12 Fire Technology, 48, 493–512, 2012

 2011 Springer Science+Business Media, LLC. Manufactured in The United States

DOI: 10.1007/s10694-011-0241-x

A Simplified Approach for Predicting

Temperature Profile in Steel Members
With Locally Damaged Fire Protection

M. M. S. Dwaikat, Industrial Design Faculty, Delft Technical University,

Delft, The Netherlands
V. K. R. Kodur*, Department of Civil and Environmental Engineering,
Michigan State University, East Lansing, MI, USA

Received: 2 September 2010/Accepted: 8 September 2011/Published online: 22 September 2011

Abstract. Steel structures in building are to be provided with external insulation to

delay temperature rise and associated strength degradation when exposed to fire.
However, due to delicateness and fragility of some insulation systems, damage might
occur in these insulation systems during their service life, and such damage can lead
to rapid rise in steel temperature and result in lower fire resistance of structural mem-
bers. Currently, there are no simple calculation methods for quantifying the effect of
insulation loss on steel temperature. In this paper, a simple approach is proposed for
tracing the temperature profile in steel members with partial loss of fire protection.
The method is developed based on modifying the existing one-dimensional finite dif-
ference solutions of the heat transfer equation. The validity of the proposed method
is established by comparing the predictions from the proposed method against tem-
peratures obtained from finite element heat transfer model that is created using
ANSYS. The comparisons indicate that the proposed method is capable of predicting
temperature in steel members with partially damaged insulation to a good degree of
accuracy over a wide range of situations. Further, the simplicity of the proposed
method makes it attractive for use in fire resistance assessment in steel structures.

Keywords: Heat transfer, Fire protection, Insulation loss, Radiation, Convection, Fire resistance

Abbreviations

Ap Insulation cross sectional area

Fp The heated perimeter of the cross section
Fy Yield strength of steel at room temperature
hc The convective heat transfer coefficient
Hp The heat stored in the insulation
Hs The heat stored in the steel
k Thermal conductivity
kp The thermal conductivity of the insulation material
‘p Equivalent length of the lost insulation area
m Interpolation coefficient for averaging temperature in insulation using fire
temperature and steel temperature
ny, nz Components of the vector normal to the boundary in the plane of the cross-section

* Correspondence should be addressed to: V. K. R. Kodur, E-mail: [email protected]

494 Fire Technology 2012

p Percentage of lost insulation per insulation area

Qcon The net heat flux delivered to the plate through convection
Qrad The net heat flux delivered to the plate through radiation
q Heat flux
qb Heat flux through the boundaries
qcon Heat flux due to convection
qrad Heat flux due to radiation
T Steel temperature
Tf Fire temperature
t Time
tp Insulation thickness
r Stefan–Boltzmann constant
@ @
r @x þ @y is the gradient operator
qc Heat capacity
ee The effective emissivity of the exposed surface
qpcp and qscs The heat capacity of insulation and steel, respectively

1. Introduction
Damage to fire insulation on steel structural members can occur due to many rea-
sons, including accidental peeling of the material, bond degradation due to envi-
ronmental factors, or fall off due to sudden impact load. Loss of insulation can
lead to a rapid rise in steel temperature in the event of fire and this can cause
drastic reduction in strength and stiffness of steel, and hence the member carrying
capacity. Also, uneven loss of insulation on one side of the protected section can
lead to significant thermal gradients to develop along the cross section, and this
thermal gradient can generate lateral deflection (thermal bowing) or bending
moments due to restraint effects.
Up to date, there is a lack of simple methods for assessing the fire resistance of
steel members with partial loss of insulation. Currently, the only method for such
analysis is using complex finite element models that require skilled expertise and a
large amount of input parameters. Recently, few studies have been carried out to
investigate the effect of partial loss of insulation (especially spray-applied fire resis-
tance materials ‘‘SFRM’’) on fire resistance of steel structures. In most of these
studies, the insulation damage was simulated by removing a certain amount of
insulation from the face of the member/section. While there is a lack of experi-
mental studies on the effect of partial loss of fire protection on fire response, there
are quite a few numerical studies on this issue. Tomecek and Milke [1] and Ryder
et al. [2] performed a 2D and 3D finite element analysis on a steel column with
partial loss of insulation, respectively. These studies concluded that a 4% loss of
insulation per surface area resulted in a 15% reduction in fire resistance of a 1 h
fire rated W10 9 49 column and a 40% reduction in fire resistance for a 2-h fire
rated column.
Fontana and Knobloch [3] and Milke et al. [4] conducted a nonlinear 3D finite
element analysis to analyze steel columns with partial insulation loss. They con-
cluded that that extent of insulation damage and the size of the column have
significant influence on the thermal response of a column regardless of insulation
thickness.
Fire Resistance Assessment in Steel Structures 495

The effect of partial loss of SFRM fire protection on the load carrying capacity
of steel columns and beams was studied by Pessiki et al. [5] and Kang et al. [6],
respectively. These studies concluded that the capacity of beams and columns is
highly dependent on the location of insulation damage and that the removal of
even a small part of insulation leads to a dramatic increase of steel temperatures
in columns and beams and severely decreases the load capacity of columns and
beams.
In the previous studies, only detailed complex numerical simulations were pre-
sented as tools to study the effect of insulation loss on the temperature and load
capacity of steel members. In this paper, a simplified method is developed for
tracing the effect of partial loss of the fire insulation on the temperature profile of
steel member exposed to fire. Also, the method developed in this study can be use-
ful in post-fire investigation, where for instance, the maximum steel temperature
need to be estimated after a fire that has been caused by an accident which resul-
ted in a partial loss of insulation. Such simple method can facilitate reliable
assessment of cross-sectional steel temperature without the need for using complex
finite element modeling that requires skilled expertise.

2. Effect of Insulation Loss

When a part of insulation is lost, heat flow through that region will increase sig-
nificantly leading to localized sudden rise in steel temperature in that region. Due
to high conductivity of steel, this local rise of temperature then spreads into the
rest of the cross section and thus the average temperature of steel rises. Therefore,
partial loss of insulation leads to faster rate of strength and stiffness degradation
of steel and the structural member.
Further, uneven loss of insulation leads to development of high temperature
gradients across the steel section. Such temperature gradients generate two
major effects; first, thermal gradient leads to thermal bowing (lateral deflection)
if the member is not rotationally restrained. However, if the member is rota-
tionally restrained, then bending moments will develop in the member due to
restraint of thermal bowing. Second, thermal gradient causes a non-uniform
distribution of steel stiffness across the section. This inhomogeneity of steel
mechanical properties leads to a shift in the center of stiffness from its original
location at the center of gravity towards the cooler regions on the cross sec-
tion. The consequences of thermal gradient that is induced by partial loss of
insulation include the development of additional bending moments due to the
local P–d effect from either thermal bowing or migration of center of stiffness,
or both effects.
Such fire induced moments are to be given due consideration in evaluating fire
response of structural members, and for this, the effect of thermal gradient that
develops in the cross section due to a partial loss of insulation is to be quantified.
In the next sections, the governing heat transfer equation will be introduced, from
which a simple method will be developed for computing temperature rise due to
partial loss of insulation.
496 Fire Technology 2012

2.1. Heat Transfer Equation

The governing partial differential equation for heat transfer within the cross sec-
tion of a structural member (beam, column, etc.) can be written as [7]:

dT
qc ¼ r  ðkrT Þ ð1Þ
dt

where qc = heat capacity, T = temperature, t = time, and r = is the spatial

gradient operator, and k = conductivity matrix.
At the fire-member (e.g. beam) interface, heat is transferred through radiation
and convection. The heat flux on the boundary due to radiation qrad is generally
governed by Stefan–Boltzmann law:
 
qrad ¼ ree T 4  Tf4 ð2Þ

where r = Stefan–Boltzmann constant = 5.67 9 10-8 (W/m2 K4), and ee is the

effective emissivity of the exposed surface and it is related to the ‘‘visibility’’ of
the surface to the fire [7, 8]. Tf is temperature of the atmosphere surrounding the
boundary (in this case it is the fire temperature). The factor ee generally varies
between 0.5 and 1.0 depending on the material and nature of the surface and the
flames of fire. For steel members that are directly exposed to indoor fire, emissiv-
ity can vary between 0.7 and 0.9. The less ‘‘severe’’ exposure conditions (i.e., situ-
ations where heat flux from radiation is less), such as exposure to external fire
outside building façade or rougher surfaces, such as insulation or concrete sur-
faces, the emissivity can be between 0.3 and 0.7 [7, 9].
Heat flux transferred through convection qcon is generally governed by Newton’s
law of cooling:
 
qcon ¼ hc T  Tf ð3Þ

where hc is the convective heat transfer coefficient and it varies between 20 W/

m2 C and 35 W/m2 C for typical fire exposures. This factor also depends on the
atmospheric conditions of the fire and the type of exposure of the surface to the
fire. If the surface is directly exposed to fire, such as the bottom flange of steel
beams, the value of hc tends to be high. If the surface is far from direct exposure,
such as the case with steel connections or gaps, values of hc can be as low as
15 W/m2 C [7].
The heat flux and temperature gradient in any medium are generally related
through Fourier’s law of conduction:

q ¼ krT ð4Þ

Using Fourier’s law, the governing heat transfer equation on the boundary of the
beam can be expressed as:
Fire Resistance Assessment in Steel Structures 497

 
@T @T
k ny þ nz ¼ qb ð5Þ
@y @z

where ny and nz are components of the vector normal to the boundary in the
plane of the cross-section. The right hand side of Equation 5 is dependent on the
type of boundary condition.

2.2. Simplified Approach

In order to account for the effect of partial loss of insulation on steel temperature,
consider a steel plate covered with uniform spray-applied fire proofing material
of thickness tp as shown in Figure 1. The heated perimeter through which the
insulation is in contact with the steel plate is denoted as Fp. If we assume that a
percentage (p) of the insulation cross-sectional area Ap is lost due to some reason,
then the length of the lost protection (‘p) is computed as (Ap/tp) 9 p as shown in
Figure 1.
The total heat that is transferred to the plate is comprised of two parts, heat
due to radiation Qrad and heat due to convection Qcon. Based on conservation of
energy, the incoming heat should equal the heat stored in the insulation Hp and in
steel Hs, that is:

Hp þ Hs ¼ Qcon þ Qrad ð6Þ

The heat stored in insulation and steel can be computed, per unit depth of the
plate, as:
 
  DTf þ DT
Hp þ Hs ¼ qp cp tp Fp  ‘p þ qs cs As DT ð7Þ
m

where qpcp and qscs are the heat capacity of insulation and steel, respectively. In
Equation 7, the temperature on the surface of the insulation was assumed to be
equal to the fire temperature. Also, it was assumed that the temperature in the

Figure 1. Cross section of steel plate with partial loss of insulation.

498 Fire Technology 2012

protection is an average of the fire and steel temperatures, i.e. (Tf + T)/m, where
m = 2 for linear variation.
The heat transferred to the plate can be computed, per unit depth of the plate, as:
h     i
Qcon þ Qrad ¼ hc ‘p þ kp =tp Fp  ‘p Tf  T þ ‘p ree Tf4  T 4 Dt ð8Þ

where kp is the thermal conductivity of the insulation material. The heat transferred
due to convection in Equation 8 is computed due to convection through exposed
steel (‘p, i.e. where insulation is damaged) only. Convection and radiation is neglec-
ted on the surface of the insulation (Fp - ‘p, i.e. where insulation is intact) simply
because the temperature on that surface is assumed to be equal to the fire tempera-
ture (Tf). Therefore, the heat flux on the boundary (Fp - ‘p) between insulation
and the steel plate is computed in Equation 8 using Fourier’s law of conduction
(Equation 4) and assuming linear gradient of temperature across the insulation.
Combining Equations 7 and 8 and solving for the change in steel temperature
leads to the following relationship:
 
Fp  ‘p =As    DT
DT ¼   1 þ hc ‘p Tf  T þ qr Dt  ð9Þ
tp / 1 þ m/
kp qs cs 1 þ m
     
q p cp t p Fp ‘p ‘p
with / ¼ and qr ¼ t =k re Tf4  T 4 :Obviously, Equation 9
q s cs As ðFp ‘p Þ p p e
assumes a single exposed plate (with lost insulation) that is not connected to other
plates of the cross section. However, it will be shown that the assumptions made
above are sufficient to arrive at a reasonable estimation of average temperature
rise in the exposed plate.
On the other hand, the temperatures in the other plates of the cross section will
be affected by the localized rise in temperature of the exposed plate. Obviously,
due to the high conductivity of steel, more heat will flow to the other plates from
the plate of concentrated temperature rise where insulation was lost. The
approach above does not incorporate thermal conductivity of steel since it consid-
ers steel to have a uniform temperature. However, to account for the flow
between the exposed plate and the other plates, a corrective factor (a) is intro-
duced and is used for the other plates. Using similar approach as the one above,
temperature change in the other steel plates can be computed as:

aFp =As   DT
DT ¼   Tf  T Dt  ð10Þ
tp
qs cs 1 þ m/ 1 þ m/
kp
n o
q c t F
with / ¼ qp cps p a Aps , and the factor a in Equation 10 is introduced to account for
s
the heat flow from the exposed (hotter) plate, where insulation is lost, to the other
plates of the cross section.
Based on intensive numerical studies, the factor a is affected primarily by the
change in thermal properties of steel with temperature and the differential in the
Fire Resistance Assessment in Steel Structures 499

heat stored in the exposed plate and the rest of the plates of the section. The ther-
mal conductivity of steel drops to half after reaching 700C, and specific heat of
steel also rises significantly after such temperature. The differential in stored
energy in the exposed plate and the other plates of the cross section can be related
to the excessive heat transferred to the exposed plate through the insulation gap
(‘p). This is justified in case of 4-side exposure, where the cross section tends to
have a uniform temperature distribution. Thus, introducing a gap in the insulation
will lead to a gradient in the heat stored in the steel plates.
Therefore, to compensate for the heat flow from the steel plate where the insu-
lation is damaged, regression analysis was conducted to estimate the correction
factor a. The results of numerical analysis on a wide range of steel cross sections,
with different ratios of damaged insulation were normalized to the results from
analysis on the same set of cross sections but with fully intact insulation. Based
on the results of numerical studies, the factor a is assumed to attain these values:
( ‘p
2:0 þ T  700 C
a¼ ðFp Þexp ð11Þ
1:0 T > 700 C

where (Fp)exp is the heated perimeter of the exposed plate (with partial loss of
insulation).
Eurocode 3 [10] provides a relation for computing temperature in fully pro-
tected steel members. Eurocode relation is given as:
!
Fp =As Tf ðtÞ  T ðtÞ  
/=10
DT ¼ tp Dt  e  1 DTf ð12Þ
kp qs cs 1 þ /3

q c F  
where / ¼ tp qp cps Aps and m = 3. The term e/=10  1 is suggested by Eurocode to
s
provide better correlation with test data. This equation is quite similar to Equa-
tion 9 but cannot be used as it is for computing temperature in steel members
with damaged or lost insulation. Therefore, Equation 12 can be adjusted by
replacing the section factor Fp/As with the reduced section
 factor (Fp - ‘p)/As and
then incorporating the heat fluxes due to convection hc ‘p Tf  T and radiation qr
as seen in Equation 9.

3. Numerical Analysis
In order to verify the validity of the approach developed above, predictions from
the proposed equations are compared to results from rigorous finite element anal-
ysis. The finite element model for thermal analysis is developed using ANSYS [11]
and is presented below.

3.1. Thermal Analysis

A steel cross section of an I-beam is modeled for heat transfer analysis using
ANSYS [11]. The thermal mesh was created using two types of elements, namely
500 Fire Technology 2012

PLANE55 and SURF151. PLANE55 is a plane element with two-dimensional

thermal conduction capability and has four nodes with a single degree of freedom,
namely temperature, at each node. The element is applicable to a two-dimen-
sional, steady-state or transient thermal analysis. SURF151 element is generally
used to account for various load and surface effect applications. In the current
analysis, SURF151 element was overlaid onto the face of PLANE55 2D thermal
solid element to simulate the effect of thermal radiation from ambient air to steel
section.
The cross-section was meshed with PLANE55 element, while the exposed
boundaries of the cross-section were meshed using SURF151 elements. Both heat
convection and radiation loads were applied at the exposed boundaries of the sec-
tion (on SURF151 elements). A convection coefficient of hc = 25 W/(m2 C) was
assumed in the analysis based on the recommendation in literature [7, 9, 12, 13]
and in codes [14]. Different values of effective emissivity factor (ee) were used
depending on the exposure boundary [7]. For the outer side surfaces of the flanges
the value of ee was 0.7 and for the internal sides of the web and flanges the
value of ee was 0.5. The Stefan–Boltzmann radiation constant of r = 5.67 9
10-8 W/(m2 K4) was used in the analysis and the ambient (bulk) temperature (T¥)
surrounding the faces of SUR151 element was assumed to be equal to the fire
temperature (Tf).

3.2. High Temperature Thermal Properties

The temperature-dependent thermal properties of steel that are used in the
thermal analysis model are according to the Eurocode constitutive models [10].
Figure 2 shows the variation of thermal conductivity and heat capacity of struc-
tural steel as a function of temperature. While the thermal conductivity of steel
drops linearly until 800C, heat capacity undergoes small increase until a sudden
increase in heat capacity occurs at around 750C. The peak in heat capacity of

Figure 2. Thermal properties of structural steel according

to Eurocode 3.
Fire Resistance Assessment in Steel Structures 501

Figure 3. Temperature dependent thermal properties of the insula-

tion material. a Thermal conductivity (kp). b Heat capacity (qpcp).

steel at 750C can be attributed to the phase change that occurs in steel at such
high temperature, which results in absorption of huge amount of energy.
For the insulation used in the thermal analysis, the thermal properties are plot-
ted in Figure 3 as function of temperature. The insulation material whose proper-
ties are presented in the figure is CAFCO300 [15], and it is a typical cementitious
fire proofing material that is spray-applied on steel members. This insulation
material has been used in the fire tests that are used for the validation of the ther-
mal analysis model as presented in the next section.
Both thermal conductivity and heat capacity of the insulation material were
directly measured at various temperatures using Hot Disk ‘‘TPS2500’’ [16] test appa-
ratus which measures these properties as per ISO/DIS22007-2 specifications [17].
Thermal conductivity and specific heat tests were carried out using Hot Disk is
connected to an electric furnace through which a specimen can be subjected to
increasing temperature up to 735C. This state-of-the-art test equipment utilizes a
flat plate sensor through transient plane source (TPS) technique to measure ther-
mal conductivity and specific heat in 20C to 735C range [17]. The sensors are
made-up of electricity-conducting double spiral shaped nickel foil, insulated with
25 lm thick layer of either kapton or mica sensors. Kapton sensor can be used
for measurements up to 200C, whereas mica sensor has to be used for measure-
ments in 100C to 735C. The sensor, when sandwiched between two halves of
specimen, acts as a heat source, as well as detector, and reads input and output
voltage signals. During high temperature thermal property tests, specimen and
sensor assembly is kept in electric furnace that is connected to the Hot Disk. The
furnace generates specified temperature in the specimen and the sensor records
thermal conductivity and specific heat at the target temperature.
Kapton sensor was used to undertake tests at room temperature (20C) and
mica sensor was used for the remaining temperature points from 100C to 735C.
The sensor, sandwiched between two halves of the specimen and the assembly, is
502 Fire Technology 2012

exposed to high temperature generated in the electric furnace connected to Hot

Disk apparatus. The test condition parameters namely, probing depth, sensor
type, initial thermal coefficient of resistance (TCR), hold time at steady state, mea-
surement power, and measurement times are to be programmed (as input) into
Hot Disk apparatus by user. For these tests, initial TCR of 0.004698/K, probing
depth of 15 mm, sensor measurement power of 1.0 W, hold time of 30 min and
measurement time of 40 s was used. For high temperature measurements, a built-
in value of TCR at each temperature as per sensor type was used. At each target
temperature, once the thermal equilibrium is attained, thermal conductivity and
specific heat are recorded. The temperature then goes to next target temperature
for new measurements.
The measured thermal conductivity is compared to data from literature [18] in
Figure 3a. Results indicate that thermal conductivity of the insulation generally
increases with temperature, probably due to the shrinkage of the SFRM mate-
rial under heating, while the heat capacity (Figure 3b) first decreases slightly to
400C, then increases again with temperature. The initial decrease in the heat
capacity is attribute to the phase changes in the material. It is worth mentioning
that the TPS method can skip the effect of the moisture evaporation on the heat
capacity of the SFRM, and thus, careful attention should be taken when using
the measured properties. This however does not affect the proposed method
since the method is incremental and can employ variable temperature-dependent
thermal properties of SFRM and/or steel. The differences between the different
measured values of thermal conductivity can be ascribed to differences in test
methods or environmental parameters, such as moisture content of the insula-
tion on the day of test, in addition to the differences in chemical composition of
different types of SFRMs. Therefore, based on that comparison, a representative
‘‘smooth’’ curve for thermal conductivity as function of temperature is adopted
for thermal analysis and is shown in Figure 3b. Similarly, the recorded heat
capacity was smoothened via a polynomial function for efficient input for the
thermal analysis model.

3.3. Modeling Insulation Damage

Since the parameters of convection and radiation heat fluxes across the region
where insulation is partially lost or damaged are not precisely known, an indirect
approach is used to model the loss of insulation. This approach is based on
replacing the insulation gap with a hypothetical material that has much higher
conductivity and much lower heat capacity than those of the original insulation
material. The surface of that material (where insulation is lost) is assumed to con-
tinue to have the same heat convection and radiation parameters as those of the
cross section.
In order to determine how much to amplify the conduction properties of the
hypothetical material, a comparison is made between the model predictions and
published data obtained from fire tests. This process is shown in the validation of
the thermal model in the next section. It is worth mentioning that this is a crude
method for modeling insulation damage/loss. This approach is justified because of
Fire Resistance Assessment in Steel Structures 503

the lack of information on the heat transfer parameters (convection/radiation) on

the area of damaged/lost insulation.

3.4. Validation of ANSYS Thermal Model

There is a lack of fire tests on steel members with damaged insulation. Recently,
fire resistance tests on steel beam-columns were carried out at Michigan State
University [19], wherein the insulation was scraped (removed) on one side of the
beam-columns. The fire resistance tests, conducted on four insulated beam-col-
umns of W8 9 48 sections, were aimed at studying the effect of thermal gradient
on beam-column response. These beam-columns were of A992 Grade 50 steel
(measured average Fy = 345 MPa) and had a length of 3.3 m. The specimen
dimension, thermal mesh and insulation scheme are shown in Figure 4. Full
details of the fire resistance experiments, including test procedure, instrumenta-
tion, and detailed measurements are presented elsewhere [19].
The applied insulation was uniform with an average thickness of 40 mm. The
spray-applied fire resistant material (SFRM) that was used is CAFCO 300 [15]
and its thermal properties (thermal conductivity and heat capacity) are plotted in
Figure 3 as function of temperature. Thermal gradients were generated by remov-
ing the insulation material at specific location as shown in Figure 4a. Figure 4b
shows the thermal mesh, where the cross section was meshed using PLANE55 ele-
ments and the exposed boundary with SURF151 elements.
The insulation gap was modeled using a higher conducting hypothetical insula-
tion material as explained earlier. The resulting temperature distribution after
60 min of fire exposure is shown in Figure 4c, which shows a high thermal gradi-
ent across the section and higher temperature near the insulation gap.
A comparison between the results of ANSYS model and measured temperatures
in the fire test is shown in Figure 5. Since the insulation loss was modeled using a
higher conducting hypothetical material, the thermal properties of such material
need to be calibrated. The thermal properties of the damaged insulation (hypo-
thetical material) were ‘‘degraded’’ to 3 and 5 times of the original thermal proper-
ties of the insulation and then the results were compared to the recorded
temperatures. Figure 5a shows that the results obtained of using (kp)Damaged =
3 9 (kp)Original for the hypothetical material led to a good agreement with the tem-
peratures recorded in the test. However, using (kp)Damaged = 5 9 (kp)Original for the
damaged insulation resulted in higher temperature in the steel section as shown in
Figure 5b. This is primarily due to the higher conduction in the region with lost
insulation. Therefore, using (kp)Damaged = 3 9 (kp)Original as a ‘‘degradation ratio’’
for the modeling the insulation gap leads to better predictions. Also, predictions
from the proposed approach (Equations 9, 10) are compared to test data in
Figure 5c. The comparison shows that the proposed approach provide conserva-
tive estimates of steel temperature, especially in the flange with damaged insula-
tion. Overall, the comparison shows that the thermal model in ANSYS and the
proposed approach are capable of predicting temperatures in steel members with
local insulation damage within a sufficient accuracy.
504 Fire Technology 2012

Figure 4. Dimensions, thermal mesh and temperature contour of the

tested specimen. a Cross section dimensions with 40 mm average
insulation thickness. b Thermal mesh, thermocouples and lost insula-
tion. c Temperature gradient at t = 60 min.
Fire Resistance Assessment in Steel Structures 505

Figure 5. Temperature in steel using different damage properties

ðk p Þ
of insulation. a ANSYS model vs. test data using k Damaged ¼ 3 and
ð p ÞOriginal
ð p p ÞOriginal
q c ð pÞ
k ð p pÞ
q c
¼ 3; b k Damaged ¼ 5 and q c Original ¼ 5; c Test data compared
ðqp cp ÞDamaged ð p ÞOriginal ð p p ÞDamaged
to predictions from the proposed method.
506 Fire Technology 2012

4. Validating the Proposed Approach

For validating the proposed approach, temperatures obtained from Equations 9
and 10 are compared with results generated from finite element analysis. Equa-
tion 9 is used to predict temperature in the exposed plate with partial loss of insu-
lation, while Equation 10 is used for the other plates. Two cases are considered.
Case one is with a part of the insulation removed from the flange, and case two is
with part of the insulation removed from the web. In each case, the ratio (p) of
the removed insulation area to the total insulation area was varied from 3% to
10%. The two cases are illustrated in Figure 6 which shows the meshed sections
with the corresponding insulation areas removed from either flange or web. The
cross section selected for analysis is a typical W14 9 311 [20] column exposed to
fire from four sides. The results of the comparison are presented in Figures 7, 8,
9, and 10.

Figure 6. Different ratios of insulation loss in steel columns. a Insula-

tion loss in the flange. b Insulation loss in the web.
Fire Resistance Assessment in Steel Structures 507

Figure 7. Steel temperature in W14 3 311 cross section with

12.7 mm insulation, 4-sided fire exposure, with insulation loss
in the flange. a p = 3%. b p = 10%.

Figure 8. Steel temperature in W14 3 311 cross section with

25.4 mm insulation, 4-sided fire exposure, with insulation loss
in the flange. a p = 3%. b p = 10%.

4.1. Insulation Loss in the Flange

Figures 7 and 8 show steel temperature developed in the column in the case of
lost insulation in the flange of the column. It can be seen in the figures that a
thermal gradient develops in the cross section due to the loss of insulation, with
higher temperature in the flange where the insulation is lost. The thermal gradient
increases with the increase of the lost insulation area. Losing 3% of the insulation
area in the flange leads to 200C thermal gradient, as shown in Figure 7a. Fur-
ther, if the lost insulation area in the flange increases to 10%, the thermal gradi-
ent in the section increases to 400C, as shown in Figure 7b.
508 Fire Technology 2012

Figure 9. Steel temperature in W14 3 311 cross section with

12.7 mm insulation, 4-sided fire exposure, with insulation loss
in the web. a p = 3%. b p = 10%.

Figure 10. Steel temperature in W14 3 311 cross section with

25.4 mm insulation, 4-sided fire exposure, with insulation loss
in the web.

The results predicted using the proposed approach (Equations 9, 10) compare
well with the results obtained from ANSYS. Similar trend can be seen in Figure 8,
which shows the effect of insulation loss on steel temperature but with increased
insulation thickness to 25.4 mm. Based on comparison between Figures 7 and 8, it
seems that the insulation thickness has small effect on the thermal gradient devel-
oped due to the insulation loss. This observation was mentioned in previous stud-
ies [3, 4]. The reason behind such behavior can be attributed to the fact that, in
case of insulation loss, the thermal gradient that develops in the section is mainly
due to the influx of temperature through the gap of insulation, and thus depends
mainly on the exposed length where insulation is removed rather than on the insu-
lation thickness.
Fire Resistance Assessment in Steel Structures 509

4.2. Insulation Loss in the Web

Figures 9 and 10 show the effect of losing insulation in the web of the cross sec-
tion on the steel temperature. In the figures, only the web and flange temperatures
are reported due to the symmetry of results in the flanges. As expected, losing
some of the insulation that covers the web leads to a rapid rise in temperature in
that region. This is seen in Figure 9a, b, where the temperature of the web is
higher with larger insulation loss. Also, comparing Figure 9a, b indicates that lar-
ger insulation loss leads to higher difference between web and flange temperature,
obviously due to higher influx of heat to the exposed web. Further, comparing the
results in Figure 9a to those in Figure 10a, which shows the same results but with
thicker insulation, leads to a conclusion that the insulation thickness has little
influence on the difference between the web and flange temperatures. This can be
attributed to the fact that the heat influx through the insulation gap is a function
of the exposed length (‘p) rather than the thickness of the insulation.
On all cases, the results obtained using the proposed approach seems to be in a
good comparison with those obtained from ANSYS. The proposed approach
tends to overestimate the temperature in the exposed plate (the web) at tempera-
tures higher than 800C. This can be attributed to the differences in the heat
transfer mechanism between the simplified approach and the finite element
method. While the finite element method approach models the insulation loss/gap
as a higher conducing material, the proposed approach accounts for radiation and
convection heat fluxes. It is hard to determine the exact values of effective emissiv-
ity (ee) and convective heat transfer coefficient (hc) in the area of lost insulation.
The values of ee and hc that were used in the proposed approach (in Equation 9
for that region) were 0.3 W/m2 C and 15 W/m2 C, respectively. The correlation
between the experimental and the theoretical values of ee and hc in case of dam-
aged/lost insulation requires intensive tests and is beyond the scope of this study.
The aim of this study is to present a simplified approach that is capable of pre-
dicting the temperature profile in steel section when it suffers a loss in fire protec-
tion.
As it was explained earlier, the partial loss of insulation have a significant
impact on the bearing capacity and hence fire resistance of structural members.
This effect will be discussed through an example in the following section, where
the applicability of the above computed steel temperatures to determine the sec-
tional capacity of steel members is illustrated.

5. Design Applicability
In order to illustrate the applicability of the proposed approach in fire resistance
assessment or design situation, an example is presented herein. A steel column of
AISC section W14 9 145 [20], with 25.4 mm uniform spray applied fire protection
material is exposed to fire from four sides and suffers a loss of 5% of fire protec-
tion in one of the flanges of the column, as shown in Figure 6a. The insulation
properties are assumed to be similar to those presented in this paper. The pro-
posed approach (Equations 9, 10) is used to compute the temperature in the three
510 Fire Technology 2012

Figure 11. Effect of insulation loss on the temperature and capacity

of steel column. a Steel temperature. b Plastic and axial capacity.

plates of the section. Figure 11a plots the temperatures obtained using the pro-
posed approach and from ANSYS for the three plates of the section. Also, the
average temperature of the section assuming no insulation loss is plotted in
Figure 11a. The faster rise in temperature of the section due to the loss of 5%
insulation can be readily seen in Figure 11a when compared to the average tem-
perature with fully intact fire protection.
Figure 11b shows the reduction in the plastic axial capacity of the column as
function Pof fire exposure time. The axial capacity is computed as
Py ðT Þ ¼ Fy ðTplate ÞAplate and using the mechanical properties of steel as specified
in Eurocode 3 [10]. As it can be seen in Figure 11b, 5% loss of insulation results
in a severe reduction in the carrying capacity of the column. The results indicate
that if the insulation is fully intact, the column retains its axial plastic capacity for
100 min of fire exposure. However, the capacity drops to 50% after 100 min of
exposure when 5% of insulation is lost.
Based on Figure 11a, if the applied axial load on the column above was 50% of
its room temperature axial capacity, the column with fully intact fire protection
will have 3 h of fire resistance, but if the same column suffers 5% insulation loss
in the flange, the fire resistance drops to 90 min.

6. Conclusions
Based on this study, the following conclusions can be drawn:

1. Partial loss of fire insulation causes rapid rise in steel temperature and develop-
ment of significant thermal gradients in steel sections exposed to fire. In turn,
the rise in steel temperature causes a severe reduction in the load carrying
capacity of steel members, and development of additional internal forces if the
member is restrained.
Fire Resistance Assessment in Steel Structures 511

2. A new approach is developed based on basic heat transfer principles for com-
puting the influence of partial loss of fire protection on the temperature profile
in steel sections exposed to fire.
3. Results from the analysis show that 5% loss of insulation can reduce fire resis-
tance of a steel column from 3 h to 90 min.
4. There is a lack of data on the heat transfer parameters (convection and radia-
tion) in the region where fire protection is damaged or lost in a protected steel
section.
5. The simplicity and versatility of the proposed approach makes it attractive for
application in fire resistance assessment in design and analysis situations.

Acknowledgements
The research presented in this paper is supported by the National Science Foun-
dation (Grant No. 0652292) and Department of Commerce/National Institute of
Standards and Technology (Grant No. 60NANB7D6120). Any opinions, findings,
and conclusions or recommendations expressed in this paper are those of the
authors and do not necessarily reflect the views of the sponsors.

References
1. Tomecek DV, Milke JA (1993) A study of the effect of partial loss of protection on the
fire resistance of steel columns. Fire Technol 29(1):4–21
2. Ryder NL, Wolin SD, Milke JA (2002) An investigation of the reduction in fire resis-
tance of steel columns caused by loss of spray-applied fire protection. J Fire Prot Eng
12(1):31–44
3. Fontana M, Knobloch M (2004) Fire resistance of steel columns with partial loss of fire
protection. In Proceedings of the IABSE symposium Shanghai 2004 metropolitan habi-
tats and infrastructure. IABSE report, vol 88. Shanghai, China, ISBN 3-85748-110-2
4. Milke JA, Ryder N, Wolin S (2003) Analyses of the impact of loss of spray-applied fire
protection on the fire resistance of steel columns. In: Fire safety science—proceedings
of the seventh international symposium, pp 1025–1036
5. Pessiki S, Kwon K, Lee BJ (2006) Fire load behavior of steel building columns with
damaged spray-applied fire resistive material. In: Fourth international workshop struc-
tures in fire. Aveiro, Portugal, pp 235–245
6. Kang Y, Hadjisophocleous GV, Khoo HA (2006) The effect of partial fire protection
loss on the fire resistance reduction of steel beams. In: Proceeding of fourth interna-
tional workshop structures in fire. Aveiro, Portugal, pp 63–74
7. Purkiss JA (2007) Fire safety engineering design of structures, 2nd edn. Butterworth-
Heinemann, Linacre House, Jordan Hill, Oxford
8. Buchanan AH (2002) Structural design for fire safety. Wiley, Chichester
9. Staggs JEJ, Phylaktou HN (2008) The effects of emissivity on the performance of steel
in furnace tests. Fire Saf J 43(1):1–10
10. EC3 (2005) EN 1993-1-2: design of steel structures. Part 1-2: general rules—structural
fire design. European Committee for Standardization, Brussels
11. ANSYS (2007) ANSYS multiphysics, version 11.0 SP1. ANSYS Inc, Canonsburg
512 Fire Technology 2012

12. Wang YC (2002) Steel and composite structures, behavior and design for fire safety.
Spon Press, London
13. Both C, Twilt L (1990) Report no. BI-89-208. TNO Building and Construction
Research, Netherlands
14. EC1 (2002) EN 1991-1-2: actions on structures. Part 1–2: General actions—actions on
structures exposed to fire. European Committee for Standardization, Brussels
15. Isolatek International (2008) CAFCO 300 guide specifications. Stanhope http://www.
cafco.com/pdfs/CAFCO%20300%20Brochure.pdf
16. Hot Disk AB (2007) Thermal constants analyzer—TPS 2500 s. Chalmers University
of Technology, Gothenburg. http://www.hotdiskinstruments.com/images/stories/pdf/
TPS2500.pdf
17. Log T, Gustafsson SE (1994) Transient plane source (TPS) technique for measuring
thermal transport properties of building materials. Fire Mater 19(1):43–49
18. Bentz DP, Prasad KR (2007) Thermal performance of fire resistive materials I. Charac-
terization with respect to thermal performance models. Report BFRL–NIST 7401.
Gaithersburg
19. Dwaikat MMS, Kodur VKR, Quiel SE, Garlock MEM (2010) Experimental behaviour
of steel beam-columns subjected to fire-induced thermal gradients. J Constr Steel Res
(in press)
20. AISC (2005) Steel construction manual, 3rd edn. American Institute of Steel Construc-
tion, Chicago