Engineering Structures 110 (2016) 293–306

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

An approach for evaluating residual capacity of reinforced concrete

beams exposed to fire
V.K.R. Kodur ⇑, Ankit Agrawal
Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI, United States

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents an approach for assessing the residual capacity of fire exposed reinforced concrete
Received 15 May 2015 (RC) beams. The approach involves capturing response of RC beams in three stages, namely, structural
Revised 16 November 2015 response at ambient conditions, thermo-mechanical response during fire exposure, and post-fire residual
Accepted 18 November 2015
response after cooling down of beams. Distinct material properties of reinforcing steel and concrete are
considered during heating and cooling phase of fire exposure and residual (after cool down) phase of
analysis. In addition, relevant load level, specific fire scenarios, boundary conditions, and plastic deforma-
Keywords:
tions that develop in a beam during fire exposure are also incorporated in evaluating residual response of
Reinforced concrete beams
Residual strength
fire exposed RC beams. The proposed approach is implemented using a detailed numerical model devel-
Finite element analysis oped in the finite element computer program ABAQUS. Predictions from the numerical model show good
Fire exposed beams correlation with the response parameters measured in experiments for evaluating residual capacity of
Residual deformations fire exposed RC beams. Also, predictions of residual capacity from the finite element analysis are com-
pared with that obtained from simplified sectional analysis based on maximum rebar temperatures con-
sideration. This comparison indicates that the finite element analysis yields more realistic predictions of
residual capacity than that predicted from simplified sectional analysis. The applicability of the proposed
approach in evaluating residual capacity of fire exposed RC beams is illustrated through a case study.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction ascertained. Unlike fire induced spalling and significant loss of

Reinforce concrete (RC) structural members generally exhibit
good fire resistance due to relatively low thermal conductivity,
high thermal capacity, and slower degradation of mechanical prop-
erties of concrete with temperature. However, under fire exposure,
RC members experience loss of strength and stiffness as a result of
increased temperatures in reinforcing steel and concrete.
In case of exposure to a severe fire, a RC member might experi-
ence significant structural damage resulting from loss of concrete
due to possible fire induced spalling, high rebar temperatures
and relatively large permanent deformations with very limited
residual load carrying capacity. Alternatively, exposure to moder-
ate fire scenarios may not result in noticeable deformations or loss
of concrete section due to spalling. In such scenarios, loss of capac-
ity in RC structural members due to fire exposure may not be sig-
nificant. Nevertheless, a fire exposed building (or structure) cannot
be opened for immediate reoccupation, even after fire is fully
extinguished, until load bearing capacity of RC members are
⇑ Corresponding author. Tel.: +1 5173539813.
E-mail addresses: kodur@egr.msu.edu (V.K.R. Kodur), agrawa36@msu.edu
(A. Agrawal).
cross section, which is a visible sign of damage, structural deterio-
ration in a RC member due to temperature induced degradation of
mechanical properties and redistribution of stresses within the
member, may not be too apparent. Thus, it is imperative to ascer-
tain the residual capacity of structural members through rational
engineering methods for facilitating re-occupancy or to develop
retrofitting measures in fire exposed concrete buildings.
The extent of strength and stiffness degradation in fire exposed
concrete members is dependent on a number of factors, including
type of fire exposure, properties of concrete and steel reinforce-
ment, load level and boundary (support) conditions. Many of these
factors are interdependent and can vary significantly in different
scenarios. Thus, residual strength evaluation of fire exposed RC
members is quite complex and the computed capacity can vary
widely depending on assumptions used in the analysis.
One approach to evaluate residual capacity of RC members is
through destructive fire tests in a laboratory environment. Such
residual capacity tests have been conducted in the past by
researchers [1–5] and this comprised of exposing RC beams to
standard or parametric fires for a predetermined duration, and
then loading the beams to failure after cooling down to room tem-
perature, if no failure occurred during fire exposure. Common

http://dx.doi.org/10.1016/j.engstruct.2015.11.047
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
294 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306
observations from these tests inferred include that significant
residual load carrying capacity remained in these members and
the extent of residual capacity varied with the duration of fire
exposure and peak rebar temperature.
Researchers have proposed nondestructive or numerical
approaches for evaluating residual capacity of RC beams following
fire exposure [6–8] as well. These approaches can be broadly cate-
gorized into two types, namely, simplified cross-sectional analysis
and, detailed finite element analysis. In the first type of approach,
modified versions of strength equations for evaluating capacity at
room-temperature are utilized accompanied with strength reduc-
tion factors based on high temperature exposure.
Hsu and Lin [6] proposed a simplified sectional approach to
evaluate residual strength of fire exposed RC beams. This approach
involves dividing critical cross section into a number of strips and
then calculating temperatures along each strip through a finite dif-
ference approach. Knowing temperatures across the cross section
and strength-temperature relations of reinforcing steel and con-
crete, residual capacity of the fire exposed concrete beam is evalu-
ated through a strain compatibility analysis. They concluded that
this approach could be applied to calculate post-fire moment
capacity, shear strength and elastic modulus of an RC beam. Also,
the rate of degradation in moment capacity, shear strength and
elastic modulus were different.
Kodur et al. [7] presented a simplified approach for evaluating
residual capacity of flexural members based on peak rebar temper-
atures experienced during a fire. As part of this method, an empir-
ical equation was proposed for predicting maximum rebar
temperatures under a specified parametric fire exposure. Knowing
maximum temperatures attained in the rebar, residual capacity
was computed using modified versions of room-temperature
strength design equations [9], but taking into account residual
properties of reinforcing steel. It was shown that the proposed
approach gives conservative predictions of residual moment
capacity of a fire damaged RC beam. Peak rebar temperature was
identified as the key parameter in ascertaining the level of damage
sustained to the beam during fire exposure.
In a more recent study by Bai and Wang [8], maximum temper-
atures across the cross section are evaluated using ANSYS [10] soft-
ware. In this approach, instead of assuming degradation in the
strength of concrete or steel reinforcement with increasing tem-
perature, the original section of concrete and rebar is reduced to
an equivalent (reduced) section to account for damage due to fire
and then post-fire residual capacity of RC members is evaluated.
Based on this study it was inferred that concrete or steel strength
has negligible influence on the rate of degradation in moment
capacity due to fire exposure.
In the second type of approach, researchers utilize a finite ele-
ment model to trace the post-fire response of RC beams. Ožbolt
et al. [11] have proposed a transient three dimensional thermo-
mechanical finite element model to simulate the behavior of RC
beams exposed to elevated temperatures. The model was validated
by comparing predictions against test data generated from post-
fire residual strength tests [2]. The load-carrying capacity and the
initial stiffness of the beam reduced with the increase in duration
of fire exposure. Moreover, it was found that on cool down of the
beam to room temperature, additional damage resulted in the
beam due to thermally induced strains.
A major drawback of the approaches proposed by Hsu and Lin
[6] and Bai and Wang [8] is that these approaches utilize temper-
ature induced degradation of mechanical properties of concrete
and reinforcing steel evaluated based on heating phase alone.
However, such properties are not representative of those observed
(residual mechanical properties) after cool down following fire
exposure. While concrete continues to display some level of
strength loss even after cooling down [12], reinforcing steel regains
most of its room temperature strength if maximum temperature
experienced during fire exposure remains below 500 °C [13]. Also,
most of the cross-sectional residual capacity evaluation
approaches discussed above do not account for strain hardening
effects in reinforcement, which can lead to under-prediction of
moment capacity by 15–25% than the actual value experienced
in real practice [7]. It is not possible to predict post-fire residual
deformations in RC beams through the current sectional
approaches. Finally, the model proposed by Ožbolt et al. [11] does
not account for distinct material properties during heating and
cooling (decay) phases of fire. Moreover, the primary focus of the
study was to study behavior of RC beams at elevated temperature
rather than evaluation of post-fire residual capacity of RC beams.
To overcome some of the above drawbacks, an approach is pro-
posed for predicting residual capacity and residual deflections of
fire exposed RC beams. The novelty of the current approach lies
in the consideration of distinct material properties of reinforcing
steel and concrete during heating and cooling phases of fire expo-
sure and residual (after cool down) phase, as well as in incorpora-
tion of plastic deformations occurring during fire exposure of RC
beams into post-fire response analysis. The proposed approach is
implemented through a detailed numerical model developed in a
finite element based computer program ABAQUS [14]. The model
is validated by comparing predictions against experimental data
generated in residual strength tests on RC beams [2,3]. Finally,
the numerical model is applied to evaluate post-fire residual
response of fire exposed simply-supported, unrestrained, rectan-
gular RC beams.
2. Approach for evaluating post-fire residual capacity
The extent of structural damage in a fire exposed RC beam is
influenced by a number of factors including load level, support
conditions, properties of concrete and reinforcement as well as fire
exposure scenario [7]. For evaluating post-fire residual response of
RC members, three stages of analysis are required to account for all
these parameters.
2.1. General procedure
The three stages of analysis for evaluating residual capacity of
fire exposed RC members comprise of, evaluating capacity at room
temperature prior to fire exposure (Stage 1), fire resistance analysis
during exposure to fire (Stage 2) and finally, post-fire residual anal-
ysis after cooling down of the member (Stage 3). A flow chart in
Fig. 1 illustrates various steps required for evaluating residual
strength of fire exposed RC beams. This analysis procedure can
be implemented using any finite element based package, such as
ABAQUS [14].
In Stage 1, room temperature capacity of the RC beam is evalu-
ated through a detailed finite element analysis by gradually incre-
menting the load on the structure till failure occurs. Alternatively,
the ultimate capacity can also be estimated using specified
strength equations in design standards [9]. The room temperature
capacity determined in Stage 1 is utilized to assess relative load
level on the beam prior to fire exposure during Stage 2. Also, the
capacity calculated during Stage 1 is used to estimate extent of
degradation in capacity once the residual capacity is ascertained
in Stage 3. It should be noted that for this stage of analysis, room
temperature mechanical properties of concrete and reinforcing
steel are utilized.
In Stage 2 of the analysis, the response of RC beam is evaluated
under a given fire exposure scenario, load level, and boundary con-
ditions. Realistic loads that are present throughout a typical fire
event are applied on the beam prior to thermo-mechanical analysis
 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306 295

START

Discretization of beam
for structural analysis

Stage 1 Evaluate response

Room under mechanical
temperature loading at room
stress-strain temperature
curves

Evaluate response
Temperature under fire exposure and
dependent thermal apply failure criteria
and mechanical
properties
Stage 2

Check failure Yes

No

Stage 3
Evaluate residual
Temperature
capacity of fire
dependent residual
exposed beam
material properties

STOP

Fig. 1. Flow chart illustrating the three stages involved in residual capacity analysis.

during fire exposure. The time–temperature curve for the entire
duration of fire exposure can be approximated using computer
models or empirically through parametric curves [15]. This stage
of the analysis is carried out at various time increments till the fail-
ure of the beam or till the fire exposure subsides. Response param-
eters from thermal and structural analysis are to be utilized at the
end of each time increment to check the state of the RC beam
under different failure limit states. In this stage, temperature
dependent thermal and mechanical properties of concrete and
reinforcing steel, that are distinct during heating and cooling
phases of fire exposure, are to be input into ABAQUS [14].
Following the cooling down of the beam, and if there is no fail-
ure of the beam in Stage 2, Stage 3 of the analysis is to be carried
out. The temperature induced residual stress and strains that exist
in the beam after fire exposure are evident in the form of residual
deformations observed when no loads are acting on the beam.
These residual deformations result from accumulation of damage
induced in the beam due to heating, structural parameters and
material properties. This state of the beam is the initial state for
Stage 3 of analysis. In this stage of the analysis, the cooled down
RC beam is loaded incrementally till failure and the structural
response of the beam is traced. The residual capacity corresponds
to maximum load that the beam can carry prior to failure. For this
analysis, residual properties of concrete and steel reinforcement
are required.
2.2. Selection of analysis parameters
The post-fire residual capacity of a RC beam depends on load
level, fire exposure scenario, structural parameters and material
characteristics both during fire and after cool down [7]. At higher
load levels and severe fire exposure conditions, concrete beams
may sustain significant loss of cross section resulting from spalling,
leading to very little residual capacity. If however, there is no (vis-
ible) sign of loss of cross section resulting from spalling, a fire
exposed RC beam might have experienced only marginal loss of
capacity. In these cases residual capacity evaluation of a fire
exposed RC beam is required for deciding on possible reuse and
retrofitting of the fire exposed structure.
296 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306
The post-fire residual properties of concrete and reinforcing
steel are influenced by maximum temperatures experienced dur-
ing the fire, time allowed for recovery after fire, as well as the
method of cooling used for quenching the fire. The residual com-
pressive strength of concrete experiencing temperatures of
220 °C or higher can decrease up to 20% of its original room-
temperature strength immediately after cooling down [16]. How-
ever, it has also been reported that with sufficient recovery time
at room temperature, concrete can regain 100% of its original
room-temperature compressive strength [17]. More recent investi-
gations [18] suggest that the ‘short term’ or ‘temporary phase’
wherein the compressive strength of concrete does not recover
can last until three years after fire exposure. After this phase, there
is noticeable recovery in concrete strength. This is referred to as
the ‘long term’ response of RC structures after fire exposure. The
cooling method (air-cooled or water-quenched) does not influence
the post-fire compressive strength of concrete as significantly as it
does to the failure strain and the elastic modulus [19]. Heated con-
crete that is cooled by quenching typically attains a higher com-
pressive strain at failure than it would if it were cooled in air.
There is no significant change in the post-fire failure strain of con-
crete as compared to its original room-temperature value for expo-
sure temperatures up to about 220 °C [20].
Generally, residual strength of concrete has smaller influence
on the residual flexural capacity of RC beams, under positive bend-
ing. This is because of a more gradual loss of concrete strength with
increasing temperatures, combined with lower temperatures
attained in farther compressive layers of concrete from the fire
exposed surface. In contrast to concrete, residual strength of rein-
forcing steel significantly influences the flexural capacity of an RC
beam after fire exposure. This is mainly due to the fact that rebars,
located closer to the fire exposed surface, not only attain much
higher temperatures, but also due to steel undergoing rapid
strength degradation with temperature. While the rate of degrada-
tion in the strength of concrete with temperature may be slightly
greater in concrete than reinforcing steel in some situations, the
overall temperatures attained in concrete than in reinforcing steel
for beams under positive bending exposed to fire from below are
relatively lower. Thus, residual strength of rebars is crucial in esti-
mating the residual capacity of an RC beam after fire exposure.
The residual strength in rebars depends mainly on the maxi-
mum temperature reached in the steel reinforcement. Provided
the temperature of hot-rolled reinforcing steel does not exceed
500 °C, reinforcing steel will recover almost 100% of its initial
room-temperature yield strength upon cooling [13]. However,
when heated to above 500 °C, reinforcing steel regains only part
of its initial strength. The ratio between ultimate strength and
yield strength which is about 1.5 at room temperature, also
decreases with increasing temperatures and becomes 1 at around
800 °C. This suggests that strain-hardening becomes less promi-
nent at very high temperatures [21].
Therefore, the choice of fire scenario and strength recovery
assumed in reinforcing steel after fire exposure are critical factors
to be considered, while evaluating residual capacity of fire exposed
RC beams.
2.3. Failure criteria
In undertaking residual strength analysis of a RC beam, different
failure criteria are to be applied at each stage of the analysis
depending on the applicable failure limit states. In Stage 1 of anal-
ysis (at ambient conditions), for evaluating capacity of the beam
prior to the fire exposure, strength limit state generally governs
the failure. In Stage 2 of the analysis (under fire exposure), a RC
beam under fire exposure experiences high temperatures and
mid-span deflections due to degradation in strength and stiffness
properties of concrete and reinforcing steel. In addition to strength
criterion, deflection limit state is a reliable performance index to
evaluate failure during fire exposure. Accordingly, the failure of
the beam is said to occur [22] when:

The maximum deflection in the beam exceeds L/20 (mm) at any
fire exposure time, or

The rate of deflection exceeds L2/9000d (mm/min).
Where, L = span length of the beam (mm) and d = effective
depth of the beam (mm).
It is important to note here that the aforementioned deflection
limit states are developed for isolated RC beams tested under stan-
dard fire conditions in the laboratory and therefore may not apply
to beams under realistic loading and fire scenarios. However, fail-
ure times obtained from the above deflection based limits would
be conservative under most practical saturations.
To evaluate residual capacity after exposure, in Stage 3 (after
cooling down of the beam), strength and extent of residual deflec-
tions generally govern failure.
3. Development of finite element model
The above methodology is applied for evaluating the residual
capacity of a fire exposed RC beam. A finite element model is devel-
oped to evaluate the response of the beams in Stages 1, 2 and 3.
3.1. General
The analysis is carried out using the finite element computer
program ABAQUS [14]. The constitutive models for concrete and
steel are defined within the framework of the software package
and the modeling of fire exposed RC beams is undertaken using
sequentially coupled thermo-mechanical analysis procedure. In
this procedure, the mechanical analysis utilizes the results (tem-
peratures) generated in the heat transfer analysis, but no reverse
dependency exists.
In Stage 1 analysis, room temperature capacity of the RC beam is
determined using detailed finite element analysis or strength equa-
tions present in design standards [9]. In Stage 2, the beam is sub-
jected to a specified load level taken to be a percentage of the
ultimate capacity of the beam (based on the estimated loads present
during fire exposure) and a realistic time–temperature curve result-
ing during fire exposure [15]. The response of the RC beam during
fire exposure in Stage 2 is traced through two sets of discretization
models, one for undertaking thermal analysis and the other for
undertaking mechanical (strength) analysis. Results from thermal
analysis are applied as thermal-body-loads on the structural model,
uniformly along the RC beam. Temperature dependent thermal and
mechanical properties of concrete and reinforcing steel are incorpo-
rated in the analysis. The load bearing capacity is evaluated by
applying relevant failure limit state as discussed in Section 2.2. Fol-
lowing cooling of the fire exposed RC beam, residual load-bearing
capacity of the beam is evaluated by undertaking Stage 3 analysis,
through loading the beam, in increments, until failure is attained.
It should be noted that the problem history can be divided into
sequential steps in ABAQUS with the response state (i.e. stresses,
strains and temperatures) of the beam being in each step of the
analysis. This enables response parameters to be transferred from
Stage 2 to Stage 3 of analysis.
3.2. Modeling assumptions
The following assumptions are made in the development of the
numerical model:
 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306

No bond-slip is assumed to occur between steel reinforcement
and concrete implying that there is perfect bond and that the
total strain in the reinforcement is equal to that in the concrete.
The assumption is quite accurate for the compression zone of
concrete where no crack occurs. On the contrary, cracks may
occur in the tensile zone causing weakness in the bond between
concrete and reinforcement, and resulting in tensile reinforce-
ment slip. However, over a length that includes several cracks
(beam segment), the average strain in both the reinforcement
and the concrete is approximately equal [23].

Fire induced spalling has not been explicitly modeled in the
analysis implying that the current model is applicable only for
normal strength concrete (NSC) with concrete compressive
strengths of 70 MPa or lower and in cases when there is mini-
mal spalling due to high temperature exposure [23].

Only short term post-fire residual mechanical properties of con-
crete are considered after cool down of the beam to room tem-
perature. The strength gain observed in compressive strength of
concrete months after fire exposure is conservatively ignored in
the analysis [18].

Transient creep strain in concrete is implicitly accounted for
through constitutive models, implying that this strain is
neglected upon cool down of the beam [24]. It should be noted
however, that in all generality it is preferable to use an explicit
transient creep model when studying concrete under natural
fire, in order not to recover this strain during cooling. However,
using an implicit model for transient creep does not lead to a
large error in deflection predictions for RC beams under natural
fires, especially during the cooling phase, since a relatively
small area of concrete is under compression. Nonetheless, it
should be noted that this can lead to slight underestimation
of residual strains in concrete and may not always be a conser-
vative assumption.

Thermal properties of concrete and reinforcing steel i.e. thermal
conductivity, heat capacity and thermal expansion have been
assumed to be completely reversible. The effect of decrease in
heat capacity due to loss of moisture, as well as residual thermal
expansion (or shrinkage), in concrete during cool down [25] is
neglected for simplicity.
3.3. Discretization of the beam
Two sub-models, namely, thermal and structural models, are
needed to carry out three stages of analysis. A structural model is
needed to carry out strength analysis in Stages 1, 2, and 3, while
Stage 2 of the analysis requires a thermal model to undertake heat
transfer calculations to compute temperatures in the RC beam.
For the structural analysis, the RC beam is discretized using
eight-noded continuum elements (C3D8) and two-noded link ele-
ments (T3D2) for concrete and reinforcing steel respectively.
C3D8 element has eight nodes with three degrees of freedom;
namely three translations in x, y, and z directions. This element
can be used for 3D modeling of solids with or without reinforce-
ment and it is capable of accounting for cracking of concrete in ten-
sion, crushing of concrete in compression, creep and large strains
[14]. T3D2 elements are used to model one-dimensional reinforc-
ing bars or rods that are assumed to deform by axial stretching
only. They are pin jointed at their nodes; only translational dis-
placements and the initial position vector at each node are used
in the discretization. When the strains are large, the formulation
is simplified by assuming that the trusses to be made of incom-
pressible material. This approach has been used effectively to
model reinforcement explicitly wherein nodes of reinforcement
are coincident with corresponding nodes of concrete [14]. The
interaction between concrete and reinforcement is achieved by
297
using the embedded region constraint, i.e. defining reinforcement
to be embedded in concrete.
For the thermal model in 3D space, concrete and reinforcement
are discretized using DC3D8 element (8 noded linear brick ele-
ment) and DC1D2 element (2 noded link element) respectively,
available in ABAQUS [14] library, having nodal temperature
(NT11) as the only active degree of freedom.
While beam elements are supposedly a more computationally
efficient choice for both the structural and thermal models, predic-
tion of an accurate temperature distribution within the member
requires the use of three dimensional elements. Moreover, similar
modeling strategies have worked well for researchers in the past
[26]. Finally, there is a possibility to extend such a model to include
complex phenomenon like concrete spalling in future studies.
A tie constraint is used to apply temperatures from concrete to
reinforcing steel bars at that location. The external surface areas
of DC3D8 elements, which are exposed to fire from three sides,
are used to simulate the surface effect of convection and radiation
that occur from fire (ambient air) to the beam. According to EN
1991-1-2 [15], the convective heat transfer coefficient is taken to
be 25 W/m2.°C on fire exposed surface. For the un-exposed surfaces,
a convective coefficient of 9 W/m2 °C is used to account for the
effects of heat transfer through radiation. The emissivity for radia-
tive heat transfer at the exposed surfaces of the concrete member is
taken as 0.8.
Since RC beams experience large deflections during fire expo-
sure, the effect of geometric non-linearity is included in numerical
model using updated Lagrangian method [14]. The Newton–Raph-
son method is employed as the solution technique with a tolerance
limit of 0.02 on the displacement norm as the convergence crite-
rion [27,28]. Also, line search function is activated to achieve rapid
convergence [29,30]. Similar analysis parameters have yielded
results with sufficient accuracy and efficiency in earlier studies
[26].
3.4. Input parameters for analysis
Various parameters such as load distribution, boundary condi-
tions, fire scenario and material properties are to be input to carry
out different stages of analysis. In Stage 1 of the analysis, the ulti-
mate capacity of the RC beam prior to fire exposure is calculated.
Based on ultimate capacity evaluated in Stage 1 a percentage of
the ultimate load is applied during fire exposure in Stage 2 of anal-
ysis. If the beam survives fire exposure during Stage 2, residual
response of the RC beam is evaluated in Stage 3.
For Stage 1 analysis, room temperature stress–strain relation-
ships for concrete and steel are utilized. The compressive stress–
strain curve of concrete is assumed to linear elastic until the stress
reaches 0.33 fc, where fc denotes the uniaxial compressive strength
of concrete. The linear portion of the curve is followed by a para-
bolic (non-linear) increase in stress representing development of
microcracks in the concrete. The post-peak softening behavior of
concrete is represented through a linear descending branch. A
damaged plasticity constitutive model [14] is used to model the
complex behavior of concrete, involving strong nonlinearity and
different failure mechanisms under compression and tension
(crushing or cracking). The stress–strain relation for concrete in
tension is represented by a bilinear relationship which is elastic
up to peak stress. The tensile strength of concrete at elevated tem-
peratures is assumed to vary as per Eurocode 2 [31] but with some
modifications [32] to avoid the conditions where the tensile
strength becomes zero at relatively low temperatures (600 °C). A
value of ect (ultimate strain in tension) that lies between 0.002
and 0.004, independent of temperature, have been assumed in lit-
erature [33]. A metal plasticity model that utilizes Mises yield sur-
face with associated plastic flow and isotropic hardening available
298 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306

in ABAQUS [14] is adopted for the constitutive modeling of rein-

Normalized Compressive Stregth

1.2
forcing steel. Elevated Compressive Strength
In Stage 2 of analysis, during heating phase, the temperature 1 Residual Compressive Strength
dependent thermal and mechanical properties of reinforcing steel
and concrete are assumed to follow as that of Eurocode 2, 3 and 0.8

(fck,T/fck)
4 provisions [12,31,34]. Moreover, variation of the Poisson’s ratio
0.6
of concrete is assumed based on existing literature [35]. The vari-
ation of mechanical and thermal properties with respect to tem- 0.4
perature, is different in cooling phase as compared to heating
phase, and depends on the maximum temperature reached during 0.2
heating phase. During the cooling phase, a linear interpolation
0
between the elevated and residual material properties after cool 0 200 400 600 800 1000 1200 1400
down is adopted [12].
Temperature (oC)
In Stage 3 of analysis, after cooling of the fire exposed RC beam,
the residual uniaxial compressive and tensile strength of concrete Fig. 3. Normalized compressive strength of concrete with maximum exposure
after cooling down to room temperature is assumed to be 10% less temperature.
than the strength attained at the maximum temperature. This
assumption is based on Eurocode 4 [12] recommendations and
has been shown to yield accurate results in predicting residual Room Temperature (20°C)
capacity of RC columns immediately after cool down [36]. The Residual (200°C)
450
Residual (500°C)
residual stress–strain relationship for reinforcing steel is calculated 400
Residual (600°C)
using degradation trends reported by Neves et al. [13] (Fig. 2). 350 Residual (800°C)
The stress–strain curves for concrete as well as reinforcing steel
Stress (MPa)
300
adopted during room temperature analysis, thermo-mechanical
250
analysis during heating (elevated) and cooling phases of fire expo-
200
sure, and residual (after cool down) stage of analysis are plotted in
150
Figs. 3 and 4 respectively.
100
50
3.5. Output results
0
0 0.05 0.1 0.15 0.2 0.25
Displacements, stresses and temperature fields are the primary Strain
output variables that are generated during different stages of anal-
ysis. In Stage 1 of analysis, failure of the beam is ascertained based Fig. 4. Residual stress–strain curves for reinforcing steel having yield strength of
on deflection limit state or when any further increment in applied 420 MPa.

load leads to instability. In Stage 2 of analysis, at each time step,

the output from the thermal analysis, namely nodal temperatures,
residual mechanical properties to be used for the residual capacity
is applied as a thermal body load on the structural elements
evaluation in Stage 3 of analysis, when necessary. In Stage 3 of
(nodes) to evaluate the structural response of RC beam under fire
analysis, the load deflection response is utilized to evaluate resid-
exposure. An identifier to ascertain if the material is under heating
ual load carrying capacity of a fire exposed reinforced concrete
or cooling phase to apply the appropriate material properties is
beam. Furthermore, the sectional capacity at failure is calculated
updated in the structural analysis using subroutine UFIELD pro-
by integrating stresses experienced within the concrete and rein-
vided in ABAQUS [14]. Also, the maximum temperature experi-
forcing steel, as generated in ABAQUS [14], to ascertain failure
enced at each node during thermal analysis is used to calculate
based on moment capacity or shear capacity criteria.

4. Model validation

Four RC beams tested by Dwaikat and Kodur [3] and Kumar and
Kumar [2] respectively were selected to validate the above devel-
oped numerical model. These tests are selected for validation
because comprehensive results have been reported by the two
authors and this facilitates finite element simulations and detailed
comparisons.

4.1. Tests by Dwaikat and Kodur [3]

Two concrete beams tested under fire exposure by Dwaikat and

Fig. 2. Normalized residual strength of hot rolled reinforcing steel [13].
Kodur [3] were analyzed by applying the above numerical proce-
dure to validate the proposed approach. The characteristics of the
beam, together with the summary of the results are tabulated in
Table 1. These beams are made of normal strength concrete (desig-
nated B1 and B2 as per Table 1) and there was no observed spalling
in these beams during fire exposure. The dimensions and reinforce-
ment details of both these beams are identical and are shown in
Fig. 5. Both beams are tested under two point loading of 50 kN
 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306 299

Table 1
Summary of test parameters and results for beams used for validation.

Study Beam designation Fire exposure Maximum rebar Residual Residual load bearing capacity (kN)
temperature (°C) deformation (mm)
Measured Model Measured Model Measured Model Simplified approacha
Dwaikat and Kodur [9] B1 ASTM E119 577 579 – – – – –
B2 SF 493 496 13.7 20.9 119.5 120.8 82.7
Kumar and Kumar [10] Reference beam None – – – – 166 168 149
Beam 1 (1 h exposure) ISO 834 Not measured 576 18 16 139 140 132
Beam 2 (1.5 h exposure) ISO 834 Not measured 648 22 20 119 120 114
Beam 3 (2 h exposure) ISO 834 Not measured 706 28 23 90 101 86
a
Simplified approach proposed by Kodur et al. [7].

(as depicted in Fig. 5a) which produced a bending moment equal to 1200
55% of the beam capacity, as per ACI 318 [9] capacity equation. In
the tests, one of the beams was subjected to ASTM E119 [37] stan- 1000
dard fire and the other beam was subjected to a Short Design Fire

Temperature (oC)
(SF) as illustrated in Fig. 6. 800
In the fire tests, the authors applied loading 30 min before the
start of the fire and this loading was maintained till no further 600
increase in deformation could be measured. This was selected as
the initial condition for the deflection of the beam. The load was 400
then maintained constant throughout the duration of fire expo- ASTM E119
sure. Beam B1 failed during fire exposure after 180 min. In case 200
Short Design fire (SF)
of beam B2, failure did not occur during fire exposure and this
beam was tested for residual capacity after cool down to ambient 0
conditions (24 h after fire exposure). The beam was subjected to 0 30 60 90 120 150 180 210 240
incremental loading at the rate of 3 kN/min till failure occurred. Time (min)
These two beams were analyzed in ABAQUS [14] by applying
Fig. 6. Time–temperature curves for fire scenarios used in fire tests on beams B1
the above discussed procedure. The validation process included
and B2.
comparison of thermal and structural response predictions from
the analysis with that reported in the two fire tests on beams B1
and B2. Temperatures in concrete and reinforcing steel in beam than in the case of beam B1 during the heating phase of the fire.
B1 exposed to ASTM E119 [37] standard fire time–temperature This can be attributed to the steep rise in the fire temperature dur-
curve rose steadily until failure. In beam B2, exposed to a Short ing early stages of fire exposure. After attaining a maximum value,
Design Fire (SF), the rate of temperature increase is slightly higher temperature start to drop in the beam due to presence of a decay

(a) Elevation and cross-section of beams ‘B1’ and ‘B2’

(b) Elevation and cross- section of beams ‘Reference beam’, ‘Beam 1’, ‘Beam 2’, ‘Beam 3’ and
‘Beam 4’
Fig. 5. Loading and reinforcement details of RC beams used for validation.
300 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306

phase in this design fire exposure. It is interesting to note that peak 0 30 60 90 120 150 180 210 240 270 300
temperatures in beam B2 occur during the decay phase of the fire
-10
exposure. This can be attributed to thermal inertia of concrete,

Mid-span deflection (mm)

which leads to a lag between fire temperatures and the tempera-
-30
tures within beam cross section. A comparison of measured and
predicted temperatures at various locations in beams B1 and B2 -50
in Fig. 7 indicates that there is close agreement between predicted
and measured values during the fire test. -70
The measured and predicted mid-span deflection response in
-90
two beams B1 and B2, during the fire test is shown in Fig. 8. It B2(Test)
can be seen that mid-span deflections increase during early stages B2(Model)
-110
of fire exposure due to degradation of strength and stiffness prop-
B1(Model)
erties of concrete and reinforcing steel with temperature. Since -130
B1(Test)
temperatures in beam B1 continue to increase steadily during
the entire duration of fire exposure, failure occurs due to signifi- -150
cant degradation in properties of concrete and reinforcing steel Time (min)
after about 180 min of fire exposure. For beam B2 however, due
Fig. 8. Comparison of the predicted and measured mid-span deflections for beams
to presence of cooling phase, the temperatures experienced in con- B1 and B2 during fire exposure.
crete and reinforcing steel of the beam are relatively lower and a
recovery in mid-span deflection is seen during the decay phase
of fire. It can be seen in Fig. 8 that the predicted mid-span deflec- The measured and predicted load–deflection response in beam
tions from the developed model agree well with measured B2, which did not fail during fire exposure, is presented in Fig. 9.
response during the fire test. There is an unrecoverable deflection during fire exposure which
was assumed to be the initial state for the residual strength analy-
sis. The predicted load deflection response by the finite element
model is very similar to the measured response during the test.
Furnace temperature (ASTM E119) These residual deformations represent the state of structural dam-
Rebar(Test)
Rebar(Model) age in the beam resulting from fire exposure and the extent of
Quarter-depth(Test) damage depending on load level, boundary conditions, and tem-
Quarter-depth(Model)
Mid-depth(Test) perature induced degradation in material properties. While deflec-
1200 Mid-depth(Model) tion prediction during fire exposure is lower than test data, the
final residual deformation predicted by the model is relatively
1000
higher than the measured value during tests. This difference can
Temperature (oC)

800 be attributed to the fact that cooling phase properties adopted in

the study are based on Eurocode 4 [5] provisions which are conser-
600 vative and lead to higher prediction of post fire residual deforma-
tions. Also, there is uncertainty regarding the adopted process in
400 unloading of the beam in experiments after fire exposure before
carrying out the residual test. Moreover, since test data is available
200 only for first 300 min, no comparison could be made with pre-
dicted deformation beyond 300 min while the beam cooled down
0
0 30 60 90 120 150 180 to room temperature. It should also be noted that he predicted
Time (min) residual capacity in is higher than room temperature capacity com-
puted as per ACI 318 [9]. This is mainly due to the effect of strain
(a) Beam B1 hardening of steel reinforcement which is not taken into account in

Furnace Temperature (SF)

Rebar (Test)
Rebar(Model) B2(Test)
Quarter-depth(Test) 180 B2(Model-without residual deformations)
Quarter-depth(Model)
1200 Mid-depth(Test) 160
B2(Model)

1000 140
Temperature (oC)

800 120
Load (kN)

100
600
80
400
60
200
40
0 20
0 30 60 90 120 150 180 210 240 270 300
0
Time (min) 0 20 40 60 80 100 120
(b) Beam B2 Deflection (mm)

Fig. 7. Comparison of predicted and measured cross sectional temperatures for Fig. 9. Comparison of the predicted and measured residual load–deflection
beams B1 and B2 during fire exposure. response for beam B2.
 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306
ACI 318 [9] strength design equations [7]. The beam failed in flex-
ure mode at 187 min during numerical simulations and this agrees
with experimentally observed failure mode.
Furthermore, to illustrate the influence of residual deformations
on post-fire response of the beam, residual response of fire exposed
beam B2 is evaluated from a single stage simplified analysis (with-
out residual deformations), and three stage analysis (with residual
deformations). Maintaining all other relevant analysis parameters
identical, the load–deflection response of the beam is calculated
using post-fire residual properties of concrete and reinforcing steel
based on its temperature history. This implies there is no residual
deflection in the beam at the start of the analysis in evaluating
residual response. Fig. 9 shows the difference between the post-
fire response of the beam without considering residual plastic
deformations as opposed to the three stage detailed analysis proce-
dure proposed in this paper. The residual load-carrying capacity of
the RC beam without incorporating residual capacity is calculated
to be (153 kN) which is significantly higher (approximately 26%
more) than that obtained using a three stage (120.8 kN) procedure.
Moreover, it is not possible to predict residual deformations occur-
ring in the beam after fire exposure using a single stage procedure.
This comparison clearly infers that accounting residual deflections
is critical in evaluating realistic residual capacity in fire exposed RC
beams.
Finally, in order to re-examine the assumption made to neglect
strains due to residual expansion (or shrinkage) in analysis, the
magnitude of residual plastic strains occurring at the critical sec-
tion (mid-span) are compared with expected strains from residual
thermal expansion or shrinkage (Fig. 10). The strains from thermal
expansion are calculated based on the maximum temperature
attained at each node of the cross section during the entire fire
regime [25]. While the strains from residual thermal expansion
(shrinkage) is significant, they are still generally lower, varying
from 0.012% to 0.6%, as compared to temperature induced residual
plastic strains ranging from 0.03% to 1.5%.
Overall, predictions from the proposed model are in good agree-
ment with the reported test data. The slight differences in deflec-
tion predictions can be attributed to minor variations in
idealization adopted in the analysis, such as stress–strain relation-
ship of steel and concrete. However, it is important to note that the
predicted load–deformation curve in Stage 3 follows a very similar
trend to the experimentally measured values. Moreover, the pre-
dictions by the numerical model are closer to experimental values
than those calculated using simplified cross sectional approaches.
4.2. Tests by Kumar and Kumar [2]
The model is further validated by comparing analysis predic-
tions against data from another four beams tested by Kumar and
Strain
0 0.005 0.01 0.015
0 1000
Residual plastic strain
0.05
Concrete depth (mm)
Residual thermal strain
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
Fig. 10. Expected residual thermal expansion (or shrinkage) strains versus residual
plastic deformations across the section at mid-span for beam B2.
301
Kumar [2]. These beams were made of normal strength concrete
and the dimensions and reinforcement details of these beams were
identical and are shown in Fig. 5b. The test parameters and results
of the chosen beams and is summarized in Table 1.
One of the beams was tested by loading to failure, without any
exposure to fire, and this served as the control or reference speci-
men. In the fire tests, four beams were subjected to ISO 834 stan-
dard fire exposure from three sides. The exposure duration varied
from 1, 1.5, 2 and 2.5 h respectively. During fire exposure no super-
imposed (external) load was applied on the beams. The heated
beams were then allowed to cool naturally to ambient temperature
and subsequently tested under four point bending (Fig. 5b) in load
control mode. All the beams were tested within six weeks after fire
exposure [2]. The load–deflection response for the control beams
and beams with 1, 1.5 and 2 h fire exposure was reported by the
authors [2]. However, the beam that was heated for 2.5 h showed
excessive deflections and failed during fire exposure. Hence, no
residual capacity test is performed on this beam.
The response of the control beam (without fire exposure) and
three beams subjected to 1, 1.5 and 2 h of fire exposure, is simu-
lated by applying the above proposed approach in ABAQUS [14].
For the thermal analysis the heating scenario is assumed to be
according to ISO 834 [38] standard fire, as in the fire tests. The
cooling of the beam is simulated through a linear decrease of air
temperature, following fire exposure, for a period of two hours.
The respective fire curves are illustrated in Fig. 11. The tempera-
ture across the beam cross section during fire exposure was not
recorded during the fire tests. In order to establish confidence
regarding the thermal analysis procedure, a beam with a cross-
section of 160  300 mm is analyzed. Temperature contours pre-
dicted by the model after 0.5, 1 and 1.5 h of fire exposure are in
good agreement with corresponding contours provided by Euro-
code 2 [31] for similar exposure and thus establishes the validity
of the thermal analysis procedure.
A comparison of measured and predicted load–deflection
response for the reference beam (with no heating) is shown in
Fig. 12a. The total load is plotted as a function of deflection mea-
sured during static four point bending. Test data shows that
post-peak response of the beam could not be captured since the
test was load controlled. In the response predicted by the numer-
ical model, the post-peak response shows sudden drop of load
which indicates imminent failure. Fig. 12a shows that for the refer-
ence beam, there is good agreement between measured and
numerically predicted load–deflection response.
The load–deflection response for the beams tested for residual
capacity after fire exposure is also presented in Fig. 12b–d.
Although no loads were applied during fire exposure, residual
deflection of 18 mm, 22 mm and 28 mm was measured in tests
for beams Beam 1, Beam 2 and Beam 3 respectively. The predicted
1200
0.02
2 h Exposure
Temprature (oC)
800 1 h Exposure
1.5 h Exposure
600
400
200
0
0 50 100 150 200 250 300
Time (min)
Fig. 11. Fire exposures for tested beams.
302 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306

1.5h Exposure(Test)
1.5 Exposure(Model-without residual deformations)
140 1.5h Exposure(Model)
180
160 120
140
100

Load (kN)
Load (kN)

120
80
100
80 60
60
40
40
No Exposure(Model) 20
20
No Exposure(Test)
0 0
0 5 10 15 20 25 30 35 0 20 40 60 80 100 120
Deflection (mm) Deflcetion (mm)

(a) No exposure (room temperature testing) (c) Residual response after 1.5 hour exposure

1h Exposure(Test) 2h Exposure(Test)
1h Exposure (Model-without residual deformations) 2 Exposure(Model-without residual deformations)
1h Exposure(Model) 120 2h Exposure(Model)
160
140 100
120
Load (kN)
Load (kN)

80
100
80 60
60
40
40
20
20
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Title Title
(b) Residual response after 1.5 hour exposure (d) Residual response after 2 hour exposure
Fig. 12. Comparison of predicted and measured post fire load–deflection response for tested beams.

values of residual deflections are in close agreement with test data In addition to above comparison, strains due to residual thermal
(refer Table 1). However, in this case, the residual deformations expansion (or shrinkage) are compared with permanent deforma-
predicted by the model are lower than that observed experimen- tions occurring due to temperature induced degradation at the
tally. This can be attributed to the fact that significant fire induced critical section (mid-span) for all three beams (refer Fig. 13). The
spalling was reported in the beams as the beam cooled down to comparison of these strains shows that the residual thermal
room temperature and before residual tests could be carried out. strains, expansion (or shrinkage), are relatively small as compared
This is especially important in the case of Beam 3, wherein exces- to the magnitude of residual plastic strains occurring at the same
sive spalling had been reported due to prolonged fire exposure. In location. The residual expansion (or shrinkage) for all the three
the case of Beam 1 (with 1 h exposure), the residual capacity was beams varies between 0.06% (shrinkage) and 0.6% (expansion)
found to be approximately 83% of the reference beam. Also, the
predicted response by the model, for the case when the beam is
exposed to 1 h of fire exposure, is in good agreement with its
experimental counterpart. It can be seen that for the cases of 1.5
and 2 h of fire exposure i.e. Beam 2 and Beam 3, the ultimate load Strain
-0.005 0 0.005 0.01 0.015 0.02 0.025
carrying capacity reduces significantly (approximately 50% of the 0
1 h Exposure(Residual thermal strain)
reference beam). This reduction occurs due to rapid degradation 1.5 h Exposure(Residual thermal strain)
in mechanical properties of reinforcing steel and concrete. The 0.05 2 h Exposure(Residual thermal strain)
1 h Exposure(Residual plastic strain)
residual capacity predicted by the model is higher than that mea-
Concrete depth (mm)

0.1 1.5 h Exposure(Residual plastic strain)

sured experimentally for Beam 2 and Beam 3. This is attributed to 2 h Exposure(Residualplastic strain)

the fact that during fire experiment, reduction in cross-section due 0.15
to spalling was reported after 1.5–2 h of fire exposure. Spalling of
concrete due to high temperature exposure is not explicitly 0.2

accounted in the currently developed numerical model. Hence, 0.25

the residual capacity predicted by the current model for Beam 2
and Beam 3 tends to be slightly higher than that measured during 0.3
tests. It should also be noted that all beams failed in flexural mode.
0.35
Moreover, the overall response of beams became more ductile with
greater duration of fire exposure as can be observed from the over- Fig. 13. Expected residual thermal expansion (or shrinkage) strains versus residual
all deflection trend plotted in Fig. 12b–d. plastic deformations across the section at mid-span for tested beams.
 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306
which is relatively lower than the magnitude of plastic strains that
are between 1.1% and 2.3% in the three beams studied. A single
stage analysis without incorporating residual deformations yields
significantly higher values of residual capacity as compared to
three stage analysis proposed in this paper (Fig. 12b–d). The
post-fire residual capacity, not accounting for residual deforma-
tions, is calculated to be 152 kN, 132 kN and 113 kN for tests car-
ried out after 1 h, 1.5 h and 2 h of fire exposure respectively.
These predictions are higher by 9%, 15% and 10% respectively when
compared with the corresponding predictions made by the pro-
posed three stage analysis that incorporates residual deformations.
Overall, good agreement between the predictions and the test
data demonstrate the validity of the proposed approach in evaluat-
ing residual capacity of RC beams. However, it should be noted that
the model overestimates the residual flexural capacity of RC beams
in all cases, especially when spalling occurred in the beams. This
can be attributed to the fact that loss of bond between reinforcing
steel and concrete, as well as fire induced spalling is not explicitly
considered in the current model. Also, the predictions by the model
are more realistic than those by simplified cross sectional analysis.
5. Case study
The applicability of the above validated finite element model to
evaluate the residual capacity of the fire exposed reinforced con-
crete beam is further illustrated through a case study.
A simply supported RC beam (designated BX1 as per Table 2)
with a 6 m span having a rectangular cross section of 300 mm by
480 mm is considered for residual strength analysis. The RC beam
has 3, u20 mm bars as tensile reinforcement and 2, u14 mm bars
as compression reinforcement. Stirrups are of u6 mm spaced at
150 mm center to center. Main reinforcing bars and stirrups have
yield strength of 420 MPa and 280 MPa respectively. Only 5 m of
the central span is exposed to fire. A uniformly distributed load
is applied on the beam and this produces a total bending moment
(including bending moment due to dead load) of about 50% of the
ultimate load carrying capacity of the beam. Also, as per ACI 216
[39], the fire resistance of this beam is 120 min (2 h). The dimen-
sions, reinforcement details, loading and support conditions are
presented schematically in Fig. 14. A summary of the case study
parameters are presented in Table 2.
The residual strength analysis is carried out on the above beam
by exposing it to a typical fire that can occur within a building
compartment. To simulate such practical fire exposures scenarios,
a typical compartment with a total area of 360 m2 was assumed to
be under fire. The fire load density for the compartment was
assumed to be 600 MJ/m2 (80th fractile fuel load for an office
building as per Eurocode 1). Also, for a medium rate of fire growth,
time for maximum gas (fire) temperature in case of fuel controlled
fire (tlim) is assumed as 20 min as per Eurocode 1 recommenda-
tions. Parametric time–temperature curves were calculated
according to recommendations of Eurocode 1 for varying opening
pffiffiffiffiffiffiffi
factors (o ¼ Av heq =At ) which is a function of area of openings
(Av) equivalent height of openings (heq) and total area (At). The
respective time–temperature curves for four different fire exposure
scenarios designated DF1, DF2, DF3 and DF4 are presented in
Fig. 15.
As part of Stage 1 analysis, the load carrying capacity of the RC
beam at room temperature is evaluated and results from Stage 1
analysis are presented in the form of load–mid-span deflection
response of the beam in Fig. 16. It can be seen that the mid-span
deflection increases linearly with load till yielding of reinforcing
steel and then the response becomes nonlinear due to the onset
of material and geometric nonlinearity that is incorporated in the
analysis. In the nonlinear range of response, the mid-span deflec-
303
tion increases at a faster pace with small increments in loading
and this is mainly due to spread of plasticity in the steel reinforce-
ment. Finally, the beam attains failure when it can no longer sus-
tain any further increase in load. It is important to note that the
predicted capacity at room temperature is higher than calculated
using ACI 318 [9], and this can be attributed to strain hardening
of the steel reinforcement, which is conservatively not accounted
for in ACI 318 design equations (see Table 2).
In Stage 2 analysis, both thermal and structural response of the
RC beam under fire exposure is evaluated. An initial loading, corre-
sponding to 50% of the ultimate load carrying capacity, is applied
on the beam to simulate realistic stress levels during fire exposure.
The peak rebar temperature depends not only on peak fire temper-
atures but the duration of heating and cooling phases of fire expo-
sure as well. The maximum rebar temperature is attained in the
parametric fire exposure DF1, wherein the peak fire temperature
is the lowest but the duration of the heating phase is the longest
compared to other fire exposure scenarios. The mid-span deflec-
tions of the RC beam under different fire exposures, from the
sequentially coupled thermo-mechanical analysis, are presented
in Fig. 16. No failure occurs in the beam under all four parametric
fire exposure scenarios and thus the beam has residual capacity
after cooling down. The mid-span deflections plotted in Fig. 16 also
indicate that largest deformations occurred in fire exposure sce-
nario DF1, which lead to higher peak rebar temperature as com-
pared to fire exposure scenarios DF2, DF3 and DF4. The fire
exposed beam recovers part of its stiffness and mid-span deflection
reverts as soon as rebar temperatures begin to reduce during the
cooling (decay) phase of fire exposure (refer Fig. 17). However, only
10–20% of the peak deflections are recovered after cool down to
room temperature which suggests that deformations experienced
during fire exposure are plastic and are not fully recoverable.
Residual response of the beam after different parametric fire
exposures is traced through incremental loading in Stage 3 of anal-
ysis. The load–deflection response from residual capacity analysis
is plotted in Fig. 18 and the results from the analysis are summa-
rized in Table 3. It can be inferred from the tabulated results that
significant residual deformations remain after fire exposure in RC
beams. Moreover, the extent of residual deformation varies with
fire severity, with larger residual deformations for more severe fire
exposure scenarios. Also, post-fire residual capacity decreases with
increased residual deformations in a RC beam. This infers that the
magnitude of residual deformations determine the residual capac-
ity in fire exposed RC beams. It should also be noted that the over-
all behavior of the beam becomes more ductile under more severe
fire exposure. The deflection at failure of beam BX1 after exposure
to design fire scenario DF1 is 114 mm as compared to only 53 mm
for the beam simulated without any fire exposure.
The influence of accounting for residual deformations on post-
fire response of the beam is clearly illustrated in Fig. 19. Maintain-
ing all other relevant analysis parameters identical, post-fire resid-
ual response of fire exposed beam BX1 under DF3 fire exposure is
evaluated from a single stage simplified analysis (without residual
deformations) and three stage analysis (with residual deforma-
tions). The single stage analysis yields residual load-carrying
capacity of the RC beam to be 263 kN, which is significantly higher
(approximately 13%) than that obtained using a three stage proce-
dure, which was calculated to be 230 kN. This comparison clearly
establishes that it is critical to account for residual deflections for
evaluating post-fire residual capacity in fire exposed RC beams.
The beam, BX1, studied in this case study, retained 60–70% of its
room temperature capacity even after exposure to various
parametric fire exposure scenarios. This confirms that significant
residual capacity is retained in fire exposed RC beams and the
beams can be rehabilitated through minor repair and retrofitting
strategies. However, significant uncertainty exists in post-fire
304 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306

Table 2
Properties and analysis results for case study beam BX1.

Beam designation Beam dimensions (mm) Support conditions Ultimate capacity Initial load (kN/m) Fire resistance (ACI 216) (min)
(kN/m)
ACI 318 Model
BX1 300  480 SS 254.3 299.8 25 120

w kN/m A Compartment
Wall

A’
5000 mm
5300 mm

6000 mm
(a) Elevation

2 Ø10 mm rebars
300 mm
25 mm (cover
to reinforcement)

Ø 6 mm srrups
480 mm 150 mm c/c

3 Ø 20 mm

25 mm (cover
to reinforcement)

(b) Cross-section A-A’

Fig. 14. Dimensions and reinforcement details of a typical RC beam selected for case study.

1200 350
DF1(O=0.02)
1000 DF2(O=0.04) 300
Temperature (oC)

DF3(O=0.09)
800 DF4(O=0.2) 250
Load (kN)

600 200

150
400
100
200
50
0
0 60 120 180 0
Time (min) 0 10 20 30 40 50 60
Mid-span deflection (mm)
Fig. 15. Parametric time–temperature curves for varying opening factors.
Fig. 16. Load–deflection response of the RC beam at ambient conditions (Stage 1).

behavior which must be investigated through rational engineering

approaches as presented in this study.
6. Future work
The approach presented here, is capable of tracing the response
of RC beams from the initial pre-fire stage to the post-fire residual
stage under any given fire and loading scenarios. This approach can
be implemented using finite element based computer program.
Using such a numerical model, a designer can rationally estimate
residual capacity of RC beams by explicitly accounting for a wide
variety of fire, material and structural parameters.
While the applicability of the approach has been demonstrated
for RC beams, it can be applied to other RC members under flexural
loading such as slabs. While some assumptions made in modeling
may not hold good for RC members under compression (columns),
the overall approach is applicable for predicting post-fire response
of RC columns as well.
 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306 305

0 100 200 300 400 500 600 700 800 BX1(Without residual deformations)
10 300
DF1 BX1(With residual deformations)
0 DF2
DF3 250
-10 DF4
200

Load (kN)
-20
Deflection (mm)

-30
150
-40
100
-50
-60 50
-70
0
-80 0 20 40 60 80 100
-90 Deflection (mm)
-100 Fig. 19. Influence of incorporating residual deformations on post-fire response of
Time (min) beam BX1 for fire scenario DF3.

Fig. 17. Mid-span deflections for different fire scenarios for beam BX1.

350 7. Conclusions
DF2
DF4 A procedure based on nonlinear finite element analysis is devel-
300
ROOM
DF3 oped for evaluating residual capacity of fire exposed RC beam. The
250 DF1 procedure is validated against experimental data following which
Load (kN)

it is applied to evaluate residual capacity of fire exposed simply-

200
supported, unrestrained, rectangular RC beams. Based on the
150
results of analysis the following conclusions can be drawn:

100
50
0
0 20 40 60 80
Deflection (mm)
Fig. 18. Residual load–displacement response after different fire exposures for
beam BX1.
In its current form, the model cannot account for damage
resulting from fire induced spalling or temperature induced deteri-
oration in interfacial bond at steel reinforcement and concrete
interface. These effects are particularly important for studying
post-fire response of RC structures made of high strength concrete.
Besides extending the current approach to incorporate these
effects, a series of parametric studies to integrate the influence of
various parameters on the post-fire response of RC beams is
needed. Moreover, the influence of factors like specific composi-
tion, cooling rate and loading needs to be quantified through fur-
ther analysis. Data generated from such studies can be utilized to
develop rational engineering based approaches to ascertain post-
fire residual capacity of RC beams. A detailed study on these
aspects is currently in progress at Michigan State University.

Three stages of analysis are required for evaluating the residual
capacity of fire exposed reinforced concrete structural mem-
bers, namely at ambient conditions, during fire exposure, and
following cool down of fire exposed beam, while taking into
100 120 140
consideration, temperature induced plastic deformations and
temperature induced degradation of material properties.

Incorporating temperature dependent strain hardening proper-
ties of steel reinforcement and tension stiffening properties of
concrete leads to more realistic predictions of residual capacity
than simplified cross sectional analysis.

Large irrecoverable plastic deformations can remain in fire
exposed RC beams, and higher residual deformations leads to
lower post-fire residual capacity in RC beams.

RC beams analyzed in the present study, retain almost 60–70%
of their ultimate room temperature capacity for a range of para-
metric fire exposure scenarios with a distinct cooling phase.
Acknowledgements
The authors wish to acknowledge the support of United States
Agency for International Development (through Pakistan-US
Science and Technology Cooperative Program grant PGA-
2000003665) and Michigan State University for undertaking this
research. Any opinions, findings, conclusions, or recommendations

Table 3
Predicted residual deformation and residual capacity in a RC beam under different fire scenarios.

Beam designation Fire exposure Maximum rebar temperature (°C) Peak deformation (mm) Residual deformation (mm) Residual load bearing capacity
(kN)
Model Simplified approacha
BX1 DF1 559 88 69 202.9 192.3
DF2 401 64 58 224.9 208.5
DF3 289 48 45 230.7 212.1
DF4 174 29 25 248.5 214.6
a
Simplified approach proposed by Kodur et al. [7].
306 V.K.R. Kodur, A. Agrawal / Engineering Structures 110 (2016) 293–306
expressed in this paper are those of the author and do not neces-
sarily reflect the views of the institution.
References
[1] Moetaz MEH, Elibiarif S, Ragabs M. Effect of fire on flexural behaviour. Constr
Build Mater 1996;10:147–50. http://dx.doi.org/10.1016/0950-0618(95)00041-
0.
[2] Kumar A, Kumar V. Behaviour of RCC beams after exposure to elevated
temperatures. Inst Eng India Civ Eng Div 2003;84:165–70.
[3] Dwaikat MB, Kodur VKR. Response of restrained concrete beams under design
fire exposure. J Struct Eng 2009;135:1408–17.
[4] Xu YY, Wu B, Jiang M, Huang X. Experimental study on residual flexural
behavior of reinforced concrete beams after exposure to fire. Adv Mater Res
2012;457–458:183–7. http://dx.doi.org/10.4028/www.scientific.net/
AMR.457-458.183.
[5] Choi EG, Shin YS, Kim HS. Structural damage evaluation of reinforced concrete
beams exposed to high temperatures. J Fire Prot Eng 2013;23:135–51. http://
dx.doi.org/10.1177/1042391512474666.
[6] Hsu JH, Lin CS. Effect of fire on the residual mechanical properties and
structural performance of reinforced concrete beams. J Fire Prot Eng
2008;18:245–74. http://dx.doi.org/10.1177/1042391507077171.
[7] Kodur VKR, Dwaikat MB, Fike RS. An approach for evaluating the residual
strength of fire-exposed RC beams. Mag Concr Res 2010;62:479–88. http://dx.
doi.org/10.1680/macr.2010.62.7.479.
[8] Bai LL, Wang ZQ. Residual bearing capacity of reinforced concrete member
after exposure to high temperature. Adv Mater Res 2011;368–373:577–81.
http://dx.doi.org/10.4028/www.scientific.net/AMR.368-373.577.
[9] ACI. ACI 318-08: building code requirements for reinforced concrete, vol.
552. Detroit (MI): ACI; 2008.
[10] ANSYS. ANSYS metaphysics. Canonsburg (PA): ANSYS Inc.; 2007.
[11] Ožbolt J, Bošnjak J, Periškić G, Sharma A. 3D numerical analysis of reinforced
concrete beams exposed to elevated temperature. Eng Struct 2014;58:166–74.
http://dx.doi.org/10.1016/j.engstruct.2012.11.030.
[12] ECS. EN 1994-1-2: design of composite steel and concrete structures. Part 1-2:
General rules-structural fire design, vol. 2. ECS, Brussels; 2005.
[13] Neves I, Rodrigues JPC, Loureiro ADP. Mechanical properties of reinforcing and
prestressing steels after heating. J Mater Civ Eng 1996;8:189–94.
[14] ABAQUS. Version 6.12 documentation. Providence (RI): Dassault Systemes
Simulia Corp.; 2012.
[15] ECS. EN 1991-1-2: actions on structures. Part 1-2: General actions – actions on
structures exposed to fire. ECS, Brussels; 2002.
[16] Drysdale DD, Schneider U, Babrauskas V, Graysin SJ. Reparability of fire
damaged structures CIB W14 report. Fire Saf J 1990;16:251–336.
[17] Lie TT. Structural fire protection. New York: American Society of Civil
Engineers; 1992.
[18] Van Coile R. Reliability-based decision making for concrete elements exposed
to fire. Ghent University; 2015.
[19] Nassif A. Postfire full stress–strain response of fire-damaged concrete. Fire
Mater 2006;30:323–32. http://dx.doi.org/10.1002/fam.911.
[20] Chowdhury EU, Bisby LA, Green MF, Kodur VKR. Residual behavior of fire-
exposed reinforced concrete beams prestrengthened in flexure with fiber-
reinforced polymer sheets. J Compos Constr 2008;12(1):61–8. http://dx.doi.
org/10.1061/(ASCE)1090-0268(2008)12:1(61).
[21] Tao Z, Wang X, Uy B. Stress–strain curves of structural and reinforcing steels
after exposure to elevated temperatures. J Mater Civ Eng 2012;25:1306–16.
http://dx.doi.org/10.1061/(ASCE)MT.1943-5533.0000676.
[22] BS 476-20. Fire tests on building materials and structures – Part 20: Method
for determination of the fire resistance of elements of construction (general
principles). BSi, UK; 1987.
[23] Kodur VKR, Dwaikat MB. A numerical model for predicting the fire resistance
of reinforced concrete beams. Cem Concr Compos 2008;30:431–43. http://dx.
doi.org/10.1016/j.cemconcomp.2007.08.012.
[24] Lu L, Yuan Y, Caspeele R, Taerwe L. Influencing factors for fire performance of
simply supported RC beams with implicit and explicit transient creep strain
material models. Fire Saf J 2015;73:29–36. http://dx.doi.org/10.1016/
j.firesaf.2015.02.009.
[25] Schneider U. Concrete at high temperatures—a general review. Fire Saf J
1988;13:55–68.
[26] Gao WY, Dai JG, Teng JG, Chen GM. Finite element modeling of reinforced
concrete beams exposed to fire. Eng Struct 2013;52:488–501. http://dx.doi.
org/10.1016/j.engstruct.2013.03.017.
[27] Wu B, Lu JZ. A numerical study of the behaviour of restrained RC beams at
elevated temperatures. Fire Saf J 2009;44:522–31. http://dx.doi.org/10.1016/
j.firesaf.2008.10.006.
[28] Rafi M, Nadjai A, Ali F. Finite element modeling of carbon fiber-reinforced
polymer reinforced concrete beams under elevated temperatures. ACI Struct J
2008;105:701–10.
[29] Crisfield M. Accelerated solution techniques and concrete cracking. Comput
Methods Appl Mech 1982;33:585–607.
[30] Schweizerhof K. Consistent concept for line search algorithms in combination
with arc-length constraints. Commun Numer Methods Eng 1993;9:773–84.
[31] ECS. EN 1992-1-2: deign of concrete structures. Part 1-2: General rules-
structural fire design. ECS, Brussels; 2004.
[32] Dwaikat MB, Kodur VKR. Hydrothermal model for predicting fire-induced
spalling in concrete structural systems. Fire Saf J 2009;44:425–34. http://dx.
doi.org/10.1016/j.firesaf.2008.09.001.
[33] Terro M. Numerical modeling of the behavior of concrete structures in fire. ACI
Struct J 1998;95:183–92.
[34] ECS. EN 1993-1-2: design of steel structures. Part 1-2: General rules-structural
fire design, vol. 2. ECS, Brussels; 2005.
[35] Elghazouli A, Izzuddin B. Analytical assessment of the structural performance
of composite floors subject to compartment fires. Fire Saf J 2001;36:769–93.
[36] Gernay T, Franssen J. Consideration of transient creep in the Eurocode
constitutive model for concrete in the fire situation. In: Sixth international
conference on structures in fire; 2010.
[37] ASTM. ASTM E119-07: standard methods of fire test of building construction
and materials. American Society of Testing Materials, West Conshohocken, PA;
2007.
[38] ISO 834-1. Fire resistance tests-elements of building construction. Part 1:
General requirement. Geneva, Switzerland; 1999.
[39] ACI 216.1-97. Standard method for determining fire resistance of concrete and
masonry construction assemblies. Farmington Hills, MI; 1997.