Professional Documents
Culture Documents
To cite this article: Behrouz Behnam & Hamid Ronagh (2013) Performance of reinforced concrete
structures subjected to fire following earthquake, European Journal of Environmental and Civil
Engineering, 17:4, 270-292, DOI: 10.1080/19648189.2013.783882
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
European Journal of Environmental and Civil Engineering, 2013
Vol. 17, No. 4, 270–292, http://dx.doi.org/10.1080/19648189.2013.783882
Fire following earthquake (FFE) is a serious threat to structures that are partially
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
1. Introduction
Buildings in urban regions need to be designed in a way that provides a “safe and
sound” environment for the inhabitants, for the design life of the building. Indeed,
being structurally sound is a fundamental necessity for buildings located in urban areas
and, as such, they are designed to meet the regulations of codes and standards. The
“life safety” criterion, therefore, is the criterion by which most (if not all) building
structures are designed. “Life safety” guarantees that even under an extreme loading
situation, structures will continue to bear gravity load, allowing occupants to safely
evacuate the building regardless of the level of damage, which is allowed to vary, based
on the level of “importance” of the structure at the time of design (Borden, 1996).
Whilst almost all codes guarantee safety under a number of load combinations repre-
senting the extreme loading possibilities, fire following earthquake (FFE) loading has
not received adequate attention in the available codes. FFE loading, however, is a real
situation, which has happened in almost every major earthquake in history. Thus, miti-
gation of the effect of FFE on buildings, in such a way that provides adequate time for
evacuating people surrounded by the fire, must be a key aspect of FFE safety strategy.
Recorded experiences have proven that FFE can potentially bring about even more
damage in comparison with the earthquake itself. For instance, in the fire that followed
the earthquake in Kanto, Japan, in 1923, more than 447000 mostly wooden houses
were destroyed. The fire alone accounted for about 77% of the total losses (Scawthorn
& Eidinger, 2005). As an another example of FFE, the district of Marina located in city
of Loma Prieta, America, can be addressed, where a large number of houses were com-
pletely destroyed as a result of FFE in 1987 (Mohammadi & Alysian, 1992). More
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
recently, large areas of the city of Kesennuma were engulfed by the FFE in Japan, 2011
(Thomson, 2011). Structural damage resulting from FFE is more severe than that caused
by fire alone, as any pre-fire earthquake loading would have weakened the structure.
From a different point of view, as earthquakes may cause serious damage to lifeline
structures, arterial roads and bridges, fire brigades would have increased difficulty in
suppressing fires. Accordingly, more time will be required to control a fire than in the
usual situations. This time may be further increased, as helping people trapped under
the rubble takes priority. Thus, more strict fire provisions must be provided in a FFE.
Using the philosophy of earthquake design based on performance (California
Seismic Safety Commission, 1996), structural members are normally designed to satisfy
various levels of performance, some of which are Operational (O), Immediate Occu-
pancy (IO), Life Safety (LS) and Collapse Prevention (CP). According to the perfor-
mance design criteria, the expected performance of structures shall be controlled by the
assignment of each structure to one of several “Seismic Use Groups”. In FEMA 450
(2003), for example, there are three “Seismic Use Groups” which are categorised based
on the occupancy of the structures within the group and the relative consequences of
earthquake-induced damage to the structures. Design codes specify progressively more
conservative strength, serviceability and detailing requirements for structures, in order
to attain minimum levels of earthquake performance suitable to the individual occupan-
cies. Structures contained in these groups are not specific to a certain seismic zone,
rather they are spread across all zones from high to low hazard and, as such, the groups
do not really relate to hazard. Rather, the groups, categorised by occupancy or use, are
used to establish design criteria intended to produce specific types of performance in
Design Earthquake events, based on the importance of reducing structural damage and
improving life safety (Figure 1).
There are some structures that contain substances which, if released into the envi-
ronment, are deemed hazardous to the public. These special occupancies, Group III, are
supposed to remain in the “Immediate Occupancy” category when subjected to the
Design Earthquake, while remaining “Operational” under all frequent earthquakes. One
class lower in terms of performance requirement are Group II structures, which either
contain a large number of occupants or contain occupants whose ability to exit is
restrained. Schools, day care centres and suburban medical facilities are examples of
this category. Group I, then, are those “ordinary” structures, such as residential
buildings, which shall remain in the “Life Safety level” when subjected to the Design
Earthquake. The various performance levels required for buildings of different
categories can implicitly be met by increasing the “Design Earthquake” by a factor
272 B. Behnam and H. Ronagh
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
Figure 1. Buildings performance levels vs. earthquake severity (FEMA 450 2003).
called the “Importance factor”. The importance factor basically adjusts the intensity of
earthquake in the design so that the required performance level under the “Design
Earthquake” is met.
In buildings designed for IO performance level, it is expected that, after an earth-
quake, only minor damage will be sustained by the structural elements. These structures
are expected to remain below the level of 1% drift. By contrast, buildings designed to
meet the LS level of performance will sustain notable damage, with the values of total
drift around 2%, if hit by an earthquake at the design level (FEMA 356, 2000). Under-
standing the structural behaviour of a building is more important when a fire occurs
after earthquake as it intensifies the level of complexity. In general, fire-resistance rating
is defined as a period of time in which the integrity of a member subjected to fire is
maintained to resist applied loads (Kodur & Dwaikat, 2007). This definition correlates
with various factors, one of which is the type of the building being designed. Indeed,
the purpose is not only to provide sufficient time to evacuate people entrapped due to
the fire, but also to reduce the possibility of any conflagration (Taylor, 2003). Although
typically fire-resistance ratings are presented in national building codes, such as NRCC
(2005) and IBC (2006), many of them are provided for fire alone and not for FFE. This
is an important point, because the vulnerability of earthquake-damaged structures
exposed to FFE is much higher than their vulnerability to fire alone. This is mostly
because earthquake excitation may produce residual lateral deformation as well as resid-
ual stresses on the members (Mousavi, Kodur, & Bagchi, 2008). Moreover, experiences
from past earthquakes have shown that both active and passive fireproofing systems,
such as sprinklers and fire control systems, may become seriously damaged and as a
result, the fire-resistance capability will decrease considerably (Collier, 2005). Thus,
evaluation of a building’s performance under FFE condition is an essential design con-
sideration that must be scrutinised carefully. The FFE resistance of a building is depen-
dent on various uncertainties, some of which are the changed structural geometry and
the degradation in stiffness resulting from the earthquake. In addition, for reinforced
concrete structures, the effects of the level of damage (including tensile cracking,
removal of rebar cover and compressive crushing) on FFE resistance have to be consid-
ered as well. Assuming ductile behaviour of RC elements, a typical moment–curvature
relation can be idealised to separate stages (Kwak & Kim, 2002). While it seems that
tensile cracking, as the first stage of cracking, has no significant effect on the FFE
resistance, major cracking under tension or crushing of concrete in compression which
European Journal of Environmental and Civil Engineering 273
will result in removal of cover will drastically reduce the FFE resistance (Ervine, Gillie,
Stratford, & Pankaj, 2011).
Aligned with the above-mentioned studies and FEMA 356 performance level
definitions, in this study, a series of numerical investigations are carried out on the FFE
resistance of conventional buildings designed for IO and LS performance levels. The
study here includes a sequential analysis comprising both earthquake and the aftermath
fire using FEMA 356 descriptive performance levels while consideration is given to the
effect of the removal of cover on the FFE resistance.
quake. Assuming elastic perfectly plastic behaviour for steel and considering P-Δ effect
with P from gravity loads and Δ from the earthquake, the fire resistance was evaluated.
Further study on steel frames was carried out by Zaharia and Pintea (2009). They inves-
tigated two different steel frames, designed for two various return periods of ground
motion, 2475 and 475 years. The seismic response of the structures was then evaluated
by a pushover analysis developed by Fajfar (1996). While the frame designed for the
2475 years return period remained elastic in the pushover analysis, the weaker frame,
designed for the 475 years return period, sustained notable inter-story drift. They then
performed a fire analysis on both frames, which confirmed that the fire resistance of the
structures considering their deformed state under earthquake is notably lower than the
structures that do not have any history of deformation prior to the application of the
fire. In 2010, Mostafaei and Kabeyasawa (2010) investigated FFE resistance of a rein-
forced concrete structure with shear wall. The model was first subjected to an equiva-
lent Kobe 1995 earthquake on a shaking table. The damage sustained by the structure
was quantified both by observation and by a method they called Axial–Shear–Flexure
Interaction (Kabeyasawa & Mostafaei, 2007). These results were then used in a numeri-
cal thermal analysis to find the temperature rise in and around both the seriously dam-
aged and the intact sections subjected to fire. Figure 2 shows the temperature
distribution in an intact section and removed cover and crushed core as a results of the
earthquake load. Fire loading was then applied to the damaged structure, considering
the effect of concrete’s degraded compressive strength. The results showed that the
ability of the structure to sustain gravity loads in the damaged components was
Recently, in 2011, Ervine et al. (2011) conducted an experimental and numerical study
of a reinforced concrete element subjected to conventional loads followed by a fire
load. Applying two concentrated vertical loads on the specimen and recording the sub-
sequent deflection, intentional cracks were forced into the member. The model was then
subjected to fire load to find the effect of the created cracks on the changes of thermal
propagation inside the section. The results showed that minor tensile cracking would
not significantly change the heat penetration inside the section. They concluded that fire
resistance of the intact specimen and of the minor damaged specimen are roughly iden-
tical; however, exposing the rebar directly to fire, i.e. in the case of crushing of the
cover, considerably changes both the thermal and the structural behaviour of the speci-
men. Another study into this issue was recently undertaken by Bhargava et al. (2010)
on the fire resistance of an earthquake-damaged RC frame. In the first stage of this
study, which is currently being continued, a nearly full-scale portal frame was first
loaded by the relevant gravity loads and then subjected to a cyclic lateral load, based
on the Indian standard, in a quasi-static style. The load-controlled mode was considered
to reach 2% drift, corresponding to the LS performance level as described in FEMA
356 code (2000). The cracks’ widths were then observed using optical tools, non-
destructive tests and ultrasonic method. A computational analysis was also performed
using the finite element method with ABAQUS (2008) for making a comparison
between the test and the analytical results. The results showed a good conformity with
FEMA 356 descriptive definitions of damage levels at various performance levels, such
as IO and LS. They suggested that the results of a quasi-static cyclic test can be used
for the subsequent fire analysis.
3. Methodology
3.1. Sequential analysis
Sequential analysis is a functional tool for considering the effect of both earthquake and
fire on a structure. Figure 3 schematically shows the stages of the non-linear sequential
analysis. The first stage of loading is the application of gravity loads, which are
assumed to be static and uniform. A pseudo earthquake load then follows in a pushover
form reaching its maximum and returning to zero in a short time. The earthquake load-
ing may result in residual displacement and damaged properties of materials, which
have to be considered at the next stage of the analysis, i.e. fire.
Clearly, during this time, gravity loads are also applied. The pattern that is chosen
for the application of the earthquake load is similar to pushover analysis, with the
European Journal of Environmental and Civil Engineering 275
difference that the structure is unloaded after reaching a certain level of load. Here, it is
assumed that the maximum level of earthquake load corresponds to the LS level of
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
FEMA 356 (2000). This assumption is in line with the seismic design philosophy, in
which the performance level of structures shall not exceed the intended level when sub-
jected to the “Design Earthquake”. Therefore, the structures are pushed to these levels
and are then unloaded.
Load duration is not important for either gravity or earthquake loads; because in this
study, long-term effects such as creep and shrinkage are not incorporated in the analy-
sis. Thus, any arbitrary load duration could be chosen for the loads’ patterns. As can be
seen in Figure 3, the fire load is applied to the structure at the last stage of sequential
loading with a pattern chosen either according to ISO fire or to natural fire. Prior to fire
loading, the properties of the structures are set to the reference temperature; but during
fire, the mechanical properties change with temperature.
In order to perform the sequential analysis, use is made of the SAFIR programme
(Franssen, 2011), which is capable of sequential analysis, that does not take into
account – for the sake of simplicity – damage accumulation due to load cycling. The
background required for this analysis is provided in Sections 3.2–3.4.
seismic performance evaluation of existing and new structures (Fardis, 2007; Isaković,
Lazaro, & Fischinger, 2008). In this method, using a specific load pattern, the structure
is pushed to arrive at a displacement called the target displacement. The target displace-
ment serves as an estimate of the global displacement that the structure is expected to
experience in a Design Earthquake, often represented by the roof displacement at the
centre of mass of the roof. In this study, a vertical distribution of loads proportional to
the shape of the fundamental mode in the direction under consideration is used.
Obviously this way, the effect of higher modes is not considered in the analysis. This
effect, however, is not important for the structure under consideration as the fundamen-
tal mode is the governing mode of vibration having a natural fundamental period of 3 s
only (Fajfar, 1999). Figure 4 shows a structural frame subjected to a lateral load pattern
and a typical base shear vs. top story displacement.
Building performance is a combination of the performance of both structural and
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
Δ
Base Shear (V)
V
Roof Story Displacement ( Δ)
and LS levels of performance according to FEMA. At this level of drift, some structural
elements might go beyond the yield point in the corresponding pushover curve, while
non-structural elements might fail to operate properly, owing to mechanical failure or to
lack of amenities, such as disconnection of electricity (Behnam, 2006). Most buildings
in urban areas, such as residential and light commercial buildings, are designed to meet
the LS level of performance. The main objective at this performance level is to limit
the level of damage in the buildings and, as a result, ensure the life safety of the inhab-
itants. For this, limiting the values of drift to around 2% is recommended by FEMA
356, as the boundary of LS level of performance and CP. At the LS level, it is expected
that, along with some residual displacement in the elements, there will be considerable
damage in both structural and non-structural elements. However, adequate capacity for
carrying the applied gravity loads should remain. Obviously, buildings designed for the
CP performance level (sometimes called Limited Safety) would sustain more damage
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
compared with the other levels of performance. At this level, it is expected that the
imposed drift is more than 4%, which in turn leads to extensive damage to structural
components.
In this study, the IO and LS performance levels are considered for the seismic anal-
ysis prior to fire loading. Clearly, in structures that experience plastic deformations,
residual deformations remain in the structure and thus, the structures do not return to
their initial condition. Using the definition of lumped plasticity, the potential locations
of high plasticity are introduced by plastic hinges in SAP2000 (2002).The moment–
rotation behavior of each plastic hinge follows FEMA definitions as shown in Figure 5.
In addition, the plastic hinge length (LP) can be found via several equations. Here, Park
and Pauley’s (1975) formula is used as a simple but accurate method, where Lp = .5H
and H is the section height. The extent of damage in the beams and columns are also
taken from FEMA 356. These definitions in a concrete cross section are required for
the post-earthquake fire analysis, because variation of temperature across the section is
highly dependent on the state of damage (Ervine, Gillie, Stratford, & Pankaj, 2012). In
FEMA 356, it is assumed that, while in IO performance level only minor cracking will
occur in the elements with insignificant effect on FFE resistance, in the LS performance
level, extensive damage will be observed in beams, while ductile columns will sustain
spalling of their cover. Whereas the interpretation of “extensive damage” is left to the
reader in FEMA, the assumption here has been the further penetration of the fire fron-
tier to behind the outermost reinforcing bars, unlike columns where the fire is assumed
to penetrate to the edge of the concrete cover as it spalls according to FEMA. The
dotted lines and the arrows in Figure 6 show the assumed pattern of the applied fire
frontier for damaged beams and columns after the pushover analysis. This assumption
is based on the authors’ interpretation of the information available in the FEMA 356
code, the Japan Building Disaster Prevention Association (JBDPA) and an experimental
study performed by Bhargava et al. All of the mentioned past studies point to the fact
that the concrete cover does not remain part of the section after earthquake. In FEMA
356, “Table C1–3 Structural Performance Levels and Damage”, the different levels of
damage in columns and beams are explained. While at the LS level of performance, the
cover of the columns is spalled, it states that more damage is observed in the beams. In
a similar manner, but rather in terms of quantity than quality, an experiment was per-
formed by Bhargava et al. (2010) on a nearly full-scale RC frame in order to find the
level of damage when the frame was pushed to a certain level of displacement. Results
showed that, while at a roof drift ratio of 1.37% flexural cracking was observed (which
corresponds to the drift ratio of the IO level of performance), at 2.11% drift ratio
278 B. Behnam and H. Ronagh
(corresponding to the drift ratio of the LS level of performance), spalling and wide
cracks in columns and beams were observed. The study does not reveal any differences
between the columns and the beams. JBDPA presents similar observations (Meada &
Kang, 2009; Nakano, Maeda, Kuramoto, & Murakami, 2004).
Considering the above, in this study, it is assumed that the damaged parts of the
structure are limited to the extent of plastic hinges, the position of which is determined
by the pushover analysis while its length is determined using the approach of Park and
Pauley (1975) as stated previously.
Figure 8. Fire pattern according to ISO 834 and ASTM E119 (ASTM, 2006; ISO 834, 1999).
Temperature
Time
Ignition Heating Cooling
4. Case study
Two reinforced concrete frames (A and B) with identical geometry but designed for
two levels of performances, IO and LS, are selected as shown in Figures 10 and 11.
Frame A is loaded as if it were a school (corresponding to the IO level of performance)
and Frame B as if it were a residential building (corresponding to the LS level of per-
formance). This assumption is then needed for calculating the fire load density. The
floor and the ceiling are made of normal weight concrete and the compartment parti-
tions are built with standard bricks. Table 1 shows the thermal characteristics of the
materials in the building and Figure 10 the opening dimensions of the case study.
European Journal of Environmental and Civil Engineering 281
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
Figure 10. The case study and the position of the openings.
The selected frames are loaded and designed for Peak Ground Acceleration of .30 g
and based on ACI 318-08 code. The properties of the designed frames are presented in
Figure 11. The selected frames are made of normal strength concrete with compressive
strength of 25 MPa and longitudinal and transverse reinforcing bars with a yield stress
of 400 MPa. The frames are dimensioned for the load combinations of 8.0 kPa for dead
load and 2.5 kPa for live load. Combinations of 100% dead load and 20% live load for
the designed frame in the LS level of performance and 40% live load for the designed
frame in the IO level of performance are used to find the required mass for calculating
the earthquake load (ACI 318, 2008). In order to consider thermal actions, two methods
previously mentioned in Section 2.5 are used here. For the thermal analysis, it is
assumed that the concrete moisture content is 2%. Moreover, the thermal expansion
coefficient of rebar and concrete are assumed to be 12 106 and 10 106/°C,
respectively. Poisson’s ratio of .2 is considered for the concrete. In order to improve
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
our understanding of the behaviour, fire analyses are also performed on the undeformed
frames, i.e. before the occurrence of earthquake.
In which qt,d (MJ/m2) is the fire load of the compartment, qf,d is the occupancy of
the fire compartment and Aroof (m2) and At (m2) are the areas of the roof and the total
area consisting of the area of walls, roof and floor, respectively.
where δq1 is the risk of fire activation, which increases while the floor area increases.
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
For this study, δq1 is 1.54, and δq2 is the type of occupancy (which for the purpose of
this study is 1.0). Combustion factor, shown by m in Equation (2), is between 0 and 1,
and is derived from experimental results. It is assumed to be .8 for most cellulosic
materials. qf,k or the so-called “characteristic fire load density” can be calculated using
the values given in Annexe E of Eurocode 1, which for residential buildings and
schools is 948 and 347 MJ/m2, respectively. In order to take into account the effect of
active fire fighting, the factor δn is applied in Equation (2). This factor is based either
on available fire-extinguishing systems established in the building, such as detection
systems and sprinklers, or on those used by professional fire brigades and rescue teams.
The proposed values are based on a normal condition and, as such, cannot be utilised
in an FFE scenario where all of the response teams are under stress. Consequently, the
mentioned factors shall be modified for this study. Here, based on the high possibility
of disturbance of both internal facilities (such as sprinkler and detection systems) and
urban facilities as a result of earthquake, such modifications are made, as shown in
Table 2.
It is worth noting that the time–temperature curve based on the natural fire method
relies on several factors, some of which are: thermal characteristics of materials, type of
occupancy, total area of compartment and size and shape of openings. Accordingly, fire
behaves as ventilation-controlled or fuel-controlled. In most small and medium-size
compartments, fire is governed by a ventilation mode, which means that the growth of
a fire is limited by the availability of air. By contrast, in large compartments, fires are
mostly fuel-controlled, which means that the growth of a fire is limited by the availabil-
ity of combustible materials (Zehfuss & Hosser, 2007). The application of the method
starts with defining several factors as follows:
• Wall factor b which accounts for the thermal properties of the enclosure
pffiffiffiffiffiffiffiffi
b¼ cqk ð3Þ
where c, ρ and λ stand for specific heat (J/kgK), density (kg/m3) and thermal conductiv-
ity (W/mk), respectively. When different materials are used in a compartment, an aver-
age b factor is used as represented in Equation (4). Where, Ai and bi are related to each
layer. Using the information given in Table 1, b factor of the compartment shown in
Figure 10 is, therefore, calculated, which is 1653.2 J/m2S.5 K.
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
284
Automatic An
water Automatic onsite Safe Normal fire Smoke
B. Behnam and H. Ronagh
Types of Fire extinguishing Water fire Automatic Automatic fire An off-site fire access fighting exhaust
occupancy situation systems supply detection fire alarm transmission brigade brigade routes devices systems Sum
δ1 δ2 δ3 δ4 δ5 δ6 δ7 δ8 δ9 δ10 Σ δn
School Normal .61 .7 1.0 1.0 .87 .61 .71 1.0 1.0 1.5 .37
condition
After EQ 1.0 1.0 1.0 1.0 .87 1.0 1.0 1.5 1.5 1.5 4.52
Residential Normal 1.0 1.0 1.0 1.0 1.0 1.0 .71 1.0 1.0 1.5 1.64
condition
After EQ 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.5 1.5 1.5 5.20
European Journal of Environmental and Civil Engineering 285
P
bi Ai
b¼ P ð4Þ
Ai
• Parameter O that accounts for the opening in the vertical walls using Equation (5)
pffiffiffiffiffiffi
O ¼ Av heq =At ð5Þ
where Av is the area of the openings, At is the total area of the enclosure (walls, ceiling
and floor including the opening) and heq is the average height of the openings and is
accounted for using Equation (6). For the compartment here, the calculated opening
factor is .212.
X X
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
C ¼ ½ðO=:04Þ=ðb=1160Þ2 ð7Þ
When Γ is more than 1, the temperature in the heating phase increases sharply and is
always more than that of ISO 834 fire curve. However, the duration of the heating
phase is often short. By contrast, for Γ smaller than 1, the temperature in the heating
phase is lower than that of ISO 834, with a longer duration. Using the above calcula-
tions, the expansion factor of 13.86 is thus accounted for.
• Defining the two parameters of the shortest and the longest possible heating
phase, which are commonly shown with tlim and tmax, respectively. The shortest
possible duration of heating phase is dependent on the fire growth rate, from
slow to fast. Here, assuming a medium fire growth a value of .333 h is adopted
as mentioned in Eurocode1. For the longest time, tmax, Equation (8) is used.
Based on different conditions – before and after earthquake – shown in Table 2, the
qt,d and tmax parameters are consequently accounted for. If tmax > tlim, then the fire is
ventilation-controlled, otherwise the fire is fuel-controlled. For the compartment consid-
ered here and using the performed calculations, tmax of the residential building and for
both situations, before and after earthquake, is higher than tlim. Therefore, the fire is
ventilation-controlled. For the school building, however, tmax is lower than tlim at nor-
mal situation, i.e. before earthquake, which means the fire is fuel-controlled. Neverthe-
less, the fire is ventilation-controlled after earthquake, as tmax is larger than the assumed
tlim. It is also worth mentioning that for the situation in which tmax < tlim, some modifi-
cations have to be made to the opening factor (O) and the expansion factor (Γ), as
explained in Eurocode1. Finally, the temperature during the heating and then cooling
phases are accounted for as explained in the mentioned code. Figure 13 shows time–
temperature curves for the case studies in two different situations, before and after
earthquake. The time–temperature based on ISO834 is also shown to draw a
comparison between the results of natural fire curves.
286 B. Behnam and H. Ronagh
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
5. Results
Sectional dimensions of the beams and columns for the frames were provided previ-
ously, in Figure 11. All material properties were also introduced in previous sections.
The sequential analysis comprises three main stages, which are: the gravity loading, the
seismic pushover and finally the FFE. In seismic analysis, the structure is subjected to a
monotonically increasing lateral load. Indeed, the structure is pushed to a certain level
of displacement. According to the philosophy of design based on performance, it is
expected that structures designed for a specific level of performance remain at the
assumed level when subjected to the “Design Earthquake”. Accordingly, for the two
different levels of performance, i.e. IO and LS, two different pushover analyses are
needed. Using the FEMA356 procedure, a target displacement is calculated for control-
ling the response of the structure for the assumed performance level. In this respect,
SAP2000 programme is used for the pushover analysis. Moreover, FEMA procedure is
used to define hinge positions and properties. The results show that none of the hinges
go beyond their defined levels of performance. The lateral forces corresponding to the
target displacement extracted from the SAP2000 programme are then input to the
SAFIR programme for performing sequential analysis. Figure 14 shows the pushover
curves for the mentioned performance levels resulting from SAP2000 and used for the
sequential analysis in SAFIR.
For the final stage of the sequential analysis, a FFE is applied to the structure. Two
different scenarios are used for the fire analysis for both damaged and undamaged
frames, using both the ISO834 curve and the natural fire curve. In case one, the undam-
aged structure is exposed to the fire load, while in case two, the damaged structure is
exposed to the fire load. In other words, in the first case, the fire follows the applied
gravity loads, but in the second case, the fire follows the gravity and the earthquake
loads. Figure 15 shows the temperature distribution due to exposure to fire on the two
European Journal of Environmental and Civil Engineering 287
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
types of applied sections, using the ISO model. It is mentionable that according to the
work performed by Ervine et al. (2012), minor cracking has no significant effect on the
heat penetration. As at the IO level of performance, only minor cracking occurs, the fire
frontiers are then applied using assumed pattern in Figure 6(a).
Figure 16(a) and (b) show the fire resistance based on the ISO model in both situa-
tions, i.e. fire alone and FFE, and for the defined performance levels. The sharp
increase and then decrease in FFE analysis is due to the structures being first pushed to
a certain level of displacement and then unloaded. Some degree of damage would now
exist in the structures. The damaged structures are then loaded with fire as a sequential
load, which arrives at the structure in its residually deformed state. To do this, SAFIR
allows a function to be written inside its computing environment that allows the import-
ing of pushover loads extracted from SAP at the target displacement. These loads then
applied inside the SAFIR environment to the structural model bring the structure to the
target displacement first, after which unlading takes place continued by reloading with
fire.
The fire resistance is defined as the time at which the displacements, either globally
(i.e. the drift of a certain point) or locally (i.e. the deformations at the middle of a
beam), go beyond the chosen thresholds. The thresholds have been identified by the
Figure 15. Distribution of temperature inside a column and using ISO 834 model 9.
288 B. Behnam and H. Ronagh
curve for displacements vs. time step merging towards the horizontal asymptote by a
1% error. In other words, a member is considered to have failed when it is unable to
resist the initially applied gravity loads (Kodur & Dwaikat, 2007), which is assumed to
be the case when the response curve becomes too flat. As shown in Figure 16, there is
a correlation between the fire-resistance rating and the performance levels, so that the
fire resistance of the designed frame at the higher performance level, i.e. IO, is more
than that of the frame designed for the lower performance level, i.e. LS. In other words,
the FFE resistance is influenced by the intensity of damage induced in the frame during
the prior earthquake. On the other hand, while the collapse time for FFE-IO level is
around 80 min, this increases considerably to about 4 h in the fire-only situation, repre-
senting a 300% difference in the fire resistance. By contrast, while the fire resistance of
the undamaged frame designed for LS level is about 4 h, it reduces significantly to
around 1 h in case of FFE, signifying a 400% reduction in the fire resistance.
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
The fire resistances found based on the natural fire model for both performance lev-
els, i.e. IO and LS, are roughly similar to those based on the ISO834 model, as shown
in Figure 17. The figure shows that the fire resistances of both models decline consider-
ably when the models have been previously subjected to lateral loading; however, a
notable difference between the results of FFE resistance based on ISO model and natu-
ral fire model is observed in the frame designed for LS performance level. While the
FFE resistance found based on the ISO model is about 1 h, it is about 40 min based on
the natural fire model.
It is worth noting that two types of collapse, global collapse and local collapse,
were observed during the analyses. While global collapse is defined as the collapse con-
figuration in which the frame fails because of considerable lateral movement of col-
umns, local collapse involves mainly failure of the beams. In the studied frames, at
both performance levels, i.e. IO and LS, and in the case of fire after earthquake, global
collapse occurred. However, local collapse occurred when the elements were subjected
to fire alone. Figure 18 schematically shows the two types of collapse failure configura-
tions mentioned above.
Figure 16. Fire resistance based on ISO 834 curve (first story lateral displacement).
European Journal of Environmental and Civil Engineering 289
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
Figure 17. Fire resistance based on natural fire curve (first story lateral displacement).
6. Conclusion
FFE is a problematic situation, which has received inadequate attention in the past.
Investigating the effects of FFE on structures classified as “Ordinary” (such as educa-
tional, commercial, residential and so on) is important, as these buildings comprise a
major part of urban-built environment. Design based on performance requires these
buildings to remain within the “Life Safety” level of response under a Design
Earthquake. These structures, however, are more vulnerable if loaded with fire after
they have been weakened to some extent by a prior earthquake. A sequential-based
non-linear analysis is proposed in this study, for the fire- following-earthquake analysis
of structures designed for IO and LS levels of performance. In this respect, two
reinforced concrete frames were selected and designed for two different occupancy
290 B. Behnam and H. Ronagh
purposes, a school and a residential building. The structures were then pushed to the
maximum allowable inter-story drift, which is assumed to satisfy the IO and LS perfor-
mance levels. Pushover curve was then extracted for use in the subsequent analysis.
Sequential loading, consisting of gravity and lateral loads followed by fire (ISO834
model and natural fire model), was a key aspect of the study conducted using SAFIR
software. In SAFIR, the P-Δ effect and the residual lateral deformation as well as degra-
dation in stiffness were considered. Defining the damaged sections (in terms of spalling
in cover and such) in the thermal analysis was an additional factor considered in the fire
analysis. The patterns of damage were drawn from the descriptive definition of
FEMA356 and other numerical and experimental studies as mentioned earlier, and for
buildings designed for different performance levels. Accordingly, the following remarks
can be made:
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
• As results show, the structures fail prior to the fire reaching the decay phase in
the natural fire model. The main issue with the ISO model not having a decay
phase is, therefore, not a weakness when compared with the Natural model here.
Considering that the assumptions made for the natural fire model are only
assumptions and not facts that can be relied upon to draw general conclusions,
choosing of ISO standard curve seems more reasonable for analyses similar to
the one made here.
• Two types of collapse mechanisms were observed during the fire analyses, these
being global collapse and local collapse. While local collapse occurred in the
beams, global collapse was represented mostly by considerable lateral movement
in the columns. Interestingly, but not unexpectedly, majority of fire-only analyses
resulted in local collapse, while all FFE analyses resulted in global collapse.
• As emphasised earlier, FFE is a significant problem, which requires that further
studies be performed, either numerically or experimentally, on different levels of
structures subjected to FFE, in order to develop a better understanding of the issue.
References
ABAQUS. (2008). 6.8. Providence, RI: Dassault Systemes Simulia Corporation.
ACI 318. (2008). Building code requirements for structural concrete (ACI 318-08) and commen-
tary. New York, NY: American Concrete Institute.
Agency, F. E. M. (2000). FEMA356, in rehabilitation requirements (p. 36). New York, NY:
Federal Emergency Management Agency and American Society of Civil Engineering.
ASCE. (2006). Minimum design loads for buildings and other structures, in SEI/ASCE 7-0.5.
New York, NY: American Society of Civil Engineers.
ASTM. (2006). Standard test methods for determining effects of large hydrocarbon pool fires on
structural members and assemblies, in ASTM E1529-06. New York, NY: American Society
for Testing and Materials.
Bamonte, P., Gambarova, P. G., & Meda, A. (2008). Today’s concretes exposed to fire – test
results and sectional analysis. Structural Concrete, 9, 19–29.
Behnam, B. (2006). Retrofitting management for residential buildings, in faculty of civil engineer-
ing (p. 195). Tehran: Tehran Polytechnic.
Bhargava, P. Sharma, U. K., Singh, Y., Singh, B., Usmani, A., & Torero, J. (2010). Fire testing
of an earthquake damaged RC frame. In Sixth International Conference, Structures in Fire,
DEStech Publications, Inc., Lancaster, Pennsylvania, USA.
Borden, F. (1996). The 1994 Northridge earthquake and the fires that followed. In Thirteenth
meeting of the UJNR panel on fire research and safety. New York, NY: National Institute of
Standards and Technology.
European Journal of Environmental and Civil Engineering 291
British Standard. (2002). Basis of structural design, in Eurocode0. Dublin: National Standards
Authority of Ireland.
Buchanan, A. (2001). Structural design for fire safety (p. 417). New York, NY: John Wiley and
Sons.
California Seismic Safety Commission. (1996). Seismic evaluation and retrofit of concrete buildings
(ATC 40), in chapter 2, overview. New York, NY: California Seismic Safety Commission.
Collier, P. (2005). Post earthquake performance of passive fire protection systems. Sydney: Build-
ing Research Levy and the Department of Building and Housing.
Della Corte, G., Landolfo, R., & Mazzolani, F. M. (2003). Post earthquake fire resistance of
moment resisting steel frames. Fire Safety Journal, 38, 593–612.
Ervine, A., Gillie, M., Stratford, T. J., & Pankaj, P. (2011). Thermal diffusivity of tensile cracked con-
crete. In International Conference Applications of Structural Fire Engineering (pp. 97–102), Prague.
Ervine, A., Gillie, M., Stratford, T. J., & Pankaj, P. (2012). Thermal propagation through tensile
cracks in reinforced concrete. Journal of Materials in Civil Engineering, 24, 516–522.
Faggiano, B., & Gregorgio D. (2010). Assessment of the robustness of structures subjected to fire
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
Mohammadi J., & Alysian S. (1992). Analysis of post-earthquake fire hazard. Earthquake
engineering, Tenth World Conference. Rotterdam.
Monti, G., & Spacone, E. (2000). Reinforced concrete fiber beam element with bond-slip. Journal
of Structural Engineering, 126, 654–661.
Mostafaei, H., & Kabeyasawa T. (2010). Performance of a six-story reinforced concrete structure
in post-earthquake fire. In 10th Canadian Conference on Earthquake Engineering, Institute for
Research in Construction, Toronto, Ontario.
Mousavi, S., Kodur, V. K. R., & Bagchi, A. (2008). Review of post earthquake fire hazard to
building structures. Canadian Journal of Civil Engineering, 35, 689–698.
Nakano, Y., Maeda, M., Kuramoto, H., & Murakami, M. (2004). Guidelines for post-earthquake
damage eveluation and rehabilitation of RC buildings in Japan. In 13th World Conference on
Earthquake Engineering, Vancouver, Canada.
National Research Council Canada. (2005). National fire code in buildings. Ottawa: National
Research Council Canada.
Park R., & Paulay. T. (1975). Reinforced concrete structures (p. 769). New York, NY: John Wiley
Downloaded by [UOV University of Oviedo] at 03:12 30 October 2014
& Sons.
Ranzo, G., & Petrangeli, M. (1998). A fibre finite beam element with section shear modelling for
seismic analysis of RC structures. Journal of earthquake engineering, 2, 443–473.
Remesh, K., & Tan, K. H. (2007). Performance comparison of zone models with compartment
fire tests. Journal of fire sciences, 25, 321–353.
SAP2000-V14. (2002). Integrated finite element analysis and design of structures basic analysis
reference manual. Berkeley, CA: CSI.
Scawthorn, C., Eidinger, J. M., & Schiff A. (2005). Fire following earthquake. New York, NY:
American Society of Civil Engineers.
Spacone, E., Filippou, F. C., & Taucer, F. F. (1996). Fibre beam-column model for non-linear
analysis of R/C frames: Part I. Formulation. Earthquake Engineering and Structural
Dynamics, 25, 711–726.
Tanaka, T. (1998). Performance-based fire safety design of a high-rise office building. Tokyo:
Building Research Institute.
Taylor, J. (2003). Post earthquake fire in tall buildings and the New Zealand building code, in
civil engineering (p. 160). Christchurch: University of Canterbury.
Thomson, A. (2011). Japan disaster: Kesennuma hit by quake, tsunami and fire. Retrieved 2011
from http://www.channel4.com/news/kesennuma-quake-tsunami-and-now-fire
Youssef, M. A., & Moftah, M. (2007). General stress-strain relationship for concrete at elevated
temperatures. Engineering Structures, 29, 2618–2634.
Zaharia, R., & Pintea, D. (2009). Fire after earthquake analysis of steel moment resisting frames.
International Journal of Steel Structures, 9, 275–284.
Zehfuss, J., & Hosser, D. (2007). A parametric natural fire model for the structural fire design of
multi-storey buildings. Fire Safety Journal, 42, 115–126.