Simple Method for Predicting Temperatures in

Reinforced Concrete Beams Exposed to a Standard Fire

W.Y. Gao, J.G. Dai* and J.G. Teng

Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong, China

(Received: September 2 2013; Received revised form: November 21 2013; Accepted: March 10 2014)

Abstract: In performance-based fire safety design, the fire performance of a structure

needs to be accurately evaluated, which requires the accurate prediction of
temperatures in the structure. While a finite-element or a finite-difference analysis
may be carried out for this purpose, structural engineers generally prefer a simpler
method. This paper therefore presents a simple, design-oriented method for predicting
temperatures in RC beams under a standard fire exposure. Results from finite element
heat transfer analysis are first examined to identify the key parameters that determine
temperature distributions in RC beams. On the basis of this knowledge, a simple
method in the form of handy formulae and diagrams is derived from regression
analysis of finite element temperature data, with due consideration of the effects of
beam geometry and fire exposure duration. The accuracy of the proposed method is
demonstrated by comparing its predictions with temperature data from both finite
element analysis and laboratory tests. The proposed method is believed to be attractive
to practicing engineers for use in the fire resistance evaluation of RC beams exposed
to a standard fire because of its simplicity and accuracy.

Key words: simple method, temperature fields, RC beams, fire, heat transfer.

1. INTRODUCTION approach in the fire resistance design of reinforced

Fire represents one of the most severe hazards to which concrete (RC) structures; the latter provides a more
reinforced concrete (RC) structures may be subjected. rational and flexible design tool. In the performance-
Fire safety is therefore an important issue in the design based design approach, the fire resistance of the structure
of RC structures. Structural engineers have traditionally needs to be accurately evaluated, which requires the
adopted a prescriptive approach in fire safety design, in accurate prediction of temperature fields in the structure.
which certain deemed-to-satisfy requirements (e.g., A generic approach for temperature field analysis is to
minimum member dimensions and minimum concrete employ the finite-element (FE) or the finite-difference
cover depths for steel reinforcement) [e.g., BS 8110-2 (FD) method using an appropriate software package.
(1985); EN 1992-1-2 (2004); FIP/CEB (2004); ACI However, structural engineers often prefer a simpler
(2007); AS 3600 (2009)] are met to achieve the required approach as they may not have the expertise to deal with
fire resistance rating. However, such a prescriptive the complexity involved in such numerical computations
approach has become widely recognized as being too within the tight timeframe of a design task. This paper is
empirical to achieve rational and realistic fire safety concerned with the development of such a simple,
design (Kodur and Dwaikat 2011). design-oriented method given in explicit, closed-form
Recent years have seen a gradual transition from the expressions for temperature fields in RC beams for use in
prescriptive approach to the performance-based fire resistance design.

*Corresponding author. Email address: cejgdai@polyu.edu.hk; Fax: +852-23346389; Tel: +852-27666026.

Advances in Structural Engineering Vol. 17 No. 4 2014 573

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

Attempts have been made in the past to develop The reliability of the proposed method is verified by
simple design-oriented methods for the prediction of comparing its predictions with experimental data as well
temperature profiles in RC members subjected to one- as FE predictions.
dimensional heat transfer (Hertz 1981; Harmathy
1993; Kodur et al. 2013) or two-dimensional heat 2. FIRE RESISTANCE DESIGN
transfer (Wickstrom 1986; Desai 1995, 1998; Abbasi Current design codes and standards provide a choice of
2003; Abbasi and Hogg 2005; Kodur et al. 2013). three methods for determining the fire resistance of RC
Wickstrom (1986) presented a method for predicting beams: (a) tabulated data (tables or charts), (b)
temperatures in fire-exposed RC structures. For the simplified method, and (c) advanced method. The
ISO 834 standard fire curve and normal weight tabulated data offered by BS 8110-2 (1985), FIP/CEB
concrete, this method needs only two input (2004), EN 1992-1-2 (2004), ACI 216.1 (2007) and AS
parameters: the fire-exposure time and the concrete 3600 (2009) provide the fastest and most direct way of
depth. Wickstrom’s (1986) method was used by determining the minimum dimensions and concrete
Eamon and Jensen (2012) to determine the 500 oC cover depth of an RC beam for a required fire
isotherm and the temperature history of steel resistance rating. However, these tables were
reinforcement in the fire resistance analysis of pre- established on the basis of empirical relationships
stressed RC beams. Desai (1995, 1998) assumed that obtained from limited fire resistance test results
the isotherms of a rectangular RC beam are parallel to (Narayanan and Beeby 2005), and as a result, many
the exposed surfaces of the beam, and then proposed a significant factors are not properly considered in this
simple equation to determine these isotherms. Desai’s prescriptive approach. More importantly, it has been
(1995, 1998) method is governed by three factors: the shown that this approach does not always provide a
fire-exposure time, the width of beam cross-section, conservative fire resistance evaluation for RC beams
and the ratio between overall beam height and beam (Kodur and Dwaikat 2011).
width. More recently, Abbasi and Hogg (2005) By contrast, both the simplified and the advanced
developed a formula to predict temperatures of fiber- methods are based on numerical calculations, and
reinforced polymer (FRP) rebars in FRP-reinforced therefore they provide a more rigorous and flexible way
concrete beams exposed to fire. Under the ISO 834 for fire resistance design. The advanced method
standard fire curve, the rebar temperature was recommended by Eurocode 2 (EN 1992-1-2 2004)
assumed to be influenced only by the fire-exposure involves detailed thermal and mechanical analyses of
time and the concrete cover thickness. Rafi (2010) and RC beams exposed to fire (i.e., a heat transfer analysis
Nigro et al. (2012) proposed best-fit curves for the to determine temperatures within the cross-section,
temperature responses of FRP bars based upon fire followed by an accurate mechanical response analysis to
tests of FRP-reinforced concrete beams and slabs. determine the strength degradation of the RC beam).
Kodur et al. (2013) proposed a simple method for The advanced method requires detailed descriptions of
evaluating temperatures over cross-sections of RC constitutive laws for various materials at elevated
members exposed to a standard fire. The method was temperatures as well as complex computations. The
based on a modification of Wickstrom’s (1986) simplified method specified in EN 1992-1-2 (2004) and
equations and derived through a regression analysis of BS 8110-2 (1985) provides a reasonably easy and
temperature data generated from FE analysis. Despite accurate analytical approach for the evaluation of fire
the existence of these methods, there is still resistance: it involves a simple structural analysis based
considerable uncertainty with the prediction of on the “500 oC isotherm method” (Anderberg 1978) or
temperatures in RC beams exposed to a standard fire: the “Zone method” (Hertz 1981, 1985). The former
(a) the above-mentioned methods have all been method is based on the assumption that the part of
validated using only a limited test database, so their concrete with temperatures exceeding 500 oC has
ability to provide close predictions of other completely lost its strength, whilst the rest of the
experimental results is uncertain; and (b) the concrete retains its full initial strength. In the latter
differences between the predictions of these models method, the cross-section is divided into several zones
are unclear and may be substantial. of equal thickness and the reduced strength of each zone
This paper presents a simple yet accurate method for is evaluated. It should be noted that, the 500 oC isotherm
the prediction of temperature fields in RC beams under method was originally devised for RC sections
a standard fire exposure as an alternative to a complex subjected to pure bending, where failure is generally
FE or FD analysis. The method consists of explicit controlled by the yielding of steel tension reinforcement
formulae and diagrams for use in design calculations. (Anderberg 1978; fib 2007). Rigberth (2000) has

574 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

however demonstrated that the simplified method temperature fields of RC beams under a standard fire
generates slightly conservative estimates of fire exposure as summarized below.
resistance than the exact numerical model (i.e., the
advanced method). 3.1. Wickstrom’s Method
The above review of current design methodology For an RC beam made of normal weight concrete and
clearly indicates that, when fire resistance is assessed exposed to the ISO 834 standard fire, Wickstrom’s
using a calculation-based method (i.e., the simplified or (1986) method predicts the temperature rise ∆T(x, y) at
the advanced method), the first step is to determine a given point (x, y) in the concrete at the fire-exposure
temperatures in the beam cross-section. Therefore, an time t by the following equation:
explicit, design-oriented method for predicting
temperatures in RC beams is always needed by design ∆T(x, y) = [nw (nx +ny – 2nxny) + nxny]∆θf (1)
engineers. The availability of such a temperature where ∆θ f is temperature rise of the standard ISO 834
prediction method, especially in combination with the fire curve; nw is the ratio between the temperature rise of
simplified method for structural analysis, facilitates a the beam surface to that of the fire, which depends on
quick yet sufficiently accurate approach for the fire the fire-exposure time and is given by:
resistance evaluation of RC beams under a standard fire
exposure. nw = 1– 0.0616t – 0.88 (2)
While the simple temperature prediction method as In Eqn 2, t (in hours) is the fire exposure time; nx (or
presented in the present paper is obviously needed in the ny) is a function of fire-exposure time and the ratio
fire safety design of RC beams, it is actually even more between the thermal diffusivity of the RC beam (i.e., α)
important in the fire resistance evaluation of concrete and a reference value ac (i.e., αc = 417 × 10–9 m2 s–1):
beams reinforced with fiber-reinforced polymer (FRP)
bars or strengthened with FRP laminates (Stratford et al.
2009). FRP composites start to deteriorate quickly when α t 
nx = 0.18 ln  × 2  – 0.81 (3)
the glass transition temperature [around 130 oC for  αc x 
commonly used glass FRP bars (Clarke 1996); about 45
oC to about 80 oC for commonly used FRP laminates where x (in m) is the distance between the point under
(fib 2001; ACI 2008)] is being approached. As the glass consideration and the fire-exposed surface in the beam
transition temperatures of FRP composites are much width direction. A similar equation is used to calculate
lower than the highest temperatures they may ny by replacing x with y in the beam height direction. It
experience in a fire, the accuracy of temperature should be noted that Eqn 3 only holds if
prediction becomes much more important in assessing
x (or y) ≥ 2h – 3.6(0.0015t)0.5 (4)
the fire resistance of FRP bar-reinforced concrete
beams. For RC beams strengthened with FRP laminates, where h (in m) is the dimension of the RC beam in the
it is very attractive to be able to demonstrate the direction under consideration.
feasibility of ensuring the fire safety of the RC beam
without any fire insulation through an accurate 3.2. Desai’s Method
evaluation of its fire resistance. When such an approach Desai (1995, 1998) adopted a simple equation to predict
is used, the contribution of the bonded FRP the temperature profiles of rectangular RC beams based
reinforcement to the mechanical resistance of the RC on his own experimental results and those of some
beam in a fire can be ignored, and as a result, an others (e.g., Lin et al. 1988; Wade 1991). The
accurate evaluation of the fire resistance of the RC beam temperature T, (oC) of a contour at a distance x (in mm)
alone is needed. from the fire-exposed surfaces of the beam was assumed
to be influenced by the following factors:
3. EXISTING METHODS FOR (a) t, the fire-exposure time (in min);
TEMPERATURE PREDICTIONS (b) b, the width of beam cross-section (in mm); and
The temperature fields of an RC beam exposed to a (c) r, the ratio between beam height and beam
standard fire depends on several factors, such as the fire- width.
exposure time, the beam width, the aggregate type and Desai’s (1998) equation is as follows:
the moisture content of concrete. Many studies have
been carried out to determine the temperature fields of
RC beams. In total, four simple methods have been D – Ax + Bx 2 – Cx 3
T= (5)
found in the published literature for predicting the r 0.25

Advances in Structural Engineering Vol. 17 No. 4 2014 575

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

where By the combined use of Eqns 10-12, the rebar

temperature in the beam subjected to the standard ISO
  100 – t   fire can be obtained as follows:
A = 3.33 3 + 0.0033t +   (6)
  b  
   7.602   
T = θ – 767 exp  –  0.001·exp  ·t  (13)
   c ′ – 23.623   
B = 0.085 (7)
As this method was derived on the basis of fire test
C = 0.000221 (8) data after 30 minutes of fire exposure, the initial
temperature calculated by Eqn 13 (i.e., t = 0) is always
7  1 equal to –747 oC, which is unreasonable. Again for
D = 475r 12 –  b – 105t 3  comparison purposes in the paper, the temperature rise
  (9)
is taken as the predicted temperature minus the initial
temperature of a standard fire (i.e., 20 oC).
The applicability of Eqn 5 is limited to beams
satisfying the following two conditions: 100 mm < b < 3.4. Kodur et al.’s Method
300 mm and 1 < r < 3. If r ≤ 1.5, then it is assumed that More recently, Kodur et al. (2013) modified
r = 1.5 in the calculation. The initial temperature Wickstrom’s (1986) equations and proposed a simple
calculated by Eqn 5 varies from negative values to method for predicting temperatures over cross-sections
positive ones depending on the beam geometry and the of fire-exposed RC members. The method was derived
distance from the fire-exposed surfaces, which is an from temperature data generated from a FE parametric
undesirable feature. In the comparisons presented later study on RC members exposed to a standard fire. Based
in the paper, the temperature rise is taken as the on a trial–and-error process as well as regression
temperature predicted using Eqn 5 minus the initial analysis, the following formulae were developed for
temperature of a standard fire (i.e., 20 oC). calculating temperatures in RC members:
1-D heat transfer (for RC slabs):
3.3. Abbasi and Hogg’s Method
Abbasi and Hogg’s (2005) method was developed only T = c1 · nx – (at n) (14a)
for predicting the temperatures of FRP rebars in beams
subjected to the standard ISO 834 fire. In this method, 2-D heat transfer (for RC beams and columns):
the difference between the rebar temperature and the fire
temperature, after 30 minutes of fire exposure, is T = c2· [–1.481·(nx·ny)+0.985·(nx+ny)+0.017]·(atn) (14b)
assumed to be in the following exponential form: where at n is an approximation of the standard fire curve;
and c1 and c2 are coefficients whose values depend on
θ – T = A′ exp(–β · t) (10)
the aggregate type (i.e., siliceous or calcareous) and the
where θ and T are the fire temperature and the rebar strength (i.e., high-strength or normal-strength) of
temperature, respectively; A′ is an empirical constant concrete. nx is a function of the fire-exposure time and
and = 767; t is the fire exposure time (in min); β, being the distance from the point under consideration to the
θ – T  fire exposure surface and is given by:
the gradient of ln   versus t curves, was derived
 A′ 
 t 
from fire test data by regression analysis and has the nx = 0.155 ln  1.5  – 0.348 x – 0.371 (15)
following exponential form: x 

 b′  where t is the fire exposure time (in hours); and x is the

β = a ′ exp  (11) distance from the point under consideration to the fire
 c ′ + d ′ 
exposure surface in the member width direction. A
where c′ (in mm) is the concrete cover thickness; a′, b′ similar equation was also proposed for determining ny in
and d ′, being empirical constants determined from the member height direction. For comparison purposes
regression of test data, are as follows: in the present paper, the temperature rises of all points
of the member section are assumed to be the predictions
a′ = 0.001, b′ = 7.602, d′ = –23.623 (12) of the model minus 20 oC.

576 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

3.5. Deficiencies of Existing Simple Methods 4.2. FE Analysis

The above four simple methods were developed based In the FE analysis, the ISO 834 standard fire curve was
on two different approaches. The first approach involves adopted as the thermal load. Heat fluxes flow to the
a regression analysis of existing experimental data of bottom and two side surfaces of RC beams and
fire-exposed RC beams (Desai 1995, 1998; Abbasi and exchange heat with them through convection and
Hogg 2005). It should be mentioned that the existing radiation, whereas heat transmission occurs within
experimental data suffer from the following limitations: concrete through conduction. The time-dependent
(a) the number of fire tests has been limited; (b) the test temperature distribution in an RC beam is described by
specimens usually had relatively small dimensions Fourier’s differential equation for heat conduction
(Dwaikat and Kodur 2009); and (c) detailed records of (Purkiss 2007):
temperature responses were not obtained for some of the
tests (Desai 1995, 1998). Obvious, the empirical models ∂  ∂T  ∂  ∂T  ∂T
derived using this approach suffers from the weaknesses  k  +  k  + Q = ρc (16)
∂x  ∂x  ∂y  ∂y  ∂t
of the test data.
The second approach involves a regression analysis where k, ρ and c denote the temperature-dependent
of a large amount of numerical temperature data thermal conductivity, density and specific heat capacity,
generated using an FE or FD software package for respectively; Q is the rate of heat generated internally
thermal analysis (Wickstrom 1995; Kodur et al. 2013). per unit volume; and t is the time variable. For heat
However, in Wickstrom’s analysis, the effects of transfer analysis of an RC beam exposed to fire, internal
aggregate type (siliceous or calcareous) and the heat generation is inactive (i.e., Q = 0) (fib 2007). To
variation of thermal diffusivity of concrete were solve the above differential equation, the initial
ignored. Kodur et al.’s (2013) method provides only a temperature distribution and proper boundary
rough estimation of temperatures in fire-exposed RC conditions are required. The initial temperature
beams and show apparent discrepancies when its distribution in the RC beam at t = 0 is described by:
predictions are compared with the experimental data of
existing fire tests. T(x, y, t)t=0 =T0(x, y) (17)
The free boundary condition is applied to the
4. FE PARAMETRIC STUDY unexposed surface (i.e., the top surface) of the beam
4.1. General specimens. The heat fluxes exchange heat with the fire-
A detailed parametric study was carried out using the exposed surfaces of the RC beam via convection and
validated FE model (Gao et al. 2013) to generate radiation, which can be depicted by means of Robin’s
temperature data for a set of beam cross-sections. This boundary condition (Purkiss 2007):
parametric study has two purposes: (a) to investigate
the influence of each parameter on the sectional ∂T
temperature distribution; and (b) to provide sufficient –k = hc (T – T f ) + φε m ε f σ
∂n (18)
temperature data for regression analysis to derive a
simple predictive method. Table 1 provides a summary (T – T )4 – (T – T )4 
 z f z 
of the cross-sections used in the FE parametric study.
Details of the FE model for heat transfer analysis of RC where n represents the outward normal direction of the
beams under a standard fire exposure can be found in beam surface; hc is the convective heat transfer
Gao et al. (2013), but a brief summary is given below. coefficient; Tf denotes the fire temperature in degree

Table 1. Cross-sections examined in the FE parametric study

Purpose Parameter Cross-section (mm × mm)

Identification of significant parameters Depth-to-width ratio 200 × 200; 200 × 300; 200 × 400; 200 × 500; 200 × 600.
Beam width 200 × 200; 300 × 300; 400 × 400; 500 × 500; 600 × 600.
Formulation of the simple method Mid-width temperatures in wide beams 600 × 600
Beam width 200 × 600; 300 × 600; 400 × 600; 500 × 600; 600 × 600.
Temperatures of corner concrete 600 × 600

Advances in Structural Engineering Vol. 17 No. 4 2014 577

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

Celsius; Tz is the absolute zero temperature and is equal (1995) because in the updated version (EN 1992-1-2
to -273.15 oC; φ is a configuration parameter; εm and εf 2004) the lower and upper limits rather than specific
are the emissivity coefficients of the exposed surfaces values are provided for thermal conductivity of concrete
and of the fire, respectively; and σ is the Stephan- (Capua and Mari 2007). The effect of the moisture
Boltzmann constant. The values recommended by EN content of concrete is taken into account by adjusting
1991-1-2 (2002) were adopted to define the boundary the specific heat capacity to represent the latent heat of
conditions and the corresponding initial temperature: water evaporation. An early study conducted by
(a) Coefficient of convective heat transfer of the Harmathy (1965) showed that the presence of moisture
exposed surfaces: hc = 25 W/(m2⋅K); in building components is beneficial to their fire
(b) Configuration factor for radiation: φ = 1.0; resistance if it is not so excessive as to trigger the
(c) Emissivity of the exposed surfaces: εm = 0.8; spalling of concrete. Therefore, a slightly lower-than-
(d) Emissivity of the fire: εf = 1.0; normal moisture content of 1.5% by weight (i.e., u =
(e) Stephan-Boltzmann constant: σ = 5.67×10–8 1.5%) was assumed in the FE analysis to ensure
W/(m2.K4); conservative predictions (Hertz 1981; Biondini and
(f) Initial temperature: T0(x, y) = 20 oC. Nero 2011). According to EN 1992-1-2 (2004), a peak
The thermal conductivity of concrete made of value (i.e., cc,peak) of 1.470 kJ/kg was adopted for the
siliceous aggregate or calcareous aggregate shown in specific heat capacity of concrete to implicitly consider the
Figure 1(a) is determined according to ENV 1992-1-2 latent heat of evaporation component [see in Figure 1(b)].

3
Eurocode 2- siliceous aggregate 1.6
Eurocode 2- calcareous aggregate Eurocode 2
Thermal conductivity (W/m.K)

2.5 (siliceous and calcareous aggregates)

Specific heat (kJ/kg.K)

1.4
2
Cc,peak

1.5 1.2
u = 1.5%

1
1

0.5

0.8
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Temperature (°C) Temperature (°C)

(a) Thermal conductivity (b) Specific heat capacity

2400
Eurocode 2
(siliceous and calcareous aggregates)

2300
Density (kg/m3)

2200

2100

2000
0 200 400 600 800 1000 1200
Temperature (°C)

(c) Density

Figure 1. Thermal properties of concrete assumed in the calculation (moisture content = 1.5%)

578 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

The thermal properties of steel are not considered in the similar temperature evolutions at a given location. By
FE analysis since the effects of steel reinforcement are contrast, when the depth-to-width ratio is kept constant
usually negligible in heat transfer analysis (Rodrigo et (e.g., r = 1.0), the beam width has a significant effect on
al. 2010; Biondini and Nero 2011). In other words, the temperature evolutions with time at all locations [Figure
temperature of internal steel reinforcement is taken to be 2(b)], particularly when the depth of the location is large
equal to that of concrete at the same location. It should [e.g., 200 mm in Figure 2(b)]. For example, when the
be noted that the thermal properties of concrete at beam width increases from 200 mm to 600 mm,
elevated temperatures and the thermal boundary the temperature rise changes from 542 oC to 368 oC at
conditions of RC beams in heat transfer analysis were the 50 mm concrete depth and from 322 oC to 27 oC at
both determined according to the appropriate Eurocodes the 200 mm concrete depth after a two-hour fire
(EN 1991-1-2 2002; EN 1992-1-2 2004). These code exposure. Based on these numerical results, the beam
provisions have been extensively validated by previous width, rather than the depth-to-width ratio, needs to be
standard fire tests in the literature. considered as a key parameter in a simple predictive
model for temperatures.
4.3. Factors Governing Temperature Fields
Obviously, the temperature response at a given point in 5. FORMULATION OF THE SIMPLE METHOD
the RC beam is affected by the position of the point and 5.1. General
the exposure time. In the existing literature, Desai The proposed method for predicting temperature fields
(1995, 1998) reported that the temperature distribution in RC beams consists of two steps: (a) determination of
over the cross-section is also influenced by the beam the temperature distribution along the mid-width
width b, (in mm) and the beam height-to-width ratio r. vertical line of beam cross-section; and (b)
To clarify the effects of these two parameters, a determination of temperature profiles in corner regions
parametric FE study was conducted to examine the of beam cross-section with due consideration of two-
temperature distributions over the mid-width vertical dimensional heat transfer from both the bottom and the
line of the cross-section for five different beam widths side surfaces. In fact, the first step can be divided into
and height-to-width ratios (Table 1). These results are two sub-steps. When a wide beam is used, the prediction
presented in Figures 2(a) and 2(b), and suggest that, for of the mid-width temperatures can in fact be
a given beam width (here b = 200 mm), the temperature approximated as a one-dimensional heat transfer
rises at the five different depths (i.e., 5 mm, 25 mm, 50 problem. The one-dimensional temperature distribution
mm, 100 mm, and 200 mm) from the beam bottom are can then be modified for other beams with due
independent of the depth-to-width ratio (varying from consideration of the beam width effect. These two sub-
1.0 to 3.0 with the beam height increasing from 200 mm steps were clearly separated in the FE parametric study
to 600 mm); the different depth-to-width ratios lead to (Table 1).

1200 b = 200 mm
1200
100mm b = 300 mm
50 mm b = 400 mm
1000 1000 b = 500 mm
25 mm b = 600 mm
Temperature rise (°C)

Temperature rise (°C)

800 800 r = 1.0

5 mm 200 mm 5 mm
600 600
50 mm
400 400
r = 1.0 200 mm
r = 1.5
200 r = 2.0 200
r = 2.5
b = 200 mm r = 3.0
0 0
0 60 120 180 240 300 360 0 60 120 180 240 300 360
Fire exposure time (min) Fire exposure time (min)

(a) Effect of depth-to-width ratio (b) Effect of beam width

Figure 2. Effects of beam dimensions on mid-with temperature rises

Advances in Structural Engineering Vol. 17 No. 4 2014 579

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

5.2. Mid-Width Temperatures in Wide Beams (e.g., BS 8110-2 1985; EN 1992-1-2 2004; ACI 216.1
FE analyses not reported in this paper due to space 2007; AS 3600 2009). θd,120 can be expressed as an
limitation revealed that, when the beam width is larger exponential function based on least-square regression of
than 500 mm, the temperature distribution over the mid- the present FE numerical results:
width vertical line of the cross-section can be obtained
from a one-dimensional heat transfer analysis. That is, θd,120 = a0·exp(a1d )+a2 (20)
the effect of heat transfer from the two beam sides on where a0 = 872.5, a1 = –1.771 × 10-2, and a2 = 4.526 for
this temperature distribution is negligible. Therefore, a siliceous aggregate concrete [Figure 4(a)]; a0 = 895.7,
beam cross-section of 600 mm in width by 600 mm in a1 = –1.881 × 10–2, and a2 = 1.882 for calcareous
depth was employed in FE analysis to generate the time- aggregate concrete [Figure 4(b)].
dependent temperature data for the one-dimensional The value of kt can be found using Figure 5, in which
temperature distribution prediction. Figure 3 shows the the temperature rises at different depths with the fire
FE predictions for the evolving temperature rises at exposure time are normalized by the reference
different concrete depths over the mid-width vertical temperature θd,120. For depths larger than 200 mm, their
line of the cross-section. It is obvious that the curves are not provided because temperatures at these
temperatures are influenced by the concrete depth d and locations were found to be low (below 100 oC) after a
the fire exposure time t. The temperature rises more three-hour standard fire exposure (see Figure 3 for more
quickly for locations with a smaller depth as expected. details). Because degradations of concrete in both
Therefore, the temperature rise (∆T, in oC) at a point strength and stiffness below 100 oC are negligible, the
located at a distance d (in mm) from the bottom surface temperature rises at concrete depths beyond 200 mm are
of the beam is governed by the following two factors: assumed to be the same as those at 200 mm for
(a) t, the fire exposure time (in min); and simplicity. This simplification is believed to have little
(b) d, the distance from the fire-exposed surface (in influence on the fire resistance prediction of RC beams.
mm). Temperature rises at locations not directly represented
Thus, the temperature rise ∆T at a point of the mid- by the curves in Figure 5 can be obtained by linear
width line can be expressed as: interpolation. In addition, FE results (not given here to
save space) showed that the curves in Figure 5 are
∆T = θd,120ktkb (19)
applicable to both siliceous and calcareous aggregate
where d,120 is the reference temperature rise at t = 120 RC beams.
min for a certain depth d and a beam width of 600 mm; The values of kt can also be alternatively
kt accounts for the effect of fire exposure time; and kb represented using an algebraic equation which is more
accounts for the effect of beam width. For wide beams, convenient for use with spreadsheets in design
kb = 1. θd,120 is chosen as the reference temperature rise calculations. It was found by trial and error that the
because a 120-minute duration represents a typical fire- Morgan-Mercer-Flodin (MMF) function (Morgan et
resistance rating of RC beams in existing design codes al. 1975) is suitable for describing time-dependent

d = 5 mm
1200 d = 10 mm 350
d = 20 mm d = 150 mm 600
d = 40 mm d = 200 mm
1000 d = 60 mm 300 d = 250 mm
d = 80 mm d = 300 mm 600
d = 100 mm d = 350 mm
Temperature rise (°C)

Temperature rise (°C)

250
800 d = 400 mm d
d = 500 mm
200
600
150
400
600 100

200 50
600
d
0 0
0 60 120 180 240 300 360 0 60 120 180 240 300 360
Fire exposure time (min) Fire exposure time (min)

Figure 3. Mid-width temperature rises in a wide beam

580 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

1000 1000
FE prediction FE prediction
Best-fit curve Best-fit curve
800 800

Temperature rise (°C)

Temperature rise (°C)

600 600

400 400

200 200

0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Distance from fire exposed surface (mm) Distance from fire exposed surface (mm)
(a) Siliceous aggregate concrete (b) Calcareous aggregate concrete

Figure 4. θd.120 defined as a function of distance from fire exposed surface

9
d = 0mm
8
Temperature rise normalized by (θd,120 = kt)

d = 25 mm
d = 50 mm
7 d = 75 mm
d = 100 mm
6 1
d = 125 mm
d = 150 mm
5 0.8
d = 200 mm
4
0.6

3
0.4
2
0.2
1
0
0 0 30 60 90 120
0 60 120 180 240 300 360

Fire exposure time (min)

Figure 5. Factor accounting for exposure time

temperature rises at all concrete depths. The MMF where the coefficients mj(j=1,2,3,4,5) were determined
function was originally developed to describe the through least-square regression analysis of the FE data
nutritional response of higher organisms and has the presented in Figure 5, and their values are provided in
following form: Table 2. The curves predicted using the above two
formulae (i.e., Eqns 21 and 22) are presented in Figure 6
for different concrete depths. The close agreement
t1t2 + t3t t4
kt = (21) between the predicted curves and the FE data
t2 + t t4 demonstrates the accuracy of Eqns 21 and 22 in
predicting the values of kt value.
where ti(i=1,2,3,4) are functions of the concrete depth. A
regression process indicated that ti(i=1,2,3,4) can be 5.3. Effect of Beam Width
expressed as polynomials: The numerical results given in Section 4.3 clearly
demonstrate that the beam width is a key parameter
ti(i=1,2,3,4) = m1+m2·d+m3·d2+m4·d3+m5·d4 (22) which needs to be probably considered in the proposed

Advances in Structural Engineering Vol. 17 No. 4 2014 581

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

Table 2. Coefficients mj determined through least-square regression analysis

0 mm ≤ d ≤ 75 mm 75 mm ≤ d ≤ 200 mm
t1 m1 = –4.335/103, m2 = –4.355/103, m2 = 1.267/104, m1 =-8.402/101, m2 = 3.685/102, m2 = –5.527/104,
m4 = –1.725/106, m5 = 8.845/109. m4 = 3.149/106, m5 = –6.543/109.
t2 m1 = 4.849 × 101, m2 = –4.487/101, m2 = 2.405/101, m1 = –6.016 × 104, m2 = 2.231 × 103, m2 = –2.936 × 101,
m4 = –5.372/103, m5 = 7.153/105. m4 = 1.630/101, m5 = –3.049/104.
t3 m1 = 1.531, m2 = 2.543/102, m2 = 1.278/104, m1 = 1.730 × 102, m2 = –6.176, m2 = 7.701/102,
m4 = –4.308/107, m5 = 9.636/109. m4 = –3.836/104, m5 = 7.086/107.
t4 m1 = 9.285/101, m2 = –1.187/103, m2 = 1.715/104, m1 = 2.176, m2 = –3.988/102, m2 = 4.309/104,
m4 = –2.031/106, m5 = 1.022/108. m4 = –1.754/106, m5 = 2.482/109.

9 5
d = 5 mm
Temperature rise normalized by (θd,120 = kt)

Temperature rise normalized by (θd,120 = kb)

FE prediction
d = 25 mm
8 Eq. 19 d = 50 mm
d = 100 mm
7 4 d = 200 mm
200 mm
6

5 3
150 mm
4
100 mm
3
2
50 mm
2

1
0 mm 1
0
0 60 120 180 240 300 360 200 300 400 500 600
Fire exposure time (min) Beam width (mm)
Figure 6. Effect of exposure time: empirical equation versus FE Figure 7. Factor accounting for beam width effect
results

method. To investigate the beam width effect, FE where the parameters b0, b1 and b3 are functions of the
analyses were carried out to compare the mid-width concrete depth as given below:
temperature distributions for five different beam widths
(the beam depth is identical, i.e., 600 mm) (Table 1). b0 = –0.1307–1.450 × 10–2d + 5.809 × 10–5d2 (24a)
Figure 7 illustrates the effect of beam width (i.e., from
200 mm to 600 mm) on temperatures at different
b1 = –0.1712–2.035 × 10–2d – 3.421 × 10–5d2 (24b)
concrete depths (i.e., from 5 mm to 200 mm) over the
mid-width line of beam cross-section at t = 120 min.
The results are all normalized by θd,120 for a 600 mm b2 = 7.388 × 10–2 + 6.593 × 10–5d – 4.116 × 10–5d2 (24c)
wide beam. As the concrete depth increases, the effect
of beam width is clearly seen to become more To validate the expression proposed for kb, the curves
significant. In addition, the temperature at a given predicted by Eqn 23 are shown in Figure 7 as continuous
concrete depth decreases as the beam width increases. lines. In the same figure, the symbols represent the FE
Using a curve-fitting software package available freely temperature data, with different symbols corresponding
at <www.curveexpert.webhop.biz>, the exponential to different concrete depths. The close agreement
function given below was found to lead to the best between the predictions of Eqn 23 and the FE results
representation of the FE temperature data (Figure 7): confirms the accuracy of the expression for kb. It is
noteworthy that both Eqn 23 and Figure 7 are obtained
  from the parametric study for RC beams with widths
 b1  b  greater than or equal to 200 mm. Therefore, the
kb = exp  b0 + + b2 ln   (23) proposed method may not be applicable to RC beams
 b  200  
with narrow widths (i.e., b<200mm), for which the
 200 
cross-sectional temperature distributions may be

582 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

simultaneously influenced by the heat transfer from the  y

beam bottom as well as two beam sides (Kodur et al. functions m(y) and n   are introduced here into
2013). Nevertheless, such a limitation of the proposed  x
method is not a concern for practical design as RC Wickstrom’s (1986) original function to account for the
beams usually have a width larger than 200 mm. interactive effect of the two distances mentioned above
Combining Eqns 19, 20 and 23 as well as Figure 5 (or on heat transfer. Both functions were obtained through
Eqn 21), the temperature distribution over the mid- regression analysis of FE temperature data for 120
width vertical line of beam cross-section can be locations in the corner regions of the beam cross-section
obtained. Figure 8 compares the predictions of the (i.e., 600 mm × 600 mm). The detailed regression
method proposed herein (using Figure 5) with those process is not presented here as it was conducted using
obtained from FE analyses for a 600 mm × 600 mm routine least-square analysis. As a result, the
cross-section. The excellent agreement between the  y
expressions for m(y) and n   were obtained as follows:
results from the two sources demonstrates that the  x
explicit formulae have properly captured the effects of
fire exposure time t, the concrete depth d and the beam m(y) = 0.7591 + 4.372 × 10–2y – 1.714 × 10–5y2 (26a)
width b.

5.4. Temperatures of Corner Concrete 2

 y  y  y
The temperatures of points in corner regions depend on n   = 1.264 – 1.323   + 0.8808   (26b)
 x  x  x
their distances from both the bottom surface and the two
side surfaces of the beam. A logarithmic function used Eqn 26 was derived from regression analysis of FE
by Wickstrom (1986) is modified to predict the two- temperature data for RC beams made of siliceous
dimensional temperature response with due aggregate. It is also taken to be accurate for RC beams
consideration of the two distances: made of calcareous aggregate as demonstrated in the
following section. The influence of aggregate type on
   θ    y   temperature fields was already considered in deriving
∆T =   ln  x + 1 + 1·θ y ·m( y)·n    kt kb (25) the equation for the reference temperature (i.e., Eqn 20).
   θ y    x 

6. VERIFICATION OF THE PROPOSED
where x is the distance between the point under APPROACH
consideration and the beam bottom surface or the nearer Four well-reported fire tests of RC beams from
beam side surface, whichever is larger, while y is the independent sources are used to verify the accuracy of
smaller one; θx and θy are calculated according to Eqn the proposed method for predicting temperatures in RC
20 by replacing d with x and y, respectively. The beams under a standard fire exposure (Lin et al. 1981;

1200 1200
FE prediction FE prediction
Proposed method Proposed method
1000 1000
5 mm 5 mm
Temperature rise (°C)

Temperature rise (°C)

800 800
25 mm 25 mm

600 600
50 mm 50 mm

400 400
100 mm 200 mm 100 mm 200 mm
200 200

0 0
0 60 120 180 240 300 360 0 60 120 180 240 300 360
Fire exposure time (min) Fire exposure time (min)

(a) Siliceous aggregate concrete beam (b) Calcareous aggregate concrete beam

Figure 8. Comparison for mid-width temperature rises

Advances in Structural Engineering Vol. 17 No. 4 2014 583

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

Dotreppe and Franssen 1985; Wu et al. 1993; Dwaikat estimation leads to a more conservative fire resistance
and Kodur 2009). Figures 9 to 12 show comparisons design for RC beams.
between predictions from the proposed method and the The proposed method is further validated by
test results. Predictions from the four existing methods comparing its predictions with experimental results of
(i.e., Wickstrom 1986; Desai 1998; Abbasi and Hogg a simply-supported RC beam tested by Dwaikat and
2005; Kodur et al. 2013) are also presented in these Kodur (2009). The RC beam was made of normal
figures for comparison. In Wickstrom’s (1986) method, strength concrete with calcareous aggregate. It had a
the thermal diffusivity of concrete (i.e., αc) is an rectangular cross section of 406 mm × 254 mm and a
important parameter. Wickstrom (1986) suggested a concrete cover of 54 mm to the tension rebars. Figures
value of 417 × 10–9 m2.s–1 for αc, and this value was 10(a)-10(e) provide the temperature comparisons at the
adopted in making predictions using Wickstrom’s corner tension rebars (i.e., 54 mm deep) and at two
(1986) method for comparison in Figures 9(a), 10(a), concrete depths (i.e., 101 mm and 203 mm) along the
11(a) and 12(a). According to Eamon and Jensen mid-width vertical line. It can be seen that for all the
(2012), the actual thermal diffusivity may vary between locations under consideration, the proposed method
0.75 and 1.5 times this recommended value, depending provides accurate temperature predictions for the
on the aggregate type, density of concrete, temperature entire range of fire exposure [Figure 10(e)]. At the
and some other factors. location of corner rebars, the temperatures predicted
Figures 9a-9e show comparisons between the test by Wickstrom’s (1986) and Kodur et al.’s (2013)
data and the predictions at two concrete depths along the methods are much higher than the measured values,
mid-width vertical line of cross-section for the RC whereas for the other two locations along the mid-
beams tested by Wu et al. (1993). This series of fire tests width vertical line, both methods lead to negative
included three RC beams made of calcareous aggregate temperature predictions over a long period of time in
concrete. All three beams were designed to have an the early stage although they capture the trend of
identical section of 200 mm × 300 mm and tested under temperature response well afterwards [Figures 10(a)
the ISO 834 standard fire exposure. The test data in and 10(d)]. The predictions obtained with the other
Figures 9(a)-9(d) were averaged from thermocouple two methods (Desai 1998; Abbasi and Hogg 2005)
readings of these three RC beams. Since the methods of show relatively large discrepancies for all the three
Wickstrom (1986), Abbasi and Hogg (2005) and Kodur locations [Figures 10(b) and 10(c)]. One possible
et al. (2013) lead to negative temperature values during reason for these discrepancies is that these two
the early stage of fire exposure, which is obviously methods were only validated using very limited fire
incorrect, only the positive values are shown in these test data.
figures [Figures 9(a), 9(c) and 9(d)]. During the later Figures 11(a)-11(b) compare the predictions of rebar
stages of fire exposure, Wickstrom’s (1986) and Kodur temperatures from the proposed method with the test
et al.’s (2013) methods seem to capture closely the trend data of Lin et al. (1981) for a simply-supported RC
of temperature response [Figures 9(a) and 9(d)], beam made of calcareous aggregate under the ASTM
whereas predictions of the other two methods (Desai E119 standard fire exposure. The beam had a section of
1998; Abbasi and Hogg 2005) deviate substantially 305 mm × 355 mm, with the bottom and the side
from the test results [Figures 9(b) and 9(c)]. The concrete covers of corner tension bars being 25 mm and
temperatures predicted by the proposed method agree 38 mm respectively. The methods of Wickstrom (1986),
closely with the test data although there is some over- Abbasi and Hogg (2005) and Kodur et al. (2013) all lead
estimation during later stages for larger concrete depths to substantial over-estimations, particularly during the
(i.e., above 100 mm from the bottom of the beam). This later stages of fire exposure [Figure 10(a)]. Desai’s
over-estimation may be attributed to the higher moisture (1998) method gives close predictions of the test data,
content of the test beams, which is about 5.5% and much probably because Desai’s (1998) method was based on
larger than the value assumed in the present analysis Lin et al.’s (1988) test results; the concrete properties
(i.e., 1.5%). In the thin layer of concrete adjacent to the and the beam dimensions of Lin et al.’s (1988) tests are
fire-exposed surfaces, the moisture can be released into similar to those of the specimen from Lin et al. (1981)
the air during the heating process, but the moisture under consideration here. The proposed method is by far
further away from the surface moves towards the inner the most accurate method among all the methods
colder region of concrete due to pore pressure-induced [Figure 10(b)], indicating its accuracy for both the
flow. As a result, the effect of moisture on temperature ASTM E119 and the ISO-834 standard fire exposures.
changes is more significant for the inner concrete. It The two standard fire scenarios are in fact similar, and
should be noted that the above-mentioned over- indeed for a long fire exposure time, their difference in

584 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

ISO 834 ISO 834

1200 Test data, d = 25 mm 1200 Test data, d = 25 mm
Test data, d = 100 mm Test data, d = 100 mm
Prediction, d = 25 mm Prediction, d = 25 mm
1000 Prediction, d = 100 mm 1000 Prediction, d = 100 mm
Temperature rise (°C)

Temperature rise (°C)

800 800

600 600

400 400

200 200

0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Fire exposure time (min) Fire exposure time (min)

(a) Wickstrom (1986) (b) Desai (1998)

ISO 834 ISO 834

1200 Test data, d = 25 mm 1200 Test data, d = 25 mm
Test data, d = 100 mm Test data, d = 100 mm
Prediction, d = 25 mm Prediction, d = 25 mm
1000 Prediction, d = 100 mm 1000 Prediction, d = 100 mm
Temperature rise (°C)
Temperature rise (°C)

800 800

600 600

400 400

200 200

0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Fire exposure time (min) Fire exposure time (min)

(c) Abbasi and Hogg (2005) (d) Kodur et al. (2013)

1200

1000
Temperature rise (°C)

ISO 834
800 Test data, d = 25mm
Test data, d = 100mm
600 d = 25mm (Based on Fig. 5)
d = 25mm (Based on Eq. 21)
d = 100mm (Based on Fig. 5)
400 d = 100mm (Based on Eq. 21)

200

0
0 20 40 60 80 100 120
Fire exposure time (min)

(e) Proposed method

Figure 9. Comparisons between predictions from simple methods and test data for Wu et al.’s (1993) beams

Advances in Structural Engineering Vol. 17 No. 4 2014 585

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

ASTM E119 ASTM E119

Test data, d = 54 mm Test data, d = 54 mm
Test data, d = 101 mm Test data, d = 101 mm
1200 Test data, d = 203 mm 1200 Test data, d = 203 mm
Prediction, d = 54 mm Prediction, d = 54 mm
Prediction, d = 101 mm Prediction, d = 101 mm
1000 Prediction, d = 203 mm Prediction, d = 203 mm
1000
Temperature rise (°C)

Temperature rise (°C)

800 800

600 600

400 400

200 200

0 0
0 30 60 90 120 150 180 0 30 60 90 120 150 180
Fire exposure time (min) Fire exposure time (min)

(a) Wickstrom (1986) (b) Desai (1998)

ASTM E119 ASTM E119

Test data, d = 54 mm Test data, d = 54 mm
Test data, d = 101 mm Test data, d = 101 mm
1200 Test data, d = 203 mm
1200 Test data, d = 203 mm
Prediction, d = 54 mm Prediction, d = 54 mm
Prediction, d = 101 mm Prediction, d = 101 mm
1000 Prediction, d = 203 mm 1000 Prediction, d = 203 mm
Temperature rise (°C)

Temperature rise (°C)

800 800

600 600

400 400

200 200

0 0
0 30 60 90 120 150 180 0 30 60 90 120 150 180
Fire exposure time (min) Fire exposure time (min)

(c) Abbasi and Hogg (2005) (d) Kodur et al. (2013)

1200

1000
ASTM E119
Temperature rise (°C)

800 Test data, d = 54 mm

Test data, d = 101 mm
Test data, d = 203 mm
600 d = 54 mm (Based on Fig. 5)
d = 54 mm (Based on Eq. 21)
d = 101 mm (Based on Fig. 5)
400 d = 101 mm (Based on Eq. 21)
d = 203 mm (Based on Fig. 5)
d = 203 mm (Based on Eq. 21)
200

0
0 30 60 90 120 150 180
Fire exposure time (min)

(e) Proposed method

Figure 10. Comparisons between predictions from simple methods and test data for Dwaikat and Kodur’s (2009) beam

586 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

severity of exposure is negligible (Harmathy et al. underestimations or overestimations of the test results to
1987). different extents [Figure 11(a)].
The last test specimen was conducted by Dotreppe
and Franssen (1985) who reported only the fire test of 7. CONCLUSIONS
one simply-supported RC beam made of siliceous Based on numerical results obtained using a validated
aggregate. The beam had a width of 200 mm and a depth finite element model, this paper has presented an
of 600 mm. The bottom concrete cover of steel rebars accurate yet simple method for predicting temperatures
was 40 mm. Comparisons between test and predicted in RC beams exposed to a standard fire. In formulating
results for the temperature responses of the central the simple method, the FE model was used to
tension rebar are given in Figures 11(a)-11(b). These investigate the effects of various parameters to identify
comparisons indicate again that the proposed method the key parameters of influence and to generate data for
provides consistent and satisfactory predictions of the regression analysis. In this simple method, the
test results throughout the entire fire exposure period, temperature at any point of a beam cross-section is
while the four existing methods lead to given as a function of its coordinates, the beam width

1200 1200
ASTM E119
Test data
1000 1000 Prediction (Based on Fig. 5)
Prediction (Based on Eq. 21)
Temperature rise (°C)

Temperature rise (°C)

800 800

600 600

400 400
ASTM E119
Test data
200 Prediction (Wickstrom 1986) 200
Prediction (Desai 1998)
Prediction (Abbasi and Hogg 2005)
Prediction (Kodur et al. 2013)
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Fire exposure time (min) Fire exposure time (min)

(a) Predictions by existing simple methods (b) Predictions by the proposed method

Figure 11. Comparisons between predictions from simple methods and test data for Lin et al.’s (1981) beam

1200 ISO 834

Test data 1200
ISO 834
Prediction (Wickstrom 1986)
1000 1000 Test data
Prediction (Desai 1998)
Prediction (Based on Fig. 5)
Temperature rise (C°)

Prediction (Abbasi and Hogg 2005)

Temperature rise (C°)

Prediction (Based on Eq. 21)

800 Prediction (Kodur et al. 2013) 800

600 600

400 400

200 200

0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Fire exposure time (min) Fire exposure time (min)

(a) Predictions by existing simple methods (b) Predictions by the proposed method

Figure 12. Comparisons between predictions from simple methods and test data for Dotreppe and Franssen’s (1985) beam

Advances in Structural Engineering Vol. 17 No. 4 2014 587

 Simple Method for Predicting Temperatures in Reinforced Concrete Beams Exposed to a Standard Fire

and the fire exposure time. Predictions from the Clarke, J.L. (1996). Structural Design of Polymer Composites –
proposed simple method were compared with numerical Eurocomp Design Code and Handbook, E & FN Spon. Ltd.,
predictions from FE analysis as well as test data. These London, UK.
comparisons showed that the proposed method provides Desai, S.B. (1995). Shear Resistance at Normal and High
much more accurate predictions of the test results than Temperatures of Reinforced Concrete Members with Links and
the four existing methods of the same kind. Due to its Central Bars, PhD Thesis[ZJ2], City University London,
explicit form, the proposed method can be easily London, UK.
implemented into a spreadsheet for direct use in Desai, S.B. (1998). “Design of reinforced concrete beams under fire
structural fire design. exposure conditions”, Magazine of Concrete Research, Vol. 50,
No. 1, pp. 75–83.
ACKNOWLEDGEMENTS Dotreppe, J.C. and Franssen, J.M. (1985). “The use of numerical
The authors are grateful for the financial support models for the fire analysis of reinforced concrete and
received from the National Basic Research Program of composite structures”, Engineering Analysis, Vol. 2, No. 2, pp.
China (i.e., the 973 Program) (Project No.: 67–74.
2012CB026201) and from the Construction Industry Dwaikat, M.M.S. and Kodur, V.K.R. (2009). “Responses of
Institute (Hong Kong)/PolyU Innovation Fund (Project restrained concrete beams under design fire exposure”, Journal of
No: 5-ZJE8). They are also grateful for a PhD Structural Engineering, ASCE, Vol. 135, No. 11, pp. 1408-1417.
studentship awarded to the first author by The Hong Eamon, C.D. and Jensen, E. (2012). “Reliability analysis of
Kong Polytechnic University. prestressed concrete beams exposed to fire”, Engineering
Structures, Vol. 43, pp. 69–77.
REFERENCES EN 1991-1-2 (2002). Eurocode 1: Actions on Structures —- Part 1-
Abbasi, A. (2003). Behavior of GFRP Rebars Reinforced Concrete 2: General Actions —- Actions on Structures Exposed to Fire,
Elements under Elevated Temperature and Fire, PhD Thesis, British Standards Institution, London, UK.
Queen Mary University of London, London, UK. ENV 1992-1-2 (1995). Eurocode 2: Design of Concrete Structures
Abbasi, A. and Hogg, P.J. (2005). “A model for predicting the —- Part 1-2: General Rules —- Structural Fire Design, British
properties of the constitutive of a glass fiber rebar reinforced Standards Institution, London, UK.
concrete beam at elevated temperatures simulating a fire test”, EN 1992-1-2 (2004). Eurocode 2: Design of Concrete Structures —
Composites Part B: Engineering, Vol. 36, No. 5, pp. 384–393. - Part 1-2: General Rules —- Structural Fire Design, British
American Concrete Institute (ACI) (2007). Code Requirements for Standards Institution, London, UK.
Determining Fire Resistance of Concrete and Masonry Federation International du Beton (fib) (2001). Externally Bonded
Construction Assemblies, ACI 216.1, American Concrete FRP Reinforcement for RC Structures, fib Bulletin 14, fib Task
Institute, Michigan, America. Group 9.3, fib, Lausanne, Switzerland.
American Concrete Institute (ACI) (2008). Guide for the Design and Federation International du Beton (fib) (2007). Fire Design of
Construction of Externally Bonded FRP Systems for Concrete Structures —- Materials, Structures and Modelling, fib
Strengthening Concrete Structures (ACI 440.2R-08), Farmington Bulletin 38, fib, Lausanne, Switzerland.
Hills, Michigan, America. Federation International de la Precontrainte-Comite Euro-International
Anderberg, Y. (1978). “Analytical fire engineering design of du Beton (FIP/CEB) (2004). FIP/CEB Report on Methods of
reinforced concrete structures based on real fire characteristics”, Assessment of the Fire Resistance of Concrete Structural Members,
Proceedings of the Eighth Congress of the Federation Cement and Concrete Association, Slough, UK.
Internationale de la Precontrainte, Concrete Society, London, Gao, W.Y., Dai, J.G. and Teng, J.G. (2013). “Finite element
UK, pp. 112–123. modeling of reinforced concrete beams exposed to fire”,
AS 3600 (2009). Concrete Structures, Standards Association of Engineering Structures, Vol. 52, pp. 488–501.
Australia, Sydney, Australia. Harmathy, T.Z. (1965). “Effect of moisture on the fire endurance of
Biondini, F.M. and Nero, A. (2011). “Cellular finite beam element building elements”, Moisture in Materials in Relation to Fire
for nonlinear analysis of concrete structures under fire”, Tests, ASTM STP 385, American Society for Testing and
Journal of Structural Engineering, ASCE, Vol. 137, No. 5, pp. Materials, USA, pp.74–95.
543–558. Harmathy, T.Z., Sultan, M.A. and MacLaurin, J.W. (1987).
BS 8110-2 (1985). Structural Use of Concrete —- Part 2: Code of “Comparison of severity of exposure in ASTM E119 and ISO 834
Practice for Special Circumstances, British Standards Institute, fire resistance tests”, Journal of Testing and Evaluation, Vol. 15,
Milton Keynes, UK. No. 6, pp. 371–375.
Capua, D.D. and Mari, A.R. (2007). “Nonlinear analysis of Harmathy, T.Z. (1993). Concrete Design and Construction Series:
reinforced concrete cross-sections exposed to fire”, Fire Safety Fire Safety Design and Concrete, Longman Scientific and
Journal, Vol. 42, No. 2, pp. 139-149. Technical, London, UK.

588 Advances in Structural Engineering Vol. 17 No. 4 2014

W.Y. Gao, J.G. Dai and J.G. Teng

Hertz, K. (1981). Simple Temperature Calculations of Fire Nigro, E., Cefarelli, G., Bilotta, A., Manfredi, G. and Cosenza, E.
Exposed Concrete Constructions (Report No. 159), Institute of (2012). “Behavior of FRP reinforced concrete slabs in case of
Building Design, Technical University of Denmark, Lyngby, fire: theoretical models and experimental tests”, Advances in
Denmark. Structural Engineering, Vol. 15, No. 4, pp. 637–652.
Hertz, K. (1985). Analyses of Prestressed Concrete Structures Purkiss, J.A. (2007). Fire Safety Engineering Design of Structures,
Exposed to Fire (Report No. 174), Institute of Building Design, 2nd Edition, Oxford, Butterworth-Heineman, Cambridge, UK.
Technical University of Denmark, Lyngby, Denmark. Rafi, M.M. (2010). Fire Performance of FRP Reinforced Concrete
Kodur, V.K.R. and Dwaikat, M.B. (2011). “Design equation for Beams: Experimental and Theoretical Studies, LAP Lambert
predicting fire resistance of reinforced concrete beams”, Academic Publishing, Saarbrücken, Germany.
Engineering Structures, Vol. 33, No. 2, pp. 602–614. Rigberth, J. (2000). Simplified Design of Fire Exposed Concrete
Kodur, V.K.R., Yu, B.L. and Dwaikat, M.M.S. (2013). “A simplified Beams and Columns, Report No.5063, Department of Fire Safety
approach for predicting temperature in reinforced concrete Engineering, Lund University, Sweden.
members exposed to standard fire”, Fire Safety Journal, Vol. 56, Rodrigo, B.C., Jaoao, B.M. and Ricardo, H.F. (2010). “Interaction
pp. 39–51. diagram for reinforced concrete sections subjected to fire”,
Lin, T.D., Gustaferro, A.H. and Abrams, M.S. (1981). Fire Engineering Structures, Vol. 32, No. 9, pp. 2832–2838.
Endurance of Continuous Reinforced Concrete Beams (RD Stratford, T.J., Gillie, M., Chen, J.F. and Usmani, A.S. (2009).
072.01B), Research and Development Bulletin, Portland Cement “Bonded fibre reinforced polymer strengthening in a real fire”,
Association, Skokie, America. Advances in Structural Engineering, Vol. 12, No. 6, pp. 867–877.
Lin, T.D., Ellingwood, B. and Piet, O. (1988). Flexural and Shear Wade, C. (1991). Method for Fire Engineering Design of Structural
Behavior of Reinforced Concrete Beams during Fire Tests (NBS- Concrete Beams and Floor Systems, BRANZ Technical
GCR-87-536), U.S. Department of Commerce, National Institute Recommendation No. 8, Building Research Association of New
for Science and Technology, Center for Fire Research, Zealand, Wellington, New Zealand.
Gaithersburg, America. Wickstrom, U. (1986). “A very simple method for estimating
Morgan, P.H., Mercer, L.P. and Flodin, N.W. (1975). “General temperatures in fire exposed concrete structures”, New
model for nutritional responses of higher order mechanisms”, Technology to Reduce Fire Losses and Costs, ed. Grayson, S.J.
Proceedings of the National Academy of Sciences of the United and Smith, D.A., Elsevier Applied Science, London, UK,
States of America, Vol. 72, No. 11, pp. 4327–4331. pp.186–194.
Narayanan, R.S. and Beeby, A. (2005). Designers’ Guide to EN Wu, H.J., Lie, T.T. and Hu, J.Y. (1993). Fire Resistance of Beam-
1992-1-1 and EN 1992-1-2. Eurocode 2: Design of Concrete Slab Specimens —- Experimental Studies, Internal Report No.
Structures. General Rules and Rules for Buildings and Structural 641, Institute for Research in Construction, National Research
Fire Design, Thomas Telford Publishing, London, UK. Council Canada, Ottawa, Canada.

Advances in Structural Engineering Vol. 17 No. 4 2014 589