thermal treatments and cooled in a controlled environment.

A muffle furnace, with a sufficient heating chamber to accommodate one core at a time, was used to heat the test cores.
The furnace, which was located in a controlled environment, was calibrated. The furnace temperature settings were
found to have a slower heating rate than that of the Standard Heating Curve as outlined in BS 476 Part 8 (1987). To
overcome this problem, the furnace settings were varied during the heating cycle to produce a heating curve as close as
possible to the standard one.

The calibration of the furnace took place with a representative core inside the heating chamber. The core was fitted
with a K-type thermocouple along its longitudinal axis inserted to half its length. Another thermocouple was placed
close to the outer surface of the concrete core. The readings of both thermocouples were recorded up to the point when
the outer surface temperature reached the test temperature. With the furnace stabilised at the test temperature, the
heating continued to the point when both thermocouples recorded the same temperature. This point will be referred to
as: the point of uniformity.

University of Portsmouth 5 Ayman Nassif


Table (1): Detailed concrete constituents and mix proportions

Block Constituents Mix Proportions Number of cores obtained and


code (cement: fine: coarse aggregate, by weight) experimental usage
10mm maximum size Approx.:30 cores used for the
river gravel aggregate, 1:1.35:3.14, w/c = 0.45 preliminary investigations. The
B1 cores were heated to the point of
OPC and well graded fcu28=52N/mm2
marine sand. uniformity and air cooled.

Approx.:30 cores obtained and


B2 heated to the point of uniformity
and air-cooled.
Approx.:30 cores heated to the
20mm maximum size
point of uniformity and maintained
B3 river gravel aggregate, 1:1.5:3.34 w/c = 0.40
at the test temperature for two
OPC and well graded fcu28= 62.9 N/mm2
hours then air-cooled.
marine sand.
Approx.:30 cores heated to the
point of uniformity and cooled by
B4
spraying with water for five
minutes.
Approx.:30 cores obtained and
B5 heated to the point of uniformity
20mm maximum size and air-cooled.
limestone aggregate, OPC 1:1.5:3.34, w/c=0.40
and well graded marine fcu28=69.9 N/mm2 Approx.:30 cores heated to the
sand. point of uniformity and cooled by
B6
spraying with water for five
minutes.

Seven test temperatures were decided upon in the light of the existing knowledge of the behaviour of concrete at
elevated temperature, namely: 217°C, 240°C, 287°C, 320°C, 378°C, 470°C and 540°C. However, cores heated to
540°C had very little structural strength and this temperature was later discarded. Every heat treatment was repeated
on three cores ( sometimes four cores). The heating and cooling regimes involved the following treatments:

1) Basic heat treatment: This treatment involved heating the core to the point when both the outer surface and the
centre reached the test temperature, i.e. the point of uniformity. At this point, the furnace was switched off and the
core was taken out to a controlled environment of 20°C where it was left to cool.

2) Two hours thermal exposure: This treatment involved soaking the core in the test temperature for two hours,
beginning from the time when the core reached the point of uniformity. At the end of the two hours, the core was taken
out of the furnace and left to cool in a controlled environment of 20°C.

3) Quenching: This treatment involved heating the core to the point of uniformity. At this point, the core was taken out
of the furnace and sprayed with tap water for a duration of 5 minutes. Uniform spraying was achieved by rotating the
core in front of atomising nozzles. Subsequently, the core was left to cool completely in a controlled environment of
20°C.

After the various heating and cooling regimes, the cores were stored in a controlled environment until the testing
programme was completed. The cores were examined visually and their Ultrasonic Pulse Velocity (UPV)
measurements were determined. Subsequently the cores were capped using sulphur compound to achieve the specified
tolerance in BS 1881 Part 120 regarding the parallelism, squareness and flatness of the core ends.

The capped cores were then mechanically tested to determine the stiffness parameters obtained by the Stiffness
Damage Test (SDT), as described in detail in references 2, 10 and 11. However, for the readers convenience, the

University of Portsmouth 6 Ayman Nassif


definitions of the Stiffness Damage Test parameters are graphically presented in Figure(1)

Every Stiffness Damage Test involved subjecting the capped test core to five low stress load-unload cycles using a
servo controlled loading machine. The load and three strain measurements were recorded using a data logger. The
following parameters were calculated and averaged for the last four cycles:

• The initial modulus (Ei) is the chord slope of the loading curve between 0.45 and 0.9 N/mm2 stress level.

• The chord modulus (Ec) is the chord slope of the loading curve between 0.45 and 4.5 N/mm2 stress level.

• The unloading modulus (Eu) is the chord slope of the unloading curve between the maximum stress and a point
0.45 N/mm2 less.

• The Damage Index (DI) is the area of hysteresis between 0.45 and 4.5N/mm2 stress level divided by the stress
range.

• The Plastic Strain (PS) is the residual plastic strain at 0.45 N/mm2 stress level.

• The Non-linearity Index (NLI) is the chord slope of the loading curve between 0.45 N/mm2 and half the maximum
stress divided by the value of the Ec

Results and discussion

Figures(2~7) show the correlation between the SDT parameters and the maximum temperature attained. The residual
stiffness parameters are plotted against the maximum temperature regardless of the method of cooling, duration of
exposure and type of the affected concrete materials. It is evident that the maximum temperature experienced is not the
only damaging effect during fires. This fact is pictorially illustrated by the width of band of results, which indicate
that:
♦ Both DI and PS are more sensitive to the characteristics of the thermal exposure and concrete
materials in the higher range of temperatures than in the lower range.
♦ Both Ei and Ec are more sensitive in the lower range of temperatures than in the higher range.
♦ Both Eu and NLI seem to have similar sensitivity over the entire range of temperatures.

This lack of strong correlation between the maximum temperature of exposure and the residual stiffness properties,
highlights the shortcomings of the existing techniques used in assessing the postfiring residual properties of concrete.
(1) (12)
The discoloration and thermoluminescence methods rely on predicting the maximum temperature attained
during fires. However, such predictions are not at all reliable, limestone coarse aggregate concrete does not show any
noticeable colour changes after heating to a temperature in the range (200~470°C). It is worth mentioning that visual
examinations for colour changes and surface cracks was the main method of assessing the degree of damage in
Kellogg’s fire(5). In addition, the prediction of the maximum temperature attained using the thermoluminescence (TL)
is severely confused by the duration of exposure yielding unreasonable misleading predictions. Even if the estimation

University of Portsmouth 7 Ayman Nassif


of the maximum temperature is satisfactorily achieved, the results presented here clearly show that the maximum
temperature is not the sole determinant of the extent of damage.

The previous researchers(13,14,15 &16) reported that the loss of elastic modulus of concrete is around 40~50% at exposure
to 300°C and around 80% at 600°C. The author believes that such statements relating the reduction of stiffness to the
maximum temperature of exposure is incomplete. The reduction of stiffness of concrete during heating (the transient
conditions) is largely of concern to the designer who attempts to design for a certain fire endurance criterion. For the
consultant practitioners, who undertake the appraisal of fire-affected structures to determine the residual load-bearing
capacity, the postfiring residual properties are of interest. Figure(2) attempts to correlate the percentage residual chord
modulus, as determined from the SDT, to the maximum temperature of the exposure regardless of the characteristics
of the exposure and the concrete material. However, it is observed that the residual stiffness after exposure to 300°C,
is in the range 68%~30% depending on the various characteristics of the thermal exposure and concrete material.

The initial stiffness was found to be very sensitive to the degree of fracture, certainly more so than the chord modulus
and the unloading modulus, as shown in Figure(8), which include the residual modulii for fired river gravel aggregate
concrete heated to uniformity and air-cooled. The initial strain includes both elastic and plastic components of
deformation and mainly represents initial crack activities such as reduction of crack width, initial propagation of
cracks and sliding of the damaged interfaces. At higher stress levels the strain better represents the elastic properties of
the concrete as a composite material. A severe reduction in the residual initial stiffness was observed, even in the
range of temperatures which is not traditionally associated with serious damage. Figure(3) attempts to correlate the
percentage residual initial modulus with the maximum temperature attained regardless of the other characteristics of
exposure or concrete materials. The severest reduction of Ei was observed for quenched limestone concrete which
retained a mere 30% of its original value after exposure to 217°C and subsequently cooled by water quenching. The
least reduction in Ei was observed in air-cooled river gravel concrete which retained around 80% of its initial value
after exposure to 217°C.

Indeed, there is hardly a temperature that can be designated as a critical temperature. However, a critical fractured
state is defined according to the internal plasticity and energy dissipation in a stress-strain cycle. In this paper, the
author introduces a new classification system of fire-damaged concrete based on accurate and direct assessment of the
internal fracture of fire-affected concrete. The new classification system caters for not only the maximum temperature
experienced but also other characteristics of the thermal exposure and type of affected concrete materials.

After a fire incident, one of the questions needing an answer is: how much reduction in the stiffness characteristics the
structural element has suffered? It is suggested that the Damage Index (DI) and the Plastic Strain (PS), measured
using the Stiffness Damage Test, can be used to estimate the percentage residual stiffness of fire-affected concrete.

Figures(9~13) show various correlation of the Damage Index and the Plastic Strain with the residual chord and initial
modulii. The parameters were calculated for river gravel and limestone concretes after firing to different temperatures
and duration. The hot concrete was either cooled by spraying with water or in dry air.

Figure(9) shows that the Damage Index correlates extremely well (R2 = 0.96) with the percentage residual initial
modulus. The correlation equation can therefore be used to determine the percentage residual initial stiffness. As the
postfiring initial modulus is also calculated in the SDT, the unfired original stiffness can be established.

The Damage Index also correlates very well with residual chord modulus (R2 = 0.93) as shown in Figure(10).
Similarly, percentage residual Ec can be established from the correlation equation.

The correlation between the Plastic Strain and the residual modulii, Figures(11&12), is very similar to that of the

University of Portsmouth 8 Ayman Nassif


Damage Index. This is hardly surprising as the Damage Index and the Plastic Strain correlate well with each other (R2
= 0.91), as shown in Figure(3). The Plastic Strain should therefore be used as a confirmation for the assessment
established by the Damage Index.

It is noteworthy that the stress-strain loading response becomes concave for any degree of fire-damage. However, the
Non-Linearity Index (NLI), measured by the SDT, seems not to exceed 0.6 regardless of the extent of damage.

In the light of the above experimental results, it is suggested that a rethink of the philosophy of assessing fire-damaged
concrete based solely on attempting to predict the maximum temperature attained during fires is desperately needed.
In addition, a new quantitative method of classification of the extent of fire-damage should be based on the
modification of the internal plasticity and internal cracking rather than on qualitative visual methods. Therefore, the
author proposes the following new classification system of fire-damaged concrete based on the strain energy
dissipated in a hystereis loop as well as the residual Plastic Strain, as outlined in Table(2).

Table (2):A new classification of the level of fire-damage based on quantification of the fractured state

Damage Index Class of fire-damage Plastic Strain, (PS) Most Probable Value of residual
(DI) µstrai stiffness.

% residual Ec % residual Ei

<5 I < 4.5 > 76 % > 73 %

5~10 II 4.5~11.4 76 ~51% 73 ~ 40%

10~15 III 11.4~20 51~ 40 % 40 ~ 28 %

15~20 IV 20~30 40 ~ 34 % 28 ~ 22 %

>20 V >30 < 34 % < 22 %

Furthermore, The available material models of predicting concrete structural behaviour during and after fire incidents
should adopt more accurate functions to reflect the modification in the mechanical properties of concrete taking into

University of Portsmouth 9 Ayman Nassif


account the characteristics of thermal exposure and type of concrete materials.

The issue of repeatability of the Stiffness Damage Test parameters was also addressed. Every heat treatment was
usually repeated on three cores and sometimes on four cores. On very few occasions the data represent the results
obtained from two cores. The coefficient of variation is calculated as the percentage of the standard deviation of the
mean value of each test group. The Most Probable Value (MPV)of the coefficient of variation of a certain parameter
is determined as:

MPV = The mean value ± the standard error

S
The standard error =
n

Where:

S = standard deviation of the mean

n = number of test groups

Table(3) shows the mean values, the standard deviation and the coefficient of variation of five test groups for all the
SDT parameters. The population consists of 30 groups representing different thermal exposures and concrete
materials.

Table (3) The average stiffness parameters, standard deviation and


coefficent of variations for a representative 5 test groups of quenched
limestone concrete.

Temperature 217°C 287°C 320°C 378°C 470°C

Average 0.82 0.75 0.73 0.70 0.70


NLI Standard deviation 0.02 0.00 0.01 0.003 0.009
Coefficient of variation 2.6% 0.7% 1.0% 0.5% 1.3%
Average, N/mm2 13,900 7,162 5,987 2,750 2,328
Ei Standard deviation 1,402 591 781 134 134
Coefficient of variation 10.1% 8.3% 13.0% 4.9% 5.8%
Average, N/mm2 19,591 13,063 11,727 6,136 5,172
Ec Standard deviation 715 1,033 1,361 258 225
Coefficient of variation 3.6% 7.9% 11.6% 4.2% 4.4%
Average, N/mm2 33,407 28,321 27,945 18,076 15,471
Eu Standard deviation 1,240 1,308 1,264 1,302 375
Coefficient of variation 3.7% 4.6% 4.5% 7.2% 2.4%
Average 15.29 27.64 32.47 72.29 86.81
DI Standard deviation 0.76 3.68 4.95 2.26 7.14
Coefficient of variation 4.9% 13.3% 15.2% 3.1% 8.2%
Average, µstrain 22.10 54.84 67.52 171.79 187.37
PS Standard deviation 5.36 9.55 18.35 9.55 23.18
Coefficient of variation 24.3% 17.4% 27.2% 5.6% 12.4%
Table (4) shows the Most Probable Values (MPV) of the coefficient of variation (Cv) of the stiffness damage test
parameters. The MPV was calculated by using the whole population without discarding the extreme values.

University of Portsmouth 10 Ayman Nassif


Table (4): The MPV of the coefficient of variation of the SDT parameters.

Stiffness Damage Test Parameters The Most Probable Value of (Cv)

NLI 2.2±0.3 %
Eu 6.6± 0.9 %
Ec 8.7±0.97 %
Ei 12.5±1.39 %
DI 14± 1.6%
PS 26.4±2.9 %

The SDT parameters have acceptable level of repeatability even though the population groups represent various
thermal exposures and concrete materials.

Conclusions
1. It is clear that all the Stiffness Damage Test parameters (Ei, Ec, Eu, DI, PS and NLI) are sensitive
to changes in the microstructure of concrete caused by elevated temperatures in the range
(217°C~470°C) regardless of the type of aggregate used. The elastic modulii (Ei, Ec and Eu) are
reduced by different amounts by heating. The initial modulus (Ei) generally exhibits the largest
reduction while Eu exhibits the smallest reduction. Both the Damage Index, as measured by the
area of hysteresis, and the internal plasticity, as assessed by the Plastic Strain, increase after
exposure to temperatures in the investigated range. The loading and unloading responses of heated
concrete are not linear and the non-linearity increases the higher the maximum temperature
experienced, regardless of the aggregate used.

2. The residual stiffness cannot be estimated solely from the knowledge of the maximum temperature
attained. This strongly indicates that the loss of residual stiffness is dependent on other
characteristics of thermal exposure as well as the maximum temperature experienced. The
available material models of predicting concrete structural behaviour during and after fire incidents
should adopt more accurate functions to reflect the modification in the mechanical properties of
concrete taking into account the effect of quenching, duration of exposure and type of the affected
concrete material.

3. The correlations between the Damage Index (DI) and the percentage residual modulii (Ei and Ec)
are remarkably strong, with the coefficient of correlation being 0.98 and 0.96 respectively. This
finding is used to introduce a new method of classification of the extent of fire-damage based on
the DI value, as shown in Table(2). The correlation holds true for the investigated siliceous and
carbonate aggregate concrete when subjected to various heating and cooling regimes.

4. The correlations between the Plastic Strain and the elastic modulii (Ei and Ec) are also very strong
with coefficients of correlation of 0.97 and 0.93 respectively. The correlation holds true for the
investigated siliceous and carbonate aggregate concrete when subjected to various heating and

University of Portsmouth 11 Ayman Nassif


cooling regimes. Indeed, the Damage Index and the Plastic Strain behave in a similar fashion and
correlate well with each other.

In the new damage classification method, It is suggested that PS is used as a confirmation to the
assessment established by DI. When the DI and PS are determined, the percentage residual stiffness can
be established. As the residual stiffness is also measured during the SDT, the original unaffected
stiffness can be readily estimated from such information. For instance, a DI value of 10, determined for
a damaged core, is most likely to be associated with a residual chord modulus of 51% of the unfired
value.

5. The repeatability of the SDT parameters is acceptable. The coefficient of variation of the Stiffness
Damage Test parameters (NLI, Ec, Eu, Ei, DI and PS) are within acceptable statistical
requirements, as shown in Table (4). However, PS exhibits the highest average coefficient of
variation (26.4±2.9%) while NLI has the lowest Cv, (2.2±0.3 %).

Further Research

The proposed classification system was based on laboratory experiments performed on uniformly heated concrete
cores. Therefore, it is essential to validate this new method on typical concrete members after exposure to real fires or
simulated ones. Indeed there are plans, at the University of Portsmouth, to carry out such validation and develop a
non-destructive in-situ test method based on the same principle as the Stiffness Damage Test. The author would very
much welcome any possible collaboration with interested organisations.

Acknowledgement

Thanks are due to Mr S.R. Rigden from Project Management and Design Services Ltd; Kent; UK, Dr W.J. French
from Geomaterial Research Services Ltd; Basildon; UK and Dr E. Burley for the fruitful discussions and
collaboration.

References
1. Concrete Society, 1990. "Assessment and repair of fire-damaged concrete structures" Technical Report

University of Portsmouth 12 Ayman Nassif


No.33.

2. Nassif, A., Burley, E., and Rigden, S.R. " The effects of rapid cooling by water quenching on the
stiffness properties of fire damaged concrete". Magazine of Concrete Research, accepted for
publication.

3. Popovics, S. sand; Popovics, J.S., 1992. "" A Critique of Ultrasonic Pulse Velocity Methods for Testing
Concrete." Proceedings of the Iinternational Cconference on the Nnon-destructive Ttesting of Cconcrete
Eelements and Sstructures., San AntonioAntoino, TX, April 1992, pp. pp.94-103.

4. Dorsch, D., 1991. "Repairing fire-damaged concrete turbine pedestal" Concrete International, Vol.13
No.6, pp.33-38.

5. Tomlinson, E., 1976. "Reinstatement of a fire damaged building" Fire Prevention Science and
Technology, No.15, pp.22-26.

6. Fleischer, C. and Chapman-Andrews, F., 1994. "Survey and analysis of bomb damaged reinforced
concrete structures" Construction Repair, Vol.8, No.2, pp.42-47.

7. Gustaferro, A.H., 1980. "Experiences from evaluating fire-damaged concrete structures" Fire Safety
of Concrete Structures, Special Publication SP80-10, ACI, USA, pp.269-277.

8. Kordina, K. and Seiler, H.F., 1976. "An examination of the effects of a big fire in some concrete
buildings" Fire Prevention Science and Technology, No.14, pp.4-18.

9. Forrest, J.C.M., 1980. "An international review of the fire resistance of lightweight concrete" 2nd
International Congress on Lightweight Aggregate Concrete, construction press, Lancaster, England,
pp.73-90.

10. Nassif, A. "Assessment of the fractured state of fire-damaged concrete". PhD Thesis, Queen Mary
and Westfield College, London University, January 1996.

11. Nassif, A.Y., Burley, E. and Rigden, S.R., 1995. "A new quantitative method of assessing fire damage to
concrete structures" Magazine of Concrete Research, Vol.47, No.172, pp.271-278.

12. Smith, L.M., 1983. "The assessment of fire damage to concrete structures". Thesis for PhD, Paisley
College of Technology, Paisley, Scotland.

13. Cruz, C.R., 1966. "Elastic properties of concrete at high temperatures" Journal of the PCA Research and
Development Laboratories, Vol.8 No.1, pp.37-45., PCA Research Bulletin 191.

14. Lie, T.T., 1968. "Fire resistance of buildings" National Research Council of Canada, Ottawa, Technical
translation No.1334.

15. Purkiss, J.A., 1972. "A study of the behaviour of concrete heated to high temperatures under restraint or
compressive load" PhD Thesis, University of London.

16. Sullivan, P.J.E., 1970. "Structural behaviour of concrete at elevated temperatures" PhD Thesis, London
University.

University of Portsmouth 13 Ayman Nassif


University of Portsmouth 14 Ayman Nassif

View publication stats