Review

Advances in Structural Engineering

1–17
Research review: Post-earthquake fire Ó The Author(s) 2017
Reprints and permissions:
assessment of steel buildings in the sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1369433217711617

United States journals.sagepub.com/home/ase

Rachel Chicchi and Amit H Varma

Abstract
Researchers and practitioners have traditionally analyzed building hazards independently; however, with the recent focus on resilience,
a multi-hazard analysis approach has emerged which considers the implications of cascading hazards. This article reviews the state of
the art for analysis and design of steel moment frame buildings subjected to fires following earthquakes, also known as post-
earthquake fires. The current design and analysis approaches for fire and seismic hazards in the United States, which follow prescrip-
tive approaches, are each explained. Performance-based methodologies for each hazard are described. A literature review pertaining
to the system behavior of buildings exposed to post-earthquake fires is provided. This includes consideration of nonstructural damage
caused from earthquakes that could in turn affect building fire performance. Finally, recommendations are made for future post-
earthquake fire analyses and design using incremental dynamic analyses and incremental fire analyses to capture building performance
when subjected to various levels of these cascading hazards. The procedure for this methodology is explained, providing a direction
for future research.

Keywords
fire, moment frame, post-earthquake fire, seismic, state of the art, steel

Introduction Each of these aspects will be discussed in the article:

(1) fire hazards will be reviewed, covering the current fire
Buildings are designed to resist earthquake loads by design approach and the state of the art in determining
dissipating energy through inelastic behavior. This can the fire scenario, conducting heat transfer analyses, and
cause structural and nonstructural damage and resi- development of incremental fire analyses; (2) seismic
dual drifts within the building. Ground motions can hazards are discussed pertaining to the current design
also lead to fire ignitions through short-circuiting, approach, typical analysis procedures, determining the
abrasions, chemical reactions, and other causes seismic hazard, and development of incremental dynamic
(Scawthorn, 2008). These fires result in strength and analyses; (3) the state of the art of PEF analyses will be
stiffness reductions in structural members due to ele- reviewed, including implications of nonstructural dam-
vated temperatures, which can further exacerbate the age; and (4) recommended future steps and research
seismic damage. The interdependencies between earth- needs will be explained. This includes a methodology for
quake and fire damage have motivated this study on assessment of steel buildings exposed to PEFs.
system behavior, analysis, and design of steel build-
ings. This article will review the current state of the art Fire hazards
in analysis and design for post-earthquake fire (PEF)
hazards subjected to steel moment frame buildings. Current fire design approach
While some important work on component and assem- Structural engineers in the United States do not usually
bly behavior will be reviewed, this article will focus pri- conduct fire analyses or design; instead, architects are
marily on system-level behavior.
An understanding of building response to earth- Lyles School of Civil Engineering, Purdue University, West Lafayette, IN,
quake and fire as separate hazards is necessary to fur- USA
ther evaluate buildings subjected to PEFs. Fire and
Corresponding author:
seismic hazards require different analysis and design Rachel Chicchi, Lyles School of Civil Engineering, Purdue University, 550
approaches and result in different potential failure W Stadium Ave, West Lafayette, IN 47907, USA.
mechanisms within the structure. Email: [email protected]
2 Advances in Structural Engineering 00(0)

typically responsible to specify the fire-resistance rating Determining the fire hazard. In the performance-based
requirements and determine the fire protection neces- approach, the fire hazard is considered a thermal load
sary to achieve that rating. The International Building applied to the structure and is referred to as a design-
Code (IBC) specifies fire-resistance ratings for struc- basis fire (Kodur et al., 2011). Design-basis fires are
tural components and assemblies based on the building typically classified as either localized or compartment
use, size, and combustibility of building materials. This fires. Localized fires do not cause flashover because of
rating is the time (in hours) that an element or system the low rate of released heat. Because this study
can be exposed to a standard fire before failure would focuses on global response, only compartment fires,
likely occur. Table 601 in IBC provides these hourly which are large, post-flashover fires, will be consid-
resistance ratings, which vary for primary members, ered. Determination of this load can be approached in
floor and floor secondary beams, roof and roof sec- a number of ways: computational fluid dynamics
ondary beams, bearing walls, and nonbearing walls (CFD) or two-zone models can be developed, time-
(IBC, 2011). Primary members are defined as columns temperature curves from a standard can be used, or
and members with direct connections to columns, as actual fire tests can be conducted.
well as bracing members necessary for stability. There CFD models involve modeling the growth and
is no distinction between gravity and lateral frames. behavior of the fire by dividing the compartment into
The strength and stiffness of steel is greatly reduced many different zones to reflect the different tempera-
at elevated temperatures, which may require passive tures throughout the space. These models are highly
fire protection measures to increase the fire-resistance complex and require a number of detailed assumptions
rating. Spray-applied fire-resistive materials (SFRM) of materials and properties within the compartment.
are commonly applied to steel structures. Fire tests can The National Institute of Standards and Technology
be conducted using ASTM E119 to determine the provides software called FDS (Fire Dynamics
required thickness of SFRM (ASTM, 2015). However, Simulator), which can be used to conduct CFD analy-
a more common, alternative approach is to reference a ses. Other programs also exist that specifically focus
database of tests that have been conducted on a limited on CFD for fires.
variety of steel shapes and assemblies. Underwriters As an alternative, time–temperature curves from
Laboratory’s database is commonly used for this ASTM E119 (ASTM, 2015), ISO 834 (ISO 834-1:1999,
approach (UL, 2016). Each wide flange beam has a 2015), and Eurocode 1 (EN 1991-1-2:2002, 2002) are
W/D value, where W is the weight per linear foot and used. As shown in Figure 1, the ASTM and ISO curves
D is the perimeter of the member exposed to the fire. only have a heating phase. These curves are commonly
The W/D for the tested beam is divided by the W/D used for fire furnace testing and are not influenced by
for the beam being designed and the thickness of the ventilation or other factors that would affect an actual
fireproofing used in the test is scaled by that amount. fire. In contrast, Eurocode parametric curves include a
The AISC Design Guide 19 also contains fire test cooling phase and vary depending on the thermal iner-
results and example problems for determining fire- tia of the enclosure (b), opening factor (O), and fire
proofing thicknesses (Ruddy et al., 2003). load density (qt,d). Varying these parameters affects the
peak fire temperature, fire duration, and rate of heat-
State of the art of fire analysis procedures ing and cooling. This cooling phase is important as it
results in thermal contraction, which can produce large
The prescriptive approach of fire-resistance ratings per
IBC, which is based on standard furnace tests of short
span members, does not necessarily translate well into
real building behavior. Following the World Trade
Center collapse in 2001, this approach has been further
scrutinized with many professionals calling for a
change to performance-based fire-resistance design and
for the structural engineer to take over the responsibil-
ity of fire-resistance design of the structure through
conducting fire analyses (Kodur et al., 2011). This tran-
sition within the design industry has been slow to
develop; nevertheless, within the realm of research, var-
ious fire analyses have been conducted. The following
sections will highlight the typical procedure for con-
ducting these analyses using the following models: fire, Figure 1. Design fires using ISO834, ASTM E119, and
heat transfer, and structural. Eurocode.
Chicchi and Varma 3

tensile forces that fail connections. The Eurocode para- fire hazard, this assumption no longer applies and 3D
metric time–temperature curves are restricted to room FEM models are used to conduct heat transfer
fires in the post-flashover phase intended for use in analyses.
spaces with rectangular enclosures, floor area less than In FEM heat transfer models, the structural mem-
500 m2, ceiling heights less than 4 m, and no ceiling ber and its corresponding fireproofing is modeled and
openings. In this post-flashover phase, it is assumed exposed to the predetermined fire hazard. Thermal
that the room contains a fully developed fire with uni- expansion, specific heat, thermal conductivity, and
form temperature throughout the compartment. This density are defined for each material in order to model
is called a one-zone approach. the thermal transfer of heat from the air to the struc-
While Eurocode assumes a one-zone approach, tural component. These models perform conduction,
designers recognize that there are actually at least two convection, and radiation calculations to determine
zones: an upper, hot zone and a lower, cooler zone. the internal temperatures of the structural member
Structural members in the lower zone (such as the floor along its cross-section. These internal temperatures are
slab below the fire) are not usually analyzed, as the determined at specific nodes, which are then applied to
change in internal temperatures in those members is the structural building model.
expected to be insignificant (Franssen and Vila Real, More simplified analytical methods, such as the
2012). ‘‘lumped mass method,’’ can also be employed, which
Pope and Bailey (2006) conducted comparisons assumes that the entire member cross-section has the
among CFD models, Eurocode, and fire tests and same temperature. This can be a valid assumption for
determined that Eurocode provides reasonable predic- steel thicknesses less than 100 mm and when exposed
tions for average compartment temperatures, though it to a sudden rise in temperature (Wang et al., 2013).
over predicts the growth phase of the fire and should The AISC Specification for Structural Steel Buildings
include a nonlinear decay rate. Additionally, CFD (ANSI/AISC 360-10, 2010) allows designers to use this
analysis results can be too complex to apply to the simplified assumption.
heat transfer structural models. For these reasons and
due to its simplicity, standard fire curves are often used
in favor of CFD analyses (Wang et al., 2013). Case studies of structural analyses. Many researchers have
Determination of active fire protection, such as focused on studying individual structural components
sprinklers, detectors, and smoke exhaust systems, must to observe behavior when subjected to a fire. In
be considered as well. Eurocode (EN 1991-1-2:2002, particular, beam-to-column connections have been
2002), for example, allows a reduction in the fire sever- studied at length (Al-Jabri et al., 2006; Fischer and
ity if sprinklers are installed; however, many designers Varma, 2017; Garlock and Selamet, 2010; Hu and
choose to not account for this reduction and conserva- Engelhardt, 2011; Mahmoud et al., 2016; Memari and
tively assume that the sprinklers are defective. Once the Mahmoud, 2014; Pakala et al., 2012; Qian et al., 2008;
fire severity is determined, the location of the fire also Sarraj, 2007; Tan and Huang, 2005; Wang et al., 2011;
needs to be decided on: the story or stories affected, Yang et al., 2009; Yu et al., 2009). Beams, columns,
full-story or compartment fires, interior or exterior and floors have also been studied in isolation
compartments, moving or stationary fires, and so on. (Agarwal et al., 2014; Agarwal and Varma, 2011,
All these factors affect the fire hazard and the building 2014; Choe et al., 2011; Dwaikat et al., 2011; Kodur
response. et al., 2013; Naser and Kodur, 2016; Selamet and
Garlock, 2012; Selden et al., 2016; Selden and Varma,
2016; Takagi and Deierlein, 2007; Zhao and Kruppa,
Heat transfer analyses. Once the fire time–temperature 1997). While the above-mentioned research informs
curve has been selected, finite element method (FEM) modeling decisions, system-level behavior will be
models can be used to conduct heat transfer analyses focused on.
to determine the temperature of each structural com- Very few full-scale fire tests have been conducted
ponent throughout its cross-section. Two-dimensional due to the associated costs. One of the most familiar
(2D) heat transfer analyses are commonly used series of experiments is the Cardington fire tests, which
because the fire time–temperature curves assume that consisted of six full-scale fire tests on an 8-story struc-
the room contains a fully developed fire with uniform ture in Bedfordshire, United Kingdom (STC, 1999).
temperature throughout the compartment; thus, there The observations of these large-scale tests helped to
is no need to use more computationally expensive inform and benchmark computational modeling of
three-dimensional (3D) modeling. When using a CFD fires. Beams experienced significant deflection, which
4 Advances in Structural Engineering 00(0)

led to catenary action but no instability or collapse. Flint et al. (2007) and Usmani et al. (2003) studied the
The bottom flange of beams were often distorted due World Trade Center collapse using 2D FEM models.
to thermal expansion and deflection when pushed They found that, as the floor deflected due to thermal
against the column. Fracture in the end-plate beam to expansion, tensile membrane action occurred. The
column connection occurred during cooling, due to exterior columns were pulled inward, forming plastic
thermal contraction causing high tensile forces in the hinges in these columns at the floor levels. Similar
connection. The top of columns, near connections responses were found when using 3D models.
where fireproofing was not present, resulted in loca- However, redistribution of loads throughout the struc-
lized buckling failure. ture was observed with the 3D models, making it a
Agarwal and Varma (2014) studied the system-level more robust model and slower to fail than the 2D
performance of a steel moment frame building using models (Flint, 2005). It is important to note that the
3D FEM models. They found that if all structural com- truss system of the World Trade Center is different
ponents were designed for the same level of fire safety, than that of traditional floor framing with conven-
gravity columns would likely fail first. Upon failure of tional hot rolled steel shapes.
the columns, catenary and flexural action redistribute In another study, vertically traveling fires were
load to the adjacent columns. Fang et al. (2011) found simulated to account for the time that it takes for
that, depending on the loading, it is even possible for actual fires to move between floors (between 6 and
loads to be redistributed to upper, ambient tempera- 30 min based on observations of actual buildings;
ture floors. Röben et al., 2010). This is different than the approach
Fischer (2015) expanded upon Agarwal’s work by of modeling multiple floor fires at once because it
studying full-story fires and varying the fire-resistance accounts for the heating and cooling response of the
ratings on the structural members. Again, gravity col- floors relative to each other. The rate of the moving
umns were the first component to fail. When these fires greatly affected the global response of the
members were protected with excess fire protection, structure.
significant deflections occurred in the beams and slab
but failure did not occur. Moving fires were also con-
sidered, but it was found that full-story fires were an Incremental fire analysis. There is a trend in fire research
appropriate conservative approach in place of model- that favors a performance-based approach that is
ing moving fires. probabilistic as opposed to deterministic (Khorasani
Memari and Mahmoud (2014) explored moment et al., 2015a; Kodur et al., 2011; Rush et al., 2014).
frames with reduced beam section connections for 3-, Due to this initiative, the incremental dynamic analysis
9-, and 20-story frames using 2D modeling. Gravity (IDA) approach that is typically used in earthquake
frames were idealized as a leaning column. They found engineering, which will be explained in a later section,
that the global stability of the structure was not com- is being adapted for fire scenarios.
promised by only a one compartment fire. At a local This type of analysis involves exposing a structure
level, beams experienced residual axial tensile forces to a hazard and incrementally increasing the intensity
and deflections. measure (IM) of the hazard with each analysis. Once
Jiang et al. (2014) conducted 2D frame analyses to the demand is determined through scaling of the IM,
observe various collapse mechanisms: heated bay col- the capacity is evaluated through a damage measure
lapse, column buckling, local lateral drift of heated (DM). This is the output from the analysis which is
floor, and global lateral collapse. This study also com- used to determine the suitability of the structure. The
pared the influence of beam sizes on the structural IM and DM are incorporated to generate response
response. Additionally, the magnitude of gravity load- curves that show how the level of damage changes as
ing was varied. They determined that local lateral drift intensity varies.
occurred at low loading levels; with increased loading, This concept is articulated by Moss et al. (2014),
column buckling occurred. As the beam sections which named the approach incremental fire analysis
increased, column failure mechanisms occurred instead (IFA) and performed preliminary studies to validate
of beam mechanisms. Additionally, edge bays were the process. Unlike IDA, which commonly uses peak
more susceptible to progressive collapse because these ground acceleration (PGA) or the spectral acceleration
bays were not able to develop adequate catenary as the IM, there does not seem to be a consensus
action. Similar analyses and findings were determined among researchers on the most indicative IM to use
by Sun et al. (2012). for fire.
Some studies have shown that thermal expansion Moss et al. studied a two-span concrete beam using
(bowing) usually controls the behavior over that of both peak room temperature and the total radiant heat
material degradation due to the elevated temperatures. energy (RHE) as the IMs. The total RHE is the
Chicchi and Varma 5

calculated area under the radiant heat flux versus time stiffness of the frame is provided through the fixed con-
curve. In this study, 16 fires were chosen using 4 differ- nections and bending rigidity of the framing members.
ent ventilation factors and 4 different fuel load densi- This system is appealing to architects as it allows for a
ties. These time–temperature curves were then scaled more open floor plan that is not inhibited by braces or
by the IMs. The maximum displacement in the beam walls. However, because the stiffness is dependent on
was recorded as the DM. RHE was found to be a bending rigidity, it tends to result in larger building
more efficient IM than the peak temperature because drifts than braced frame or shear wall systems.
it had less dispersion of values; however, RHE is not a The seismic-force-resisting system dissipates energy
simple value to calculate, as it involves radiant heat generated by the ground motion through inelastic
transferring back and forth between the enclosure and behavior. Steel moment frames are designed to form
the members inside. plastic hinges at beam ends, acting as fuses to minimize
Devaney (2014) considered different IMs for a additional damage to the remainder of the structure.
performance-based fire study. He considered Ingberg’s Depending upon the post-yield capacity of the mem-
equal area concept from 1928, which was a rough bers, these fuses may need to be replaced after an
method for measuring fire severity. It calculated the earthquake. Locating the hinges in the beams and pre-
area under a temperature–time curve, which constitu- venting hinging in the columns is called the strong
tes no numerical significance in terms of units. This column-weak beam philosophy, which is required for
concept also does not consider heat transfer or the dif- special moment frame systems in the seismic provisions
ference between a fast, hot fire and a slow, cool one. of AISC (ANSI/AISC 341-16, 2010). This philosophy
Other suggested IMs included maximum steel tem- not only provides a more efficient way to dissipate
perature (but this does not account for varying fire energy but also minimizes the potential of collapse due
protection levels), rate of temperature increase (but to a soft-story mechanism (Bruneau et al., 2011).
this does not consider fire peak or duration), and peak Reduced beam sections are sometimes employed to
compartment gas temperature, which was the chosen ensure the strong column-weak beam concept. The
IM. Devaney used Monte Carlo simulation to deter- flanges of the beam section near the ends are reduced
mine the range of realistic results. The engineering to a dog-bone shape, which enables the fuse (hinge) to
demand parameter for the beam was midspan deflec- form in this location. Haunched connections and
tion. In another study, Lange et al. (2014) also used flange rib connections are other approaches to move
peak compartment temperature as the IM. the hinge away from the column and connection.
A concrete column study was conducted using the Lately, there has been a push to further limit damage
maximum temperature within the column cross-section and increase resilience using self-centering, rocking
as the IM. A total of 27 fire scenarios were studied by frames; with this design, post-tensioning strands are
varying the compartment size, fuel load, and ventila- anchored at the beam-to-column connections to pull
tion. A clear correlation between opening factor and the frame back to plumb without causing inelastic
the residual strength index of the column was found, damage (Lin et al., 2013).
as the column capacity was affected by both the peak The Northridge Earthquake of 1994 brought to
temperature and the duration of the fire (Rush et al., light some potential issues with moment frame connec-
2014). tions. Fractures were found at or near the beam flange
In another study, fire load in a compartment groove welds, proving that moment frames were not as
(MJ/m2) was used as the IM (Gernay et al., 2016). At ductile as researchers and practitioners originally
least one compartment was studied per story. The believed. Although these failures did not result in col-
damage states used were flexural resistance of beams lapse, extensive retrofitting of the connections was
(local failure) and maximum resistance of columns required. There were a number of contributing factors
(could lead to collapse). to these failures which include low fracture toughness
of the weld metal in the beam flange to column con-
nection, poor quality of welding due to limited access,
Seismic hazards and stress concentrations from the backing bar/weld
tab. Following Northridge, minimum toughness
Current seismic design approach requirements for weld metal were improved and qual-
Moment-resisting frames (often referred to as moment ity control was more closely monitored, among other
frames) are the focus of this study. These frames con- improvements. The Kobe Earthquake, which occurred
sist of beams and column members connected with in Japan 1 year after Northridge, also resulted in
rigid beam-to-column connections that are designed to severely damaged buildings due to brittle fractures of
resist the lateral loads on the structure. The lateral beam to column connections (Bruneau et al., 2011).
6 Advances in Structural Engineering 00(0)

These earthquakes represented a milestone for the the building and plotting the force versus displace-
improvement of steel seismic-resistant structures. ment. Linear dynamic procedure (LDP) assumes elas-
As the damage from these two earthquakes illu- tic properties of the material and applies the loads
strated, the beam to column connections of the dynamically, accounting for higher modes. An exam-
moment frames are critical to the performance of the ple of this is response spectrum analysis, which uses
structure. The panel zone, which is the area of the col- the equations of motion to develop mode shapes and
umn web where the beam frames in, can experience spectral accelerations. Nonlinear dynamic procedure
very high shear forces and must be designed to prevent (NDP) incorporates both the nonlinearity of the mate-
column web yielding or crippling and flange distortion. rial and the dynamic effects of the building response
Recognizing the importance of connection design, through a time history record. This procedure is more
AISC developed a list of prequalified moment connec- computationally expensive, but it can be used for all
tions that have been tested (ANSI/AISC 358-10, building types and most closely represents true build-
2010). These include standard connections, such as ing behavior.
welded unreinforced flange-welded web connections,
but also some proprietary connections using untradi-
Determining the seismic hazard. In order to perform
tional methods, such as SidePlate, that have been
advanced analyses using the NDP, it is critical to
extensively tested and certified. Newly proposed con-
model the seismic hazard through adequate selection
nection types must undergo rigorous testing.
and scaling of ground motion records. Chapter 16 of
ASCE 7-10 requires three or more appropriate ground
motion records. When using 3D analyses, each individ-
Typical seismic analysis procedures
ual record should have orthogonal pairs of horizontal
As outlined in ASCE 41 (ASCE, 2013), there are four ground motion accelerations. These records must be
primary procedures to analyze buildings subjected to selected from events with similar ‘‘magnitudes, fault
seismic loads: linear static, nonlinear static, linear distance, and source mechanisms’’ as the maximum
dynamic, and nonlinear dynamic. Static procedures do considered earthquake (MCE) (ASCE, 2010). Online
not consider dynamic effects so they should only be databases, such as PEER (2016) and COSMOS (2016),
used with regular structures and when higher mode provide earthquake records based on station readings
effects are considered insignificant. Linear static proce- from actual events. Synthetic ground motions can also
dure (LSP) applies pseudo seismic forces to each story be created but are discouraged in favor of actual
through the equivalent lateral force procedure devel- records.
oped in ASCE 7 (ASCE, 2010). Nonlinear response is Each component of the ground motion record has a
then accounted for through the use of the response resulting 5% damped response spectrum. The orthogo-
modification coefficient (R) and the deflection amplifi- nal components must result in a square root of the sum
cation factor (Cd), also provided in the code and vary- of the squares (SRSS) with a mean value of the records
ing based on the type of lateral system. Nonlinear that must be greater than the design response spectrum
static procedure (NSP) incorporates nonlinearity in the in the range of 0.2–1.5T, as shown in Figure 2, where T
analysis itself. An example of this is static pushover, is the period of the fundamental mode of the structure.
which involves incrementally applying a static force to If at least seven records are used, design forces and

Figure 2. Structural response spectrum (in bold) with ground motion records overlaid.
Chicchi and Varma 7

drifts can be averaged. With less than seven records, risk. These analyses require nonlinear models that are
the worst case controls. reliable and calibrated based on experimental data that
National Institute of Standards and Technology appropriately represent the deterioration of strength
(NIST) provides its own guidelines through work with and stiffness in the structural components. Work by
the ATC-82 project (NIST, 2011). This initiative aims Lignos and Krawinkler (2013) explains the approach
to provide improved and clearer guidance on ground and calibration of such models. A number of agencies
motion selection and scaling, as there seems to be a provide guidelines for the performance-based seismic
lack of general consensus in the earthquake engineer- design approach: Pacific Earthquake Engineering
ing community. It supports ground motion selection Research Center Tall Building Initiative (PEER, 2010),
based on the conditional spectrum, which is a compu- FEMA 445 (ATC, 2006), and the Los Angeles Tall
tational method for selecting a ground motion spec- Buildings Structural Design Council (LATBSDC,
trum that has properties of a naturally occurring 2014).
ground motion at the site. Other approaches are uni- Incremental dynamic analyses, used for assessment
form hazard spectra and conditional mean spectra. of building performance under seismic loads, will be
FEMA P-58 (Federal Emergency Management Agency explained later, as it pertains to recommendations for
(FEMA), 2012) also provides ground motion scaling PEF design and assessment procedures.
guidance.

Gravity frame contribution to lateral stiffness of

Case studies of structural analyses. Chi et al. (1997) building. Traditionally, the gravity system is neglected
explored different modeling techniques for steel in lateral-force-resisting system design; however, stud-
moment frame buildings subjected to ground motions ies have shown that gravity frames can offer additional
through pushover analyses and a time history analysis. stiffness and energy dissipation to the building system
A 17-story building was analyzed using 2D frames that is not negligible (Flores et al., 2012; Foutch and
(with and without second-order effects) and 3D build- Yun, 2002). Because gravity frames experience essen-
ings. 3D-B is a basic model which includes all lateral tially the same drift as the lateral frames, gravity shear
frames and gravity columns which are connected connections can experience large rotations. Also, grav-
rigidly through kinematic restraints. The 3D-F model ity frames typically constitute a significant portion of a
is the full model which includes the gravity beams and building, as moment frames are more costly and usu-
shear connections. Results showed that the 3D models ally limited to the exterior. Tests of simple shear con-
produced more improved building response. It was nections were conducted with and without the floor
found that the stiffness contribution due to the gravity slabs (Liu and Astaneh-Asl, 2000). Findings showed
frames could be attributed primarily to the gravity col- that large rotations of 0.14 rad could be achieved and
umns, which provided additional lateral strength to that approximately 15%–20% of the beam plastic
critical stories in the lower region of the building. moment capacity could be transferred in the connec-
When a time history analysis was performed, the maxi- tion. With the floor slab contribution, the maximum
mum story drift ratio was about 0.025 for the 2D lateral load resistance increased by nearly two.
model, 0.023 for the 3D-B model, and 0.016 for the Flores et al. (2012) studied 2-, 4-, and 8-story
3D-F model. buildings with and without the gravity framing. The
Foutch and Yun (2002) compared modeling tech- gravity connections were modeled as partially
niques for steel moment frames subjected to seismic restrained assuming 0%, 35%, 50%, or 70% of the
loads. The report considered elastic (linear centerline plastic moment capacity of the beam. When sub-
models) as well as nonlinear centerline models, with jected to design basis earthquake (DBE) and MCE
and without panel zones modeled. Panel zones were loading, residual deformations reduced as the per-
idealized as a scissor model using the approach devel- centage of gravity connection resistance increased.
oped by Krawinkler (2000). Results showed that the In fact, for the 8-story building, soft-story collapse of
modeling decisions listed above can significantly affect the building was prevented by including the gravity fram-
building response. Refer to the original work for ing contribution. This influence is also reflected in the
detailed results of the pushover analyses. reduced interstory drift ratios.
Earthquake engineering has been transitioning from In seismic analyses, even if the stiffness contribu-
the traditional prescriptive approach to a performance- tions of the gravity system are ignored, the engineer
based approach of analysis and design that compares cannot ignore the inherent P-delta effects. This P-delta
designated DMs to the anticipated level of damage for effect is often modeled with a leaning column rigidly
a given hazard level as a method for evaluating seismic attached to the lateral-force-resisting system that has
8 Advances in Structural Engineering 00(0)

gravity loading applied so as to simulate the destabiliz-

ing load of gravity frame imperfections (Chi et al.,
1997; Foutch and Yun, 2002). When the lateral contri-
bution of gravity framing is not included in the design,
the leaning column is assigned zero flexural stiffness.
When its contribution is to be included, the equivalent
lateral stiffness of the columns can be modeled through
an equivalent bay that is rigidly attached to the lateral
frame. Similarly, gravity beam stiffnesses are modeled
as the equivalent stiffness of all the gravity frame
beams in that direction. Rotational springs can be used
to model equivalent strength and stiffness of the grav- Figure 3. IDA curves of a T = 1.8 s, 5-story steel braced
ity connections (Flores et al., 2012), using assumptions frame subjected to four different records (Vamvatsikos and
Cornell, 2002).
for elastic rotation based on connection tests by Liu
and Astaneh-Asl (2000). A similar procedure is out-
lined by Foutch and Yun (2002). The equivalent grav-
Examples of the DM can be maximum base shear,
ity frame reduced maximum interstory drifts, though
node rotations, peak roof drift, interstory drift ratios,
the contribution could sometimes be considered negli-
and so on. Results of the analysis can show a progres-
gible. In the 20-story building, these effects were more
sion of behavior with varying results including soften-
pronounced due to the greater P-delta effects.
ing, hardening, and weaving, as shown in Figure 3
Gupta and Krawinkler (1999) accounted for gravity
(Vamvatsikos and Cornell, 2002). Results of IDA can
contributions in their work using an equivalent bay
then be used to generate fragility curves.
rigidly attached to the lateral load resisting frame.
Each of the two equivalent gravity columns had a
moment of inertia equal to half of the gravity columns PEF hazards
at that level. The out-of-plane resistance of the ortho-
gonal moment-resisting frame columns was also con- The 1906 San Francisco earthquake and the 1923
sidered. It was determined that the lateral stiffness Tokyo earthquake are classic examples that brought to
contribution of the gravity frame was most influenced light the potential damage of fire following earth-
by the gravity columns and a smaller influence was quakes. The San Francisco earthquake resulted in
with the gravity connection. 80% of the total damage occurring due to PEFs
Judd et al. (2016) suggested implementation of the (Scawthorn, 2011). The Tokyo earthquake resulted in
dual system design concept for integrating the response 77% of the total losses due to the fire, causing 140,000
of lateral and gravity frames. This approach would be lives lost and 447,000 homes destroyed (Mousavi et al.,
similar to the dual systems already specified in ASCE 2008). This has continued to be a problem in more
7-10 (ASCE, 2010), but would require that 10% of the recent years, as the Kobe earthquake in 1995 resulted
seismic forces would need to be resisted by the gravity in 108 fires. Because of its dense urban setting and due
framing system. Work funded through the National to thousands of breaks in the underground water dis-
Science Foundation is currently ongoing to further tribution system caused by the earthquake, fires spread
explore the role of gravity framing on the seismic per- following the Kobe earthquake (Scawthorn, 1996).
formance (NSF, 2013). Scawthorn et al. (2005) and Botting (1998) summar-
ized the historical cases of fires following earthquakes
and the community level impacts to utilities, communi-
Incremental dynamic analyses. As mentioned previously, cations, roadways, and buildings. Most conflagrations
IDA follows the same principle as IFA. IDA is com- occur in low- or mid-rise timber buildings. Although
monly used in conjunction with the NDP to create the threat to high-rise steel buildings is unlikely, the
a parametric study of building behavior to seismic risk would be very great, as adequate time would be
loads. Each ground motion record can be scaled needed to allow for safe evacuation of inhabitants
using an IM to generate response curves. According to (Taylor, 2003).
Vamvatsikos and Cornell (2002), some possible IM Earthquakes can cause structural damage and resi-
variables are PGA, peak ground velocity (PGV), 5% dual drifts, potentially preventing people from safely
damped spectral acceleration, and the R-factor. Once exiting the building during a fire. In addition, non-
the demand is determined through scaling of the IM, structural components, such as sprinklers and pipe-
the capacity must be evaluated through a DM. lines, may be damaged, which could increase the
Chicchi and Varma 9

duration and intensity of the fire. Additionally, first xade damage is unlikely, even with drifts up
detailed, fac
responders may be slow to react to fires because they to 0.04 rad (Carpenter, 2004; Okazaki et al., 2007).
are already preoccupied responding to earthquake-
related issues. Ground motion from earthquakes has
State of the art of PEF analyses
been known to ignite fires through short-circuiting,
abrasions, chemical reactions, among other causes Case studies of structural analyses. Researchers have
(Scawthorn et al., 2005). These issues together inflame shown that, in the event of a gravity member failure
the potential for damage due to PEFs. from fire, the lateral system can prevent the collapse of
a building through load redistribution (Agarwal and
Varma, 2014; Fischer, 2015). However, if many lateral
Potential nonstructural damage members have already yielded in the earthquake, the
post-yield behavior may not be adequate to accept
Nonstructural damage can occur as a result of large
load redistribution from failing gravity members dur-
drifts and accelerations due to seismic loads. This dam-
ing a fire. The implications of this hazard are explored
age can be detrimental to the fire resistance of the
in the following case studies.
structure. While nonstructural damage will not be
Della Corte et al. (2003) classified earthquake dam-
explored in depth, some potential implications of this
age as ‘‘geometrical’’ and ‘‘mechanical.’’ Geometrical
damage on subsequent fire hazards will be considered.
damage is defined as residual deformations caused by
For instance, when steel yields during an earth-
plasticity in the structure. Mechanical damage is
quake, the adherence of the SFRM can be affected.
‘‘degradation of mechanical properties’’ due to plastic
This was studied by Braxtan and Pessiki (2011). Steel
deformation. A simple single-bay, single-story portal
plates were tension yielded and the adhesive and cohe-
frame was studied to show that the buckling critical
sive strengths of the SFRM were tested. Debonding, load of the column frames was significantly lower than
cracking, and spalling were all possible failure mechan- the Euler buckling load when considering the effects of
isms of SFRM in cyclically loaded beam-to-column seismic damage. Multi-bay, multi-story 2D frames
moment connections (Keller and Pessiki, 2012). At an were then analyzed. The study found a 10% reduction
interstory drift ratio of 3%, debonding and cracking in fire resistance for a design-level seismic event but,
of the insulation occurred. Detachment was more fre- for very rare earthquakes, the contribution of earth-
quent for dry-mix (DM) than wet-mix (WM) fire- quake damage to fire resistance was much more
proofing, while cracking was more likely in WM than significant.
DM. Using computer modeling, they found a 20%– Pantousa and Mistakidis (2014) conducted analyses
30% reduction in flexural capacity of these connec- of PEFs by considering the nonstructural damage
tions when fireproofing had spalled and the steel was caused by the earthquake, namely, the functionality of
exposed to elevated temperatures. the sprinkler system and the breakage of windows. As
Sprinkler systems may also be damaged in an earth- the seismic loads increased, so did the assumed non-
quake event, which would increase the duration of a structural damage. CFD was used to study scenarios
subsequent fire. Fragility curves were developed based where broken windows affect the ventilation. The
on physical tests conducted at the University of structural system was found to fail at the heated beams
Buffalo. These curves show the likelihood of leaking where restrained thermal expansion and catenary
for different components of the sprinkler system when action occurred. They found a 14% reduction in fire-
subjected to peak floor accelerations (Soroushian resistance time for the design earthquake when non-
et al., 2015). Interstory drift can also affect piping per- structural damage was assumed.
formance. In another study (Zaghi et al., 2012), hospi- Behnam and Ronagh (2014b) analyzed a 10-story
tal piping was tested for seismic loading, and it was moment frame building using 2D frame analyses and
found that restrained welded assemblies could with- accounted for earthquake effects through stiffness
stand interstory drift ratios of 4.34% without any degradation and residual deformations. Three different
damage or leaking; however, threaded assemblies fire scenarios were used: fire at the first, fourth, and
could only withstand drifts of 2.2% before leaking seventh floors, respectively, with both 5- and 25-min
occurred. Unrestrained piping fared worse with only delays before spreading the fires between floors. The
1.08% of drift causing leakage. application of the fire (time delay and story level) chan-
Cladding failures could also occur. Excessive drifts ged both the fire-resistance time and failure shape of
could lead to breakage of windows. Both of these sce- the building. In some cases, one level was in the cooling
narios would in turn affect the opening factor of exter- phase while the upper level was heating. For fast-
ior compartment fires. However, studies have shown moving vertical fires, collapse occurred during heating,
that if the cladding and its connections are properly but for slower spreading fires, collapse occurred during
10 Advances in Structural Engineering 00(0)

the cooling phase. Sway mechanisms were observed for to compare the effects of fire intensity and fireproofing
fast-moving fires but beam mechanisms were observed with the acceptance criteria and collapse time. A
for slower moving fires. performance-based design approach was suggested.
Khorasani et al. (2015b) compared a fire-only and
PEF scenario using OpenSees, an open-source analysis
software from UC Berkeley. They found that the earth-
Research needs and future directions
quake decreases the time to form a plastic hinge due to Further research is needed to advance the understand-
fire at the beam to perimeter column interface. Column ing of steel building performance when subjected to
drifts of 1.7% were achieved in fire following earth- PEF hazards. In particular, very few experimental tests
quake, due to thermal expansion and the residual drift have been conducted on steel components and assem-
of the earthquake. blies subjected to PEF (Braxtan, 2010; Huang et al.,
Memari et al. (2014) studied moment-resisting 2012). Instead, most work on PEF has been done com-
frames with reduced beam section connections in low-, putationally. Advancements in understanding of PEF
medium-, and high-rise structures, using 2D frame behavior can be achieved through experimental testing
analyses. The leaning column approach was used to of beam-to-column assemblies for both gravity and
represent P-delta effects and stiffness of the gravity moment frames. Researchers can determine through
columns. Panel zones in the beam to column connec- computational modeling which other components or
tions were modeled using the scissor model, with rigid assemblies should be experimentally tested. These tests
links and a rotational spring to capture the moment– can in turn be used to verify and benchmark the com-
rotation of the connection. Life safety performance putational models.
was determined in 80% of the analyses, while collapse As mentioned previously, IDA has been widely used
prevention consisted of the remaining 20%. PEFs by researchers and practitioners and it follows an
tended to produce lower interstory drift ratios than the established procedure that has been well documented.
earthquake scenarios and system-level collapse was IFA, however, is a relatively new procedure, which has
not imminent. Large tensile forces were developed in not yet garnered consensus among researchers regard-
the beams during the cooling phase, while axial com- ing the most indicative IMs and engineering damage
pressive force-bending (caused during heating because parameters to use to understand building behavior to
of thermal expansion and restraint) tended to control fire hazards. IFA conducted thus far has focused on
the beam design. individual components, such as beams and columns.
Zaharia and Pintea (2009) used pushover frame Additional research is needed to study IFA at a build-
analysis and both ISO 834 and natural fire curves to ing system level.
compare three different frames. The frames that were Additionally, a methodology needs to be established
designed for higher seismicity levels appeared to have for analyzing and evaluating the performance of struc-
reserve fire resistance. Also, the fire-resistance time of tures subjected to PEF. Faggiano and Mazzolani
the structure was affected by its level of damage, with (2011) made notable strides toward assessing robust-
undamaged structures resisting fire loads for longer ness. In their study, 2D frame analyses were conducted
prior to collapse; however, in some cases, this differ- using pushover analysis. Fire loads were applied to
ence is very minimal (roughly 1 min). Two primary two bays in each of the two stories. Seismic perfor-
methods of collapse were observed: a global (struc- mance levels were benchmarked by interstory drift
tural) sway mechanism and a beam mechanism. ratios and plastic hinge rotations according to FEMA
Typically, the same mode of collapse was observed in 356 (FEMA, 2000). Similar criteria were established
the frames, whether or not the structure was damaged for fire: Operational Fire, Life Safety fire, Local
in the earthquake. Collapse fire, Section Collapse Fire, and Global
Behnam and Ronagh (2014a) proposed a post- Collapse fire, based on yielding, plastic hinging, beam
earthquake factor to be applied to the equivalent static mechanisms, failure of the cross-section, and a global
equation for calculating base shear due to seismic mechanism, respectively. A performance chart was
loads. In this, VPEF = CPEF(t)Cs 3 W. This CPEF(t) generated which compared seismic performance levels,
value would be evaluated iteratively through redesign fire performance levels, and fire resistance in a 3D bar
of the frame until it achieves a satisfactory perfor- chart. This approach can be further advanced by
mance level when subjected to the PEF. developing 3D fragility curves to determine the prob-
Quiel and Marjanishvili (2012) studied the effect of ability of collapse for various hazard severities. These
damage on the fire resistance of steel buildings, focus- 3D surface plots can be used to show the interrelation-
ing on fire following blast or impact scenarios. They ship between the level of each hazard and the corre-
found that the structure was very susceptible to global sponding probability of collapse. The methodology for
instabilities and that further studies would be needed this approach is explained in the following sub-section.
Chicchi and Varma 11

Outline of PEF methodology 6. Generate IDA response curve. After each anal-
ysis is run, the maximum story drift ratio for
Based on the current state of the art, the following
each IM (PGA) should be recorded and used
methodology is proposed for evaluating steel buildings
to develop a plot of PGA versus story drift
exposed to PEFs. The methodology is outlined in
ratio. This will show a progression of the
Figures 4 and 5. Figure 4 includes steps (1) through (6)
building response as the intensity of the seis-
described below, including the design of the structure,
mic event increases.
selection of the seismic hazard, and implementation of
7. Select fire locations. Fires should be consid-
incremental dynamic analyses. Figure 5 illustrates
ered at strategic locations throughout the
steps (7) through (13) described below, which include
building. At a minimum, fire analyses should
the process of determining fire hazards, conducting
be conducted at a mid and upper level.
incremental fire analyses, and developing fragility
Corner, exterior, and interior compartments
curves. Each of these steps is explained in more detail
should be studied as well. Each bay of the
below:
structure could be considered one compart-
ment. Full-story fires may also be considered.
1. Design of the structure using applicable build- 8. Select fire time–temperature curves. Eurocode
ing codes. Standard design procedures should parametric time–temperature curves should be
be used to determine the structural framing used to represent compartment fires. Multiple
members, connection details, and fireproofing (three or more) fire curves are recommended
requirements. Structures that are regular and to accompany the seven seismic time histories
symmetric in shape are usually designed using required by ASCE 7. Opening factor, thermal
2D LSPs outlined in ASCE 7 for seismic and inertia of the enclosure, and fire load density
wind analyses. should be varied to produce distinct curves,
2. Development of a 3D FEM model. A detailed which vary in peak fire temperature and rate
nonlinear, inelastic 3D finite element model of of heating and cooling.
the complete building structure should be 9. Conduct 2D heat transfer. 2D heat transfer is
developed. This includes consideration of the necessary to determine the internal tempera-
gravity framing contribution, such as the com- tures of all of the structural members exposed
posite slab and gravity connections. The to the fire. This should be performed using
model should account for inelastic deforma- FEM or other numerical analysis software.
tions, instability failures, and connection dam- The fireproofing should be modeled at the
age at elevated temperatures. It should also design thickness, or the thickness determined
incorporate the effect of temperature on mate- using the prescriptive approach. Thermal
rial properties. Separate seismic and fire mod- properties for each material should be taken
els may need to be developed, with the ability as per Eurocode where applicable.
to import the results of the seismic analyses 10. Apply temperatures to building model. The
into the fire model. internal temperatures, which are determined
3. Selection of ground motions. Ground motions through heat transfer analyses, should be
may be selected using the ASCE 7 procedure assigned to the members in the building model
explained previously, or by other means. This that are exposed to the compartment fire.
includes selection of at least seven ground Five-minute increments are recommended.
motions which are scaled to fit the design 11. Incrementally scale fire time–temperature curves.
response spectrum of the building. Ground The fire time–temperature curves should be
motion records scaled from actual seismic scaled in a manner similar to the scaling of
events should be used. PGA used in IDA. The curves may be scaled
4. Apply ground motion to base of building. The by peak fire temperature, as shown in Figure 5.
selected time histories should be applied to the 12. Generate IFA response curve. The recom-
building base using either acceleration or dis- mended damage parameter for IFA of building
placement records. Ground motions should be systems is the story deflection ratio, which mea-
applied in both orthogonal directions. sures the maximum vertical deflection divided
5. Incrementally scale ground motion. Ground by the story height. This damage parameter or
motions must be scaled by a selected IM, another representative value should be recorded
which can include PGA, PGV, or Sa, among for each fire time-temperature curve using the
others. The procedure outlined in Figure 4 results of the incremental analyses conducted in
suggests scaling PGA after each analysis. the previous step.
12 Advances in Structural Engineering 00(0)

Figure 4. Part 1 of proposed PEF methodology.

Chicchi and Varma 13

Figure 5. Part 2 of proposed PEF methodology.

13. Develop fragility curves. The IDA and IFA at different IMs. Fragility curves are deter-
results should be used to generate fragility mined using the following equation
curves, which identify the probability of fail-
ure to occur. Fragility curves have been used  x

ln u
extensively in seismic analyses to calculate the PðCjxÞ ¼ u ð1Þ
probability of collapse or failure of a structure b
14 Advances in Structural Engineering 00(0)

where PðCjxÞ is the probability of collapse or failure at ASTM (2015) Standard Test Methods for Fire Tests of Build-
an IM of x. uðÞ indicates a normal cumulative distri- ing Construction and Materials (ASTM E119–E115). West
bution function. u is the mean of the fragility function Conshohocken, PA: ASTM.
and b is the standard deviation. The results from IDA Behnam B and Ronagh HR (2014a) An engineering solution
and IFA should be plotted relative to one another to improve post-earthquake fire resistance in important
reinforced concrete structures. Advances in Structural
to create a 3D surface plot, which is illustrated in
Engineering 17(7): 993–1009.
Figure 5. This plot will indicate the probability of col-
Behnam B and Ronagh HR (2014b) Behavior of moment-
lapse of the structure with varying severities of earth- resisting tall steel structures exposed to a vertically travel-
quake and cascading fire hazards. This will evaluate ing post-earthquake fire. Structural Design of Tall and
the vulnerability of steel moment frame buildings sub- Special Buildings 23(14): 1083–1096.
jected to fire following earthquake hazards. Botting R (1998) The impact of post-earthquake fire on the
This methodology has been implemented recently urban environment. Fire Engineering Research Report 98/1,
by the authors to evaluate the design of a 10-story steel June. Christchurch, New Zealand: University of Canterbury.
office building located in Chicago, IL. The results from Braxtan NJL (2010) Post-earthquake fire performance of steel
this evaluation are beyond the scope of this article and moment frame building columns. PhD Thesis, Lehigh Uni-
the subject of a future publication. versity, Bethlehem, PA.
Braxtan NL and Pessiki S (2011) Bond performance of
SFRM on steel plates subjected to tensile yielding. Jour-
Declaration of Conflicting Interests nal of Fire Protection Engineering 21(1): 37–55.
The author(s) declared no potential conflicts of interest with Bruneau M, Uang C-M and Sabelli R (2011) Ductile Design
respect to the research, authorship, and/or publication of this of Steel Structures. New York: McGraw Hill.
article. BS EN 1363-1:2012 (2012) Fire resistance tests—part 1: gen-
eral requirements.
Carpenter LD (2004) High-rise building cladding drift
Funding accommodation. Structural Design of Tall and Special
The author(s) received no financial support for the research, Buildings 13(5): 439–456.
authorship, and/or publication of this article. Chi W-M, El-Tawil S, Deierlein GG, et al. (1997) Inelastic
analyses of a 17-story steel framed building damaged dur-
References ing Northridge. Engineering Structures 20(4–6): 481–495.
Choe L, Varma AH, Agarwal A, et al. (2011) Fundamental
Agarwal A and Varma AH (2011) Design of steel columns at
behavior of steel beam-columns and columns under fire
elevated temperatures due to fire: effects of rotational
loading: experimental evaluation. Journal of Structural
restraints. Engineering Journal 48: 297–314.
Engineering 137(9): 954–966.
Agarwal A and Varma AH (2014) Fire induced progressive
COSMOS (2016) Strong-motion virtual data center (VDC).
collapse of steel building structures: the role of interior
Available at: http://strongmotioncenter.org/vdc
gravity columns. Engineering Structures 58: 129–140.
Della Corte G, Landolfo R and Mazzolani FM (2003) Post-
Agarwal A, Choe L and Varma AH (2014) Fire design of
earthquake fire resistance of moment resisting steel
steel columns: effects of thermal gradients. Journal of Con-
frames. Fire Safety Journal 38(7): 593–612.
structional Steel Research 93: 107–118.
Devaney S (2014) Development of software for reliability
Al-Jabri KS, Seibi A and Karrech A (2006) Modelling of
based design of steel framed structures in fire. PhD Thesis,
unstiffened flush end-plate bolted connections in fire.
The University of Edinburgh, Edinburgh.
Journal of Constructional Steel Research 62(1–2): 151–159.
Dwaikat MMS, Kodur VKR, Quiel SE, et al. (2011) Experi-
American Society of Civil Engineers (ASCE) (2010) Mini-
mental behavior of steel beam–columns subjected to fire-
mum Design Loads for Buildings and Other Structures
induced thermal gradients. Journal of Constructional Steel
(ASCE7–10). Reston, VA: ASCE.
Research 67(1): 30–38.
American Society of Civil Engineers (ASCE) (2013) Seismic
EN 1991-1-2:2002 (2002) Eurocode 1: actions on structures,
Evaluation and Retrofit of Existing Buildings (ASCE/SEI
part 1–2: general actions—actions on structures exposed
41–13). Reston, VA: ASCE.
to fire.
ANSI/AISC 341-16 (2010) Seismic provisions for structural
Faggiano B and Mazzolani FM (2011) Fire after earthquake
steel buildings.
robustness evaluation and design: application to steel
ANSI/AISC 358-10 (2010) Prequalified connections for spe-
structures. Steel Construction 4(3): 183–187.
cial and intermediate steel moment frames for seismic
Fang C, Izzuddin BA, Elghazouli AY, et al. (2011) Robust-
applications.
ness of steel-composite building structures subject to loca-
ANSI/AISC 360-10 (2010) Specification for structural steel
lised fire. Fire Safety Journal 46(6): 348–363.
buildings.
Federal Emergency Management Agency (FEMA) (2000)
Applied Technology Council (ATC) (2006) Next-Generation
Prestandard and commentary for the seismic rehabilitation
Performance-Based Seismic Design Guidelines: Program
of buildings. FEMA 356, November. Washington, DC:
Plan for New and Existing Buildings (FEMA 445). Red-
Federal Emergency Management Agency.
wood City, CA: ATC.
Chicchi and Varma 15

Federal Emergency Management Agency (FEMA) (2012) subjected to seismic loads. In: Proceedings of the eighth
Seismic performance assessment of buildings. Volume 1— international workshop on connections in steel structures,
methodology. FEMA P-58-1, September. Washington, Boston, MA, 24–26 May. Chicago, IL: American Insti-
DC: Federal Emergency Management Agency. tute of Steel Construction.
Fischer EC (2015) Fire behavior of simple (shear) connections Keller WJ and Pessiki S (2012) Effect of earthquake-induced
in steel-frame buildings. PhD Thesis, Purdue University, damage to spray-applied fire-resistive insulation on the
West Lafayette, IN. response of steel moment-frame beam-column connec-
Fischer EC and Varma AH (2017) Fire resilience of compo- tions during fire exposure. Journal of Fire Protection Engi-
site beams with simple connections: parametric studies neering 22(4): 271–299.
and design. Journal of Constructional Steel Research 128: Khorasani NE, Gardoni P and Garlock M (2015a) Prob-
119–135. abilistic fire analysis: material models and evaluation of
Flint G (2005) Fire induced collapse of tall buildings. PhD steel structural members. Journal of Structural Engineer-
Thesis, The University of Edinburgh, Edinburgh. ing 141(12), [04015050]. DOI: 10.1061/(ASCE)ST.1943-
Flint G, Usmani A, Lamont S, et al. (2007) Structural 541X.0001285.
response of tall buildings to multiple floor fires. Journal Khorasani NE, Garlock MEM and Quiel SE (2015b) Model-
of Structural Engineering 133(12): 1719–1732. ing steel structures in OpenSees: enhancements for fire
Flores FX, Jarrett JA and Charney FA (2012) The influence and multi-hazard probabilistic analyses. Computers &
of gravity-only framing on the performance of steel Structures 157: 218–231.
moment frames. In: Proceedings of the 15th world confer- Kodur VKR, Garlock MEM and Iwankiw N (2011) Struc-
ence on earthquake engineering, Lisbon, Portugal, 24–28 tures in fire: state-of-the-art, research and training needs.
September. Fire Technology 48(4): 825–839.
Foutch DA and Yun S-Y (2002) Modeling of steel moment Kodur VKR, Naser M, Pakala P, et al. (2013) Modeling the
frames for seismic loads. Journal of Constructional Steel response of composite beam–slab assemblies exposed to
Research 58: 529–564. fire. Journal of Constructional Steel Research 80:
Franssen J-M and Vila Real P (2012) Fire Design of Steel 163–173.
Structures. Eurocode 1: Actions on Structures Part 1–2 Krawinkler H (2000) State of the art report on systems perfor-
General Actions: Actions on Structures Exposed to Fire. mance of steel moment frames subject to earthquake ground
Eurocode 3: Design of Steel Structures Part 1–2—General shaking. FEMA-355C, September. Washington, DC:
Rules: Structural Fire Design. Hoboken, NJ: John Wiley FEMA.
& Sons. Lange D, Devaney S and Usmani A (2014) An application
Garlock M and Selamet S (2010) Modeling and behavior of of the PEER performance based earthquake engineering
steel plate connections subject to various fire scenarios. framework to structures in fire. Engineering Structures 66:
Journal of Structural Engineering 136: 897–906. 100–115.
Gernay T, Khorasani NE and Garlock M (2016) Fire fragi- Lignos DG and Krawinkler H (2013) Development and utili-
lity curves for steel buildings in a community context: a zation of structural component databases for
methodology. Engineering Structures 113: 259–276. performance-based earthquake engineering. Journal of
Gupta A and Krawinkler H (1999) Seismic Demands for Per- Structural Engineering 139(8): 1382–1394.
formance Evaluation of Steel Moment Resisting Frame Lin Y-C, Sause R and Ricles JM (2013) Seismic performance
Structures (SAC Task 5.4.3, The SAC Joint Venture). of steel self-centering, moment-resisting frame: hybrid
Stanford, CA: John A. Blume Earthquake Engineering simulations under design basis earthquake. Journal of
Center. Structural Engineering 139(11): 1823–1832.
Hu G and Engelhardt M (2011) Investigations on the beha- Liu J and Astaneh-Asl A (2000) Cyclic testing of simple con-
vior of steel single plate beam end framing connections in nections including effects of slab. Journal of Structural
fire. Journal of Structural Fire Engineering 2(3): 195–204. Engineering 126: 32–39.
Huang Y, Bevans WJ, Xiao H, et al. (2012) Experimental Los Angeles Tall Buildings Structural Design Council
validation of finite element model analysis of a steel frame (LATBSDC) (2014) An Alternative Procedure for Seismic
in simulated post-earthquake fire environments. In: Pro- Analysis and Design of Tall Buildings Located in the Los
ceedings of the SPIE 8345: sensors and smart structures Angeles Region. Los Angeles, CA: LATBSDC.
technologies for civil, mechanical, and aerospace system, Mahmoud H, Ellingwood B, Turbert C, et al. (2016)
San Diego, CA, 11 March. Bellingham, WA: Society of Response of steel reduced beam section connections
Photographic Instrumentation Engineering. exposed to fire. Journal of Structural Engineering 142(1):
IBC (2011) ICC IBC (2012): International Building Code. 04015076.
ISO 834-1:1999 (2015) Fire resistance tests—elements of Memari M and Mahmoud H (2014) Performance of steel
building construction—Part 1: general requirements. moment resisting frames with RBS connections under fire
Jiang J, Li G-Q and Usmani A (2014) Progressive collapse loading. Engineering Structures 75: 126–138.
mechanisms of steel frames exposed to fire. Advances in Memari M, Mahmoud H and Ellingwood B (2014) Post-
Structural Engineering 17(3): 381–398. earthquake fire performance of moment resisting frames
Judd JP, Charney FA and Flores FX (2016) The influence of with reduced beam section connections. Journal of Con-
gravity framing on the performance of steel buildings structional Steel Research 103: 215–229.
16 Advances in Structural Engineering 00(0)

Moss P, Abu A and Dhakal R (2014) Incremental fire analy- Rush D, Bisby L, Ioannou I, et al. (2014) Towards fragility
sis (IFA) for probabilistic fire risk assessment. In: Pro- analysis for concrete buildings in fire: residual capacity of con-
ceedings of the 23rd Australasian conference on the crete columns. In: Proceedings of the 8th international confer-
mechanics of structures and materials (ACMSM23), ence on structures in fire, Shanghai, China, 11–13 June 2014.
Southern Cross University, Lismore, NSW, Australia, 9– Sarraj M (2007) The behaviour of steel fin plate connections in
12 December. fire. PhD Thesis, The University of Sheffield, Sheffield.
Mousavi S, Bagchi A and Kodur VKR (2008) Review of Scawthorn CR (1996) Fires following the Northridge and
post-earthquake fire hazard to building structures. Cana- Kobe earthquakes. In: Thirteenth meeting of the UJNR
dian Journal of Civil Engineering 35(7): 689–698. panel on fire research and safety (NISTIR 6030), 13–20
Naser MZ and Kodur VKR (2016) Factors governing onset March, pp. 325–335. Gaithersburg, MD: NIST.
of local instabilities in fire exposed steel beams. Thin- Scawthorn CR (2008) The Shakeout Scenario Supplemental
Walled Structures 98: 48–57. Study: Fire Following Earthquake. Pasadena, CA: United
National Institute of Standards and Technology (NIST) State Geological Survey; Sacramento, CA: California
(2011) Selecting and scaling earthquake ground motions for Geological Survey.
performing response-history analyses (NEHRP Consul- Scawthorn CR (2011) Fire following earthquake aspects of
tants Joint Venture). NIST GCR 11-917-15, November. the Southern San Andreas fault Mw7.8 earthquake sce-
Gaithersburg, MD: NIST. nario. Earthquake Spectra 27(2): 419–441.
NSF (2013) NEESR planning: the role of gravity framing in Scawthorn CR, Eidinger JM and Schiff AJ (2005) Fire follow-
the seismic performance of steel buildings. Available at: ing earthquake. Technical Council on Lifeline Earthquake
https://www.nsf.gov/awardsearch/showAward?AWD_ID Engineering, Monograph no. 26. Reston, VA: American
=1344592 Society of Civil Engineers.
Okazaki T, Nakashima M, Suita K, et al. (2007) Interaction Selamet S and Garlock ME (2012) Predicting the maximum
between cladding and structural frame observed in a full- compressive beam axial force during fire considering local
scale steel building test. Earthquake Engineering & Struc- buckling. Journal of Constructional Steel Research 71:
tural Dynamics 36(1): 35–53. 189–201.
Pacific Earthquake Engineering Research Center (PEER) Selden KL and Varma AH (2016) Composite beams under
(2010) Tall Building Initiative: guidelines for Performance- fire loading: numerical modeling of behavior. Journal of
Based Seismic Design of Tall Buildings. Berkeley, CA: Structural Fire Engineering 7(2): 142–157.
PEER. Selden KL, Fischer EC and Varma AH (2016) Experimental
Pacific Earthquake Engineering Research Center (PEER) investigation of composite beams with shear connections
(2016) PEER ground motion database. Available at: subjected to fire loading. Journal of Structural Engineering
http://ngawest2.berkeley.edu/ 142(2): 04015118.
Pakala P, Kodur V, Selamet S, et al. (2012) Fire beha- Soroushian S, Zaghi AE, Maragakis EM, et al. (2015) Ana-
vior of shear angle connections in a restrained steel lytical seismic fragility analyses of fire sprinkler piping
frame. Journal of Constructional Steel Research 77: systems with threaded joints. Earthquake Spectra 31:
119–130. 1125–1155.
Pantousa D and Mistakidis E (2014) Fire resistance of a steel STC (1999) The Behaviour of Multi-Storey Steel Framed Build-
structure under different fire-after-earthquake scenarios ings in Fire (A European Joint Research Programme).
considering both structural and non-structural damage. South Yorkshire: British Steel plc. Swinden Technology
In: Proceedings of the 8th Hellenic national conference of Centre.
steel structures, Tripoli, 2–4 October 2014. Sun R, Huang Z and Burgess IW (2012) Progressive collapse
Pope ND and Bailey CG (2006) Quantitative comparison of analysis of steel structures under fire conditions. Engineer-
FDS and parametric fire curves with post-flashover ing Structures 34: 400–413.
compartment fire test data. Fire Safety Journal 41(2): Takagi J and Deierlein GG (2007) Strength design criteria for
99–110. steel members at elevated temperatures. Journal of Con-
Qian ZH, Tan KH and Burgess IW (2008) Behavior of steel structional Steel Research 63(8): 1036–1050.
beam-to-column joints at elevated temperature: experi- Tan K-H and Huang Z-F (2005) Structural responses of axially
mental investigation. Journal of Structural Engineering restrained steel beams with semirigid moment connection in
134(5): 713–726. fire. Journal of Structural Engineering 131(4): 541–551.
Quiel SE and Marjanishvili SM (2012) Fire resistance of a Taylor J (2003) Post earthquake fire in tall buildings and the
damaged steel building frame designed to resist progres- New Zealand building code. Fire Engineering Research
sive collapse. Journal of Performance of Constructed Facil- Report 03/6, February. Christchurch, New Zealand: Uni-
ities 26(4): 402–409. versity of Canterbury.
Röben C, Gillie M and Torero J (2010) Structural behaviour UL (2016) UL Online Certifications Directory: fire-Resis-
during a vertically travelling fire. Journal of Constructional tance Ratings—ANSI/UL263. Available at: http://www.
Steel Research 66(2): 191–197. ul.com/database
Ruddy JL, Marlo JP, Ioannides SA, et al. (2003) AISC Design Usmani AS, Chung YC and Torero JL (2003) How did the
Guide 19-Fire Resistance of Structural Steel Framing. Chi- WTC towers collapse: a new theory. Fire Safety Journal
cago, IL: American Institute of Steel Construction. 38(6): 501–533.
Chicchi and Varma 17

Vamvatsikos D and Cornell CA (2002) Incremental dynamic Yu H, Burgess IW, Davison JB, et al. (2009) Experimental
analysis. Earthquake Engineering & Structural Dynamics investigation of the behaviour of fin plate connections in
31(3): 491–514. fire. Journal of Constructional Steel Research 65(3): 723–736.
Wang Y, Burgess I, Wald F, et al. (2013) Performance- Zaghi AE, Maragakis EM, Itani A, et al. (2012) Experimen-
Based Fire Engineering of Structures. Boca Raton, FL: tal and analytical studies of Hospital Piping assemblies
CRC Press. subjected to seismic loading. Earthquake Spectra 28(1):
Wang YC, Dai XH and Bailey CG (2011) An experimental 367–384.
study of relative structural fire behaviour and robust- Zaharia R and Pintea D (2009) Fire after earthquake analy-
ness of different types of steel joint in restrained steel sis of steel moment resisting frames. International Journal
frames. Journal of Constructional Steel Research 67(7): of Steel Structures 9(4): 275–284.
1149–1163. Zhao B and Kruppa J (1997) Fire Resistance of Composite
Yang K-C, Chen S-J and Ho M-C (2009) Behavior of beam- Slabs with Profiled Steel Sheet and of Composite Steel
to-column moment connections under fire load. Journal of Concrete Beams. Part 2: Composite Beams. Luxembourg:
Constructional Steel Research 65(7): 1520–1527. European Commission.