Received: 7 June 2016 Revised: 21 March 2017 Accepted: 6 May 2017

DOI: 10.1002/tal.1399

RESEARCH ARTICLE

Effect of vertical structural irregularity on seismic design of tall

buildings
Aman Mwafy | Sayed Khalifa

United Arab Emirates University, Al‐Ain,

United Arab Emirates Summary
Correspondence Many tall buildings are practically irregular as an entirely regular high‐rise building rarely exists.
Aman Mwafy, Department of Civil and This study is thus devoted to assessing the approach and coefficients used in the seismic design
Environmental Engineering, United Arab of real‐life tall buildings with different vertical irregularity features. Five 50‐story buildings are
Emirates University, P.O. Box 15551, Al‐Ain,
selected and designed using finite element models and international building codes to represent
United Arab Emirates.
Email: amanmwafy@uaeu.ac.ae the most common vertical irregularities of reinforced concrete tall buildings in regions of medium
Funding information seismicity. Detailed fiber‐based simulation models are developed to assess the seismic response
United Arab Emirates University, Grant/Award of the five benchmark buildings under the effect of 40 earthquake records representing far‐field
Number: 31N132 and 31N227
and near‐source seismic scenarios. The results obtained from a large number of inelastic pushover
and incremental dynamic analyses provide insights into the local and global seismic response of
the reference structures and confirm the inferior local response of tall buildings with severe ver-
tical irregularities. Due to the significant impacts of the severe irregularity types on the seismic
response of tall buildings, the conservative code approach and coefficients are recommended
for design. It is also concluded that although the design coefficients of buildings with moderate
irregularities are adequately conservative, they can be revised to arrive at more consistent safety
margins and cost‐effective designs.

KEY W ORDS

dynamic response, seismic design codes, seismic design coefficients, tall buildings, vertical
irregularity

1 | I N T RO D U CT I O N

The structural irregularity is widely observed in buildings as a result of the architectural and service requirements in the design process, errors and
modifications during the construction phase, and changes in the building use throughout its service life. Modern seismic design codes differentiate
between the plan and vertical irregularity.[1,2] The plan (horizontal) irregularity occurs as a result of several reasons such as when the structure is
significantly subjected to torsion or exhibits a discontinuity in the lateral force resisting system (LFRS) out of its plane (out‐of‐plane offset). The
vertical irregularity occurs due to significant changes in stiffness, strength, mass, or dimensions, or as a result of an in‐plane discontinuity in LFRS.
There is also a common trend in previous studies to distinguish between the plan and vertical irregularities.[3] The growing interest in studying the
seismic behavior of irregular buildings has been observed in the scientific literature, particularly for vertical irregularity (e.g., these previous
studies[4–10]). However, the impact of different types of vertical irregularity on the seismic design of buildings has not been covered systematically
in the literature, particularly the severe irregularities of real‐life high‐rise buildings, as subsequently described.
As per the U.S. design standards and guidelines,[1,11] a building exhibits an extreme soft story vertical irregularity if a story lateral stiffness is (a)
less than 60% of that in the story above, or (b) less than 70% of the average stiffness of the three stories above. When a story mass is more than
150% of that of an adjacent story, the mass irregularity takes place. The vertical geometric irregularity occurs when the horizontal dimension of
LFRS in a story exceeds 130% of that in an adjacent story. A building exhibits an in‐plane discontinuity if there is an in‐plane offset of a vertical
lateral force resisting member that exceeds the length of this member. The extreme weak story irregularity takes place if a story lateral strength
is less than 65% of that of the story above. Additionally, the European design codes and guidelines[2,12] mainly categorize structures into regular
and nonregular. A building is to be vertically regular if the following conditions are valid: (a) all LFRSs run without any interruption from their base

Struct Design Tall Spec Build. 2017;e1399. wileyonlinelibrary.com/journal/tal Copyright © 2017 John Wiley & Sons, Ltd. 1 of 22
https://doi.org/10.1002/tal.1399
2 of 22 MWAFY AND KHALIFA

up to the roof; (b) the lateral stiffness and mass of all stories remain constant or reduce gradually along the building height; (c) in framed buildings,
the ratio of the actual story resistance to that required by analysis should not vary significantly between adjacent stories; (d) setbacks of each side
of the building do not exceed 10% of the parallel dimension at the story below, and for unsymmetric setbacks, the total setbacks at all stories
should not exceed 30% of the plan dimension at the ground floor; and (e) for a single setback within the lowest 15% of the building height, it should
not exceed 50% of the dimension at the base. The building exhibits vertical irregularity as per Eurocode‐8[2] if it does not satisfy the above
conditions.
Table 1 summarizes the definitions of international codes for different types of vertical irregularity. It is shown that design codes categorize the
vertical irregularity to five types: (a) stiffness, (b) mass, (c) geometry, (d) in‐plane discontinuity of a vertical force resisting element, and (e) discon-
tinuity in the LFRS strength. Unlike Eurocode‐8, ASCE/SEI‐7 divides the stiffness irregularity into (a) soft story and (b) extreme soft story, and the
discontinuity of lateral strength is divided into (a) weak story and (b) extreme weak story.[1,2] These detailed definitions of the soft story and weak
story irregularities reflect the importance of imposing different design requirements according to the level of severity of these two irregularities.
This study is directed towards the evaluation of the force‐based seismic design procedure adopted by existing building codes, particularly its
applications to vertically irregular high‐rise buildings. An analytical model that accounts for all sources of stiffness, P‐delta effects, and the inelastic
response is the most accurate approach for the seismic design of structures. Development of such an analytical model is costly, and hence, the
inelastic seismic response is accounted for in force‐based design approaches through the use of the response modification factor, R, deflection
amplification factor, Cd, and overstrength factor, Ωo.[1,2,13] Figure 1 describes these factors, which are termed the seismic design response factors
in this study. FEMA‐P695[13] proposed an approach to quantify these design factors, in which the R factor was related to the ratio of the spectral
acceleration of the maximum considered earthquake at the period of the structural system, SMT, and the seismic response coefficient, Cs. The Ωo
factor was related to the ratio of the ultimate strength of the structure, Smax, to the Cs coefficient. The R and Cd factors were considered to be equal
in this approach based on the equal displacement rule, which is a reasonable assumption for most systems with approximately 5% effective
damping. Detailed descriptions of the abovementioned parameters are provided by FEMA‐P695.[13] Moreover, the results of incremental dynamic

TABLE 1 Definition of vertical irregularity according to international seismic design codes[1,2]

Design code
Type of vertical irregularity U.S. design standards and guidelines[1],[11] European design code[2]

Stiffness–soft story Ki < 70% Ki+1 Ki ≠ Ki+1

Stiffness–extreme soft storya Ki < 60% Ki+1 N/A
Mass Mi > 150% Mi±1 Mi ≠ Mi+1
a
Geometry Li > 130% Li±1 Li > 120–150% L i+1
In‐plane discontinuitya Lo > Lb When discontinuity exists
Discontinuity in lateral strength–weak story Stri < 80% Stri+1 Stri ≠ Stri±1
Discontinuity in lateral strength–extreme weak storya Stri < 65% Stri+1 N/A

Note. Ki = stiffness of the soft story; Ki+1 = stiffness of the floor above the soft story; Mi = mass of a story; Mi±1 = mass of adjacent story; Li = length of a
story; Li±1 or Li+1 = length of the story adjacent to or above the irregular one; Lo = vertical element offset; Lb = vertical element length in the story below the
irregular story; Stri = lateral strength of weak story; Stri±1 or Stri+1 = lateral strength of adjacent story or the story above the weak one. ≠ Indicates a signif-
icant change.
a
Irregularity types investigated in the present study.

FIGURE 1 Description of seismic design response factors

MWAFY AND KHALIFA 3 of 22

analysis (IDA) and inelastic pushover analysis (IPA) were utilized to evaluate the seismic design factors of regular structures in previous studies (e.g.,
Elnashai and Mwafy, Kim and Choi, Mwafy, and Mwafy[14–17]) The R factor was calculated as follows: R = (PGAc/y) Ωfy, where PGAc/y is the ratio
of the peak ground acceleration (PGA) near collapse to that at the first indication of yielding, and Ωfy is the overstrength factor at the first indication
of yielding (ratio of first yielding to design strength, Vfy/Vs, as described in Figure 1). The deflection amplification factor was considered to be equal
to IDRc/y, which is the ratio of the maximum interstory drift ratio (IDR) corresponding to the collapse prevention (CP) limit state to that at the first
indication of yielding. The seismic design factors were compared with the code values, which proved that the code coefficients are conservative for
regular buildings. Additional discussions related to the of the latter approach for evaluating the seismic design factors are provided by
Mwafy, Mwafy and Elnashai, and Mwafy[16–18]. Several methods may be suggested to quantify the seismic design response factors. However, it
is beyond the scope of this study to cover all approaches.
The abovementioned brief review related to vertical structural irregularity and the effort made in previous studies to assess the design coef-
ficients of regular structures highlights the pressing need to assess the seismic design approach and coefficients recommended by building codes
for different types of irregular high‐rise buildings and verify their relative safety margins systematically. The objective of the current study is thus to
assess the impacts of different types of vertical irregularity on the existing seismic design procedure using a wide range of input ground motions
and rational performance criteria aiming at providing practical recommendations for the design. The study contributes to recent activities for
verifying and revising the seismic design approaches of tall buildings aiming at having different structures with comparable reliability and cost
effectiveness.

2 | S E L E C T I O N A N D D E S I G N O F V E R T I C A L LY I R R E G U L A R ST R U C T U RE S

One of the main tasks of the current study is to select representative vertically irregular buildings. The reference buildings are chosen based on a
concise survey of the common types of irregular high‐rise structures in the UAE, a medium seismicity region that is selected as a case study due to
its rapid rate of high‐rise building construction. Abrupt changes in the stiffness, geometric dimensions, or strength of the LFRS along the building
height due to architectural and service requirements represent the most common vertical irregularities in the case study area. Five 50‐story rein-
forced concrete (RC) high‐rise buildings are thus selected for the purpose of the current study. The selected buildings are denoted B1‐REG, B2‐SST,
B3‐GEO, B4‐DIS, and B5‐WST, which characterize a regular structure, extreme soft story irregularity, geometric irregularity, irregular building with
in‐plane discontinuity, and extreme weak story irregularity, respectively, as shown in Table 2. The regular building is used for comparison with other
irregular structures. The definitions of the selected irregularities are as per the ASCE‐7 provisions,[1] as explained in Table 1. Table 2 summarizes the
main structural characteristics of the selected reference structures, whereas Figure 2 depicts their layouts, configurations, and LFRSs. It is impor-
tant to note that, in the present study, the term “moderate irregularities” is adopted to describe those represented by B2‐SST and B3‐GEO, and the
term “severe irregularities” describes those represented by B4‐DIS and B5‐WST.
As shown in Figure 2, B2‐SST represents the extreme soft story irregularity. The increased height of the ground story (6.5 m), which is more
than double the height of the story above, causes a significant reduction in stiffness. This irregularity is practically shown in many multistory build-
ings. The ratio between the calculated initial stiffness of the ground story to that of the first story is less than 60%, which results in an extreme soft
story irregularity.[1] Figure 2(b,g) shows that B3‐GEO exemplifies the vertical geometric irregularity according to ASCE/SEI‐7.[1] The footprint
dimensions of the basement stories are 42.0 m × 45.5 m, and the layout dimensions of the ground and typical stories are 42.0 m × 29.2 m. The
ratio between the LFRS length at the basement and ground stories is thus more than 130%. Hence, the building has a vertical geometric
irregularity.[1]
As shown in Figure 2(a,c,h), B4‐DIS represents buildings with an in‐plane discontinuity of LFRS. A transfer slab at the first story level is intro-
duced to support the planted walls of typical stories. This transfer slab is supported by RC cores and columns, as shown in Figure 2(a,c). The transfer
slab is much thicker and heavier than the typical story slabs, as noted in the surveyed buildings and previous studies (e.g., Li et al.[19]). As shown in
Figure 2(d,i), B5‐WST represents the extreme weak story irregularity. Due to the significant changes of the LFRS at the basement and ground
stories (i.e., replacing shear walls with columns), the lateral strength of the lower stories considerably decreases. The flexural and shear strength
values are calculated for different vertical structural members at the ground and first stories using the design code approach.[20] The ratio of the
flexural/shear strength at the ground story to that at the first story is less than 65% of the story above. Therefore, the building exhibits an extreme
weak story irregularity as per the ASCE/SEI‐7[1] definition.

TABLE 2 Characteristics of benchmark irregular buildings

Building Typical story height Ground story height First basement height Total height
reference Building irregularity type (m) (m) (m) (m)

B1‐REG Regular building 3.2 3.2 3.2 160

B2‐SST Stiffness/extreme soft story irregularity 3.2 6.5 3.2 163.2
B3‐GEO Geometric irregularity 3.2 3.2 3.2 160
B4‐DIS In‐plane discontinuity irregularity 3.2 4.7 4.7 163
B5‐WST Discontinuity in lateral strength/weak story irregularity 3.2 3.2 3.2 160
4 of 22 MWAFY AND KHALIFA

FIGURE 2 Description of five reference structures: (a) typical layout of structures except at the irregularity levels; (b–d) layouts of B3‐GEO, B4‐
DIS, and B5‐WST at the irregularity levels, respectively; and (e–i) configuration and LFRSs of reference structures

Shear walls are mainly used as the lateral force resisting elements because they are efficient in controlling the lateral deformations
developed by wind or earthquake loads (e.g., Ali and Moon[21]). Flat slabs with marginal beams are employed as horizontal diaphragms to
transfer the gravity loads to vertical elements. In buildings B4‐DIS and B5‐WST, columns are used in the lower stories because of their
irregularities, as shown in Figure 2(c,d). The five reference structures are designed for the purpose of this study using the ACI‐318[20]
and ASCE/SEI‐7[1] codes. Three‐dimensional (3D) simulation models are developed using ETABS,[22] which is widely used for the design
of the multistory buildings. Modal response spectrum analysis is employed to estimate the lateral seismic forces using the design code spec-
trum.[1] It is noteworthy that the selected earthquake records for this assessment study represent two seismic scenarios pertinent to the
case study area (Dubai, UAE): far‐field events and short source‐to‐site earthquakes. The far‐field records fit the design response spectrum
in the long period range, and the near‐field seismic events match the design spectrum in the short period range, as subsequently discussed.
The 3D ETABS models account for the accidental torsion under both the seismic and wind loads as well as the stiffness and strength values
of structural members as per the design code.[20] The concrete strength, fc`, varies throughout the height of vertical element starting from
48 MPa (cube strength, fcu, of 60 MPa) at the foundation to 32 MPa (fcu of 40 MPa) at the roof. Cube concrete strength of 40 MPa is used
for all slabs and beams. The yield strength, fy, of reinforcing steel bars is 460 MPa for flexural design and 420 MPa for shear design.[20]
Permanent loads include the self‐weight of structural members with superimposed dead load of 4.0 kN/m2. Live loads are 2.0, 4.8, and
3.0 kN/m2 for the residential areas, corridors and staircases, and basements (parking areas), respectively.[1] The case study area (Dubai,
UAE) represents a region of medium seismicity. The spectral response accelerations at 0.2 and 1.0 s are 0.83 and 0.24 g, respectively,
and the site class is “C” (very dense soil).[23] The R factor is 4.0, Ωo factor is 2.5, and seismic design category, SDC, is “C”.[1] The adopted
seismic design criteria result in a design base shear of 20,408, 20,279, 25,988, 22,373, and 19,179 kN for B1‐REG, B2‐SST, B3‐GEO, B4‐
DIS, and B5‐WST, respectively.
Both ultimate limit state and serviceability load combinations are considered in the design process. Service load combinations are employed to
verify the vertical and lateral deformations while the structural elements are designed using the ultimate load combinations.[1,20] The ultimate load
MWAFY AND KHALIFA 5 of 22

combinations, including the vertical and horizontal seismic load effects, are as follows: (1.2 + 0.2 SDS) D + ρ QE + L and (0.9–0.2 SDS) D + ρ QE,
where SDS is the design spectral response acceleration at 0.2 s, ρ is the redundancy factor, D is the dead load, and QE is the horizontal seismic force.
The above combinations are adopted for the design of the regular structure as well as the buildings with stiffness and geometric irregularities. As
per the design code, the in‐plane discontinuity of LFRS and the discontinuity in the lateral strength of LFRS (weak story irregularity) should be
designed using special cases of loading.[1] The Ωo factor is utilized for the design of the irregularity introduced in buildings B4‐DIS and
B5‐WST.[1] This is conducted by adding the Ωo factor to the abovementioned load combinations, as follows: (1.2 + 0.2 SDS) D + Ωo ρ QE + L,
and (0.9–0.2 SDS) D + Ωo ρ QE. The special seismic loads that are factored by the Ωo coefficient are used during the design process for the structural
members of the “weak” stories and those supporting the discontinuous shear walls in buildings B5‐WST and B4‐DIS, respectively. This ensures that
these critical structural members are capable of resisting a total seismic force equal to Ωo times the seismic design force. It is noteworthy that for
the Seismic Design Category “C” and vertical irregularities considered in this study, ASCE/SEI‐7[1] permits the seismic design forces to be applied
independently in each of two orthogonal directions.
Table 3 shows a comparison between the uncracked periods of vibration of the five reference buildings obtained from the finite element (FE)
models used in design as well as the fiber‐based (FB) models developed for seismic assessment, as subsequently discussed. Figure 3 also shows the
fundamental modes of vibration and corresponding uncracked periods obtained from the ETABS 3D models of the reference structures. It is impor-
tant to note that, as per the design code,[20] the 3D structural models used in the design of the reference structures incorporate realistic estimates
of stiffness through the use of section modifiers, taking into account the expected damage under the design earthquake. The ETABS uncracked
periods of vibration given in Table 3 and Figure 3 are only presented for the sake of comparisons with those obtained from the FB models. Except
for building B3‐GEO, which has slightly shorter period, the fundamental periods of the irregular buildings in the transverse direction are longer than
the regular counterpart (B1‐REG). The longer periods are attributed to the reduced stiffness of the lower stories in buildings B2‐SST, B4‐DIS, and
B5‐WST. On the other hand, the footprint of the lower stories of building B3‐GEO is larger than those of the regular structure, and hence, the
stiffness increases and period decreases.
Straining actions generated by gravity and lateral loads are considered in the slab design. The floor slabs are designed using the structural
design software SAFE.[24] The floor slab models are exported from ETABS to SAFE with all load combinations and deformations from gravity
and lateral loads. The comprehensive design process results in member sizes and reinforcement details of the five reference structures, including
reinforcement schedules of vertical structural members and floor slabs. Figures A1 and A2 show samples of slab reinforcement details. The
design of vertical elements (columns, shear walls, and core walls) is fully automated using ETABS. Although the boundary elements of shear
walls and core walls are not required by the design code for SDC “C”,[20] they are utilized in design to enhance the seismic performance, as
shown in Figure A3. The complete design results of the five reference structures, including the reinforcement details of shear walls and floor
slabs, are discussed in more detail by Khalifa.[25] The design results are utilized to develop the FB models used in the inelastic analysis, as
discussed hereafter.

TABLE 3 Comparison of the uncracked periods of vibration for the reference buildings in the transverse directions obtained from finite element
and fiber‐based models
Modeling approach B1‐REG B2‐SST B3‐GEO B4‐DIS B5‐WST
Finite element (FE) models T1 4.688 4.822 4.603 4.815 5.021
T2 1.326 1.368 1.300 1.317 1.431
Fiber‐based (FB) models T1 4.540 4.673 4.280 4.561 4.835
T2 1.220 1.240 1.204 1.230 1.300

FIGURE 3 Fundamental modes of vibration and corresponding uncracked periods (s) obtained from the 3D ETABS models: (a) transverse direction,
and (b) longitudinal direction. Note: Uncracked periods are only presented for comparisons with those obtained from the fiber‐based models,
whereas proper section modifiers are used for design as per ACI‐318[20]
6 of 22 MWAFY AND KHALIFA

3 | F I B E R ‐ B A S E D M O DE L I N G OF I RR E G U LA R B U I LD I N G S A N D S E LE C T I O N OF E A R T H Q U A K E
RECORDS

Structures behave in a nonlinear manner during strong earthquakes, and hence, the seismic assessment of buildings should be performed using
inelastic dynamic time‐history analysis, (e.g., FEMA[11]). The nonlinear analysis platform Zeus‐NL is employed in the current study to conduct a large
number of inelastic analyses.[26] Several verifications have been conducted for this analysis platform against full‐scale tests carried out in Europe
and the United States.[27–29] For instance, comparisons of the full‐scale test results for a three‐story RC irregular building with those obtained from
Zeus‐NL confirmed the realistic prediction of this analysis platform.[28] Moreover, the Zeus‐NL modeling approach and key modeling parameters
adopted in the present study were verified by comparisons with the nonlinear dynamic response of a full‐scale seven‐story wall building slice tested
at the University of California, San Diego.[27,30] Large research projects covering multispan bridges and high‐rise buildings were also conducted
using the adopted analysis platform (e.g., Mwafy, Mwafy and Elnashai, and Mwafy et al.[16,31,32]). Simulation models are developed for the five ref-
erence buildings using Zeus‐NL to represent the LFRSs in the transverse direction. It is assumed in the present study that each reference building
consists of four comparable LFRSs in the transverse direction, as depicted in Figures 2 and 4. Each of the idealized framing systems in the trans-
verse direction resists gravity loads in addition to the lateral seismic forces generated due to 25% of the total mass of the building. It is assumed that
the exterior structural members only support gravity load, and hence, the seismic forces are entirely resisted by the internal LFRSs. In the longitu-
dinal direction, a single framing system resists the whole seismic forces in addition to gravity loads, while other structural members only support
gravity loads.
It is noted that the transverse direction of the reference buildings is in principle more vulnerable than the longitudinal direction. Although the
seismic forces in the transverse direction are equally resisted by four framing systems, each of these frames consists of two external shear walls and
an internal core. The width of these frames is also 30% less than that of the framing system in the longitudinal directions. The LFRS in the longi-
tudinal direction has more vertical members, namely, all internal cores and four external shear walls, and hence, it is somehow less vulnerable than
its counterpart. It is noteworthy that the inelastic response of the LFRSs in both the longitudinal and transverse directions of a set of buildings with
a layout comparable to B1‐REG was previously investigated.[33] It was concluded that the LFRSs in the two orthogonal directions were virtually
equivalent. The abovementioned discussion in addition to the conclusions of the latter study support conducting the inelastic analysis in the trans-
verse direction of the reference structures. This enables reducing the number of inelastic analyses by focusing on a single direction of the buildings.
Additionally, certain irregularities are introduced in the transverse direction (i.e., B3‐GEO), and hence, performing the nonlinear analysis is essential
in this direction.
Using the equivalent frame approach and considering the flat slabs and vertical structural members as a system of planar frames, the RC slab is
modeled in this study as a continuous beam strip bounded laterally by the centerline of the panel on each side of the walls/columns. Since no inter-
nal beams frame into the vertical elements, the slab width is considered from the centerline of the panel, as shown in Figures 2 and 4. The same
modeling concept is adopted for idealizing the transfer slab of the B4‐DIS building (refer to Figures 2 and A2). Other approaches were proposed in
previous studies for estimating the slab width (e.g., Luo and Durrani[34]). However, considering the full slab width in the FB numerical models devel-
oped in this study is supported by the fact that the slabs are modeled using cubic elastoplastic elements, taking into consideration the represen-
tation of all steel bars, confined and unconfined concrete, as subsequently discussed. Therefore, the concrete cracking and steel yielding
expected under earthquake loads are effectively accounted for in the developed models for inelastic analysis. It is also noteworthy that the

FIGURE 4 Modeling approach of reference structures for inelastic analysis: (a) B3‐GEO layout; (b) Zeus‐NL model; (c) geometrical modeling of
horizontal and vertical elements; and (d) fiber‐based modeling approach
MWAFY AND KHALIFA 7 of 22

straining actions and deformations considered in the design of the flat slabs are calculated using the 3D ETABS FE models of the reference
buildings under the effect of both gravity and lateral loads. The floor slab models are exported from ETABS to the structural design software SAFE
with all load combinations from gravity and lateral loads.[22,24] It is noted that using this approach, which accounts for seismic forces, in the design
of floor slabs results in a significant increase in the slab reinforcement compared with the case with gravity load design only, particularly at the
connections between the slab and vertical elements (shear walls, columns, and core walls). This design approach renders the floor slabs to be more
effective in resisting lateral loads and reduces inelasticity under cyclic loading. For the sake of brevity, the design drawings and detailed reinforcing
steel schedules of the five reference structures are presented in Khalifa.[25]
RC flexural wall, hollow rectangular, rectangular, and T‐sections are used to idealize shear walls, core walls, columns, and slabs, respectively.
The shear walls and core walls are modeled as wide columns using frame elements, which are located at the centroid of the wall cross‐section
and connected with the slab using rigid links. The rigid arm length is the distance between the centerline and the face of the vertical elements,
as presented in Figure 4(c). The frame element is assigned RC flexural wall and hollow rectangular sections for the external shear walls and internal
core walls, respectively. Although this modeling approach for the core walls neglects the effects of restrained warping, it can provide sufficient
accuracy when the torsional effect is of minor importance as compared to the flexural response. For the five reference buildings investigated in
the present study, the effect of torsion is limited. Therefore, modeling the wall cross‐section using a single cubic elastoplastic element placed
through the centroid provides a balance between efficiency and accuracy and enables utilizing the adopted analysis platform Zeus‐NL, which
was mainly developed for frame analysis. This modeling concept was adopted in several previous studies, and it has been recently verified against
the dynamic response of a full‐scale seven‐story wall building (e.g., Alwaeli et al. and Alwaeli et al.[27,35]).
Three cubic elastoplastic frame elements are used to idealize each horizontal and vertical structural member (slab, column, shear wall, and core
wall). This idealization allows utilizing different cross‐sections for each structural member, one at each member ends and another one at the
midspan. These sections help to accurately model different reinforcement profiles of structural members according to the design. Figure 4(d) shows
the Zeus‐NL cubic elastoplastic element, which includes two Gauss sections, as well as the concrete and reinforcing steel fibers. This modeling
approach effectively represents the spread of inelasticity within the cross‐section and along the member length.[26,27] Reinforcing steel, confined
concrete, and unconfined concrete are idealized using this fiber modeling approach.
A uniaxial constant confinement concrete model and a bilinear elastoplastic steel model with kinematic strain hardening are used in the Zeus‐
NL models.[26] Fully confined concrete is used in columns and the boundary elements of shear walls and core walls. Partially confined concrete is
used in the web of shear walls and core walls, whereas unconfined concrete is used to model the concrete cover. The Zeus‐NL models incorporate
realistic estimates of material strength. Therefore, the expected material properties, as opposed to nominal or specified properties, are utilized to
assess the seismic response of the reference structures.[36]
Hysteretic damping is accounted for in the elastoplastic fiber element used to model structural members. The nonhysteretic damping is caused
by many sources such as the effect of nonstructural components. The latter type of damping is considered by utilizing stiffness‐based Rayleigh
damping.[37,38] The stiffness‐proportional damping is calculated for each reference building using the equivalent period of each structure as pro-
posed by Alwaeli et al.[35] Eigenvalue analysis is used to verify the uncracked vibration periods and deformed shapes of the reference buildings.
The comparison shown in Table 3 for the uncracked periods of vibration obtained from the 3D ETABS models (refer to Figure 3) used in the design
with those obtained from the Zeus‐NL FB models shows the minor reduction in the period of vibrations obtained from the latter models. This small
reduction in period is due to the realistic modeling of rebar in Zeus‐NL, which increases the stiffness of structural elements, unlike the ETABS
models. The abovementioned results and discussion lend weight to the developed Zeus‐NL models for the assessment of the seismic design
approach and coefficients of the reference buildings using IPA and IDA.
The selection of input ground motions for the seismic assessment of high‐rise buildings is a critical task due to the wide range of vibration
periods of significance. Several previous studies concluded that the seismological parameters such as the earthquake magnitude and distance have
significant effects on the dynamic analysis results.[29,39] In the current study, the seismological and site parameters, including the record magnitude,
epicentral distance, soil class, ratio of PGA to peak ground velocity (a/v), and PGA are considered in the selection of earthquake records to repre-
sent the seismic scenarios expected in the case study region (Dubai, UAE). Three approaches for seismic performance assessment of buildings are
recommended by NEHRP[40]: intensity‐based assessment, scenario‐based assessment, and risk‐based assessment. The selection of the seismic
records depends on the implemented type of assessment. In the current study, a scenario‐based assessment is performed as per the recommen-
dation of previous studies for the case study region (e.g., Mwafy et al. and Aldama‐Bustos et al.[39,41]). The employed seismic scenarios represent
(a) severe events with a long epicentral distance and (b) moderate earthquakes with a short distance from the epicenter. For far‐field events, a mag-
nitude (Mw) range of 6.93 to 7.64, epicentral distance range of 91 to 161 km, stiff and very dense soil classes, low a/v ratio (<0.8 g/m∙s −1), and a
PGA range of 0.9 to 2.39 m/s2 are considered in the record selection. Furthermore, for the near‐field records, a magnitude range of 5.14 to 6.04,
epicentral distance range of 2.86 to 29.9 km, stiff and very dense soil classes, high a/v ratio (>1.2 g/m∙s −1), and a PGA range of 0.85 to 4.96 m/s2
are considered in the selection of earthquake records.
Two databases are used to select the input ground motions, which include the Pacific Earthquake Engineering Research center database[42]
and the Internet site for European Strong‐motion Database.[43] From the selected databases, 20 far‐field and 20 near‐field natural records are cho-
sen to represent the earthquake scenarios in the study region. Table A1 shows characteristics of the far‐field input ground motions. For the sake of
brevity, the information related to the near‐source earthquake records is provided by Khalifa.[25] The far‐field records fit the design response spec-
trum in the long period range, and the near‐field seismic events have high amplifications and match the design spectrum in the short period range,
8 of 22 MWAFY AND KHALIFA

as shown Figure 5.[1] The abovementioned two seismic scenarios and selected two sets of earthquake records account for the uncertainty of input
ground motions and represent many regions of medium seismicity. Before applying to the reference building models, the selected earthquake
records are scaled to a PGA of 0.16 g, which represents the design PGA for 10% probability of exceedance in 50 years (e.g., Mwafy et al.[39])
and its multiples.
Different intensity measures such as PGA and spectral acceleration can be used to conduct IDA. However, it is difficult to select an ideal scaling
approach for high‐rise buildings, which are considerably influenced by higher modes. Moreover, the 40 input ground motions selected in the present
study represent two specific seismic scenarios. Scaling the input ground motions as per their spectral intensities modifies their characteristics (e.g.,
Alwaeli et al.[27]). Normalizing input ground motions using PGA directly relates the seismic forces to the input accelerations and maintain the main
spectral characteristics of earthquake records. This simple scaling approach is also in agreement with seismic design codes, and hence, it was
employed in several previous studies and in the present work (e.g., Ji et al.[44]).
The IDA procedure involves conducting several response history analyses of a structural model under sets of input ground motions that are
scaled to different earthquake intensities. The scaling approach is carefully selected in the present study to force the reference structures through-
out the whole seismic behavior range, from elastic to dynamic instability. For the long period records, each input ground motion is scaled from a
PGA of 0.08 g (half of the design PGA) up to 1.12 g, with an increment of 0.08 g, as shown from the IDA results presented in subsequent sections.
Due to the less significance of short period records on the seismic response of the reference buildings, each input ground motion in this set is scaled
from a PGA of 0.32 g up to 4.48 g, with a scaling factor of 0.32 g.

4 | PERFORMANCE LIMIT STATES

The local and global inelastic response of the reference structures and the results of previous experimental studies are used in the current study to
select the IDRs corresponding to the limit states needed to assess the seismic design response factors. It is important to note that although IDAs
are conducted for the five reference structures using the 40 input ground motions that represent the far‐field and near‐source seismic scenarios,
as previously discussed, the seismic design response factors are mainly evaluate using the 20 long period records. This is attributed to the higher
impact of the long‐period earthquake records on the seismic response and CP limit states of the reference structures. This set of ground motions
has high amplifications, particularly in the period range corresponding to the second and third modes of the benchmark buildings (i.e., 1.2–1.3 s
and 0.53–0.59 s, respectively). The amplification of higher modes results in earlier damage and exceeding the CP limit state at much lower PGAs
compared with those observed under the effect of near‐source earthquake records. It is noteworthy that far‐field input ground motions were also
adopted in previous studies related to seismic design response factors for collapse performance evaluation.[13] The comprehensive dynamic
response results of the five reference structures under the effect of the far‐field and near‐source earthquake scenarios are presented by
Khalifa.[25]
For the local response, the IDR corresponding to the first indication of yielding in reinforcing steel represents the immediate occupancy (IO)
limit state, whereas the first indication of crushing in the confined concrete of vertical structural elements corresponds to the CP limit state.
Figure 6 shows sample IPA results of monitoring the plastic hinge (PH) distributions in horizontal and vertical structural members of the reference
structures. For the B1‐REG building, PHs are shown for all horizontal members, and the first PH in horizontal members is only highlighted in other
buildings for the sake of simplicity. All PHs of vertical structural members for the five reference buildings are mapped in Figure 6 due to their
significance on the overall building response.
It is noted that the first PH in horizontal members is recorded at the middle of the building height except for B4‐DIS members due to the early
yielding in the transfer slab. In the case of the extreme soft story structure (B2‐SST), the number of PHs is comparable to those of the B1‐REG
building. The number of PHs in the vertical members of the B3‐GEO building is more than those in other buildings. This observation is due to
the increased stiffness of this building, which results in attracting higher lateral forces. The results presented in Figure 6 show that increasing

FIGURE 5 Response spectra of 40 earthquake records representing far‐field and near‐source events along with the design response spectra of site
classes “C” and “D”
MWAFY AND KHALIFA 9 of 22

FIGURE 6 Sequence of plastic hinge distributions in (a) horizontal and vertical structural members of the regular structure and (b–e) all vertical
members and the first plastic hinge in horizontal elements of the irregular buildings

the lateral design forces leads to increasing the number of PHs. It is noteworthy that, no PHs are recorded in the vertical members of B4‐DIS and
B5‐WST at the lower stories since they were designed with an overstrength factor (Ωo) as per the design code.[1]
The crushing of confined concrete is assumed when the concrete compressive strain exceeds the ultimate confined concrete strain. The
concrete model proposed by Mander et al.[45] is adopted to evaluate the ultimate confined concrete strain. The confined concrete strain in vertical
structural members is monitored throughout the multistep inelastic analyses of the five reference structures, and any indications of concrete
crushing are recorded. Figure 7 shows the distributions of concrete crushing in the vertical structural members of the five reference buildings
obtained from IPA. It is noted that the member failure distributions of the B1‐REG and B2‐SST buildings are comparable. The number of member
failure cases in B3‐GEO is higher than that in B1‐REG. Again, the higher stiffness of the former structure at lower stories leads to attracting higher
lateral load, and hence increasing the number of member failure cases at the irregularity levels. Despite the enhanced global response of building
B3‐GEO due to increasing its footprint at the lower stories, the local response clearly shows the disadvantages of the geometric irregularity. No
indications of confined concrete crushing are recorded in the vertical members of B4‐DIS and B5‐WST at the lower stories as a result of using
the Ωo factor in their design.
Moreover, the shear demands are obtained from the results of inelastic dynamic analysis and compared with the shear strength to provide
insights into the shear failure potential of the main structural members. The experimentally verified shear strength model proposed by Priestley
et al.[46] and Kowalsky and Priestley[47] is adopted to estimate the shear capacity of the structural member. The IDA results of both the long period
and short period earthquake records are used for the shear response assessment. Under the effect of the long period earthquake scenario, shear
failure is not observed in any structural member of the five reference buildings before reaching the CP limit state corresponding to crushing in
confined concrete. This is attributed to the impact of the long period events on high‐rise buildings in which the seismic response is controlled
by flexure rather than shear. On the other hand, the shear assessment indicates that the short period earthquake scenario has a significant impact
on the shear response of the irregular structures. For this earthquake scenario, the IDR corresponding to the CP limit state decreases as a result of
detecting shear failure earlier than the crushing in confined concrete. Nevertheless, this brittle failure mode occurs only at significantly high ground
10 of 22 MWAFY AND KHALIFA

FIGURE 7 Distributions of concrete core crushing in (a) the vertical structural members of the regular structure and (b–e) the vertical members of
the irregular buildings

motion intensity levels. Hence, the seismic design response factors are controlled by the seismic response under the far‐field earthquake records.
Sample results are presented in Figure 8 depicting comparisons between the shear demand and shear supply using both the ACI‐318[20] and
Priestley et al.[46] approaches for selected elements of the B4‐DIS building at the CP limit state. The comprehensive shear assessment results of
the five reference structures are presented by Khalifa.[25]
Furthermore, IDAs are carried out in order to define the global limit states from the IDA curves of the five reference buildings. The equivalent
inelastic (cracked or elongated) periods of the five reference structures are calculated and used to obtain the corresponding spectral acceleration of
the selected input ground motions. Several approaches are recommended by design codes and guidelines and in the literature for estimating the
effective stiffness and cracked periods (e.g., Elnashai and Mwafy, ACI‐318, and LATBSDC[14,20,36]). To account for the expected concrete cracking
and steel yielding, the inelastic periods of vibration are estimated in the present study by considering the effective stiffness of structural elements
as recommended by the design code.[20] The equivalent inelastic periods are calculated based on the first three inelastic periods weighted by the
mass participation ratios.[35] The IDA results are used to develop the relationship between the maximum IDRs and spectral accelerations, as shown

FIGURE 8 Comparisons between the shear demand, Vd, and shear supply using both the ACI, V(ACI), and Priestley et al.,[46] Vpr, approaches for
selected elements of the B4‐DIS building at the collapse prevention limit state
MWAFY AND KHALIFA 11 of 22

from the sample results presented in Figure 9 for the B1‐REG, B3‐GEO, and B5‐WST buildings. The IO limit state is defined as the first deviation
from the elastic response, and the CP limit state is determined when the stiffness significantly reduces and reaches 20% of the elastic value. The
IDR corresponding to the IO and CP performance criteria are conservatively estimated at the 16 percentile of the lognormal distribution, as shown
in Figure 9. Following the abovementioned approach, the IO limit states of B1‐REG, B2‐SST, B3‐GEO, B4‐DIS, and B5‐WST are 0.49%, 0.48%,
0.51%, 0.27%, and 0.44%, respectively. These values are consistent with those obtained from the local response results. For the B1‐REG building,
the IDR corresponding to the IO limit states (0.49%) is also consistent with the value recommended by design guidelines and previous studies (e.g.,
ASCE/SEI‐41 and Lehman et al.[48,49]).
The IDRs corresponding to the CP limit state of B1‐REG, B2‐SST, B3‐GEO, B4‐DIS, and B5‐WST are 4.97, 4.56, 6.08, 2.17, and 3.61, respec-
tively. It is noted that the CP limit states calculated using the IDA curves are higher than those recommended in previous experimental studies
and those obtained in the present study from monitoring local response. It is noteworthy that the significant reduction in the slope of the IDA
curve that characterizes the CP limit state is linked to the numerical instability, which takes place due to the nonconvergence of the time‐inte-
gration scheme, as shown in Figure 9. This dynamic instability mainly occurs due to the crushing in the confined concrete of shear walls. The
numerical instability may be linked with the global collapse of ductile moment‐resisting frames in which a substantial load redistribution in the
columns of a story may occur while the capacity is sustained.[50] In RC wall buildings, the ductility of shear walls at the lower stories is limited
due to the high stiffness and axial force. The RC bearing wall system also has a limited possibility for load redistribution at one story due to the
limited number of vertical structural elements. When a shear wall fails, a collapse mechanism is likely to be initiated since other walls will be
overloaded. Hence, the collapse of well‐designed RC wall buildings is not likely to be associated with a significant reduction in the slope of
the IDA curve.[35]
The most conservative IDR corresponding to the CP limit state proposed in previous experimental studies related to regular shear wall struc-
tures (2.27%) is therefore adopted in the present study.[30,49,51] The previous experimental studies on RC slender walls concluded that extensive

FIGURE 9 Incremental dynamic analysis (IDA) curves obtained from 20 long‐period earthquake records showing the first yielding and collapse
points as well as their lognormal distributions that are fitted to the maximum interstory drift ratios: (a) B3‐REG; (b) B3‐GEO; and (c) B5‐WST
12 of 22 MWAFY AND KHALIFA

crushing in confined concrete and buckling in longitudinal reinforcement cause failure and loss of load‐carrying capacity. This conclusion is consis-
tent with the above discussion related to the nonconservative CP limit state obtained from IDA curves. Hence, failure modes that are not explicitly
modeled and detected by IDA curves are accounted for using conservative CP limit states selected from available experimental data. The selected
CP limit state is slightly higher than that proposed by ASCE/SEI‐41.[48] Moreover, due to the insufficient experimental studies and the lack of code
recommendations for the CP limit state of irregular high‐rise buildings as well as the dispersion of the results observed from the IDA results, the
most conservative CP limit states are adjusted using the CP value adopted for the regular building based on test results (i.e., 2.27). The aforemen-
tioned approach results in CP limit states of 2.26%, 2.39%, 1.18%, and 1.38% for B2‐SST, B3‐GEO, B4‐DIS, and B5‐WST, respectively. The com-
parable results obtained from IPA and IDA of B1‐REG, B2‐SST, and B3‐GEO, as subsequently discussed, lend weight to the CP values adopted for
these three buildings (i.e., 2.27, 2.26, and 2.39, respectively).
It is also noteworthy that the CP limit state observed in the comprehensive analytical study of Alwaeli et al.[35] for a regular shear wall building
with flexure‐controlled response was associated with an IDR of 2.07%, which is also close to the selected value for B1‐REG. Moreover, it is
important to note that the selected CP performance criteria of B4‐DIS and B5‐WST are conservative and in line with those recommended in
the literature. Among the scarce experimental data related to the RC high‐rise buildings and their LFRS, few test results are available for irregular
high‐rise structures (e.g., these previous studies[19,30,49,51,52]). The CP limit states observed in the experimental study of Li et al.[19] and Lee and
Ko[52] for vertically irregular shear wall buildings with a discontinuity and a weak story were associated with an IDR of 1.25% and 1.57%, respec-
tively. The abovementioned values are also consistent with those adopted in the present study. Finally, it is emphasized that, although the adopted
IDRs to represent the performance limit states of the set of regular and irregular high‐rise buildings is based on the values observed from the inelas-
tic analyses carried out in the current study as well as those recommended in the literature, this topic is still a significant research issue in the
structural engineering (e.g., Alwaeli et al. and Kinali and Ellingwood[35,53]).

5 | I M P A C T S OF V E RT I C A L I R R E G U L A R I T Y O N S E I S M I C D E S I G N

Although IDA is mainly employed to assess the seismic design response factors under a wide range of earthquake records, IPA is first conducted to
obtain an initial estimation of the lateral capacity and to estimate the Ωo factor of the five reference buildings. IPA was also useful for tracing the
development of PH and crushing in concrete. A number of previous studies concluded that the IPA accuracy was not significantly depreciated even
for well‐designed irregular structures while other studies concluded that the uniform lateral load distribution can be used to conservatively esti-
mate the initial stiffness and lateral capacity of high‐rise buildings.[39,54] In the current study, IPA with a lateral load pattern representing the mass
distribution throughout the building height is thus deployed to (a) compare between the capacity curves of the regular and irregular structures and
(b) initially assess the local response of structural elements and map it with the adopted global damage measure. It is important to note that IPA was
not employed in the present study to estimate seismic demands. These demands are exclusively predicted using IDAs, as explained below.
Pushover analysis is performed until the FB models exhibit a significant lateral strength loss, which causes numerical instability in the Zeus‐NL
inelastic analysis. The IPA results presented in Figure 10 show rebar yielding in slabs, followed by gradual stiffness degradation due to the spread of
yielding, as also illustrated in Figure 6. Yielding in walls/cores is consistent with the global yielding, which is estimated from the idealized response.
In the regular structure and those with moderate irregularities (i.e., B1‐REG, B2‐SST, and B3‐GEO), the confined concrete crushing in core walls
occurs first, and then followed by the exceedance of concrete crushing strain in external shear walls at the foundation level (support), as shown in
Figure 7. For buildings with severe irregularities (i.e., B4‐DIS and B5‐WST), special seismic loads are used in the design of the structural members of
the “weak” stories and those supporting the discontinuous shear walls, which protected these elements from yielding and crushing. Hence, a
significant lateral strength loss occurs when the concrete crushing in the shear walls above the stories causing the irregularity increases, as shown
in Figures 7 and 10.
Comparisons between the capacity curves of B1‐REG, B2‐SST, and B4‐DIS are shown in Figure 10. The ultimate strength obtained from IPA
represents a conservative estimate of the lateral capacity of a building, as previously discussed. The global yielding is evaluated from an
elastoplastic idealization of the real capacity curve (e.g., Elnashai and Mwafy[14]). For the sake of brevity, only sample results are presented in
Figure 10, whereas the idealized capacity envelopes of the five reference buildings obtained from the actual capacity curves are depicted in
Figure 11. It is observed from the IPA results that the ultimate strength, initial stiffness, and ductility (ultimate‐to‐yield displacement) of the building
with the extreme soft story irregularity (B2‐SST) are slightly lower than those of the regular structure (B1‐REG). These minor differences between
the response of B2‐SST and B1‐REG validate the design code approach regarding the design of this type of irregularity (i.e., no special requirements
are needed in the design to SDC “C”).[1] In contrast, the abovementioned characteristics (i.e., strength, stiffness, and ductility) of B3‐GEO are slightly
higher than those of the B1‐REG building due to the increase in the footprint of the former structure. Again, this observation validates the code
approach towards this type of irregularity. For B4‐DIS and B5‐WST, the ultimate capacity and initial stiffness are much higher than those of
B1‐REG, while the ductility significantly decreases, as shown in Figures 10(c) and 11. These differences in response are mainly due to the use of
the Ωo factor in the design of buildings B4‐DIS and B5‐WST at the irregularity levels as required by the design code.[1] Although the initial stiffness
and ultimate strength of the latter two buildings are improved, the ductility reduction and inferior local response support the code conservative
approach toward the design of these types of vertical irregularity, particularly regarding the use of special load cases in design.
MWAFY AND KHALIFA 13 of 22

FIGURE 10 Sample results presenting comparisons of the capacity envelopes of the irregular and regular structures: (a) B1‐REG; (b) B2‐SST versus
B1‐REG; and (c) B4‐DIS versus B1‐REG

FIGURE 11 Idealized capacity envelopes obtained from the actual capacity curves of the five reference buildings along with the design base shear
(also refer to Figure 10)
14 of 22 MWAFY AND KHALIFA

FIGURE 12 Comparisons between the distributions of interstory drift ratios at ultimate strength obtained from inelastic pushover analysis for
different vertically irregular structures and the regular building

Figure 12 compares between the distribution of IDRs throughout the building height at the ultimate strength for B1‐REG with those for the irreg-
ular buildings. As indicated in Figure 12, these results are obtained from IPA. It is observed that the IDR distributions of the first three building are
comparable. The observed discrepancies of IDRs at the lower stories of B4‐DIS and B5‐WST are attributed to the significant change in LFRS and irreg-
ularity at these levels. The presented results in Figure 12 support the design code provisions related to the design requirements of different types of
irregularity. The IDRs corresponding to the first yielding and crushing in structural members as well as the global yielding are mapped on the capacity
curves. These results are summarized for the five reference buildings in Figure 13(a). The results presented in Figure 13(a) also support the
abovementioned observations related to the comparable results of the first three structures and significant impact of severe irregularity on IDRs.
The actual strength is influenced by several parameters such as the design safety factors, material characteristics, and structural system.[14] The
structural overstrength is the actual‐to‐design strength, as shown in Figures 1, 10, and 11. The overstrength factor is calculated in the present study

FIGURE 13 Summary of interstory drift ratios (IDRs) and overstrength factors of the reference buildings obtained from inelastic pushover analysis:
(a) IDRs at the first indication of member yielding and crushing; and (b) overstrength factors at the first indication of yielding, Ωfy; global yielding,
Ωgy; and at ultimate capacity, Ωu (refer to Figures 1 and 10)
MWAFY AND KHALIFA 15 of 22

FIGURE 14 Incremental dynamic analysis results of five reference buildings using 20 long‐period earthquake records: (a–e) collapse‐to‐yield PGA
ratios, PGAc/y; and (f) summary of the median PGAc/y ratios

at different strength levels, including the first indication of yielding, global yielding, and ultimate capacity. Figure 13(b) depicts the overstrength
factors at the first PH, Ωfy; global yielding, Ωgy; and ultimate capacity, Ωu, of the reference buildings. These three performance levels are almost
consistent with the IDRs depicted in Figure 13(a), particularly for the regular or moderately irregular buildings. As previously discussed for buildings
with severe irregularities, the ultimate strength corresponding to a significant lateral strength loss occurs when the concrete crushing in the shear
walls above the stories causing the irregularity increases, as shown in Figures 7 and 10. Hence, the ultimate strength used to calculate Ωu is slightly
higher than the value observed at the first wall crushing. It is shown that the overstrength factors of the B1‐REG and B2‐SST buildings are
comparable, which confirms the marginal effect of the soft story irregularity on lateral capacity. It is noteworthy that ASCE/SEI‐7[1] does not
impose special seismic loads for the design of buildings exhibiting the stiffness/soft story irregularity owing to its lower impact on seismic response
compared with the irregularities associated with the discontinuity or weak story. Moreover, it is unlikely that the well‐designed B2‐SST building to
16 of 22 MWAFY AND KHALIFA

FIGURE 15 Incremental dynamic analysis results of five reference buildings using 20 long‐period earthquake records: (a–e) collapse‐to‐yield IDRs,
IDRc/y; and (f) relationship between PGAc/y and IDRc/y

develop a soft story mechanism since its LFRS consists of a system of stiff shear walls and core walls, which effectively control lateral drift even
with the increased story height of B2‐SST. The marginal impact of the soft story irregularity on the calculated Cd factor is also shown from the
results presented hereafter. Due to the higher design base shear of building B3‐GEO, the overstrength factors of this structure are lower than
the regular one. The overstrength factor at the first plastic hinge of building B4‐DIS is significantly lower than that of other buildings due to the
early yielding of the transfer slab, which supports the heavy vertical load from typical stories (refer to Figure 2). The overstrength factors of
the B4‐DIS building at global yielding and ultimate capacity are slightly lower than those of the regular building. The overstrength factors of the
B5‐WST building are higher than those of other structures due to the use of an Ωo factor in design.[1]
The R and Cd factors of the regular and irregular benchmark structures are estimated following the approach previously discussed. The R
factor is the collapse‐to‐yield PGA ratio, PGAc/y, times the Ωfy factor, whereas the Cd factor is considered equal to the collapse‐to‐yield IDR,
MWAFY AND KHALIFA 17 of 22

FIGURE 16 Overstrength factors obtained from inelastic pushover analysis (IPA) and incremental dynamic analysis (IDA)

FIGURE 17 R and Cd factors of five reference buildings obtained from incremental dynamic analyses using 20 long period input ground motions

IDRc/y.[16,17] The IDA results using the selected 20 long period records are employed to evaluate the R and Cd factors as a result of their higher
impact on the response of the reference structures.[25] The IDRs and PGAs ratios at yield and collapse for the five reference buildings are depicted
in Figures 14 and 15, whereas Figures 14(f) and 15(f) depict the median values. It is shown that PGAc/y is larger than IDRc/y for the five reference
structure. These differences reflect the margin of safety when Cd is considered equal to R.[13]
The overstrength factors at the first indication of yielding for the five reference buildings are calculated using IPAs and IDAs, as shown in
Figure 16. Because IDA is more reliable than IPA for irregular buildings, considering the capability of IDA to account for the input ground motion
uncertainty and higher mode effects, the overstrength factors calculated using IDA are adopted to evaluate the seismic design factors. Figure 17
depicts a comparison between the calculated R and Cd factors of the five benchmark buildings with the code recommended values. The R and Cd
factors of the B2‐SST and B3‐GEO buildings are comparable to the regular structure (B1‐REG). On the other hand, the R and Cd factors of the
B4‐DIS and B5‐WST buildings are lower than those of B1‐REG. This is attributed to the significant irregularity and the use of overstrength factor
(Ωo) in the design of the lower stories of the latter two buildings. It is shown from Figure 17 that the code recommended factors are conservative
for the five reference systems.
The results of the present study confirm that the impacts of different irregularity types on the seismic design response factors vary. The
discontinuity of the LFRS and the weak story irregularity, which are represented by B4‐DIS and B5‐WST, have the highest impact on the seismic
design response factors and local seismic response. The results indicate that the R factors of the regular structure and buildings with moderate
irregularity (i.e., B2‐SST and B3‐GEO) can be initially increased by approximately 10%, whereas the Cd factor can be decreased by the same
ratio. As per the current precision adopted by ASCE 7 for the seismic design response factors of most LFRSs, the R factors of the investigated
structural systems with moderate or no irregularity can be increased to 4.5, while the Cd factor is decreased to 3.5. Further adjustment is pos-
sible after a careful response assessment of the structures designed using the suggested reduction in seismic design forces. Due to the signif-
icant impact of irregularity on the local and global response of buildings with a discontinuity in LFRS and weak story, the conservative R and Cd
factors of the design code are recommended to be maintained. The work presented in the current study supports ongoing activities for revising
the seismic design approaches and coefficients of tall buildings aiming at having different structures with comparable reliability and cost‐
effectiveness.
18 of 22 MWAFY AND KHALIFA

6 | C O N CL U S I O N S

The aim of this study was to assess the impacts of the most common vertical irregularity types on the existing seismic design approach and
coefficients of high‐rise buildings. Five 50‐story RC buildings denoted B1‐REG, B2‐SST, B3‐GEO, B4‐DIS, and B5‐WST were selected to represent
well‐designed regular and irregular high‐rise buildings in a medium seismicity region represented by Dubai, UAE. The five reference buildings were
designed and detailed for the purpose of this study using 3D FE models according to international building codes. Inelastic FB simulation models
were developed and verified for the five benchmark structures. Two earthquake scenarios, applicable to several seismic regions, were selected to
represent the seismicity of the case study area: (a) far‐field earthquakes with a medium‐to‐high magnitude and long distance from the epicenter and
(b) near‐field events with a low‐to‐medium magnitude and a short site‐to‐source distance. The ground motion uncertainty was accounted for using
40 earthquake records representing the abovementioned two seismic scenarios. A large number of IDAs were performed to assess the seismic
performance of the five reference structures at different performance levels. The performance limit states were selected based on the inelastic
analysis results and the values recommended in previous experimental studies related to shear wall structures and irregular buildings. The following
conclusions were drawn based on the findings of this study:

• Although the seismic design code recommends the use of an overstrength factor for the building with a discontinuity in LFRS (B4‐DIS) and the
structure with extreme weak story (B5‐WST), the local response results of these irregular buildings reflected the need for imposing reduction
limits on the special seismic forces (or Ωo), and consequently on the cross‐sections and steel ratios for the stories above the irregularity levels
to avoid any sudden changes in stiffness and strength, and hence adverse impacts on local response.

• Except for the B4‐DIS building, the calculated overstrength factors using IDAs were more than those recommended by the design code. The
relatively less satisfactory safety margin of the B4‐DIS building was confirmed by the observed low level of reserve strength.
• For the structures with moderate irregularity (i.e., B2‐SST and B3‐GEO), the response modification factors could be safely increased. An
initial increase in the R factor to 4.5 is proposed while a further adjustment is possible after a careful assessment of the structures designed
using the suggested reduction in seismic design forces. Due to the significant impacts of the irregularities related to discontinuities in LFRS
or a weak story on the local and global seismic response of high‐rise buildings, the conservative code design forces are recommended to be
retained.

• The calculated deflection amplification factors for the reference irregular structures were significantly lower than those recommended by the
design code. The code Cd factor could be initially decreased to 3.5 for the efficient shear wall structural system while a more reduction is
possible after the assessment of the structures designed using the proposed reduction in deflection amplification factors.

While this study provided general insights into the existing seismic design approach of real‐life tall buildings with different types of vertical
irregularity, it also recommended revised design coefficients for structures with moderate irregularities and highlighted the deficiencies of struc-
tures with severe irregularities. The study supports ongoing activities for revising the seismic design approach of high‐rise buildings aiming at
having different structures with comparable reliability and cost‐effectiveness. Finally, it is emphasized that, notwithstanding the systematic assess-
ment approach adopted, the conclusions drawn were based on the adopted range of irregular high‐rise buildings, modeling assumptions, and per-
formance limit states. There is a pressing need to extend the research to study the implications of considering multiaxial earthquake loading on the
seismic design approaches of irregular high‐rise buildings using 3D numerical models and considering different deformation and capacity‐based
damage indices. Benchmarking the adopted approach for quantifying the seismic design response factors against other approaches such as that
proposed by FEMA P695 is also needed. The outcomes and claimed generality of this study are, strictly speaking, applicable to vertically irregular
high‐rise buildings, noting that every effort was made to design and model representative benchmark structures, and adopt verified assessment
tools and a wide range of input ground motions.

ACKNOWLEDGMENTS
This work was supported by the United Arab Emirates University under research grants 31N227 and 31N132. The authors would also like to
acknowledge the helpful comments and suggestions provided by anonymous reviewers on a previous version of the paper.

RE FE R ENC E S
[1] ASCE/SEI‐7, Minimum Design Loads for Buildings and Other Structures, ASCE Standard ASCE/SEI 7–10, American Society of Civil Engineers, Reston, VA
2010.

[2] CEN, Eurocode 8: Design of Structures for Earthquake Resistance. General Rules, Seismic Actions and Rules for Buildings. EN 1998–1:2004, CEN, European
Committee for Standardization, Brussels 2004.

[3] M. De Stefano, B. Pintucchi, Bulletin of Earthquake Engineering 2008, 6(2), 285.

[4] C. Athanassiadou, Engineering structures 2008, 30(5), 1250.

MWAFY AND KHALIFA 19 of 22

[5] C. Chintanapakdee, A. K. Chopra, Journal of Structural Engineering, ASCE 2004, 130(8), 1177.
[6] S. Das, J. M. Nau, Earthquake Spectra 2003, 19(3), 455.
[7] X. Lu, N. Su, Y. Zhou, The Structural Design of Tall and Special Buildings 2013, 22(7), 593.
[8] J. P. Moehle, Journal of Structural Engineering 1984, 110(9), 2002.
[9] P. Rajeev, S. Tesfamariam, Structural Safety 2012, 39, 1.
[10] S. Varadharajan, V. Sehgal, B. Saini, The Structural Design of Tall and Special Buildings 2014, 23(18), 1345.
[11] FEMA, NEHRP Recommended Seismic Provisions for New Buildings and Other Structures, FEMA P‐1050, Part 1: Provisions and Part 2: Commentary, Federal
Emergency Management Agency, Washington, D.C. 2015.
[12] FIB, Tall Buildings: Structural Design of Concrete Buildings up to 300m Tall, Fib Bulletin No. 73, The International Federation for Structural Concrete,
Lausanne, Switzerland 2014.
[13] FEMA, Quantification of Building Seismic Performance Factors, FEMA P695, Federal Emergency Management Agency, Washington, D.C. 2009.
[14] A. S. Elnashai, A. M. Mwafy, The Structural Design of Tall Buildings 2002, 11(5), 329.
[15] J. Kim, H. Choi, Engineering Structures 2005, 27(2), 285.
[16] A. M. Mwafy, Earthquake Engineering and Engineering Vibration 2011, 10(1), 115.
[17] A. M. Mwafy, A. S. Elnashai, Journal of Earthquake Engineering 2002, 6(2), 239.
[18] A. M. Mwafy, Structures and Buildings, The Institution of Civil Engineers (ICE) 2013, 166(6), 282.
[19] C. Li, S. Lam, M. Zhang, Y. Wong, Journal of structural engineering 2006, 132(11), 1732.
[20] ACI‐318, Building Code Requirements for Structural Concrete and Commentary, American Concrete Institute, Detroit, Michigan 2011.
[21] M. M. Ali, K. S. Moon, Architectural Science Review 2007, 50(3), 205.
[22] CSI, ETABS—Integrated Building Design Software, Computers and Structures, Inc., Berkeley, California 2011.
[23] DMA, Abu Dhabi International Building Code, Department of Municipal Affairs, Abu Dhabi, UAE 2013.
[24] CSI, SAFE—Design of Slabs, Beams and Foundations—Reinforced and Post‐Tensioned Concrete, Computers and Structures, Inc., Berkeley, California 2011.
[25] E. Khalifa, Assessment of Multi‐story Building Seismic Design Factors With Structural Irregularity, MSc, United Arab Emirates University, UAE 2015.
[26] A. S. Elnashai, V. Papanikolaou, D. Lee, Zeus‐NL—A System for Inelastic Analysis of Structures—User Manual, Mid‐America Earthquake Center, University
of Illinois at Urbana‐Champaign, Urbana, IL 2012.
[27] W. Alwaeli, A. M. Mwafy, K. Pilakoutas, M. Guadagnini, Bulletin of Earthquake Engineering 2017, 15(5), 2035. https://doi.org/10.1007/s10518‐016‐
0056‐8.
[28] S. H. Jeong, A. S. Elnashai, Journal of Earthquake Engineering 2005, 9(1), 95.
[29] O. S. Kwon, A. S. Elnashai, Engineering Structures 2006, 28(2), 289.
[30] M. Panagiotou, J. I. Restrepo, J. P. Conte, Journal of Structural Engineering 2011, 137(6), 691.
[31] A. M. Mwafy, A. M. Elnashai, Engineering Structures 2008, 30(6), 1590.
[32] A. M. Mwafy, O.‐S. Kwon, A. S. Elnashai, Engineering Structures 2010, 32(8), 2192.
[33] A. M. Mwafy, The Structural Design of Tall and Special Buildings 2012, 21(11), 824.
[34] Y. Luo, A. Durrani, ACI Structural Journal 1995, 92(1), 115.
[35] W. Alwaeli, A. M. Mwafy, K. Pilakoutas, M. Guadagnini, Earthquake Engineering and Structural Dynamics 2017, 46(8), 1325. https://doi.org/10.1002/
eqe.2858.
[36] LATBSDC, An Alternative Procedure for Seismic Analysis and Design of Tall Buildings Located in the Los Angeles Region, 2014 Edition with 2015 Supplements,
Los Angeles Tall Buildings Structural Design Council, Los Angeles, CA 2015.
[37] PEER, Guidelines for Performance‐based Seismic Design of Tall Buildings, Pacific Earthquake Engineering Research Center (PEER), University of California,
Berkeley 2010.
[38] M. J. N. Priestley, D. N. Grant, Journal of earthquake engineering 2005, 9(2), 229.
[39] A. M. Mwafy, A. S. Elnashai, R. Sigbjornsson, A. Salama, The Structural Design of Tall and Special Buildings 2006, 15(4), 391.
[40] NEHRP, Selecting and Scaling Earthquake Ground Motions for Performing Response‐History Analyses, National Institute of Standards and Technology,
Redwood City, California 2011.
[41] G. Aldama‐Bustos, J. J. Bommer, C. H. Fenton, P. J. Stafford, Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards 2009,
3(1), 1.
[42] PEER, PEER NGA database. Pacific Earthquake Engineering Research Center, University of California, Berkeley, California 2015. Retrieved from http://
peer.berkeley.edu/nga [Accessed June 2014].
[43] N. N. Ambraseys, P. Smit, J. Douglas, B. Margaris, R. Sigbjörnsson, S. Olafsson, P. Suhadolc, G. Costa, Bollettino di Geofisica Teorica ed Applicata 2004,
45(3), 113.
[44] J. Ji, A. S. Elnashai, D. A. Kuchma, Engineering Structures 2007, 29(12), 3197.
[45] J. B. Mander, M. J. N. Priestley, R. Park, Journal of Structural Engineering 1988, 114(8), 1804.
[46] M. J. N. Priestley, R. Verma, Y. Xiao, Journal of Structural Engineering 1994, 120(8), 2310.
[47] M. J. Kowalsky, M. N. Priestley, ACI Structural Journal 2000, 97(3), 388.
[48] ASCE/SEI‐41, Seismic Rehabilitation of Existing Buildings, Standard ASCE/SEI 41–06, American Society of Civil Engineers, Reston VA 2007.
[49] D. E. Lehman, J. A. Turgeon, A. C. Birely, C. R. Hart, K. P. Marley, D. A. Kuchma, L. N. Lowes, Journal of Structural Engineering 2013, 139(8), 1371.
[50] D. Vamvatsikos, C. A. Cornell, Earthquake Engineering & Structural Dynamics 2002, 31(3), 491.
20 of 22 MWAFY AND KHALIFA

[51] K. Beyer, A. Dazio, M. Priestley, Journal of earthquake engineering 2008, 12(7), 1023.
[52] H.‐S. Lee, D.‐W. Ko, Engineering structures 2007, 29(11), 3149.
[53] K. Kinali, B. R. Ellingwood, Engineering Structures 2007, 29(6), 1115.
[54] A. M. Mwafy, N. Hussain, K. El‐Sawy, Advances in Structural Engineering 2015, 18(12), 2017.

Aman Mwafy holds a PhD and a Diploma of Imperial College London in Structural Engineering. He is an associate professor in the Department
of Civil and Environmental Engineering at the UAE University. He was a senior research fellow at the University of Illinois at Urbana‐Champaign
from 2004 to 2007. He has been a faculty member in Egypt since 2001, and he is currently a professor of Structural Engineering at Zagazig
University (on leave). He is also an Adjunct Professor at the University of British Columbia. He has research interests in seismic behavior
and design of buildings and bridges, response of structures to extreme loading conditions, and advanced concepts for rehabilitation. He
authored and coauthored 80 publications, which accumulated 1000 citations. He was the recipient of the “Outstanding 2002 Journal Paper
Award” from Los Angeles Tall Buildings Structural Design Council for his research work on overstrength and force reduction factors, which
was published in The Structural Design of Tall Buildings. His journal papers in Engineering Structures, Elsevier, were ranked among the top
papers published in 2009, 2010, and 2012. He has taught graduate and undergraduate courses at the University of Illinois, UAE University,
and Zagazig University on fundamentals of earthquake engineering, earthquake risk management, design of wind and earthquake resistant
structures, and design of reinforced and prestressed concrete structures. He also supervised several students on their PhD, MSc, and gradua-
tion projects. He is a certified consulting engineer and an active member of a number of professional organizations. He has conducted several
consultancies on the design and assessment of high‐rise buildings, major bridges, and industrial structures in the USA, UK, UAE, and Egypt.

Sayed Khalifa is a Senior Structural Engineer. He obtained his BSc in Civil Engineering with honor in 2004 from Al Azhar University, Egypt, and
earned his MSc in 2015 from the United Arab Emirates University, UAE. He served many international consultants and has a wide experience in
the design of high‐rise structures in the UAE and Egypt. He has research interests in seismic design and assessment of high‐rise buildings.

How to cite this article: Mwafy A, Khalifa S. Effect of vertical structural irregularity on seismic design of tall buildings. Struct Design Tall
Spec Build. 2017;e1399. https://doi.org/10.1002/tal.1399

APPENDIX

FIGURE A1 Slab reinforcement details of B1‐REG and B2‐SST at all levels, and B3‐GEO, B4‐DIS, and B5‐WST at typical story levels
MWAFY AND KHALIFA 21 of 22

FIGURE A2 Reinforcement details of the transfer slab for B4‐DIS

FIGURE A3 Typical reinforcement details used in the design of shear walls and core walls
 22 of 22

TABLE A1 Characteristics of far‐field input ground motions

Mag. Site Ep. Dist. Duration a/v
Ref Earthquake Station Comp. Date (Mw) class (km) (s) PGA (m/s2) a/v g/ms−1 clas.

RL1 Bucharest Building res. Institute EW 04–03‐1977 7.53 Stiff 161 18 1.73 0.60 Low
RL2 Chi‐Chi CWB 99999 ILA013 EW 20–09‐1999 7.62 V. dense 135 117 1.36 0.52
RL3 Loma Prieta Emeryville 260 18–10‐1989 6.93 V. dense 96.5 39 2.45 0.57
RL4 Loma Prieta Golden Gate Bridge 270 18–10‐1989 6.93 V. dense 100 38 2.29 0.61
RL5 Hector mine Indio, Coachella Canal 0 16–10‐1999 7.13 Stiff 99 60 0.90 0.70
RL6 Izmit Ambarli‐Termik EW 17–08‐1999 7.64 Stiff 113 150 1.80 0.60
RL7 Izmit Istanbul‐Zeytinburnu NS 17–08‐1999 7.64 Stiff 96 129.24 1.08 0.77
RL8 Kocaeli Bursa Tofas E 17–08‐1999 7.51 Stiff 95 139 1.06 0.49
RL9 Kocaeli Hava Alani 90 17–08‐1999 7.51 V. dense 102 106.615 0.92 0.46
RL10 Loma Prieta Alameda Naval Air Stn Hanger 270 18–10‐1989 6.93 Stiff 91 29 2.39 0.73
RL11 Loma Prieta Berkeley LBL 90 18–10‐1989 6.93 V. dense 98 39 1.15 0.65
RL12 Loma Prieta Oakland‐Outer Harbor Wharf 0 18–10‐1989 6.93 Stiff 94 40 2.75 0.67
RL13 Manjil Abhar N57E 20–06‐1990 7.42 Stiff 91 29.49 1.30 0.62
RL14 Manjil Tonekabun N132 20–06‐1990 7.42 V. dense 131 40 1.22 0.76
RL15 Chi‐Chi TAP005 E 20–09‐1999 7.62 Stiff 156 134 1.34 0.49
RL16 Chi‐Chi TAP010 E 20–09‐1999 7.62 Stiff 151 144 1.19 0.50
RL17 Chi‐Chi TAP021 E 20–09‐1999 7.62 Stiff 151 125 1.15 0.47
RL18 Chi‐Chi TAP032 N 20–09‐1999 7.62 V. dense 144 90 1.13 0.64
RL19 Chi‐Chi TAP090 E 20–09‐1999 7.62 Stiff 156 125 1.28 0.41
RL20 Chi‐Chi TAP095 N 20–09‐1999 7.62 Stiff 158 123 0.96 0.52

Note. a/v: PGA/PGV, a/v classification (<0.8 low and >1.2 high), Vs30 of very dense soil = 360–760 m/s, and for stiff soil = 180–360 m/s
MWAFY AND KHALIFA