Fire Safety Journal 61 (2013) 65–71

Contents lists available at ScienceDirect

Fire Safety Journal

journal homepage: www.elsevier.com/locate/firesaf

Uncertainties in steel temperatures during fire

Dilip K. Banerjee n
Materials and Structural Systems Division, Engineering Laboratory, National Institute of Standards and Technology, Gaithersburg, MD 20899, USA

art ic l e i nf o a b s t r a c t

Article history: In order to determine the fire resistance of steel members, steel temperatures must be estimated with a
Received 1 November 2012 high confidence. There can be considerable uncertainty in temperatures of both protected and
Received in revised form unprotected steels during fire exposure. This is due to uncertainty in the thermal boundary conditions
9 May 2013
and thermophysical properties. In this study, uncertainties in both unprotected and protected steel
Accepted 4 August 2013
Available online 2 September 2013
temperatures are estimated with the use of a Monte Carlo method in conjunction with a “Lumped Heat
Capacity” approach for estimating steel temperatures. Computed data are compared with experimental
Keywords: measurements obtained during Cardington fire tests (bare steel) and National Institute of Standards and
Uncertainty Technology (NIST) World Trade Center (WTC) tests (protected). Reasonable agreement was achieved.
Heat transfer model
Published by Elsevier Ltd.
Monte Carlo method
Fire
Steel
Fire resistance tests

1. Introduction affect the overall uncertainties in steel temperatures. Density tests

Both spatial and temporal variations of temperatures need to
be accounted for when evaluating the fire resistance of steels.
The ability to predict with high confidence time-varying tempera-
ture profiles in structural members is necessary for a robust
performance-based approach to the fire resistance design of
structures. Therefore, uncertainties in steel temperatures must
be accurately estimated. The focus of this study is to demonstrate a
simple approach for estimating uncertainty in the predicted
thermal response of both unprotected and protected steels during
a fire event.
Temperature profiles in a steel section during fire exposure
depend upon the temperature-dependent thermophysical proper-
ties of steel, the thermophysical properties of fireproofing (spray
applied fire resistive material, SFRM) for protected steel and the
convective and radiative heat transfer parameters associated with
fire. However, there can be considerable uncertainty in estimates
typically used for these parameters. For example, although SFRM
thickness measurements are reported according to the ASTM E
605 standard, individual thickness measurements (as required by
the standard) can vary, while an average measurement is reported.
In most cases, the SFRM thickness will be greater than the
stipulated value as overspray is normally not penalized. Uncer-
tainties in SFRM thickness can result in increased uncertainty in
steel temperatures [1]. The variability in SFRM density can also
n
Tel.: þ 1 301 975 3538; fax: þ1 301 869 6275.
E-mail addresses: Dilip.Banerjee@nist.gov, banerjee_dilip@yahoo.com
are performed following the ASTM E605 standard. Ref. [4] showed
air dry density variability in the range of 10–20% for typical floor
truss systems. The steel temperatures are influenced because of
the effect of density variability on the volumetric heat capacity of
SFRM. A sensitivity study can be conducted to determine which of
these parameters (thermophysical and heat transfer) most sig-
nificantly influence the thermal response of the steel. Influential
parameters can be used to quantify the uncertainty in the
predicted temperatures.
Uncertainties can be broadly classified into two basic types:
aleatoric (random) and epistemic (systematic). Aleatoric uncer-
tainties are due to inherent randomness and cannot be removed
by further analysis or testing. For example, fuel load density
(MJ/m2) can be classified as inherently random. On the other
hand, epistemic (also known as knowledge-based) uncertainties
can be reduced by using improved models or algorithms. Estima-
tion of both aleatoric and epistemic uncertainties can provide a
confidence interval for time-varying estimates of structural tem-
peratures during a fire event.
Uncertainties in measured temperatures of a steel section
during fire exposure can be attributed to (a) inherent measure-
ment uncertainty associated with measuring devices such as
thermocouples, (b) uncertainties associated with thermophysical
properties of steel due to variability associated with steel compo-
sition (e.g., steel web diagonals used in trusses can be sourced
from different vendors or from a vendor using various heats for
producing steels), (c) statistical randomness associated with true
gas temperatures in fire in the vicinity of a measuring device,
(d) uncertainties in heat transfer parameters such as emissivity

0379-7112/$ - see front matter Published by Elsevier Ltd.

http://dx.doi.org/10.1016/j.firesaf.2013.08.012
66 D.K. Banerjee / Fire Safety Journal 61 (2013) 65–71

of a steel surface and convective heat transfer coefficients 56 57 58

(for example, emissivity of protected steel will vary at a location
x x x
if the quality of fireproofing degrades due to unexpected or abrupt x 55
variation in gas temperatures). For steel sections, measured x 54
temperatures can be reported as mean temperatures with their 144 145
uncertainty bounds. For example, one can report measured tem- x x x
53
peratures at top flange, bottom flange, and web as mean tempera-
tures along with uncertainty bounds for each.
Simplified analytical models are often used for modeling heat
transfer in structural members in fire. For example, the “Lumped 52
x
Heat Capacity Method” is widely used for modeling heat transfer
in steel members in fire [2]. The lumped heat capacity method is
appropriate for steel because of its high thermal conductivity. It is
useful to develop a simple and practical approach for computing x 51
uncertainties in structural temperatures when such simplified
analytical approaches are used. This paper discusses a simplified
approach, e.g., “Lumped Heat Capacity Method”, for modeling
temperatures for both unprotected and protected steel and the 142 143
x x x x x x x
use of the Monte Carlo method for computing uncertainties in
steel temperatures. Computed uncertainties are compared with 46 47 48 49 50
results of experimental measurements for validation of models for
Fig. 1. Schematic representation of the location of the thermocouples used to
prediction of uncertainties in unprotected and protected steel collect steel temperatures during the Cardington Test 1 [3].
temperatures during fire exposure. The following test data were
used for validating computational approach for steel temperatures
and prediction of uncertainties:

1. Cardington Test 1 for unprotected steel [3].

2. NIST fire resistance Test 4 for protected steel [4].

In the following section, a brief description is provided first

about experimental measurements for both unprotected and
protected steels. Then, the “Lumped Heat capacity” approach is
described for computation of steel temperatures. Finally, com-
puted steel temperatures and uncertainties in steel temperatures
are compared with experimental measurements for both unpro-
tected and protected steels. Fig. 2. Floor system of the WTC towers [6].
The validation of the computational approach for determining
uncertainties in steel temperatures in fire will allow for reasonable
prediction of uncertainties in temperatures when similar steel Mean web temperatures and standard deviations were com-
members are exposed to an unknown fire as long as uncertainties puted using the five sets of web temperature data collected along
in key parameters such as gas temperatures in fire are known. the length of the beam. Furnace temperatures (e.g., thermocouples
142 through 145 at this section in Fig. 1) were used to yield mean
fire temperatures and standard deviations as functions of time.
2. Experimental data These mean furnace temperatures and standard deviations were
used in computing uncertainties in steel web temperatures as
2.1. Unprotected steel explained later in the text.

Steel temperatures were taken from Cardington Test 1 [3].

Cardington Test 1 is a restrained beam test in which a 305 
165  40 UB beam (British Universal Beam) was heated with a gas
fired furnace over the middle 8 m of its 9 m length. The beam was
instrumented with a number of thermocouples at the top flange,
web, and bottom flange (see Fig. 1). Five sets of thermocouples
were positioned along the length of the beam for this test.
Temperature measurement data collected at the beam web were
used in this study. This is because the heating of the web can be
considered to be uniform and the influence of the floor slab
(positioned above the beam) on the web temperatures was
presumed to be minimal. This is necessary since measurement
data are compared with those computed using a simplified
approach that requires assuming no internal temperature gradient
across steel section. This is described later in the text in more
detail.
Fig. 1 shows the positions of the thermocouples at a beam cross
section. Note that thermocouples 51 through 55 represent web
temperatures at this section.
2.2. Protected steel
Protected steel temperature measurement data were taken
from the NIST fire resistance Test 4 [4]. As part of its investigation
into the World Trade Center (WTC) disaster, NIST conducted four
standard fire tests of composite floor systems. Two full-scale tests
(Test 1 and 2; span 35 ft. (10.7 m)) were conducted at the Under-
writers Laboratories (UL) fire testing facility at Toronto, Canada
and the other two (reduced scale; Test 3 and 4; span 17 ft. (5.2 m))
were conducted at Northbrook, IL. The UL test furnace was heated
by 80 individual floor mounted burners following the American
Society for Testing and Materials (ASTM) E119 standard time-
temperature curve [5], and furnace temperatures were monitored
at 16 locations in the furnace [4]. Time-temperature data were
collected at specific locations along the truss near the top chord, at
mid height of the web, and at the bottom chord.
The floor system used in the test consisted of a lightweight
concrete floor slab supported by steel trusses. Fig. 2 shows a picture
 D.K. Banerjee / Fire Safety Journal 61 (2013) 65–71
Locations of Temperature
N
Measurement on a Main
Truss
0.7 m Bridging Trusses
West Main
Truss
East Main
Truss
2.0 m
5.4 m
2.0 m
Approximately 0.8m x 0.4
m of concrete slab was
0.7 m dislodged due to spalling
Fig. 3. Thermocouple locations in the main truss in the UL test [4].
of a typical floor system tested in UL furnace. The temperature
measurement locations on the steel truss are shown in Fig. 3.
The main composite trusses, which were used in pairs, had a
nominal clear span of 17 ft. (5.2 m). The steel trusses were
fabricated using double-angles for the top and bottom chords,
and round bars for the webs. The web members protruded above
the top chord in the form of a “knuckle” which was embedded in
the concrete slab to develop composite action. The floor system
also included bridging trusses perpendicular to the main trusses
(Fig. 3). Fireproofing (SFRM) was applied directly on steel trusses
to provide passive fire protection. The applied thickness of fire-
proofing was 0.5 in. (12.7 mm) for this test. The diameter of the
web members was 0.92 in. (23.4 mm).
Steel temperatures were measured at 5 locations along the
length of the two main trusses. At each location along the main
truss, eight thermocouples were used to record temperatures. Two
thermocouples were positioned on the upper chords, four thermo-
couples were at the bottom chords, and two were at the mid-
height of the webs (Fig. 3).
Twenty temperature measurements at mid height of the web
along the two main trusses (North and South) were used to obtain
mean web temperatures and standard deviations. The standard
deviations in web temperatures (obtained as a function of time)
were taken as experimentally observed uncertainties in steel web
temperatures. Sixteen furnace thermocouple readings were used
to obtain furnace mean temperatures and uncertainties in furnace
temperatures.
3. Simplified approach for computation of steel temperatures
The “Lumped Mass” or “Lumped Heat Capacity” method is valid
when there is no temperature gradient in a member (e.g., member
temperatures are uniform). This is an idealized case because a
temperature gradient must be present in a member for heat to
conduct into or out of a body. In general, the smaller the physical
size of a member, the more realistic is the assumption of a uniform
temperature throughout the member. The method is valid if the
following inequality is maintained [7]:
hðV=FÞ
o 0:1 ð1Þ
k puting steel temperatures [8]. When the volumetric heat capacity
where h is the heat transfer coefficient, k is the thermal con-
ductivity, V is the volume, and F is the surface area of the member.
67
The left hand side of Eq. (1) is also called the dimensionless Biot
number. Steel members with typically high thermal conductivity
are suitable candidates for the use of this method for computing
temperatures. The following section describes how temperatures
are computed for unprotected and protected steels when exposed
to fire.
3.1. Lumped heat capacity method for unprotected steel
The following heat balance equation is used to derive a simple
expression for the change in steel member temperature during
exposure to fire:
Heat entering ¼ heat used to raise temperature
q_ } FΔt ¼ ρs cs VΔT s ð2Þ
where q_ } is the heat flux at the member surface (W/m2), Δt is the
time increment, ΔTs is the change in steel temperature, ρs is
the steel density, and cs is the gravimetric steel heat capacity.
Since the heat transfer to the exposed member occurs by both
convection and radiation during a fire event, Eq. (2) can be
rearranged as follows by expanding q_ } :
F 1 h i
ΔT s ¼ hc ðT f T s Þ þ sεðT 4f T 4s Þ Δt ð3Þ
V ρs cs
where hc is the convective heat transfer coefficient, s is the Stefan–
Boltzmann constant, ε is the steel effective emissivity, Tf is the
gas temperature in fire environment, and Ts is the bare steel
temperature.
Spreadsheets are often used for calculating steel temperatures
for a certain fire exposure. Gamble proposed such a method [8]
with a time increment (Δt) of 5 min. EC3 [9] suggested a
maximum time step of 30 s and a minimum value of section
factor (F/V) of 10 m  1. Kay et al. [10] reported a very good
prediction of steel temperatures in standard fire resistance tests
using this Lumped Heat Capacity method.
If both the convective and radiative heat transfer described in
Eq. (3) are combined to result in an effective heat transfer
coefficient, Eq. (3) can be integrated to yield the following
equation for steel temperature as a function of time:
   
T s T f hef f F
¼ exp  t ð4Þ
T i T f ρs cs V
where heff is the effective heat transfer coefficient, t is the elapsed
time and Ti is the initial steel temperature. Either Eq. (3) or Eq. (4)
can be used to compute steel temperatures.
3.2. Lumped heat capacity method for protected steel
The calculation approach for protected steel is similar to the
one discussed above for unprotected steel. It is assumed that
the temperature at the exposed surface of the insulation on steel is
the same as the gas temperature in fire. It is also assumed that the
internal surface temperature of the insulation equals that of
the steel member. Heat transfer coefficients are not required in
this approach as the method assumes that there is practically no
resistance to heat transfer at the exposed surface. The following
equation is obtained by algebraic manipulation [2]:
 
F ki ρs cs
ΔT s ¼ ðT f T s ÞΔt ð5Þ
V di ρs cs ðρs cs þ ðF=VÞdi ρi ci =2Þ
where ci, ρi, and di are the gravimetric heat capacity, density, and
thickness of the insulation respectively. Gamble proposed a similar
equation and formulated a spreadsheet-based approach for com-
of the insulation (product of ρi and ci) is low, both ECCS [11] and
Malhotra [12] suggest omitting the term in square brackets in Eq. (5).
68 D.K. Banerjee / Fire Safety Journal 61 (2013) 65–71

This omission is acceptable if the following inequality holds:
ρs cs A=2 4ρi ci Ai
where Ai is the cross-sectional area of the insulating material. This is
typically true for low thermal capacity insulation.
4. Mean steel temperatures and uncertainties in temperatures
4.1. Unprotected steel
Mean steel temperatures and uncertainties in steel tempera-
tures were computed using the Monte Carlo simulation method.
Monte Carlo simulation uses a sequence of random numbers to
simulate the statistics of a real experiment. In this method, inputs
are randomly generated from probability distributions to simulate
the process of sampling from a real population. It is therefore
necessary to choose distributions for the input parameters that
most closely match available data, or best represent the state of
current knowledge about them. The output data generated from
the simulation are represented as probability distributions or
converted to error bars or confidence intervals.
A MATLAB1 script was written for the Monte Carlo simulation.
Eq. (3) was used to compute steel temperatures. Note that, if the
Lumped Heat Capacity Method is used, there are four uncertain
parameters in Eq. (3): furnace temperature, steel volumetric heat
capacity, convective heat transfer coefficient, and steel effective
emissivity. Note that temperature dependent volumetric heat
capacity values for steel were used in the simulation [1], which
included effects of metallurgical transformation in steel at around
727 1C. Normally steel heat capacity increases with temperature
while thermal conductivity decreases with increase in tempera-
ture. Steel density is considered to be independent of temperature.
Steel thermal conductivity values were not required in the lumped
mass approach. Geometric variability in the steel beam is ignored.
A comparison of the computed mean web temperatures
(“Lumped Heat Capacity Mean”) and those obtained from the
ð6Þ
Cardington Test 1 is shown in Fig. 4. Note that “Cardington Mean”
represents the mean temperatures of all the thermocouples of the
beam web as shown in Fig. 1 and “Cardington Mean_1” represents
mean temperatures of all thermocouples at the beam web except
for the two thermocouples in the vicinity of the top chord (e.g., 54
and 55 in Fig. 1). It can be seen that the Lumped Heat Capacity
method predicts the heating trend reasonably well, although it
over-predicts steel temperatures during the heating cycle. The
temperatures match very well during the cooling phase.
The uncertainties of steel temperatures are discussed next.
Uncertainties in experimental web temperatures are obtained by
taking the standard deviations of the web temperatures in
Cardington Test 1. Note that the standard uncertainty u(y) of a
measurement y is defined as the estimated standard deviation of y
[14]. The computed uncertainties in steel temperatures were
obtained by the Monte Carlo simulations in which 1000 random
samples of the uncertain parameters were used to obtain a
distribution of computed web temperatures based on Eq. (3).
The mean and standard deviation of this steel temperature
distribution can be computed by MATLAB at each time during
the simulation. The standard deviations obtained from this dis-
tribution were taken as the computed uncertainties in web
temperatures as a function of time. Fig. 5 compares the uncertain-
ties in temperatures. Note that uncertainties in steel emissivity
have minimal influence on the overall uncertainties in steel
temperatures. Steel emissivity has first order influence on steel
temperatures, while fire temperature has fourth order influence
on steel temperatures as evident from Stefan–Boltzmann's
law. This is also clear from Eq. (3). It is clear from Fig. 5 that
1000
800
Temperature, 0C

In the first simulation, only furnace temperatures and steel

600
volumetric heat capacity were considered to be uncertain para-
meters. This is because of a lack of proper knowledge of the
distribution of the convective heat transfer coefficient and steel 400
effective emissivity as a function of temperature. The steel effec- Lumped Heat Capacity Mean
tive emissivity depends on many factors including the emissivity 200 Cardington Mean
of the flames, the compartment walls, and the steel itself. Note Cardington Mean_1
that heat transfer in most fires is dominated by radiative heat 0
0 50 100 150 200 250 300
exchange especially at elevated temperatures and the influence of
the convective heat transfer is not significant [2]. Paloposki and Time, minutes
Liedquist [18] reported that the standard deviations of emissivity Fig. 4. Plots of mean web temperatures obtained from the Cardington Test 1 and
can be as much as 20% of mean values at elevated temperatures. those computed from the Lumped Heat Capacity method.
Therefore, one Monte Carlo simulation was also conducted by
assuming the steel effective emissivity values were normally
distributed with its mean value of 0.7 [17] and standard deviation 70
of 20% of the mean value. The fire temperatures were considered Experimental uncertainty
Uncertainty in Temperature, 0 C

to be normally distributed with means and standard deviations 60 Computed Uncertainty

that are computed from the Cardington Test 1. The base value of 50 Computed uncertainty including steel
convective heat transfer coefficient was taken 25 W/m2/K [16]. The emissivity uncertainty

mean and standard deviations of the steel volumetric heat 40

capacity are taken from [6], where standard deviations are 30
approximately 2% of the mean values [13]. Again, the steel
volumetric heat capacity is considered to be normally distributed. 20
Normal distributions are typically used for measurement of
10
quantities in physical and engineering applications.
0
0 20 40 60 80 100 120
1
Certain commercial software or materials are identified to describe a Time, minutes
procedure or concept adequately. Such identification is not intended to imply
recommendation, endorsement, or implication by NIST that the software or Fig. 5. Uncertainty in the beam web temperatures obtained from the Cardington
materials are necessarily the best available for the purpose. Test 1 and those obtained using the Monte Carlo simulation (Eq. (3)).
 D.K. Banerjee / Fire Safety Journal 61 (2013) 65–71 69

uncertainties in steel effective emissivity did not have much
influence on uncertainties in steel temperatures. It can be seen
that the general trend obtained from the Monte Carlo approach
matches that of the experimental curve quite well. The maximum
uncertainty in temperatures was in the range of 40–50 1C. How-
ever, the computed values are somewhat lower than those
obtained from experimental measurements. This discrepancy
may be attributed to: (a) the exclusion of uncertainty in convective
heat transfer coefficient and (b) the influence of thermal mass
of the concrete slab on web temperatures, especially those near
the top chord (thermocouples 54 and 55 in Fig. 1), (c) the
approximate solution and error associated with the Lumped Heat
Capacity method. Nevertheless, this initial study shows a promis-
ing agreement.
4.2. Protected steel
Monte Carlo methods were also used to compute uncertainties
in the temperatures of the web members used in the NIST
fire resistance test described above. Expressions shown in Eq. (5)
were used to compute temperatures and the inequality in
(6) holds. The constant density and heat capacity values used for
testing this inequality are: steel density 7856 kg/m3, steel heat
capacity 658 J/kg/K, SFRM density 219 kg/m3, SFRM heat capacity
1098.2 J/kg/K. These values correspond to those at an average
temperature of 500 1C [1]. The following parameters were con-
sidered to be uncertain:
a. Fire temperatures.
b. Volumetric heat capacity of steel.
c. Thickness of fireproofing or SFRM.
d. Thermal conductivity of SFRM.
SFRM is a general category of materials used for steel fire-
proofing that includes an array of proprietary products such as
Blazeshield D C/F, Monokote etc. Each of these products has
The output temperatures were normally distributed. The standard
deviations of the computed temperatures were compared with
experimental measurements for uncertainty in steel temperatures.
Fig. 6 shows a comparison of mean temperatures for the beam
web. It can be seen that the agreement is very good until about
60 min, beyond which the computed temperatures become
increasingly greater than the experimentally recorded tempera-
tures. At about 55 min into the test, a very loud noise was heard
when a large piece of concrete fell to the lower part of the furnace
as a result of spalling [4]. See Fig. 3 for the approximation location
of this spalling. It is presumed that this resulted in significant loss
of heat from the furnace. Ceramic fiber insulation was placed over
the opening in the concrete floor in order to protect the hydraulic
equipment and allow the test to continue. Web temperatures
recorded during the full scale tests (Test 1 and 2 with 0.75 in.
(19 mm) fireproofing) were much higher than those for the
reduced scale test (e.g., Test 3 with 0.75 in (19 mm) fireproofing).
Large buildup of SFRM was noticed at the intersections of web and
chord members during the post-test observations [4]. Note that
this buildup is not a time-dependent phenomenon and NIST's
2 reduced scale tests showed a large buildup of SFRM at the
intersections of web and chord sections (much more than that was
noticed for the other 2 tests, which were full-scale tests). This is
different from the traditional overspray. The buildup was because
of the difficulty of ensuring a constant thickness at the intersection
of chord and web members because of constraint faced by the
operator while spraying. This buildup of SFRM could also have
contributed in resulting in lower measured web temperatures in
reduced scale tests.
Fig. 7 shows a comparison of uncertainties obtained in the test
and those obtained from the Monte Carlo simulation using two
1000
NIST Test Mean Temperatures
800
Lumped Heat Capacity Model
Temperature, 0C

varying thermal properties that are engineered for providing

600
passive fire protection. Here, “SFRM” is used only as a generic
identifier of a fire resistance product type, and consequently,
400
thermal properties may vary significantly among different com-
mercial products available in this category. Thermal properties of
SFRM vary with increase in temperature. SFRM densities initially 200
decrease with temperature until about 600 1C, beyond which they
increase rapidly. Thermal conductivity of SFRM gradually increases 0
0 20 40 60 80 100 120
with increase in temperature. SFRM products often undergo an
endothermic “dehydration spike” that results in very significant Time, minutes
energy absorption within a relatively narrow temperature range. Fig. 6. A plot of mean web temperatures obtained from the NIST test and those
This is reflected in the heat capacity or enthalpy vs. temperature computed from the Lumped Heat Capacity method for protected steel.
curve. A comprehensive finite element analysis employing these
temperature dependent thermal properties can provide a better
50
insight into the fire resistance performance of a SFRM product. Experimental uncertainty
Data used for volumetric heat capacity of steel are same as those
Uncertainty in temperature, 0C

Computed Uncertainty, COV of di=0.17

used for the case of unprotected steel. Mean fire temperatures and 40
Computed Uncertainty, COV of di=0.27
standard deviations were obtained from the measurements col-
lected by the 16 furnace thermocouples. The distribution of fire 30
temperatures was considered to be normal. Mean thermal con-
ductivity data for the SFRM were taken from [6]. The uncertainties
20
in thermal conductivity are about 6% of mean values [1]. Thermal
conductivity of SFRM is assumed to be normally distributed. The
10
SFRM thickness is lognormally distributed [15] with mean value
of 0.514 in. (13.1 mm) [4]. The COV (coefficient of variation) of
SFRM thickness is between 0.17 and 0.27 [1]. Since precise COV 0
values have not been reported in this test, both COV values of 0.17 0 20 40 60 80 100 120 140

and 0.27 were used for obtaining the standard deviations of SFRM Time, minutes
thickness. A MATLAB script was written to compute the distribu- Fig. 7. Uncertainties in beam web temperatures obtained from the NIST test and
tion of steel temperatures as a function of time using Eq. (5). those obtained using the Monte Carlo simulation for protected steel.
70 D.K. Banerjee / Fire Safety Journal 61 (2013) 65–71
different values of standard deviations of SFRM thickness (di)
corresponding to COV values of 0.17 and 0.27. The uncertainties
in fire temperature, steel volumetric heat capacity, and thermal
conductivity of SFRM were included in the simulation as discussed
in the previous paragraphs. Computed uncertainties in steel web
temperatures obtained for the COV values of 0.17 for di match
reasonably well with those from experimental measurements
until about 60 min, beyond which there is an abrupt increase in
experimentally obtained uncertainties in steel web temperatures
(Fig. 7). Computed uncertainties for the COV values of 0.27 for di
provide an upper bound. It appears from these plots that the COV
value for di in the experiment was probably close to 0.17. Concrete
spalling and SFRM buildup at the intersection of web and chord
members (as explained in the previous paragraph) appear to have
contributed significantly to the enhanced measured uncertainty
beyond 60 min into the fire test. The maximum uncertainty in
computed results was about 20 1C and 30 1C respectively for the
COV values of 0.17 and 0.27 for the SFRM thickness. The Monte
Carlo simulations show another interesting behavior. Uncertainties
in steel temperatures quickly increase to a maximum value within
about 20–25 min into the test and then start decreasing after
about 60 min into the test. This is because the absolute values of
sensitivity coefficients, ∂T=∂K i (Ki is an uncertain parameter), have
much higher magnitudes during initial transients and these
coefficients decrease toward the end of the test as temperatures
in the furnace become more uniform.
5. Summary
A Monte Carlo based approach was used to compute uncer-
tainties in temperatures in both unprotected and protected steel
members during fire resistance tests. The well-known “Lumped
Heat Capacity” method was used to obtain equations required for
the computation of steel temperatures. Computed time-dependent
mean steel temperatures and uncertainties in steel temperatures
were compared with those obtained from fire resistance tests
(Cardington Test 1 data were used for unprotected steel and NIST
World Trade Center fire resistance Test 4 data were used for
protected steel).
Computed time-dependent mean temperatures and uncertain-
ties in steel temperatures agreed reasonably well with those
obtained from experimental measurements. The maximum uncer-
tainties in unprotected steels during fire resistance tests were
higher than those obtained in protected steels. Loss of heat from
the furnace due to concrete spalling and SFRM buildup at web/
chord member intersection for the reduced-scale NIST fire test
possibly resulted in somewhat lower steel temperatures and
increased uncertainties in steel temperatures during the later part
of the test.
It is quite clear from this study that it is difficult to predict
uncertainties in protected steel temperatures (when steel beams
are used in fire tests as part of composite floor systems). This is
because of many unknowns such as (a) proper knowledge of the
quality and thickness of steel fireproofing at different locations
(e.g., it is difficult to ascertain the quality of adherence of fire-
proofing on steel as test progresses), (b) changes in thermal
properties of fireproofing as fireproofing degrades possibly due
to rapid and uneven heating, (c) inadequate knowledge of gas
temperatures near the vicinity of structural members, and
(d) concrete spalling that could possibly change macroscopic heat
flow in the furnace and impact the quality of SFRM bonding on the
steel etc. In addition, it is well known that during a standard
furnace test, SFRM is first applied on steel members and then the
majority of the design load is applied via load actuators. This
approach may induce SFRM micro-cracking during the onset of the
fire resistance test. The possible existence of such variable SFRM
micro-cracking at the beginning of the test may become amplified
during the subsequent elevated temperature furnace exposure as
the SFRM dehydrates, and may lead to increased and varying rate
of member heating. This phenomenon could add to the variability
of SFRM thickness distribution on steel members during fire
exposure in standard fire tests.
It is demonstrated that the Monte Carlo method can be
effectively used to compute uncertainties when the computational
approach is simple as is the case here. Further research is needed
to understand how the uncertain parameters considered in this
study could influence the overall uncertainties in steel tempera-
tures with more advanced models such as FE (finite element)
models. Use of probabilistic models such as Monte Carlo method
in conjunction with robust FE model will be needed to accurately
determine the uncertainties in steel temperatures in fire especially
for protected steels.
Data from a number of similar structural fire tests employing
different furnaces can be used to properly construct the uncer-
tainty bounds and distribution of member temperatures during
elevated temperature exposure. Such information can be used to
properly develop temperature dependent load and resistance
factors that are used by design engineers. Uncertainties in struc-
tural temperatures have effects on both the load and resistance
factors. There is ongoing active research in determining the load
and resistance factors at elevated temperatures and the present
paper is directed toward this end.
Acknowledgments
The writer wishes to gratefully acknowledge helpful exchanges
with E. Simiu concerning the application of the Monte Carlo
approach to the problems discussed in this paper.
Thanks are also extended to F. Lombardo for his help in
implementing a few MATLAB functions in the script developed
to compute uncertainties.
References
[1] NIST (National Institute of Standards and Technology), Federal building and
fire safety investigation of the World Trade Center disaster: final report on the
collapse of the World Trade Center Towers, NIST NCSTAR 1-5G, Gaithersburg,
Maryland, 2005.
[2] A.H. Buchanan, Structural Design for Fire Safety, John Wiley & Sons, U.K., 2001.
[3] British Steel plc., The behavior of multi-story steel framed buildings in fire,
Swinden Technology Institute, Rotherham, U.K., 1999.
[4] NIST (National Institute of Standards and Technology), Federal building and
fire safety investigation of the World Trade Center disaster: final report on the
collapse of the World Trade Center Towers, NIST NCSTAR 1-6B, Gaithersburg,
Maryland, 2005.
[5] ASTM, Standard test methods for fire tests of building construction and
materials, ASTM E119-73, ASTM International, Conshohocken, PA, 1973.
[6] NIST (National Institute of Standards and Technology), Federal building and
fire safety investigation of the World Trade Center disaster: final report on the
collapse of the World Trade Center Towers, NIST NCSTAR 1-6, Gaithersburg,
Maryland, 2005.
[7] J.P. Holman, Heat Transfer, sixth Ed., McGraw-Hill, New York, 1986.
[8] W.L. Gamble, Predicting protected steel member fire endurance using spread-
sheet programs, Fire Technology 25 (1989) 256–273.
[9] EC3, Eurocode 3: design of steel structures, ENV 1993-1-2: general rules –
structural fire design, European Committee for Standardization, Brussels,
Belgium, 1995.
[10] T.R. Kay, B.R. Kirby, R.R. Preston, Calculation of the heating rate of an
unprotected steel member in a standard fire resistance test, Fire Safety Journal
26 (1996) 327–350.
[11] ECCS, Design manual on the European recommendations for the fire safety of
steel structures, European Commission for Constructional Steelwork, Brussels,
Belgium, 1985.
[12] H.L. Malhotra, Design of Fire Resisting Structures, Surrey University Press, U.K.,
1982.
[13] NIST (National Institute of Standards and Technology), Federal building and
fire safety investigation of the World Trade Center disaster: final report on the
 D.K. Banerjee / Fire Safety Journal 61 (2013) 65–71
collapse of the World Trade Center Towers, NIST NCSTAR 1-3E, Gaithersburg,
Maryland, 2005.
[14] B.N. Taylor, C.E. Kuyatt, Guidelines for evaluating and expressing the uncer-
tainty of NIST measurement results, NIST Technical Note 1297, Gaithersburg,
Maryland, 1994.
[15] NIST (National Institute of Standards and Technology), Federal building and
fire safety investigation of the World Trade Center disaster: final report on the
collapse of the World Trade Center Towers, NIST NCSTAR 1-6A, Gaithersburg,
Maryland, 2005.
71
[16] ECI, Eurocode 1: Basis of design and design actions on structures, Part 2-2:
action on structures exposed to fire, ENV 1991-2-2, European Committee for
Standardization, Brussels, Belgium, 1994.
[17] S. Lamont, A.S. Usmani, D.D. Drysdale, Heat transfer analysis of the composite
slab in the Cardington frame fire tests, Fire Safety Journal 36 (2011) 815–839.
[18] T. Paloposki, L. Liedquist, Steel emissivity at high temperatures, VTT Research
Notes 2299, VTT Technical Research Centre of Finland, Vuorimiehentie,
Finland, 2005.