Journal of Earthquake Engineering

ISSN: 1363-2469 (Print) 1559-808X (Online) Journal homepage: http://www.tandfonline.com/loi/ueqe20

Nonlinear Static Seismic Performance Assessment

of Plan-Irregular Steel Structures

Rita Peres, Rita Bento & José Miguel Castro

To cite this article: Rita Peres, Rita Bento & José Miguel Castro (2018): Nonlinear Static Seismic
Performance Assessment of Plan-Irregular Steel Structures, Journal of Earthquake Engineering,
DOI: 10.1080/13632469.2018.1469438

To link to this article: https://doi.org/10.1080/13632469.2018.1469438

Published online: 02 May 2018.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=ueqe20
JOURNAL OF EARTHQUAKE ENGINEERING
https://doi.org/10.1080/13632469.2018.1469438

Nonlinear Static Seismic Performance Assessment of

Plan-Irregular Steel Structures
Rita Peresa, Rita Bento a
, and José Miguel Castro b

a
Department of Civil Engineering, Architecture and Georesources, CERIS, Instituto Superior Técnico,
Universidade de Lisboa, Lisboa, Portugal; bFaculty of Engineering, University of Porto, Porto, Portugal

ABSTRACT ARTICLE HISTORY

This paper focuses on the evaluation of a set of Nonlinear Static Received 16 February 2017
Procedures (NSPs), applied to groups of plan-regular and irregular Accepted 7 November 2017
steel structures with three stories. The NSPs selected are the conven- KEYWORDS
tional and the Extended N2, the Capacity Spectrum Method and the Pushover Analysis; Capacity
Adaptive Capacity Spectrum Method. In terms of translational response, Spectrum Method; N2
all NSPs lead to adequate estimations of results. Concerning the tor- Method; Plan-Irregular Steel
sional response, the results of the Extended N2 are the most appro- Buildings
priate to the studied irregular steel structures, except for torsionally stiff
concentrically braced frame structures. An improvement version of the
Extended N2 method is proposed for these structures.

1. Introduction
Structural analysis in earthquake engineering problems is a complex issue because the seismic
action is dynamic and typically leads to nonlinear structural responses. In the last two decades,
Nonlinear Static Procedures (NSPs) have become a powerful tool for assessing the seismic
performance of structures, becoming commonplace in engineering practice to estimate
seismic demands. In fact, several seismic codes (e.g., FEMA 440 [ATC, 2005] and Eurocode
8 [CEN, 2004] recommend them for the seismic performance assessment of structural
systems. Although it is generally accepted that seismic demands are best estimated using
incremental nonlinear time-history (TH) analyses, NSPs are frequently used in regular
engineering applications to avoid the inherent complexity and additional computational effort
required by the former.
The first NSPs proposed were the Capacity Spectrum Method (CSM) [Freeman et al.,
1975] and the N2 method [Fajfar and Fischinger, 1988; Fajfar, 1999]. An increasing
number of updated formulations of the different available NSPs have been observed in
the recent years, such as the refined of CSM version presented in FEMA440 [Applied
Technology Council, 2005], the Modal Pushover Analysis (MPA) [Chopra and Goel, 2002],
the Adaptive Modal Combination Procedure (AMCP) [Kalkan and Kunnath 2006], and the
Adaptive Capacity Spectrum Method (ACSM) [Pinho and Casarotti 2007] that improve
their performances. However, most of the methods were developed for regular structures,
and their extension to irregular structures is still not straightforward. Recent attempts have
been made to develop NSPs formulations capable of predicting the torsional response of

CONTACT Rita Bento rita.bento@tecnico.ulisboa.pt Instituto Superior Técnico, Av. Rovisco Pais 1, 1049-001
Lisbon, Portugal
© 2018 Taylor & Francis Group, LLC
2 R. PERES ET AL.

plan-irregular structures. Advanced and innovative methods have been suggested, for
example, the methods proposed by Chopra and Goel [2004], Fajfar et al. [2005], Bento
et al. [2010], and Bosco et al. [2012]. Nevertheless, to the best of the authors’ knowledge,
the performance of these procedures was not tested for all types of plan-irregular steel
structures, despite their peculiar characteristics. Thus, this work aims to investigate the
performance of a selected set of NSPs procedures when applied to different types of plan-
irregular steel buildings composed by moment-resisting frames (MRF) and/or concentri-
cally braced frames (CBF). To evaluate the torsional phenomenon that results from plan
irregularities, two types of structures were considered: torsionally unrestrained and tor-
sionally restrained irregular structures.
The NSPs examined in this work are the original N2 method recommended by Eurocode 8
[CEN, 2004], the Extended N2 method proposed by Fajfar et al. [2005], the Capacity Spectrum
Method (CSM) prescribed by ATC40 [ATC, 1996] with the improvements presented in FEMA
440 Report [Applied Technology Council, 2005] and the Adaptive Capacity Spectrum Method
(ACSM) proposed by Pinho and Casarotti [2007]. When selecting the methods, the authors
chose to study two reference conventional methods (the N2 method and the CSM) and a more
advanced method, adaptive, the ACSM. Different from the reference methods, the ACSM
consists of using a displacement-based adaptive pushover analysis (instead of the conventional
pushover analysis, which is based on the application of forces) to then derive an equally
adaptive single degree of freedom (SDOF) capacity curve that is also calculated step by step
using actual deformed pattern of the multiple degree of freedom (MDOF) at any given instant.
This method accounts for higher modes and stiffness degradation. The Extended N2 method
was added to the group because it considers, in a simplified way, the torsional response of
building structures (which is indirectly recommended by the EC8 for the estimation of the
torsional effects for torsionally flexible structures or structures with a predominately torsional
second mode of vibration) and higher mode effects.
The accuracy of these procedures was assessed by comparing the results with those
obtained with nonlinear time-history (TH) analyses, which is considered to be the analysis
that provides the most reliable estimates of seismic response.
In this paper, the seismic design of the studied structures was performed following the
Extended Improved Forced Based Design method, proposed by the authors [Peres et al.,
2016]. After introducing the structural configurations, the seismic action, the numerical
modeling issues and types of analyses performed are described. Afterwards, the NSPs are
applied to two groups of eight structures, and their accuracy and reliability are examined
through comparison of the structures’ seismic performance results with those determined
by nonlinear time-history (TH) analysis.
The examination of the results leads to a proposal of an improved version of the
Extended N2 method to be adopted in the seismic assessment of torsionally stiff and plan-
irregular structures composed by concentrically braced frames.

2. Studied Cases
2.1. Description of the Structures
The study comprises the assessment of the response of a group of eight, three stories high steel
structures with plan configurations defined to target torsionally restrained and unrestrained
systems (TR or TU, respectively), as well as plan-regular (PR) and plan-irregular (PI) structures.
 JOURNAL OF EARTHQUAKE ENGINEERING 3

The seismic lateral resisting systems chosen are moment-resisting frames (MRF) and
concentrically braced frames (CBF). These have been selected to consider different
hysteretic behaviors. The plan configuration of the torsionally restrained and unrestrained
structures depends on the location of the lateral resisting systems on the outside or inside
perimeter of the structures. The regularity in plan is achieved through the location of the
lateral resisting systems in such a way that the structure’s centers of mass and stiffness are
coincident. The irregularity is achieved by placing an additional moment-resisting frame
in one of the horizontal plan direction to shift the structure’s center of stiffness so that it
does not coincide with the center of mass in that direction.
The plan dimensions of the structures are 30 m × 18 m with a 6 m × 4 m core in which
the elevator and stairs are located. Regarding the elevation profile, the story height is equal
to 4.5 m in the first story and 3.5 m in the second and third stories. The plan layouts and
the elevation views are presented in Figs. 1–4. These structural configurations can be
associated with real industrial buildings. The structures are identified in Table 1.

2.2. Structural Design

The structures are initially designed for gravity loads following the Eurocode 3 [CEN,
2005] provisions. The loads comprise the self-weight of the structure, which are the slab
self-weight (2.93 kN/m2), the weight of the external walls (13.3 kN/m), the finishings and
partitions (1 kN/m2), and an imposed load of 2 kN/m2. European IPE and HEB sections
are used for the beam and column sections, respectively; whereas, hot finished structural
hollow sections are adopted for braces. The steel grades considered are S275 for beams
and columns and S355 for braces and gusset plates.

Figure 1. Plan layout of the MRF structures: (a) MRF-TU-PR, (b) MRF-TU-PI, (c) MRF-TR-PR, (d) MRF-TR-PI.
 4 R. PERES ET AL.

Figure 2. Elevation view of the MRF structures.

Figure 3. Plan layout of the CBF structures: (a) CBF-TU-PR, (b) CBF-TU-PI, (c) CBF-TR-PR, (d) CBF-TR-PI.

The seismic design is performed following the Extended Improved Forced Based
Design method proposed by Peres et al. [2016]. This procedure consists of a more rational
sequence of the design steps prescribed in Eurocode 8 [CEN, 2004] involving the evalua-
tion of the behavior factor in the two plan directions, instead of the selection of an
empirical value based on the ductility class and the lateral resisting system adopted.
 JOURNAL OF EARTHQUAKE ENGINEERING 5

Figure 4. Elevation view of the CBF structures.

Table 1. Structure designations.

Lateral resisting system Torsional characteristics Designation
Moment-resisting frames Torsionally unrestrained and plan-regular structures (CM = CR = CV) MRF-TU-PR
(MRF) Torsionally unrestrained and plan-irregular structures (different locations for MRF-TU-PI
CM, CR and CV)
Torsionally restrained and plan-regular structures (CM = CR = CV) MRF-TR-PR
Torsionally restrained and plan-irregular structures (different locations for MRF-TR-PI
CM, CR and CV)
Concentrically braced Torsionally unrestrained and plan-regular structures (CM = CR = CV) CBF-TU-PR
frames (CBF) Torsionally unrestrained and plan-irregular structures (different locations for CBF-TU-PI
CM, CR and CV)
Torsionally restrained and plan-regular structures (CM = CR = CV) CBF-TR-PR
Torsionally restrained and plan-irregular structures (different locations for CBF-TR-PI
CM, CR and CV)
TR—torsionally restrained; TU—torsionally unrestrained; PR—plan-regular; PI—plan-irregular;

The seismic action is evaluated as recommended by EC8 [CEN, 2004], assuming a

Type 1 response spectrum and soil type B for a peak ground acceleration of 0.3 g. The
summary of the selected sections resulting from the seismic design is presented in
Table 2.
The static and seismic design of the gusset plates is made according to the International
Normative CIDECT DG1 [Wardenier et al., 2010], as described by Araújo [2012]. The
dimensions (a and b), the thickness (tw), and the effective width (bw) of the gusset plates
are parameters required to model the gusset plates themselves and their connections.
These parameters are illustrated in Fig. 6 and listed in Tables 3 and 4.
6 R. PERES ET AL.

Table 2. Summary of the selected sections resulting from the seismic design and fundamental periods
of vibration for the structures.
Floor Beams Internal T(s)
Designation level x z External columns columns x z
MRF-TU-PR 3 IPE 330 IPE HEB 550/HEB 340/HEB 300 HEB 550 1.08 1.15
300
2 and 1 IPE 500 IPE
330
MRF-TU-PI 3 IPE 330 IPE HEB 550/HEB 340/HEB 300 HEB 550 1.03 1.07
300
2 and 1 IPE 500 IPE
330
MRF-TR-PR 3 IPE 300 IPE HEB 400/HEB 340/HEB 300 HEB 400 0.99 1.04
300
2 and 1 IPE360/IPE330 IPE
330
MRF-TR-PI 3 IPE 300 IPE HEB 400/HEB 340/HEB 300 HEB 400 0.99 0.95
300
2 and 1 IPE360/IPE330 IPE
330
Beams Braces
Floor
Designation Level x z x z Columns
CBF-TU-PR 3 IPE 330 IPE C H S 139.7 × C H S 139.7 × 3.2 HEB 140 0.80 0.87
300 3.2
2 IPE 360 C H S 139.7 × C H S 139.7 × 6.3 HEB 220
1 6.3 HEB 300
CBF-TU-PI 3 IPE 330 IPE C H S 139.7 × C H S 139.7 × 3.2 HEB 140 0.81 0.66
300 3.2
2 IPE 360 C H S 139.7 × C H S 139.7 × 6.3 HEB 220
1 6.3 HEB 300
CBF-TR-PR 3 IPE 330 IPE C H S 139.7 × C H S 139.7 × 3.2 HEB 300 0.88 0.84
300 3.2
2 IPE 360 C H S 139.7 × C H S 139.7 ×
4.0 6.3
1 C H S 139.7 × C H S 139.7 ×
5.0 8.0
CBF-TR-PI 3 IPE 330 IPE C H S 139.7 × C H S 139.7 × 3.2 HEB 300 0.83 0.72
300 3.2
2 IPE 360 C H S 139.7 × C H S 139.7 ×
4.0 6.3
1 C H S 139.7 × C H S 139.7 ×
5.0 8.0

Table 3. Gusset plate dimensions/parameters (CBF-TU-PR and CBF-TU-PI structures).

Frames in the AA′ direction Frames in the BB′ direction
Story a b tw bw a b tw bw
3 0.70 0.59 0.020 0.505 0.51 0.49 0.010 0.505
2 0.73 0.61 0.020 0.538 0.54 0.52 0.015 0.518
1 0.60 0.55 0.020 0.535 0.50 0.55 0.015 0.526

Table 4. Gusset plate dimensions/parameters (CBF-TR-PR and CBF-TR-PI structures).

Frames in the AA′ direction Frames in the BB′ direction
Story a b tw bw a b tw bw
3 0.70 0.59 0.020 0.505 0.51 0.49 0.010 0.505
2 0.76 0.65 0.020 0.538 0.54 0.52 0.015 0.518
1 0.62 0.57 0.020 0.535 0.51 0.52 0.015 0.526
 JOURNAL OF EARTHQUAKE ENGINEERING 7

3. Numerical Modeling and Analysis Procedures

3.1. Seismic Input
The seismic input for the nonlinear time-history analysis consisted of 15 records (Table 5)
obtained from real earthquake events. The selection of the ground motion records was
conducted with the SelEQ tool [Macedo and Castro, 2017] following the recommenda-
tions of Araújo et al. [2016]. The records were first pre-selected from the PEER’s database
based on geophysical parameters, considering both horizontal components, and after
scaled (both components of the record with the same scale factor) in order that the
average of the group has a good match with the Eurocode 8 spectrum (Type 1; Soil type B,
PGA = 0.30g) in the period range of interest (0.2T1 and 2T1 in accordance to EC8, where
T1 is the fundamental period of the structure), as shown in Fig. 5.
The nonlinear time-history analyses were conducted with the group of records pre-
viously mentioned. The two horizontal record components are considered acting

Table 5. Ground motion records considered.

Earthquake name Earthquake ID Station name Scaling factor
Tabas, Iran 0046 Dayhook 1.28
Imperial Valley-06 0050 Superstition Mtn Camera 6.17
Victoria, Mexico 0064 SAHOP Casa Flores 5.79
Irpinia, Italy-01 0068 Torre Del Greco 7.86
Coalinga-01 0076 Parkfield—Stone Corral 3E 4.33
N. Palm Springs 0101 Anza Fire Station 7.00
Chalfant Valley-02 0103 Tinemaha Res. Free Field 9.00
Whittier Narrows-01 0113 La Habra—Briarcliff 8.00
Loma Prieta 0118 Woodside 4.00
Northridge-01 0127 LA—N Westmoreland 2.83
Kocaeli, Turkey 0136 Mecidiyekoy 8.58
Chi-Chi, Taiwan 0137 HWA038 7.69
Chi-Chi, Taiwain-03 0172 TCU053 10.0
Chi-Chi, Taiwain-05 0174 CHY087 5.63
Chi-Chi, Taiwain-06 0175 TCU068 6.00

Figure 5. Average and target elastic response spectrum of the group of 15 ground motion records.
8 R. PERES ET AL.

simultaneously, first on the x and z directions (AA′ and BB′ of Fig. 1, respectively). After
that, the two components are rotated by 90° and applied on the z and x directions,
resulting in a final set of 30 nonlinear dynamic analyses.
The viscous damping is modeled using the Rayleigh damping formulation, considering
a damping matrix proportional to the tangent stiffness. A damping coefficient equal to
2.5% was assumed for the first mode [Peres, R., 2010].

3.2. General Modeling Assumptions

Both nonlinear static (pushover) and nonlinear time-history analyses are conducted with the
finite element analysis package OpenSEES [Mazzoni et al., 2009]. The material nonlinear
behavior is considered through a fiber modeling approach for both types of structures (MRF
and CBF). Force-based elements are employed, considering the use of a single element for
each beam and column members. The braces are modeled with two force-based elements,
allowing the consideration of an initial geometrical imperfection with values within the range
of 0.1–1.5% of the brace length. The number of integration points varied from four to ten,
depending on the element and model that was being analyzed.
The modeling approach described earlier allowed for the yielding of cross sections as
well as the flexural buckling of the braces, which are the main factors contributing to
stiffness and strength degradation effects in CBFs.
The connections in the MRF were not explicitly modeled because they were considered
to be non-dissipative components in the design process. A centerline modeling approach
has been considered and hence panel zones were not discretely modeled.
Regarding the material model, a simplified bilinear stress–strain constitutive rela-
tionship is assumed for all elements of the MRF structures and beams and columns of
the CBF structures. The Giuffre–Menegotto–Pinto steel material with isotropic strain
hardening, designated as Steel 02 in OpenSEES, is adopted for the braces and gusset
plates of the CBF structures. The strain hardening ratio considered was 1% for both
material models.
Geometrical nonlinearities are also considered in the analyses through the application
in the columns of the P-Delta transformation available in OpenSEES. Gravity loads were
computed and assigned to the model considering the influence areas of the columns.
Conventional pushover analyses are performed assuming a lateral force load vector
following the first mode of vibration of the structures; whereas, adaptive pushover analyses
are conducted through the application of a lateral displacement load vector that is
dependent on the actual deformed pattern calculated at each step of the analysis [Pinho
and Casarotti 2007]. The loads are applied independently in the two horizontal directions
and with positive/negative signs, resulting in four different nonlinear static analyses. The
initial lateral displacement load vector is a three-component displacement vector, which
explicitly incorporates the torsional characteristics of the structures through the addition
of the orthogonal and rotational load components in the direction of the analysis
[Adhikari, 2010]. The consideration of all component sign combinations requires the
conduction of 16 adaptive pushover analyses.
Regarding the application of the NSPs, an elastic–plastic perfectly plastic (EPP)
bilinear representation of the capacity curve is assumed. The CSM iterative process
applied corresponds to procedure B [Applied Technology Council, 2005] and the
 JOURNAL OF EARTHQUAKE ENGINEERING 9

choice of the expressions to compute the ACSM equivalent damping and the spectral
reduction factor is based on the work of Pinho and Casarotti [2007], [2013] and
Monteiro et al. [2014]. The equivalent damping is computed using the approach
proposed by Gulkan and Sozen [1974], which is based on the Takeda model without
hardening; whereas, the spectral reduction factor is obtained using the Lin and Chan
[2003] relationships.

3.3. Modeling Assumptions Related with the Gusset Plate Connection

The gusset plates are initially modeled considering two different approaches: the gusset
plate considered as a force-based element (Model 1—FB element) and the gusset plate
considered as a pin that allows out-of-plane rotation (Model 2—Pin), as illustrated in
Fig. 6. In the first approach, the gusset plate is modeled as a force-based element where the
element cross-section is defined by the thickness (tw) and effective width (bw) of the plate
and the length of the element is taken equal to 2* tw. When the gusset plate is modeled as a
pin (Model 2), a zero-length element is defined in OpenSEES. This element is defined by
two coincident nodes that are connected by a linear elastic spring. In both models, the
rigidity of the gusset plate, the gusset-to-beam, and gusset-to-column connections are
modeled using rigid elastic elements as proposed by Hsiao et al. (2012; 2013).
To decide the model that is more adequate to represent the real behavior of the gusset
plate connection, nonlinear static and time-history analyses were performed and the
results were compared.
The pushover curves presented in Figs. 7 and 8 illustrate the nonlinear static results
obtained with the modal pushover pattern, which is the load pattern that provides a
conservative estimate of the seismic demand. The first figure presents the CBF-TU-PR and
CBF-TU-PI pushover curves in the x and z directions; whereas, the second figure

Figure 6. Representation of the gusset plate connection.

10 R. PERES ET AL.

Figure 7. Pushover curves of CBF-TU-PR and CBF-TU-PI structures in x and z directions.

Figure 8. Pushover curves of CBF-TR-PR-3S and CBF-TR-PI-3S structures in x and z directions.

represents CBF-TR-PR and CBF-TR-PI pushover curves also in the x and z directions
(direction AA′ and BB′ shown in Fig. 3, respectively) for the two modeling approaches.
The analysis of the pushover curves indicates that the torsionally restrained structures
(CBF-TR-PR and CBF-TR-PI) exhibit higher resistance in the x direction than in the z
direction, while the torsionally unrestrained structures (CBF-TU-PR and CBF-TU-PI)
exhibit similar resistance in both directions, despite the differences in behavior in the
case of the irregular structure (CBF-TU-PI). Regarding the models of the gusset plates, it
can be observed that Model 1 provides higher capacity curves for all structures analyzed,
except for the pushover curves obtained in z direction for the torsionally restrained
structures. However, in general, both gusset plate models appear to be suitable for the
seismic assessment of the CBF structures.
To study the nonlinear dynamic behavior, three accelerograms were selected from the
pool of 15 records (records ID 0064, 0174, and 0175) and applied to the models. The three
records were selected to match, as far as possible, the spectrum of each record with the
EC8 target spectrum. Figs. 9–12 illustrate the results of these analyses by comparing the
performance of the two gusset models in both plan horizontal directions (x and z
directions).
 JOURNAL OF EARTHQUAKE ENGINEERING 11

Figure 9. Axial force–deformation relation at brace mid length at first story level for record ID 0175 in x
and z directions (CBF-TU-PR).

Figure 10. Axial force–deformation relation at brace mid length at first story level f or record ID 0174 in
x and z directions (CBF-TU-PI).

Figure 11. Axial force-deformation relation at brace mid length at first story level for record ID 0175 in
x and z directions (CBF-TR-PR).
12 R. PERES ET AL.

Figure 12. Axial force–deformation relation at brace mid length at first story level for record ID 0175 in
x and z directions (CBF-TR-PI).

The analysis of the results indicates, as expected, that the resistance in compression of the
braces is higher when the gusset plate is modeled with a force-based element (Model 1) than
when it is modeled as a pin (Model 2). This is due to the lower buckling length of the brace
in Model 1 that results from the stiffness associated to the gusset plate connections.
Recent studies based on experimental results [Hsiao et al., 2012] of gusset plate/brace
connection models reported that, when the connection is considered fixed, the model
tends to overestimate the compression resistance of the braces; whereas, when it is
considered flexible, the resistance of the braces under compression tends to be under-
estimated. Thus, the use of Model 1, which corresponds to an intermediate solution to
conventional connections, appears to be a more adequate approach to estimate the local
response of the bracing elements, especially when buckling is expected to occur. For this
reason, Model 1 is the model selected to perform all the nonlinear static and time-history
analyses conducted in this work.

4. Results of the Analyses

The results obtained with the nonlinear static and time-history analyses are presented
hereafter for the two orthogonal directions: x direction (direction AA′ shown in Figs. 1
and 3) and z direction (direction BB′ indicated in Figs. 1 and 3).
The global behavior of the structures is qualitatively assessed through the comparison
of the pushover curves obtained using the conventional and adaptive pushover analyses
with the results from incremental time-history analysis. The positive and negative signs
considered in the conventional analyses result in coincident curves; therefore, no distinc-
tion is presented in this paper regarding the sign considered in the analysis. However, the
adaptive pushover curves, which are obtained with a three components vector, depend on
the sign combinations considered in each displacement vector, resulting in a variety of
curves. These curves are coincident in the case of regular structures, but differ when plan-
irregularities are present.
The TH results are presented in terms of the mean of the maximum displacements
obtained at the center of mass at the roof level and the corresponding values of base shear:
 JOURNAL OF EARTHQUAKE ENGINEERING 13

(i) obtained at time step t corresponding to the occurrence of the maximum top displace-
ment, Vdmax, and (ii) the maximum value of base shear, Vmax _dmax, found on the interval
[t − 0.5s, t + 0.5s], where t is the time step in which the maximum top displacement is
attained [Antoniou and Pinho, 2004]. The maximum base shear values observed and the
corresponding displacement at the center of mass were also computed; however, it was
observed that these values occur for displacements lower than the maximum. Moreover,
for this reason, these results are not represented in the figures.
The seismic responses of the structures are estimated for four levels of peak ground
acceleration, namely 0.15 g, 0.3 g, 0.45 g, and 0.60 g and not for the range of top
displacement values that were considered in pushover analyses. The intensity level of 0.30
g corresponds to the design intensity, and the 0.15 g level corresponds to the serviceability
limit state (where the structural response is largely linear elastic). Values of 0.45 g and 0.60 g
were selected to assess the structural response in the nonlinear range. Nevertheless, owing to
numerical convergence problems related with the values of the geometric imperfections
adopted, it was not possible to get results for some structures, specifically for the concen-
trically braced frames and for the 0.45 g and 0.60 g intensity levels.
To evaluate the nonlinear static procedures (the conventional and Extended N2
method, CSM, and ACSM), inter-story drifts and torsional responses are evaluated for
the same levels of seismic intensity considered in the nonlinear time-history analyses.

4.1. Pushover Curves

The capacity (pushover) curves and the results from the incremental nonlinear time-
history analyses obtained for the structures designed with moment-resisting frame (MRF)
systems are plotted in Figs. 13–16. Figs. 17–20 show the corresponding results obtained
for the concentrically braced frame (CBF) structures.
The pushover curves obtained with the conventional and adaptive pushover methods
for the MRF structures (Figs. 13–16) suggest that, for regular structures, all curves are
coincident and provide a conservative (or accurate) estimation of the lateral capacity; for
irregular structures, the conventional pushover curves provide an upper bound estimate
while the adaptive pushover curves provide lower bound estimates of lateral strength.

Figure 13. Pushover curves and time-history results obtained for the MRF-TU-PR structure in x and z
directions.
14 R. PERES ET AL.

Figure 14. Pushover curves and time-history results obtained for the MRF-TU-PI structure in x and z
directions.

Figure 15. Pushover curves and time-history results obtained for the MRF-TR-PR structure in x and z
directions.

Figure 16. Pushover curves and time-history results obtained for the MRF-TR-PI structure in x and z
directions.
 JOURNAL OF EARTHQUAKE ENGINEERING 15

Figure 17. Pushover curves and time-history results obtained for the CBF-TU-PR structure in x and z
directions.

Figure 18. Pushover curves and time-history results obtained for the CBF-TU-PI structure in x and z
directions.

Figure 19. Pushover curves and time-history results obtained for the CBF-TR-PR structure in x and z
directions.
16 R. PERES ET AL.

Figure 20. Pushover curves and time-history results obtained for the CBF-TU-PI structure in x and z
directions.

Furthermore, the adaptive pushover analysis always leads to the lowest pushover curves,
being the difference between the pushover curves resistance more significant for the
torsionally unrestrained structures (more flexible structures).
The results obtained for the CBF structures (Figs. 17–20) indicate that the adaptive
pushover analysis is able to capture the dynamic behavior of the structures, which mainly
depends on the response of the braces. It is interesting to note that the expected softening
response, owing to brace instability, is captured much better by the adaptive pushover
analysis. Regarding the different adaptive pushover curves obtained, the results are
identical to those obtained for the MRF structures, since the curves are coincident in
the case of regular structures and the variations are observed when the structures are
irregular. Similarly, the variations are significant for the case of the CBF-TU-PI structure
(flexible structure).

4.2. Inter-Story Drifts

To assess the performance of the NSPs, the target displacements and the correspond-
ing inter-story drifts, measured at the center of mass, are evaluated for the same
levels of seismic intensity considered in the TH analyses, and compared with the
inter-story drifts obtained with the latter. The target of the TH results is given by the
mean of the maximum top displacements and corresponding inter-story drifts (TH);
whereas, the values of the mean plus and minus the standard deviation (TH ± SD)
are given to provide the reader additional information about the dispersion of the
results.
Figures 21–24 illustrate the inter-story drifts obtained for the MRF structures (MRF-
TU-PR, MRF-TU-PI, MRF-TR-PR, and MRF-TR-PI), for the intensity levels of 0.30 g and
0.60 g. The analysis of the figures indicates that, for both intensities, all the NSPs over-
estimate the seismic response in both directions, both the N2 methods and the CSM being
more accurate than the ACSM. An exception to this trend is the set of results obtained for
the design intensity level in the z direction, where the three procedures provide very close
estimates of the inter-story drifts.
 JOURNAL OF EARTHQUAKE ENGINEERING 17

Figure 21. Inter-story drifts obtained for the MRF-TU-PR structure.

Figure 22. Inter-story drifts obtained for the MRF-TU-PI structure.

18 R. PERES ET AL.

Figure 23. Inter-story drifts obtained for the MRF-TR-PR structure.

Figure 24. Inter-story drifts obtained for the MRF-TR-PR structure.

 JOURNAL OF EARTHQUAKE ENGINEERING 19

Figure 25. Inter-story drifts obtained for the CBF-TU-PR structure.

Figure 26. Inter-story drifts obtained for the CBF-TU-PI structure.

Figure 27. Inter-story drifts obtained for the CBF-TR-PR structure.

Figures 25–28 show the inter-story drifts obtained for the CBF structures (CBF-TU-PR,
CBF-TU-PI, CBF-TR-PR, and CBF-TR-PI) for the intensity level of 0.30 g. The results
obtained indicate that, in general, both N2 methods and the CSM provide very accurate
estimates of the seismic response of the studied frames in both directions. Regarding the
20 R. PERES ET AL.

Figure 28. Inter-story drifts obtained for the CBF-TR-PI structure.

effectiveness of the ACSM, it provides accurate approximations at the first story levels (in
some cases better than the estimated obtained with the N2 methods and the CSM) but
overestimates the results at the second and third story levels for the regular structures and,
in the x direction, for the irregular structures. In the z direction for the irregular
structures, the ACSM provides conservative results. Furthermore, the ACSM is the only
method that provides a good estimate of the distribution of inter-story drifts for the CBF-
TU-PI structure.

4.3. Torsional Responses

The torsional responses are estimated through the normalization of the edge displacements at
the roof level with respect to the corresponding displacements at the center of mass.
The results obtained for the irregular structures (MRF-TU-PI, MRF-TR-PI, CBF-TU-
PI, and CBF-TR-PI), which are the structures that exhibit torsional displacements in the
direction of the eccentricities (z direction), are plotted in Figs. 29–32.
The inspection of the results obtained for the MRF-TU-PI structure (Fig. 29) indicates
that, for the design intensity level, the N2 and the CSM methods predict lower deformations
than the TH results at both edges of the structure, although these deformations are very close
to the TH results at the flexible edge. On the other hand, for the 0.60 g intensity level, the N2
and the CSM methods underestimate the displacements at the stiff edge of the structure
(right edge frame) and slightly overestimate the displacements at the flexible edge (left edge
frame). The ACSM significantly underestimates the displacements at the stiff edge and
overestimates the displacements at the flexible edge, for both intensity levels considered.
In general, one can conclude that the application of the Extended N2 method results in
conservative estimates of structural deformations at both edges and for the two seismic
intensities considered in this study.
Regarding the MRF-TR-PI structure (Fig. 30), it can be concluded that the N2 and the
CSM methods underestimate the displacements at the flexible edge of the structure (left
edge), for the two intensity levels, whereas the ACSM overestimates the edge displace-
ments. Regarding the stiff edge (right edge), both the N2 and the CSM methods provide
an accurate estimation of the displacements, while the ACSM clearly underestimates the
 JOURNAL OF EARTHQUAKE ENGINEERING 21

Figure 29. Torsional displacements obtained for the MRF-TU-PR structure.

same deformations. The Extended N2 method provides a better estimation of the dis-
placements at the flexible edge in comparison to the conventional N2 and CSM methods.
However, it overestimates the deformations at the stiff edge.
Concerning the CBF-TU-PI structure (Fig. 31), all the nonlinear static procedures
overestimate the displacement at the flexible edge (left edge) and highly underestimate
the displacements at the stiff edge (right edge), except for the Extended N2 method, which
provides a closer and safe estimation at the stiff edge.
As for the CBF-TR-PI structure (Fig. 32), both conventional and Extended N2 meth-
ods, as well as the CSM, underestimate the displacements at the flexible edge (left edge)
and overestimate the displacements at the stiff edge (right edge). For this structure, the
ACSM underestimates the displacements at both stiff and flexible edges.
In conclusion, the Extended N2 method appears to be appropriate to accurately
estimate the torsional response of the irregular steel structures considered in this study,
except for the CBF-TR-PI structure at the flexible edge, which exhibits higher torsional
displacements in the nonlinear range than in the linear range. This is due to the
amplification of displacements that results from the concentration of plasticity on the
flexible side of the structure, combined with the degradative hysteretic behavior
associated to brace buckling. Fig. 33 shows the hysteretic response of the first story
braces located at the right (flexible) and left (stiff) sides of the CBF-TR-PI structure,
obtained for record ID 0175. Similar conclusions have been reported by Erduran and
22 R. PERES ET AL.

Figure 30. Torsional displacements obtained for the MRF-TR-PI structure.

Figure 31. Torsional displacements obtained for the CBF-TU-PI structure.

Ryan (2011) in their study on the effect of torsion on the behavior of peripheral steel-
braced frames.
It should be noted that the results presented herein depend on the level of inelasticity
experienced by the structures for each intensity level analyzed, especially that pertaining to
torsional response. The torsional response is highly affected by the level of inelasticity
attained by the structures lateral resisting systems and their location in plan. In most cases,
 JOURNAL OF EARTHQUAKE ENGINEERING 23

Figure 32. Torsional displacements obtained for the CBF-TR-PI structure.

Figure 33. Hysteretic response of the first story braces located at the right (flexible) and left (stiff) side
of the CBF-TR-PI structure (record ID 0175) .
24 R. PERES ET AL.

the increase of inelastic demand leads to a reduction of the torsional response. However, it
is observed that, for the case of concentrically braced frames torsionally stiff and plan-
irregular structures, the increase of inelastic demand leads to the amplification of the
torsional displacements, especially at the flexible edge of the structure.

5. An Improvement of the Extended N2 Method for Plan-Asymmetric and

Torsionally Stiff Structures
To overcome the limitation identified above regarding the application of the Extended N2
method to torsionally stiff structures with concentrically braced frames (CBF-TR-PI), an
improvement of the Extended N2 method is herein proposed. It consists on the amplifica-
tion of the demands obtained at the flexible edge through the application of a corrective
factor that is a function of the linear torsional response of the equivalent torsional flexible
structure. The equivalent torsionally flexible structure should be defined as a structure that
has the same number of lateral resisting systems in both directions, but positioned in a
way that the structure becomes torsionally flexible instead of torsionally stiff. This can be
achieved by placing the lateral resisting systems outside of the structure’s perimeter.
Additionally, a torsionally flexible structure is characterized by having a torsional funda-
mental period of vibration. In this study, the equivalent torsional flexible structure is
coincident with the CBF-TU-PI structure.
The result of the application of the linear corrective factor to the torsionally stiff
structure (CBF-TR-PI) is presented in Fig. 34, which indicates that the improvement
to the Extended N2 method provides a conservative estimate of the torsional
response of the CBF-TR-PI structure, although it is preferable to the underestimated
response obtained with the ACSM method, even if the latter is closer to the TH
response.
The improvement to the Extended N2 method is to be applied to torsionally stiff and
irregular structures, whenever the structural response in the nonlinear range exhibits

Figure 34. Torsional displacements obtained for CBF-TR-PI-3S structure after the application of the
corrective factor.
 JOURNAL OF EARTHQUAKE ENGINEERING 25

higher torsional displacements than in the linear range, which typically happens at the
flexible edge of structures with braced frames. The improvement can be summarized in
the following steps:

(1) The steps are the same as for the original N2 method until the target displacement
is reached. The pushover analysis should be performed with positive and negative
signs in both plan directions of a 3D model. For each analysis, the target displace-
ment is computed, the larger value being chosen in each direction.
(2) Afterwards, a linear analysis of the 3D model is performed in the two horizontal
directions.
(3) Evaluation of the Extended N2 corrective factors. The correction factors are defined
as the ratio between the normalized roof displacements obtained from the elastic
analysis and the pushover analysis. The normalized roof displacement is the roof
displacement at an arbitrary location divided by the roof displacement at the CM.
The correction factors should be defined for each horizontal direction separately.
The amplification due to torsion is obtained by multiplying the relevant results of
pushover analysis, determined in the beginning of the procedure, by these correc-
tion factors, while the reduction of the demands due to torsion is not considered.
(4) Definition of the equivalent torsionally flexible structure and linear analysis of the
corresponding 3D model in the two horizontal directions. Evaluation of the linear
torsional displacements (or linear corrective factors) through the normalization of
the roof edge displacements with respect to the CM displacements. These linear
torsional corrective factors must be evaluated for both horizontal directions, simi-
larly to the Extended N2 method.
(5) The amplification of the demands at the flexible edge of the plan-asymmetric
torsionally stiff structure is determined through the application of the linear
corrective factor, obtained in the previous step, to the demands obtained at the
end of step 3. At the stiff edge, the Extended N2 method already provides a good
estimation of the edge displacements since the method does not consider the
reduction of seismic demand due to torsional effects.

It is worth highlighting that the same study was also performed on one-story structures
with the same plan configuration [Peres R., 2017]. Similar conclusions were achieved
regarding the effectiveness of the NSPs and also concerning the application of corrective
factors to the Extended N2 method to evaluate the torsional response of torsionally stiff
and plan-irregular structures composed by concentrically braced frames.

6. Conclusions
This paper reports on the application of a group of nonlinear static procedures (conven-
tional and Extended N2 methods, CSM with the features of FEMA440 and ACSM) to a set
of eight, three stories high steel structures, consisting of MRF and CBF as lateral seismic
resisting systems, with regular and irregular plan configurations. The main objective was
to assess the effectiveness of the NSPs through a comparison with the results obtained by
nonlinear time-history (TH) analysis.
26 R. PERES ET AL.

The global behavior of the structures was evaluated based on the comparison of the
pushover curves obtained using the conventional and adaptive pushover methods with the
results from TH analyses.
The NSPs were assessed in terms of translational and torsional responses through the
evaluation of the inter-story drifts and normalization of the edge displacements with
respect to the center of mass displacements, respectively.
In terms of translational response, the results indicate that, for structures with MRF as
lateral seismic resisting systems, all the NSPs overestimate the response in both directions.
The N2 and CSM methods are more accurate than the ACSM, except for the results
obtained for the design intensity level in the z direction, where the three procedures
provided very close estimates of the inter-story drifts. Regarding the structures with CBF
as lateral seismic resisting systems, the results indicate that, in general, the N2 and the
CSM methods provide very accurate estimates of the structural response in both plan
directions. The effectiveness of the ACSM method provides accurate approximations at
the first story levels, in some cases better than the N2 and the CSM methods. However, the
method overestimates the response at the second and third story levels.
With reference to the torsional behavior, it was found that, as expected, torsional
displacements were observed in the direction of the eccentricities (z direction) in the
case of plan-irregular structures. The analysis of the results indicates that, depending on
the type and location of the lateral resisting systems, different conclusions can be drawn.
However, in general, the Extended N2 method is shown to be appropriate to accurately
estimate the torsional response of the studied irregular steel structures, except for the
CBF-TR-PI structure at the flexible edge. To overcome this limitation, an improvement of
the Extended N2 method is proposed for torsionally stiff structures and plan-irregular
structures composed by concentrically braced frames. It consists on the amplification of
the displacements obtained at the flexible edge through the application of a corrective
factor that is a function of the linear torsional response of the equivalent torsional flexible
structure.
Given the limited number of cases studied, further analyses on torsionally stiff plan-
irregular structures should be conducted to extract definite conclusions regarding the
validity of this corrective factor. Nevertheless, this work contributes to the progress
beyond the current state of the art, taking a step further on the application of pushover
analysis to plan-irregular structures.

ORCID
Rita Bento http://orcid.org/0000-0002-6503-0644
José Miguel Castro http://orcid.org/0000-0001-9732-9969

References
Adhikari, G. (2010), “The adaptive capacity spectrum method for reinforced concrete buildings”,
PhD Dissertation, Università degli Studi di Pavia, Pavia, Italy.
Antoniou, S. and Pinho, R. (2004). Advantages and limitations of adaptive and non-adaptive force-
based pushover procedures. Journal of Earthquake Engineering, 8(4), 497–522. doi: 10.1080/
13632460409350498.
 JOURNAL OF EARTHQUAKE ENGINEERING 27

Applied Technology Council (1996) Seismic evaluation and retrofit of concrete buildings, vols.1 and
2, Report No. ATC-40, Redwood City, CA.
Applied Technology Council (2005) Improvement of nonlinear static seismic analysis procedures,
FEMA 440 Report, Redwood City, CA.
Araújo, A. (2012) “Study of the behaviour of tubular members to gusset plate connections”, MSc
Dissertation, University of Porto, Porto, Portugal (in Portuguese).
Araújo, M., Macedo, L., Marques, M. and Castro, J. M. (2016). “Code-based record selection
methods for seismic performance assessment of buildings”. Earthquake Engineering and
Structural Dynamics, 45, 129–148. doi: 10.1002/eqe.2620.
Bento, R., Bhatt, C. and Pinho, R. (2010). Using nonlinear static procedures for seismic assessment
of 3D irregular SPEAR building. Earthquakes and Structures, 1(2), 1–19. doi: 10.12989/
eas.2010.1.2.177.
Bosco, M., Ghersi, A. and Marino, E. M. (2012). “Corrective eccentricities for assessment by the
nonlinear static method of 3D structures subjected to bidirectional ground motions”.
Earthquakes Engineering and Structural Dynamics, 41, 1751–1773. doi: 10.1002/eqe.2155.
CEN. (2004). EN1998-1-3, Eurocode 8: Design of Structures for Earthquake Resistance- Part 1:
General Rules, Seismic Actions and Rules for Buildings. Brussels, Belgium: European Committee
for Standardization.
CEN. (2005). EN1998-1-1, Eurocode 3: Design of Steel Structures- Part 1: General Rules, Seismic
Actions and Rules for Buildings. Brussels, Belgium: European Committee for Standardization.
Chopra, A. K. and Goel, R. K. (2002). A modal pushover analysis procedure for estimating seismic
demands for buildings. Earthquake Engineering and Structural Dynamics, 31, 561–582. doi:
10.1002/eqe.144.
Chopra, A. K. and Goel, R. K. (2004). A modal pushover analysis procedure to estimate seismic
demands for unsymmetric-plan buildings. Earthquake Engineering and Structural Dynamics, 33,
903–927. doi: 10.1002/eqe.380.
Erduran, E. and Ryan, K. L. (2011). “Effects of torsion on the behaviour of peripheral steel-braced
frame systems”. Earthquake Engineering and Structural Dynamics, 40, 491–507. doi: 10.1002/
eqe.1032.
Fajfar, P. (1999). Capacity spectrum method based on inelastic demand spectra. Earthquake
Engineering and Structural Dynamics, 28, 979–993. doi: 10.1002/(SICI)1096-9845(199909)
28:9<979::AID-EQE850>3.0.CO;2-1.
Fajfar, P. and Fishinger, M. (1988), “N2 – A method for non-linear seismic analysis of regular
buildings”, Proc. of the 9th World Conference on Earthquake Engineering, Tokyo-Kyoto,
Japan.
Fajfar, P., Marus̆ Ić, D. and Perus̆ , I. (2005). Torsional effects in the pushover-based seismic
analysis of buildings. Journal of Earthquake Engineering, 9(6), 831–854. doi: 10.1080/
13632460509350568.
Freeman, S. A., Nicoletti, J. P., and Tyrell, J. V. (1975), “Evaluation of existing buildings for seismic
risk - A case study of Puget Sound Naval Shipyard, Bremerton, Washington,” Proceedings of U.S.
National Conference on Earthquake Engineering, Berkeley, USA, pp. 113-122.
Gulkan, P. and Sozen, M. (1974). Inelastic response of reinforced concrete structures to reinforced
concrete structures. American Concrete Institute Journal, Vol 71, 604–610.
Kalkan, E. and Kunnath, S. K. (2006). Adaptive modal combination procedure for nonlinear static
analysis of building structures. Journal of Structural Engineering, 132(11), 1721–1731. doi:
10.1061/(ASCE)0733-9445(2006)132:11(1721).
Lin, Y. and Chang, K. C. (2003). A study on damping reduction factors for building under
earthquake ground motion. ASCE Journal of Structures Engineering, Vol.129(2), 206–214. doi:
10.1061/(ASCE)0733-9445(2003)129:2(206).
Macedo, L. and Castro, J. M. (2017), SelEQ: An advanced ground motion record selection and
scaling framework, Advances in Engineering Software. 10.1016/j.advengsoft.2017.05.005.
Mazzoni, S., McKenna, F., Scott, M. H. and Fenves, G. L. (2009). Open system for earthquake
engineering simulation user command-language manual—openSees version 2.0, Pacific
Earthquake Engineering Research Center, University of California, Berkeley, CA.
28 R. PERES ET AL.

Monteiro, R., Marques, M., Adhikari, G., Casarotti, C. and Pinho, R. (2014). Spectral reduction
factors evaluation for seismic assessment of frame buildings. Engineering Structures, 77, 129–142.
doi: 10.1016/j.engstruct.2014.07.041.
Peres, R. (2010), Comparison of European and American approaches for consideration of P-Δ
effects in seismic design, MSc dissertation, ROSE School, Pavia, Italy
Peres, R. (2017), “Performance based seismic design and assessment of irregular steel structures”,
PhD Dissertation, Instituto Superior Técnico, Universidade de Lisboa, Portugal.
Peres, R., Castro, J. M. and Bento, R. (2016). An extension of an improved forced based design
procedure for 3D steel structures. Steel Composite Structures, 22(5), 1115–1140. doi: 10.12989/
scs.2016.22.5.1115.
Pinho, R. and Casarotti, C. (2007). “An adaptive capacity spectrum method for assessment of
bridges subjected to earthquake motions”. Bull Earthquake Engineering, 5, 377–390. doi: 10.1007/
s10518-007-9031-8.
Pinho, R., Marques, M., Monteiro, R., Casarotti, C. and Delgado, R. (2013). Evaluation of nonlinear
static procedures in the assessment of buildings frames. Earthquake Spectra, 29(4), 1459–1476.
doi: 10.1193/100910EQS169M.
Po-Chien, H., Lehman, D. E. and Roeder, C. W. (2012). Improved analytical model for special
concentrically braced frames. Journal of Constructional Steel Research, 73(21012), 80–94. doi:
10.1016/j.jcsr.2012.01.010.
Po-Chien, H., Lehman, D. E. and Roeder, C. W. (2013). “A model to simulate special concentrically
braced frames beyond brace fracture”. Earthquake Engineering and Structural Dynamics, 42, 183–
200. doi: 10.1002/eqe.2202.
Wardenier, J., Kurobane, Y., Packer, A. J., Van Der, V. G. J. and Zhao, X. L. (2010), Design Guide
for circular hollow section (CHS) joints under predominantly static loading, Comité
International pour le Développement et l‟Étude de la Construction Tubulaire.