Fragility Analysis of Hillside Buildings Designed For Modern Seismic Design Codes

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Received: 21 January 2018 Revised: 7 March 2018 Accepted: 21 April 2018

DOI: 10.1002/tal.1500

RESEARCH ARTICLE

Fragility analysis of hillside buildings designed for modern


seismic design codes
Mitesh Surana1 | Yogendra Singh1 | Dominik H. Lang2

1
Department of Earthquake Engineering,
Indian Institute of Technology Roorkee, Summary
Roorkee, India The structural configurations of hillside buildings are significantly different than those
2
Department of Earthquake Hazard and Risk,
observed on flat terrain. To study the effect of often observed structural configura-
NORSAR, Kjeller, Norway
Correspondence
tions of hillside buildings, collapse fragility of regular “flat land (FL)” and irregular
Yogendra Singh, Department of Earthquake “split‐foundation (SF)” and “step‐back (SB)” buildings is studied using incremental
Engineering, Indian Institute of Technology
Roorkee, Roorkee, India.
dynamic analysis. The effects of building height, seismic zone, and near‐ and far‐field
Email: yogendra.eq@gmail.com sites on collapse fragility are investigated. It is observed that SF and SB hillside build-
ings exhibit significant torsional effects at the storey just above the uppermost foun-
dation level. In case of FL buildings, collapse occurs due to the flexural failure of
beams and columns. On the other hand, in SF and SB buildings, the collapse occurs
due to the combined effects of shear failure of short columns and flexural failure of
beams and columns in the storey just above the uppermost foundation level. In gen-
eral, SF buildings are observed to be the most vulnerable, whereas FL buildings are
the least vulnerable. It is observed that high‐rise SF and SB buildings show unaccept-
ably high probability of collapse at maximum considered earthquake, in Seismic Zone
V and for the near‐field site in Seismic Zone IV.

KEY W ORDS

collapse probability, fragility analysis, hillside buildings, incremental dynamic analysis, split‐
foundation building, step‐back building

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

During the past decades, the Indian subcontinent has witnessed rapid urbanization and a dramatic increase in population. Due to these factors,
building development not only has been extensive in and around the metropolitan cities but also has been spreading out to the hilly regions of
India resulting in the construction of a large number of multi‐storey reinforced concrete (RC) frame buildings on steep slopes (hereafter referred
to as “hillside buildings”). In some cases, building development in hilly regions is also taking place because of the scenic view and the pleasant
climatic conditions during the summer season. In hilly regions characterized by very mild slopes, buildings similar to those existing in flat areas
can be observed[1] and are referred to as “flat land (FL)” buildings, in this study. In case of mild to steep slopes, the structural configurations of
hillside buildings are significantly different than for buildings constructed on FL,[1–5] in order to adapt to the slope's geometry. Foundation levels
of hillside buildings generally follow the natural slope of the ground, known as “step‐back (SB)” buildings.[1] In case of very steep slopes, hillside
buildings with “split‐foundation (SF)” can be frequently observed.[1] These hillside buildings have significant structural irregularities in plan and
in elevation due to the presence of shorter columns on the uphill side and variations in mass, strength, and stiffness of successive storeys along
the height, respectively. Both these features result in complex dynamic behaviour under earthquake excitation[1–5] increasing the structure's
susceptibility to suffer greater damage.
The Sikkim earthquake of September 18, 2011 was one of the major earthquakes affecting a hilly region of India, which exposed the seismic
vulnerability of multi‐storey RC frame buildings in hills. A significantly poor performance of hillside buildings has been reported in various damage

Struct Design Tall Spec Build. 2018;e1500. wileyonlinelibrary.com/journal/tal Copyright © 2018 John Wiley & Sons, Ltd. 1 of 13
https://doi.org/10.1002/tal.1500
2 of 13 SURANA ET AL.

reports of the 2011 Sikkim earthquake.[6–9] Figure 1 shows the damage caused by the 2011 Sikkim earthquake to an SF hillside building. The
storeys nearest to the uppermost foundation level experienced the highest damage. Although this earthquake of moment magnitude Mw 6.9
exhibited only moderate ground‐motion levels (a peak ground acceleration = 0.18 g was recorded in Gangtok city, which was located at a distance
of 68 km south‐east of the epicentre), the observed damage extent was fairly high. Many buildings suffered severe damage, whereas a few build-
ings even collapsed.[6,7,9] The hilly state of Sikkim is situated in Seismic Zone IV as per the current seismic zoning map of India, and an effective
peak ground acceleration (i.e., EPGA or zone factor) of 0.24 g is assigned to this region.[10] (The actual maximum considered earthquake (MCE)
level design EPGA is 0.36 g, as the Indian code also applies a load factor of 1.5 to the earthquake load.) As the observed ground‐motion amplitude
was much lower than the design EPGA in the code, the disproportionate damage observed during the earthquake may be attributed to a number
of factors, which include the lack of earthquake‐resistant design features, faulty construction practices, irregular configurations of buildings due to
the constraints posed by the slope topography, slope instability, and seismic ground‐motion amplification caused by local topography.[6,8,9]
Recently, the Indian seismic design codes[10,11] underwent significant changes in the design requirements for special moment‐resisting
frame buildings. The major changes include a provision for the strong column–weak beam (SCWB) design, which has already been identified to
affect the collapse fragility of RC frame buildings to a significant extent.[12] The revised Indian code is at par with other major national codes
of the world, and hence, the buildings designed following the latest Indian seismic design codes[10,11] are hereafter referred as “modern code‐
designed buildings.”
A significant number of research efforts have already been made by various researchers to evaluate the collapse fragility of modern code‐
conforming regular and irregular reinforced concrete moment‐resisting frame buildings resting on FL.[12–16] On the other hand, studies addressing
SF and SB structural configurations are very limited.[1–3] Further, the earlier studies[1–3] have been conducted with a specific focus on precode and
moderate‐code buildings located at far‐field sites. However, most of the areas in the Indian Himalayan region fall under the near‐field site cate-
gory. Further, none of the current seismic design codes provide any specific guidelines for hillside buildings. Therefore, assessment of collapse fra-
gility of hillside buildings that are designed following modern codes is of prime importance.
This study aims to investigate the effects of building height, seismic zone, and near‐ and far‐field sites on the collapse fragility of RC frame
hillside buildings of FL, SF, and SB structural configurations using incremental dynamic analysis (IDA) procedure. The obtained results from IDA
are used to further study the governing collapse mechanism and to develop the collapse fragility functions. The developed collapse fragility curves
are used to compute the collapse probabilities of FL, SF, and SB hillside buildings for the maximum considered earthquake (MCE) hazard level. The
derived collapse fragilities are further important (a) to assess the adequacy of modern (current) code provisions and (b) to further develop seismic
design guidelines, with the goal to achieve acceptable collapse performance. The hillside buildings are also often subjected to slope stability
problems as well as topographic amplification effects, leading to secondary effects and an increased seismic demand, respectively. However, these
effects have not been considered in this study.

2 | P A S T S TUD I E S ON C OL L A P SE A S SE SSM E N T OF RC F R A M E BU I L D I N G S

The simulation of structural collapse under earthquake excitations completely relies on rigorous analytical models calibrated with experimental
evidences. The analytical methods of structural collapse assessment essentially consist in the simulation of the non‐linear seismic response and
estimation of capacity of the structure with associated variabilities. Earlier attempts of vulnerability analysis were based on the capacity spectrum
approach using non‐linear static analysis,[17] whereas more recent studies[1,12–16] utilize either IDA[18] or multiple stripe analysis[19] to obtain the
dynamic collapse capacity of a structure. All of these procedures require reliable estimation of a structure's strength, stiffness, and deformation
capacity accounting for the effects of load reversal while being subjected to an earthquake.
The lumped‐plasticity models are widely adopted in collapse assessment studies[1,12–15] as these can be easily calibrated with experimental
results, and the procedure is less sensitive to the detailing of RC sections, compared with the continuum and fibre models.[19] However,

FIGURE 1 Failure of the storey near to the uppermost foundation level of a split‐foundation building observed during 2011 Sikkim earthquake:
(a) Front view; and (b) Side view
SURANA ET AL. 3 of 13

lumped‐plasticity models are either calibrated for individual sections or predetermined generic quantities and therefore may result in errors in
the assessment procedure.[18–20] Some of the recent studies[19,20] utilized fibre element‐based models for collapse capacity assessment. Due
to the different opinions among researchers,[18–23] the use of a particular modelling approach in collapse assessment is still significantly
dependent on the user's choice. In this study, the lumped‐plasticity model is used for the simulation of collapse. In lumped‐plasticity models,
the force–deformation behaviour of an element is confined by an envelope, known as the “backbone curve.” These backbone curves are obtained
through experimental studies and can have two different definitions based on how these curves are derived: (a) monotonic backbone curve and (b)
cyclic backbone curve. Haselton et al.[22,23] developed monotonic backbone curve parameters on the basis of experiments conducted on 255 rect-
angular columns. Therefore, the simulation of collapse using these parameters requires explicit simulation of stiffness and strength degradation
effects at large deformations in the analytical model.[23] On the other hand, backbone curves adopted in ASCE 41[24] are based on cyclic tests[25]
and inherently include the cyclic strength deterioration effects. Therefore, the backbone curve parameters of ASCE 41 can be used in conjunction
with a hysteretic model capable of simulating the stiffness degradation and energy dissipation effects at large deformations. In addition, explicit
simulation of the postpeak behaviour (strain softening) is also important in case of simulation of collapse and therefore should also be included
in the analytical model.[14,22,23]
Shear failure of RC members is also a potential failure mode in RC buildings, especially in hillside SF and SB buildings having short columns.
In earlier developments of performance‐based earthquake engineering, shear failure was considered to be a brittle mode of failure, leading to
the loss of the axial load‐carrying capacity in columns at very small interstorey drifts.[26–28] Earlier studies[15] for structural collapse capacity
assessment of RC buildings considered shear failure as a nonsimulated collapse mode. In this procedure, the column drifts obtained from
experimental evidences are utilized to detect the shear failure in the analytical model.[15,29] Later, experimental evidences revealed that the
shear failure mode has certain postcapping plastic rotation capacity, which degrades with increase in axial loads.[25,30] On the basis of these
findings, ASCE 41[24] provides guidelines to model shear failure, in which the collapse due to shear failure of the component is considered
corresponding to the full loss of the vertical load‐carrying capacity.[24,25] However, ASCE 41 guidelines are based on test results from 274
rectangular RC columns,[25] only two of which being short columns,[30] which are particularly prone to shear failure in case of hillside buildings.
Li and Hwang[30] compiled test data for 33 short columns with height‐to‐depth ratios less than 2 and proposed a model to simulate the shear
failure of short columns. Their model has two basic differences when compared with the ASCE 41[24] model. The model considers shear
cracking of the concrete prior to reaching the strength point, and after reaching the strength point, the shear strength degrades gradually
to 0 at the point of axial failure.[30] The point of axial failure (full loss of vertical load‐carrying capacity) can be computed on the basis of
recommendations given in ASCE 41.
In the IDA, the recorded earthquake ground‐motions are incrementally scaled in amplitude and applied to the structure till the structure
reaches collapse. The appropriate selection and scaling of the ground‐motion records is also an important issue in the collapse assessment of
structures.[13,31] FEMA P695[32] recommends a selection criterion for general ground‐motion records to be used for the estimation of collapse
capacity of buildings and has identified sets of 28 (14 no pulse‐like ground‐motion records and 14 pulse‐like ground‐motion records) and 22
pairs of ground‐motion records on the basis of the recommended criterion, representative of near‐ and far‐field ground‐motion record suites,
respectively. In this study, the same near‐ and far‐field ground‐motion record suites, as recommended in FEMA P695,[32] are used.
Past research[31,33–35] shows that rare ground‐motions (MCE ground‐motions with return period ≈ 2,475 years) able to cause structural col-
lapse exhibit significantly different spectral shape as compared with the scaled ground‐motions of shorter return periods. This effect of spectral
shape alone has significant influence on the estimated median collapse capacity of a structural model. The scaled ground‐motions may consider-
ably underpredict (in case of positive “ε”) or overpredict (in case of negative “ε”) the median collapse capacity of the buildings at the site of interest
depending on the value of “ε” at the site of interest. The parameter “ε” is defined as the number of standard deviations by which the value of lnSa
differs from the mean predicted lnSa, for the corresponding magnitude and distance, using a certain ground‐motion prediction equation, corre-
sponding to the period of the considered building.[31] The most ideal way to consider spectral shape effects is the selection of ground‐motion
records having the targeted (corresponding to the rare earthquakes expected at the site) spectral shape, which can be obtained from either the
conditional mean spectrum (CMS[36]) or the conditional spectrum[37]. However, the selection of ground‐motion records based on CMS and
conditional spectrum is a cumbersome task as it requires different ground‐motion record suites for buildings of different natural periods.[31]
Alternatively, selection of an intensity measure (IM) insensitive to the effects of spectral shape of the used ground‐motion records can be a
preferred choice for structural collapse assessment of a class of buildings of varying building heights.
Recent research[34,35] showed that an IM consisting of average spectral accelerations Sa,avg (0.2T–3T, 5%) representing the geometric mean of
the 5% damped spectral acceleration over the period range between 0.2T and 3T (equally spaced at an interval of 0.01 s) can capture the spectral
shape effects, with a significantly reduced dispersion in IM corresponding to collapse, for regular symmetric buildings subjected to unidirectional
excitations.[34,35] The better capability of Sa,avg (0.2T–3T, 5%) in capturing the spectral shape effects as compared with Sa (T, 5%; duly adjusted for
the spectral shape effects through “ε”) lies in the fact that the seismic response of a non‐linear multiple‐degree‐of‐freedom system is significantly
affected by the higher mode periods (periods shorter than T) as well as the elongated period (periods longer than T) of the structure. This effect is
due to the significantly inelastic response of the building, particularly near the collapse. Whereas Sa (T, 5%) includes any information of the
spectral ordinates neither at higher mode periods nor at elongated periods and is thereby considered to be relatively inefficient in capturing
the spectral shape effects as compared with Sa,avg (0.2T–3T, 5%).[34,35] Further, this IM was observed to be very efficient even in case of torsionally
irregular hillside buildings designed for moderate seismic design codes and subjected to bidirectional excitations.[1]
4 of 13 SURANA ET AL.

All of the previous studies on collapse capacity assessment are limited to regular FL buildings and are mainly based on two‐dimensional
models subjected to a unidirectional earthquake excitation. The hillside SF and SB buildings have significantly different dynamic characteristics
and behaviour as compared with FL buildings.[1–4] Due to their irregular structural configurations, hillside buildings are also subjected to torsional
effects in the cross‐slope direction. Therefore, this study investigates the collapse fragility of modern‐code designed hillside buildings
(three‐dimensional [3D] models with FL, SF, and SB structural configurations) subjected to bidirectional earthquake excitations.

3 | N U M E R I C A L ST U D Y

For the numerical study, a generic building plan is chosen on the basis of extensive field surveys conducted in the Indian Himalayan regions.[1]
Figure 2 shows plan and elevations of the various building models investigated in this study. Although the low‐ and mid‐rise buildings are quite
popular in hilly regions, high‐rise buildings are very rare, because the building bye‐laws permit buildings only up to 12 m height.[1] Accordingly,
three different building heights representative of low‐ (two‐storey), mid‐ (four‐storey), and high‐rise (eight‐storey) buildings are considered. In case
of hillside SF and SB buildings, the number of storeys has a slightly different connotation than on FL. It has already been observed that the
dynamic characteristics of hillside buildings are controlled by the number of storeys above the uppermost foundation level.[1–4] Therefore,
the same height classification consistent with earlier studies[1–4] is considered in this study. In all the investigated building models, the number
of storeys below the uppermost foundation level is kept equal to 4. The storey height is kept constant equal to 3.3 m, which was also most
commonly observed during the field surveys.[1] For buildings of SF and SB configurations, their larger dimension (denoted as the “length,”
hereafter) is kept parallel to the direction of the slope. In case of buildings of SB configuration, a slope gradient of 2H:1V (corresponding to
50% steepness or 30° slope angle) is considered, which was most predominantly observed for the surveyed region.[1] All building models are
designed for the combined effects of gravity and earthquake loads following the provisions for “special moment‐resisting frames” according to
Indian Standards IS 1893[10] and IS 13920.[11] The seismic design and performance of the buildings are influenced to a great extent by the capping
on the design period.[38] In the seismic design of all the considered building configurations (FL, SF, and SB), the upper‐bound design period based
on the building height above the uppermost foundation level is considered.[10]

FIGURE 2 Typical structural configurations investigated in this study: (a) generic plan of the investigated building models and elevation of a four‐
storey (b) flat land (FL), (c) split‐foundation (SF), and (d) step‐back (SB) building. In case of split‐foundation buildings, the short columns are 1.1 m
high, whereas in case of step‐back buildings, the heights of neighbouring short columns resting on the slope are of 1.1 and 2.75 m (refer to (c) and
(d), respectively) All dimensions are in meters
SURANA ET AL. 5 of 13

Eighteen building models (three different structural configurations, viz., FL, SF, and SB, three different heights, that is, two, four, and eight
storeys above the uppermost foundation level, and two different seismic zones, that is, Zone IV and Zone V, according to Indian seismic zoning
map) are considered. Indian seismic design codes[10,11] do not provide any specific design guidelines for the seismic design of buildings located
at near‐field sites. Therefore, the seismic design of buildings located at near‐ and far‐field sites does not differ. The buildings in Seismic Zone
IV are assessed for the near‐field suites with and without velocity pulse as well as for the far‐field ground‐motion record suite, whereas the
buildings in Seismic Zone V are assessed for the far‐field ground‐motion record suite only.
The 3D building models are generated in the proprietary building analysis and design software ETABS 2016.[39] Beams and columns are
modelled using 3D frame elements, whereas slabs are defined as rigid diaphragms. The cracked section properties of beams and columns
are derived following ASCE 41[24] guidelines. Dead loads and live loads on the buildings are assigned according to IS 875 Part 1[40] and IS 875
Part 2,[41] respectively. The buildings are designed for Zone IV (EPGAMCE = 0.36 g, SDS = 0.45 g, and SD1 = 0.18 g), or for Zone V (EPGAMCE = 0.54 g,
SDS = 0.68 g, and SD1 = 0.27 g) on soil type I (i.e., hard soil/rock). The Indian Seismic Zones IV and V can be considered as equivalent to seismic
design category C and Dmax, respectively, as specified in FEMA P695. All building models confirm to the SCWB design criteria as per IS 13920,[11]
which suggests a SCWB ratio of 1.40. P‐delta effects are considered in the analysis as well as in the design. The periods of vibration (arithmetic
mean of the periods of vibration in the two principal directions of the building), design forces, and obtained member sizes for the considered
building models are summarized in Table 1.
The member sizes are chosen such that the reinforcement in columns is between 1.5% and 4% and in beams between 0.75% and 1.5% on
each face, also consistent with the typical design practice in India. It can be observed from Table 1 that the average period of vibration of SF
and SB buildings of the same height above the uppermost foundation level is quite close to the corresponding FL buildings. This observation
has been found to be in agreement with earlier studies on hillside buildings.[1–4] Accordingly, the design base shear coefficients (controlled by
the upper‐bound design period) of all buildings of the same height above the uppermost foundation level are identical.

3.1 | Non‐linear modelling and analysis


Uniaxial moment (M3) plastic hinges and P–M2–M3 interaction hinges are assigned at both ends of the beams and columns, respectively. The
backbone curve parameters for beams and columns are derived following ASCE 41[24] guidelines. The strength degradation effects are inherently
considered in the ASCE 41 backbone curves (as these backbone curves are derived from cyclic envelopes), whereas the degradation of stiffness
and energy dissipation capacity is modelled explicitly using an energy‐based degrading hysteresis model in ETABS 2016.[39] This hysteresis model
requires three parameters to capture the energy dissipation and stiffness degradation effects, namely, energy factor (f1), which represents the ratio
of the energy dissipated in a degraded loop to nondegraded loop for deformations within the pre‐capping range, energy factor (f2), which repre-
sents the ratio of the energy dissipated in a degraded loop to a nondegraded loop for deformations in the postcapping range, and stiffness
weighting factor (s) to account for stiffness degradation effects. In this study, the parameters f1, f2, and s calibrated for RC elements conforming

TABLE 1 Period, design base shear, and member sizes of the considered hillside building models (all dimensions are in millimetres)
Building model Ta (s) V/Wb Beams Columns Short columns

2‐FL‐IV 1.00 0.090 300 × 300 350 × 350 —


4‐FL‐IV 1.50 0.066 300 × 400 350 × 350 —
8‐FL‐IV 3.35 0.041 300 × 400 300 × 300, 350 × 350 —
2‐SF‐IV 1.05 0.090 300 × 300 350 × 350 400 × 400
4‐SF‐IV 1.51 0.066 300 × 400 350 × 350 400 × 400
8‐SF‐IV 3.38 0.041 300 × 400 300 × 300, 350 × 350 400 × 400
2‐SB‐IV 1.07 0.090 300 × 300 350 × 350 600 × 600, 450 × 450
4‐SB‐IV 1.52 0.066 300 × 400 350 × 350 650 × 650, 450 × 450
8‐SB‐IV 3.36 0.041 300 × 400 300 × 300, 350 × 350 650 × 650, 450 × 450
2‐FL‐V 1.00 0.135 300 × 300 350 × 350 —
4‐FL‐V 1.50 0.098 300 × 400 350 × 350 —
8‐FL‐V 3.35 0.060 300 × 400 300 × 300, 350 × 350 —
2‐SF‐V 1.03 0.135 300 × 300 350 × 350 450 × 450
4‐SF‐V 1.49 0.098 300 × 400 350 × 350 450 × 450
8‐SF‐V 3.30 0.060 300 × 400 300 × 300, 350 × 350 500 × 500
2‐SB‐V 1.00 0.135 300 × 300 350 × 350 750 × 750, 450 × 450
4‐SB‐V 1.49 0.098 300 × 400 350 × 350 800 × 800, 450 × 450
8‐SB‐V 3.32 0.060 300 × 400 300 × 300, 350 × 350 800 × 800, 500 × 500
a
Arithmetic mean of the periods in the two orthogonal direction of the considered model.
b
Design base shear coefficient.
6 of 13 SURANA ET AL.

to Indian code are adopted from an earlier study on collapse assessment of RC frame buildings.[1] The details of calibration and validation of
the degrading hysteresis model can be found in Surana et al.[1] Shear failure is modelled following ASCE 41[24] guidelines considering it a displace-
ment‐controlled action, whereas the shear strength is computed on the basis of section and transverse reinforcement details. On the basis of
the recommendations from Li and Hwang,[30] a gradual reduction in the postpeak strength is considered, and shear failure is assumed
when the displacement exceeds the allowable limit specified in ASCE 41.[24] To model damping effects, a Rayleigh damping of 5% is assigned
at periods corresponding to the lowest (fundamental) mode and the mode resulting in 95% cumulative mass participation.
In this study, Sa,avg (0.2T–3T, 5%) is used as the scaling parameter as well as the IM. It is important to consider that in case of SF and SB
structural configurations, torsional effects in the cross‐slope direction are observed due to plan irregularity. Therefore, bidirectional excitations
(i.e., simultaneous application of two horizontal components of the ground‐motions) are used in the IDA. For each ground‐motion record, two
analyses have been performed by interchanging the two horizontal components of the ground‐motion along two principal orthogonal axes. To
scale the bidirectional ground‐motion records, the geometric mean of Sa,avg (0.2T–3T, 5%) of individual components is used. As the building models
have different periods in the two principal directions, the arithmetic mean of the periods in the fundamental translational modes in the two
orthogonal directions is used for the computation of Sa,avg (0.2T–3T, 5%), which is also consistent with the provisions of FEMA P58.[42] The
collapse is defined on occurrence of any of the following limit states: (a) a slight increment in IM, Sa,avg (0.2T–3T, 5%), causes a large increase
in damage measure (i.e., the maximum interstorey drift ratio in this study); (b) maximum interstorey drift exceeds 10%; and (c) axial failure of
columns, leading to loss of vertical (gravity) load‐carrying capacity. The ground‐motion intensity at which any of these limit states is reached is
recorded for each of the ground‐motion pair.

4 | D Y N A M I C CA P A C I T Y CU R V E S

Figure 3 presents the dynamic capacity curves for the four‐storey hillside SF and SB buildings, obtained using the far‐field ground‐motion record
suite. Similar results are also obtained for the remaining building models investigated in this study but are not presented here for the sake of
brevity. The IDA results are presented in terms of IM (Sa,avg [0.2T–3T, 5%]) versus damage measure (maximum interstorey drift ratio) curves for
individual records, as well as the 16th percentile, 50th percentile (median), and 84th percentile dynamic capacity curve.
Table 2 presents the median collapse capacities (in terms of Sa,avg) of all the investigated hillside building models, for near‐ (with and without
velocity pulse) and far‐field ground‐motion record suites. It can be observed that, although all hillside buildings of the same height above the
uppermost foundation level, in a particular seismic zone, were designed for identical base shear coefficients (Table 1), the SF buildings exhibit
the least median collapse capacity, whereas FL buildings have the highest median collapse capacity, except in case of the four‐storey building
designed for Zone IV. It is interesting to note that the hillside building models in Zone V were designed for 50% higher base shear coefficients;
however, the increase in median collapse capacity is observed up to 10% only (when compared with corresponding values in Seismic Zone IV),
because a larger portion of collapse capacity is contributed by ductility capacity of the building, which remains more or less the same. Further,
the median collapse capacities of the same building (designed for Seismic Zone IV) obtained using different ground‐motion record suites (far‐field
and near‐field) have some difference. The difference is observed only up to 10%, in case of low‐ and mid‐rise buildings (Table 2), and up to 25%, in
case of high‐rise buildings.
To further investigate the relatively lower median collapse capacity in case of SF buildings in comparison with the corresponding SB hillside
buildings, the variation of the normalized eccentricity (e/b, where e is the eccentricity between centre of mass, CM, and centre of rigidity, CR, and
b is the plan dimension perpendicular to the direction of the earthquake excitation) for the four‐storey SF and SB buildings along the height of the
building is shown in Figure 4. In case of multi‐storey buildings, two different definitions of CR are available.[43] In the first case, the CR is defined
by a set of points located on each floor through which the application of the lateral load profile would not cause rotation of any floor. In the

FIGURE 3 Typical dynamic capacity curves for four‐storey buildings designed for Seismic Zone IV, subjected to the far‐field ground‐motion
record suite: (a) split‐foundation building and (b) step‐back building. IDA = incremental dynamic analysis
SURANA ET AL. 7 of 13

TABLE 2 Median collapse capacities of all the considered buildings for NF and FF ground‐motion record suites

Median collapse capacity, Sa,avg


Ground‐motion Building
record suite height FL (g) SF (g) SB (g)
NF (NP) Zone IV 2‐storey 0.65 0.48 0.61
4‐storey 0.31 0.29 0.38
8‐storey 0.08 0.06 0.07
NF (P) Zone IV 2‐storey 0.63 0.54 0.64
4‐storey 0.29 0.32 0.35
8‐storey 0.08 0.07 0.07
FF Zone IV 2‐storey 0.66 0.49 0.60
4‐storey 0.32 0.29 0.38
8‐storey 0.10 0.08 0.09
FF Zone V 2‐storey 0.66 0.52 0.63
4‐storey 0.36 0.32 0.37
8‐storey 0.11 0.07 0.10

Note. NF = near‐field; FF = far‐field; NP = ground‐motion suite with no pulse; P = ground‐motion suite with pulse; FL = flat land; SF = split‐foundation;
SB = step‐back.

FIGURE 4 Variation of the normalized floor eccentricity for a four‐storey split‐foundation (SF) and step‐back (SB) buildings. The grey‐shaded
area shows the building portion below the uppermost foundation level

second case, the CR is represented for a single floor such that the application of lateral load at this point would not cause any rotation of that floor,
whereas the other floors may undergo rotation.[43] In this study, the single floor definition of CR is adopted to compute the eccentricity of the
considered SF and SB structural configurations.
It can be observed from Figure 4 that, for the elastic response, the SB hillside building has a greater normalized eccentricity than the corre-
sponding SF hillside building. Because, in the code‐based design procedure, the buildings are designed using elastic response, the SB buildings are
designed for higher torsional effects. However, with the progress in yielding of the elements at the flexible side (i.e., the downhill side columns),
the relative eccentricity in case of SF building increases more rapidly than the SB building, especially at the floor just above the uppermost foun-
dation level. This phenomenon leads to significantly increased torsional effects in SF buildings in the inelastic range (due to shifting of centre of
rigidity towards the uphill side), for which SF hillside buildings were not originally designed. This effect leads to a reduction in median collapse
capacity for SF hillside buildings, as compared with the SB hillside buildings, even though both were designed for identical base shear coefficients.
The observed trends for normalized floor eccentricity and reduction in median collapse capacity are quite consistent among all the investigated SF
and SB building models. This observation on the effect of eccentricity on median collapse capacity is contrary to what was observed in case of
precode buildings, investigated in an earlier study.[1] This difference in observation can be attributed to the fact that the precode buildings were
not designed for torsion at all, and as a result, in those cases, torsional effects were observed to be more pronounced in SB buildings as compared
with the SF buildings.
The hinge patterns (at collapse) for the four‐storey FL, SF, and SB buildings are presented in Figure 5. Because the hinge pattern varies with
ground‐motion record, the presented failure mechanisms correspond to the most frequently observed hinge patterns, in the analysis for a suite of
22 far‐field ground‐motion records. The FL buildings show flexural failure of beams and columns of the first storey (Figure 5a), whereas the SF
buildings show shear failure of short columns at the uppermost foundation level as well as flexural failure of beams and columns on the downhill
side (Figure 5b). As the downhill side is the flexible side of the building, the columns on the downhill side are subjected to larger displacements, due
to torsional effects. In case of the SB building, a totally different failure mechanism is observed. In this building, shear failure of short columns
occurs at the uphill side and propagates in lower storeys towards the downhill side. This failure sequence is also known as “zippering effect” as
explained by Welsh‐Huggins et al.[5] In addition, the flexural failure of members at the storey just above the uppermost foundation level can also
be observed (Figure 5c). In SF and SB buildings, the storey just above the uppermost foundation level is found to be the most susceptible to
8 of 13 SURANA ET AL.

FIGURE 5 Hinge patterns at collapse for the four‐storey buildings: (a) flat land building, (b) split‐foundation building, and (c) step‐back building

damage due to larger torsional effects and also due to transfer of large shear forces to the short columns. The observed trends are consistent
among all the investigated low‐ and mid‐rise buildings. In case of high‐rise buildings, the beam failure in multiple storeys is observed when com-
pared with low‐ and mid‐rise buildings, where beam failure is observed in either one or two storeys. Similar damage patterns have also been
observed for the near‐field ground‐motion record suites, except in case of few pulse‐like ground‐motions, where the damage pattern was steered
by column failure at upper storeys. The observed column failure in upper storeys can be attributed to travelling wave effect in case of pulse‐like
ground‐motions and also observed to be consistent with an earlier study on FL buildings.[44]
While comparing SF and SB buildings, it is observed that the short columns in the SB buildings are more susceptible to shear failure, as
compared with the SF buildings. The reason for this observation can be attributed to the fact that the storey shear at the uppermost foundation
level, in case of SF building, is distributed in a relatively larger number of short columns (16 in case of SF building and 4 in case of SB building, in
the considered models). The failure pattern obtained from IDA in the analytical investigations of SF building can also be corroborated with the
damage observed in a SF hillside building after the Sikkim earthquake of 2011, which resulted in the failure of a storey just at the uppermost
foundation level (Figure 1).

5 | FRAGILITY ANALYSIS

The results obtained from the IDA are further postprocessed following the methodology used in earlier studies[1,12] for the purpose of generating
the corresponding fragility curves. Fragility curves are log‐normal distributions obtained from median collapse capacity and associated uncertainty.
The uncertainty in collapse capacity has four major components[32]: (a) record‐to‐record uncertainty (βRTR), (b) design requirement‐related uncer-
tainty (βDR), (c) test data‐related uncertainty (βTD), and (d) modelling uncertainty (βM). The record‐to‐record uncertainty (βRTR) is directly obtained
from the IDA results. On the basis of the United States building dataset, Liel et al.[45] showed that the combined epistemic uncertainty (βCE) due to
βDR, βTD, and βM can be taken as 0.50. This value is also consistent with the recommendations of FEMA P695,[32] considering design for modern
codes and good confidence in results. In the absence of more reliable estimates of the uncertainty in simulation of collapse of modern Indian code‐
designed buildings, the combined epistemic uncertainty, βCE from previous studies[12,45] is adopted for this study. The uncertainties βRTR and βCE
are combined using the square‐root‐of‐sum‐of‐squares method[12,32,45] in order to obtain total uncertainty (βT).
Table 3 presents βRTR for the hillside building models investigated in this study, subjected to near‐ and far‐field ground‐motion record suites. A
comparison of values for βRTR (in terms of Sa,avg [0.2T–3T, 5%]) as derived in this study, with the typical values of βRTR (in terms of Sa [T, 5%]) pre-
sented in earlier studies,[2,3,13,32] indicates that considering Sa,avg (0.2T–3T, 5%) as the IM results in much lower βRTR. From the same far‐field
record suite, βRTR is found out to vary between 0.36 and 0.46 with an average value of 0.40, in case of regular buildings subjected to unidirectional
earthquake excitations, when Sa (T, 5%) was chosen as the collapse IM.[12,13] A significant reduction (of the order of 40%) in βRTR is observed when
Sa,avg (0.2T–3T, 5%) is chosen as the collapse IM, and the corresponding βRTR varies between 0.14–0.43 and 0.15–0.43 in case of near‐ and
far‐field ground‐motion record suites, respectively, with an average value of 0.25, in case of both near‐ as well as far‐field ground‐motion record
suites. This reduction has not only been observed in case of FL buildings and also in case of torsionally irregular SF and SB buildings, even when
subjected to bidirectional earthquake excitation. The presented results clearly show the superiority of Sa,avg (0.2T–3T, 5%) as the collapse IM when
compared with Sa (T, 5%).
In order to compute the collapse probability, a site‐specific seismic hazard analysis, estimating the shape of the governing earthquake, is
necessary. In this study, results of the site‐specific seismic hazard for two sites in the Indian Himalayas, at Mussoorie (Uttarakhand state, north
India) and Shillong (Meghalaya state, north‐east India), are used. The Mussoorie site is a near‐field rock site (NEHRP site Class B) located in Zone
IV, whereas Shillong is a far‐field rock site (NEHRP site Class B) located in Zone V. In case of the Mussoorie site, seismic hazard analyses in terms
of Sa (T, 5%) are conducted considering two different ground‐motion attenuation models: (a) Campbell and Bozorgnia[46] (neglecting near‐source
SURANA ET AL. 9 of 13

TABLE 3 Record‐to‐record uncertainty in collapse capacity of the considered buildings, using NF and FF ground‐motion record suites

Record‐to‐record uncertainty, βRTR


Ground‐motion Building
record suite height FL SF SB
NF (NP) Zone IV 2‐storey 0.17 0.21 0.14
4‐storey 0.29 0.33 0.25
8‐storey 0.27 0.43 0.23
NF (P) Zone IV 2‐storey 0.22 0.23 0.19
4‐storey 0.21 0.16 0.19
8‐storey 0.30 0.33 0.31
FF Zone IV 2‐storey 0.21 0.21 0.21
4‐storey 0.15 0.25 0.15
8‐storey 0.27 0.43 0.26
FF Zone V 2‐storey 0.19 0.21 0.23
4‐storey 0.17 0.24 0.20
8‐storey 0.32 0.33 0.33

Note. NF = near‐field; FF = far‐field; NP = ground‐motion suite with no pulse; P = ground‐motion suite with pulse; FL = flat land; SF = split‐foundation;
SB = step‐back.

effect) and (b) Campbell and Bozorgnia[47] (including near‐source effect), whereas for the Shillong site, the seismic hazard analysis is conducted
using the Campbell and Bozorgnia[47] attenuation model only. The “ε” values at each site and period of interest (fundamental period of the building
under consideration) are derived from the deaggregation of seismic hazard. The “ε” values at other spectral periods are obtained using cross‐
correlation coefficients developed for the Pacific Earthquake Engineering Research Center Next Generation Attenuation (NGA) ground‐motion
records by Baker and Jayaram.[48] The mean expected response spectrum (accounting for the expected spectral shape), also called as CMS, at
the site and period of interest is obtained using the procedure recommended by Baker.[36] The obtained CMS is used to compute the Sa,avg
(0.2T–3T, 5%) demand corresponding to the MCE hazard level (average return period of 2,475 years). This way of computing Sa,avg (0.2T–3T,
5%) is an indirect procedure because the hazard analysis is conducted in terms of Sa (T, 5%). However, this procedure leads to identical median
Sa,avg (0.2T–3T, 5%) values when compared with the case in which the seismic hazard analysis is performed in terms of Sa,avg (0.2T–3T, 5%) using
the direct procedure.[49] Both the sites considered in this study are located in hilly topography, and it is well known that the ground‐motion inten-
sity can vary depending on the location of a building with respect to a hill feature. However, as clarified earlier, this effect has not been considered
in this study.
Figure 6 shows a comparison of the 5% damped elastic design response spectra (as per IS 1893[10]), which were actually used for design of
buildings, with the CMS obtained for three different periods of interest corresponding to the two‐, four‐ and eight‐storey buildings, for MCE
(although the periods of buildings of same height but different structural configurations differ slightly, the CMS are shown for 1.00, 1.50, and
3.35 s, which represent average values of period for the three structural configurations) for the far‐field sites (in Zones IV and V) and the
near‐field site (in Zone IV). It is clear from the comparison that the design response spectrum of the Indian code is on the conservative side, when
compared with the CMS for the far‐field sites, for most of the period range. On the other hand, for the near‐field site, the design response spec-
trum is nonconservative in the short‐period range. The nonconservativeness of the design seismic hazard specified in the Indian seismic design
code can be majorly attributed to the fact that it does not include near‐source effects. However, at the fundamental periods of the considered
buildings, the design response spectrum is conservative than the site‐specific response spectra, for all the sites in both the zones.

FIGURE 6 Comparison of the 5% damped elastic response spectra obtained from Indian seismic design code[10] with site‐specific conditional
mean spectra: (a) far‐field site in Seismic Zone IV, (b) far‐field site in Seismic Zone V, and (c) near‐field site in Seismic Zone IV. The three
vertical lines represent the fundamental building periods corresponding to two‐, four‐ and eight‐storey buildings, consecutively
10 of 13 SURANA ET AL.

Figures 7 and 8 present the collapse fragility curves for the investigated building models. The investigated building models, even of the same
building height with different structural configurations, have slightly different periods of vibration, resulting in different spectral ordinates (IM) for
a given level of hazard. Therefore, the collapse fragility curves have been plotted with the IM normalized using the IM at MCE. This normalization
eliminates the effect of different periods of vibration of buildings in collapse fragility curves, and therefore, fragility curves for different buildings
can be compared directly. Figure 7 shows the comparison of the collapse fragility curves of the buildings in Zone IV for near‐ and far‐field sites.
The buildings have almost identical capacities for the near‐ and far‐field ground‐motion record suites (Table 2), but the seismic fragility of the
buildings located at the near‐field sites is much higher (Figure 7), as the hazard increases significantly, when the near‐field effects are considered.
Figure 8 shows that the buildings in the higher seismic zone (V) are more vulnerable as compared with their counterpart buildings designed for
Zone IV, despite being designed using the same code. The reason for this observation lies in the fact that collapse capacity of a special
moment‐resisting frame building is majorly contributed by its ductility capacity. The buildings designed for Zone V have higher strength, but
the ductility capacity of buildings designed for both the seismic zones is almost identical, as the reinforcement detailing and capacity design
requirements in both the zones are the same. Further, the effects of near‐field site conditions and seismic zone are more pronounced in case
of high‐rise buildings, as compared with the low‐ and mid‐rise buildings.
Table 4 presents the collapse probabilities of the investigated building models at near‐ and far‐field sites, computed for two different response
spectra, namely, site‐specific CMS and site‐specific CMS scaled to the same Sa (T, 5%) as the MCE response spectrum in IS 1893,[10] and originally
used in the design of buildings. Choosing these two response spectra in comparison of collapse fragility enables to consider the effect of spectral
shape and the hazard level, separately. In case of site‐specific CMS, for all the investigated hillside building models in Seismic Zones IV and V, the
collapse probability conditioned on the occurrence of MCE is found out to be well within 10% for the near‐ as well as far‐field sites (except for the
case of eight‐storey buildings with SF and SB configurations, in near‐field site in Zone IV and far‐field site in Zone V). It is interesting to note that
in the case when site‐specific CMS is scaled to the design MCE, the collapse probabilities are almost identical for the near‐field and far‐field sites,
in Seismic Zone IV. The slight difference observed in the collapse probability values can be attributed to the slight differences in the median
collapse capacities and record‐to‐record variabilities obtained using near‐ and far‐field ground‐motion record suites. Further, these collapse
probabilities are observed to be unacceptably high (more than the limit of 10% specified in FEMA P695) in case of mid‐ and high‐rise buildings,
at both near‐ and far‐field sites. In case of buildings in Seismic Zone V, these collapse probabilities increase further and are observed to be

FIGURE 7 Effect of site (Seismic Zone IV) on collapse fragility of considered flat land (FL), split‐foundation (SF), and step‐back (SB) structural
configurations: (a) two‐storey buildings, (b) four‐storey buildings, and (c) eight‐storey buildings. To plot the fragility curves using the normalized
intensity measure, the Sa,avg (maximum considered earthquake [MCE]) has been obtained from site‐specific seismic hazards analyses. FF = far‐field;
NF = near‐field

FIGURE 8 Effect of seismic zone on collapse fragility of considered flat land (FL), split‐foundation (SF), and step‐back (SB) structural
configurations: (a) two‐storey buildings, (b) four‐storey buildings, and (c) eight‐storey buildings. To plot fragility curves using normalized
intensity measure, the Sa,avg (maximum considered earthquake [MCE]) has been obtained from site‐specific seismic hazards analyses
SURANA ET AL. 11 of 13

TABLE 4 Collapse probabilities (at design MCE level hazard) for all the considered buildings for NF and FF sites

Collapse probability
Site‐specific hazard Design code hazard
Building
Site height FL (%) SF (%) SB (%) Average (%) FL (%) SF (%) SB (%) Average (%)
NF Zone IV 2‐storey 01.95 07.24 02.32 03.84 01.95 07.24 02.32 03.84
4‐storey 07.77 10.66 03.26 07.23 12.04 15.74 05.62 11.13
8‐storey 18.51 36.76 21.56 25.61 42.67 60.88 48.00 50.52
FF Zone IV 2‐storey 00.13 00.66 00.22 00.34 02.64 08.19 03.87 04.90
4‐storey 00.31 00.84 00.10 00.42 11.63 16.92 06.25 11.60
8‐storey 00.91 04.43 01.20 02.18 41.90 56.42 47.64 48.65
FF Zone V 2‐storey 00.65 02.30 01.06 01.34 14.53 27.57 17.60 19.90
4‐storey 01.77 03.66 01.96 02.46 27.68 36.93 27.23 30.61
8‐storey 18.63 45.09 25.79 29.84 68.84 89.19 76.12 78.05

Note. The values in bold are higher than the FEMA P695 acceptance criterion. NF = near‐field; FF = far‐field; FL = flat land; SF = split‐foundation; SB = step‐
back.

significantly higher even in case of low‐rise buildings. This indicates that the considered hillside buildings have unacceptably high collapse
probability at the same level of hazard for which these were designed.
It can also be observed from the results that, in general, the SF buildings are the most vulnerable, whereas the FL buildings are observed to be
the least vulnerable. The higher vulnerability of SF buildings can be attributed to increased torsional effects in SF buildings in inelastic range (as
also discussed in Section 4). The presented results clearly suggest the need for special design provisions for SF and SB buildings, particularly in
case of high‐rise buildings and higher seismic zone (V). Further, the seismic fragility in case of the near‐field sites is higher due to the site‐specific
seismic hazard being higher than the design hazard specified in the code (Figure 6). The current Indian seismic design code does not distinguish
between near‐ and far‐field sites and provides identical design spectra for seismic design, irrespective of site location relative to active seismic
sources. The presented estimates of collapse fragility suggest a clear need to consider the effects of structural configuration and near‐field site
conditions in seismic design of hillside buildings.

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

Seismic capacity and fragility analyses have been performed for code‐conforming reinforced‐concrete special moment‐resisting frame hillside
buildings designed for the Indian Seismic Zones IV and V, covering the Indian Himalayan regions. The often observed FL, SF, and SB structural
configuration have been studied. The effects of building height, seismic zone, and near‐ and far‐field site conditions on collapse fragility have also
been studied. In general, the SF buildings have the lowest median collapse capacity, whereas the FL buildings have the highest collapse capacity,
even though all buildings (FL, SF and SB) with the same height above the uppermost foundation level were designed for identical base shear coef-
ficients. Further, the comparison of the seismic capacity of buildings designed for Zones IV and V indicates that the increase in the median collapse
capacity of buildings is much lower than the corresponding increase in the design base shear.
The use of Sa,avg (0.2T–3T, 5%) as the IM resulted in collapse capacity of a building almost independent of the ground‐motion record suite with
a significantly reduced record‐to‐record variability. This has been observed not only in case of FL buildings but also in case of torsionally irregular
SF and SB buildings and subjected to bidirectional earthquake excitations.
The hinge patterns at collapse indicate that the storey just above the uppermost foundation level is the most vulnerable location in the SF and
SB hillside buildings, which is in good agreement with what has already been observed after the 2011 Sikkim earthquake, in case of an SF building.
A comparison of the collapse probabilities of the buildings with different structural configuration indicates that the current code provisions are
adequate only for low‐ and mid‐rise buildings in Seismic Zone IV (having moderate seismicity), but these are inadequate for high‐rise buildings in
Seismic Zone IV and for all the buildings in Seismic Zone V. The study provides important insight into the seismic behaviour of the irregular hillside
buildings and highlights the need for further development of design codes considering the effect of structural configuration, building height, level
of seismicity, and near‐field effects.
As identified in an earlier study,[1] the building configurations other than those investigated in this study (e.g., step‐ and set‐back buildings;
buildings on stilts) could also be observed in the hilly regions. The adequacy of seismic design code provisions needs to be assessed for these
structural configurations as well. This study has been conducted on RC frame buildings without infill walls. The presence of infill walls can signif-
icantly alter the collapse fragility of hillside buildings. Further, the slope stability and topographic amplification effects are also important, which
should be considered in the future research.

ACKNOWLEDGEMEN TS
The first author's fellowships granted by the Ministry of Human Resource and Development, Government of India, and the Royal Norwegian
Embassy to India (New Delhi) allowed the conduct of this study. Parts of the investigations were carried out under the Indo‐Norwegian
12 of 13 SURANA ET AL.

collaboration project EQRisk. The support received from both agencies is gratefully acknowledged. The ground‐motion records used in this study
were obtained from the PEER NGA WEST 2 database. The authors would also like to express their thanks to the two anonymous reviewers for
their valuable comments, resulting in the significant enhancement in the manuscript.

ORCID
Yogendra Singh http://orcid.org/0000-0001-6722-8956

RE FE R ENC E S
[1] M. Surana, Y. Singh, D. H. Lang, Nat Haz Rev 2017, 19. https://doi.org/10.1061/(ASCE)NH.1527‐6996.0000275 04017024
[2] Surana M, Singh Y, Lang DH.. Proc. of SECED—An International Conference on Earthquake Risk and Engineering Towards Resilient World, July 2015,
Homerton College, Cambridge, United Kingdom.
[3] Surana M, Singh Y, Lang DH. Proc. of Structural Engineering Convention—2016, December 2016, Advances in Structural Engineering, Indian Institute of
Technology, Madras.
[4] Singh Y, Gade P, Lang DH, Erduran E. Proc. of the 15th World Conference on Earthquake Engineering, 2012, Lisbon, Portugal.
[5] Welsh‐Huggins SJ, Rodgers JE, Holmes WT, Liel AB.. Proc. of the 16th World Conference on Earthquake Engineering, 2017, Santiago, Chile.
[6] Murthy C V R, Menon A, Goswami R, Vijyanarayanan A R, Gandhi SR, Kalidindi SN, Raghukanth ST G, Jaiswal A, Seth A. Proc. of the 15th World
Conference on Earthquake Engineering, Lisbon, Portugal 2012.
[7] Sharma ML, Sinhval A, Singh Y, Maheshwari BK. Damage Survey Report, 2012, Department of Earthquake Engineering, Indian Institute of Technology
Roorkee, Roorkee, India.
[8] Vijyanarayanan AR, Goswami R, Murthy C V R. Proc. of the 15th World Conference on Earthquake Engineering, Lisbon, Portugal 2012.
[9] EERI (Special Earthquake Report). EERI Report, Earthquake Engineering Research Report, 2012, Oakland, United States.
[10] Bureau of Indian Standards (BIS). Part 1: General Provisions and Buildings (Sixth Revision). Bureau of Indian Standards, New Delhi, India, 2016.
[11] Bureau of Indian Standards (BIS). Bureau of Indian Standards, New Delhi, India, 2016.
[12] Haselton CB, Deierlein GG (2007). PEER Report 2007/08, PEER Center, 2007, University of California, Berkeley, CA.
[13] C. B. Haselton, A. B. Liel, G. G. Deierlein, B. S. Dean, J. H. Chou, J Struct Eng 2011, 137(4), 481.
[14] Ibarra LF, Krawinkler H. Global Collapse of Frame Structures Under Seismic Excitations.John A. Blume Earthquake Engineering Center, PEER Report
2005/06, PEER Center, 2005, Department of Civil and Environmental Engineering, Stanford University, United States.
[15] A. B. Liel, C. B. Haselton, G. G. Deierlein, J Struct Eng 2011, 137(4), 492.
[16] S. Manie, A. S. Moghadam, M. Ghafory‐Ashtiany, The Structural Design of Tall and Special Buildings 2015, 24(8), 607.
[17] A. J. Kappos, G. Panagopoulos, C. Panagiotopoulos, G. Penelis, Bull Earthq Eng 2006, 4(4), 391.
[18] D. Vamvatsikos, C. A. Cornell, Earthq Eng and Struct Dyn 2002, 31, 491.
[19] M. E. Koopaee, R. P. Dhakal, G. MacRae, Earthq Eng and Struct Dyn 2017, 46, 1875. https://doi.org/10.1002/eqe.2891
[20] D. Feng, C. Kolay, J. M. Ricles, J. Li, The Structural Design of Tall and Special Buildings 2016, 25(12), 578.
[21] I. Kazaz, P. Gulkan, A. Yakut, Eng Struct 2012, 43, 105.
[22] Haselton CB, Liel AB, Lange ST, Deierlein GG.. PEER Report 2007/03, PEER Center, 2007, University of California, Berkeley, CA.
[23] C. B. Haselton, A. B. Liel, S. T. Lange, G. G. Deierlein, ACI Struc J 2016, 113(6), 1141.
[24] American Society of Civil Engineers (ASCE). ASCE Standard, ASCE/SEI 41‐13, Reston, Virginia, United States, 2013.
[25] K. J. Elwood, A. B. Matamoros, J. W. Wallace, D. E. Lehman, J. A. Heintz, A. D. Mitchell, M. A. Moore, M. T. Valley, L. N. Lowes, C. D. Comartin, J. P.
Moehole, Earthq Spec 2007, 23(3), 493.
[26] Federal Emergency Management Agency (FEMA). FEMA 273, Washington, D.C., United States, 1997.
[27] Federal Emergency Management Agency (FEMA). FEMA 356, Washington, D.C., United States, 2000.
[28] American Society of Civil Engineers (ASCE). ASCE Standard, ASCE/SEI 41‐06, Reston, Virginia, United States, 2007.
[29] Aslani H. Ph.D. Dissertation, Stanford University, Stanford, California 2005.
[30] Y. Li, S. Hwang, J Struct Eng 2016, 143. https://doi.org/10.1061/(ASCE)ST. 1943‐541X.0001656
[31] C. B. Haselton, J. W. Baker, A. B. Liel, J Struct Eng 2011, 137(3), 332.
[32] Federal Emergency Management Agency (FEMA). FEMA P695, Washington, D.C., United States, 2009.
[33] J. W. Baker, C. A. Cornell, Earthq Eng and Struct Dyn 2006, 35(9), 1077.
[34] L. Eads, E. Miranda, D. G. Lignos, Earthq Eng and Struct Dyn 2015, 44, 2057.
[35] L. Eads, E. Miranda, D. G. Lignos, Earthq Eng and Struct Dyn 2016, 45, 1643.
[36] J. W. Baker, J Struct Eng 2011, 137(3), 322.
[37] T. Lin, C. B. Haselton, J. W. Baker, Earthq Eng and Struct Dyn 2013, 42, 1847.
[38] V. N. Khose, Y. Singh, D. H. Lang, Earthq Spect 2012, 28(3), 1047.
[39] CSI. ETABS 2016 Integrated Building Design Software, Version 16.0.2, Computers and Structures, Inc., Berkeley, U.S.A, 2016.
[40] Bureau of Indian Standards (BIS). Indian Standard—Code of Practice for Design loads (other than earthquake) for Buildings and Structures (Dead
Loads), IS 875‐Part 1, New Delhi, India, 1987.
SURANA ET AL. 13 of 13

[41] Bureau of Indian Standards (BIS). Indian Standard—Code of Practice for Design loads (other than earthquake) for Buildings and Structures (Live Loads),
IS 875‐Part 2, New Delhi, India, 1987.
[42] Federal Emergency Management Agency (FEMA). Seismic Performance Assessment of Buildings‐Volume 1 Methodology. FEMA P58, Washington,
D.C., United States, 2012.
[43] D. Basu, S. K. Jain, Earthq Eng and Struct Dyn 2007, 36(7), 965.
[44] C. Champion, A. B. Liel, Earthq Eng and Struct Dyn 2012, 41, 1391.
[45] A. B. Liel, C. B. Haselton, G. G. Deierlein, J. W. Baker, Struct Saf 2009, 31(2), 197.
[46] K. W. Campbell, Y. Bozorgnia, Earthq Spect 2008, 24(1), 139.
[47] K. W. Campbell, Y. Bozorgnia, Earthq Spect 2014, 30(3), 1087.
[48] J. W. Baker, N. Jayaram, Earthq Spect 2008, 24(1), 299.
[49] M. Kohrangi, S. R. Kotha, P. Bazzurro, Bull Earthq Eng 2017, 16, 45. https://doi.org/10.1007/s10518‐017‐0216‐5

Mitesh Surana is currently working as Project Fellow at the Department of Earthquake Engineering, Indian Institute of Technology Roorkee,
Roorkee, India. He has completed his M.Tech. and PhD in 2013 and 2018, respectively. His PhD research was focused on seismic vulnerability
of hill buildings. He has published more than 10 articles on seismic fragility analysis of buildings and seismic design of nonstructural
components.

Yogendra Singh is currently working as Professor and Head of Department of Earthquake Engineering, Indian Institute of Technology
Roorkee, Roorkee, India. He has published more than 100 articles on topics such as seismic vulnerability and risk assessment, performance‐
based seismic design of buildings and bridges, and slope–building interaction.

Dominik H. Lang is currently working as Head of the Department of Earthquake Hazard and Risk, NORSAR, Kjeller, Norway. He has
developed the earthquake loss estimation tool SELENA and published more than 100 articles on seismic vulnerability and risk assessment
for different cities in various parts of the world.

How to cite this article: Surana M, Singh Y, Lang DH. Fragility analysis of hillside buildings designed for modern seismic design codes.
Struct Design Tall Spec Build. 2018;e1500. https://doi.org/10.1002/tal.1500

You might also like