Professional Documents
Culture Documents
4.1 INTRODUCTION
accurately estimated using the empirical formula given in design codes for regular
buildings. This chapter presents a modification to the code equation to estimate the
frame.
to stepped building is presented here, and it is shown there methods fail to predict the
base shear, roof displacement capacity and target displacement for stepped building
stepped buildings by proposing a new load pattern and an empirical equation for target
displacement.
A fixed lateral load pattern is developed using the properties of the first three elastic
modes. The resulting pushover curve and the hinge distribution in the stepped buildings
are found to be in close agreement with the nonlinear time history analysis results.
85
4.2 ESTIMATION OF FUNDAMENTAL TIME PERIOD FOR STEPPED
BUILDING
As explained in the Section 2.2, most of the design codes (IS 1893:2002, ASCE 7:2005,
buildings, with the base shear scaled up to the value corresponding to the fundamental
period as per the code specified empirical formulas. These formulas, however, have been
developed for regular buildings and are a function of building height only. For example,
T = 0.075h0.75 (4.1)
For a stepped building, the height is not constant. It varies from one side to the other side
of the frame. The building height at the lower side is likely to under-estimate the actual
time period, and consequently over-estimates the base shear. Conversely, if we consider
the overall building height, we will get an unconservative estimate of the base shear.
Table 3.2 (Chapter 3) shows that generally the fundamental period decreases with the
increased irregularity when the overall building height is the same. Most of the stepped
buildings are high-rise multi-storeyed and their period fall in the constant velocity region
of the response spectrum, where the spectral acceleration value is very sensitive to the
fundamental period (Fig. 4.1). A slight reduction in the fundamental period in this region
results in a considerable increase in the design base shear. In this study, an attempt has
been made to improve the code-based empirical equation for estimating the fundamental
86
3.0
Spectral Acceleration/g...
Stepped Frame
2.0
Regular Frame
1.0
0.0
0 1 2 3
Period (s)
Fig. 4.1: Shift in time period and spectral acceleration due to stepped irregularity (design
spectrum, IS 1893)
In a stepped building, both mass and stiffness decrease with decreasing regularity index
(η). Reduction of mass reduces the building fundamental period, whereas reduction of
stiffness increases the fundamental period. So, the fundamental period does not have a
monotonic relation with the regularity index. Ratio of the fundamental period of stepped
frame (T) to that of similar regular frame without step (Tref) obtained from the modal
analysis of 78 stepped frames with varying regularity indices, height and number of bays
are plotted and shown in Fig. 4.2. It is seen that the building fundamental period depends
on the building height and its regularity index. Number of bays has only minor effect on
Based on a polynomial fit for these data, the following correction to the code empirical
87
T = 0.075h 0.75 × κ (4.2)
T
κ= = [1 − 2(1 − η )(2η − 1)] for 0.6 ≤ η ≤ 1.0 (4.3)
Tref
1.1
0.9
κ = T/Tref
0.8
0.7
0.6 T/Tref
Curve fit (T/Tref)
0.5
0.6 0.7 0.8 0.9 1
Regularity Index (η)
Fig. 4.2 shows that the correction factor (κ) initially decreases with decreasing regularity
index (i.e., increasing irregularity) and it shows minimum value of correction factor when
the regularity index (η) reaches 0.75 and then it increases with decreasing regularity
index. It is to be noted that the correction factor in Eq. 4.3 is based on the stepped
88
building frames with regularity index ranging from 0.6 to 1.0, which generally covers the
1.00
0.90
0.80
κ = T/Tref
0.70
η = 0.98
η = 0.89
0.60
η = 0.76
0.50
3 4 5 6
Number of bays
The correction factor calculated using Eq. 4.3 matches closely with the exact value
obtained from the modal analysis results. Fig. 4.4 presents the correlation between the
predicted correction factor using Eq. 4.3 and the actual time period ratio obtained from
the modal analysis. The average ratio of predicted to actual correction factor for 78
samples is 0.993 and the correlation coefficient between the actual and predicted
89
1.0
0.9
Predicted (T/Tref )
0.8
0.7
0.7 0.8 0.9 1.0
Actual (T/T ref )
Fig. 4.4: Correlation between ‘predicted’ and ‘exact’ correction factor (κ)
To accurately evaluate the inelastic response of the structure, the prescribed earthquake
load profiles must be able to describe the actual dynamic force profiles which produce
maximum design values (peak values of inter-storey drift, story shear, and member
because various load profiles can develop during the nonlinear time history response of a
Since it is difficult to predict inelastic load profiles, most studies have focussed on the
90
combination methods, such as the square root of square sum (SRSS) and complete
quadratic combination (CQC) are used to address the effect of higher dynamic modes in
the response spectrum analysis of a structure. These modal combination methods can
predict the absolute maximum value of each design parameter by combining the
In the present study, potential earthquake load profiles for the nonlinear pushover
analysis of stepped buildings were developed, based on SRSS combination. The basic
concept is taken from the upper-bound load profile proposed by Jan et. al. (2004). A
study was performed to investigate the effect of higher dynamic modes in elastic time
where {u} is the floor displacement vector relative to the ground, [m] , [c ] , and [k ] are
the mass, classical damping, and lateral stiffness matrices of the system.
The solution of the above differential equation governing the response of a MDOF
system to an earthquake ground motion can be expressed, using the mode superposition
method, as:
N
{u (t )} = ∑ {φ n }qn (t ) (4.5)
n =1
{φ } [ m]{1} D (t )
T
qn (t ) = Γ n Dn (t ) = n T (4.6)
{φn } [ m]{φn }
n
91
where q n (t ) is the modal coordinate, Γn is the modal participation factor of the nth mode
and Dn (t ) is governed by the following equation of motion for a SDOF system, with nth
+ ( 4π T ) ξ D + ( 2π T )2 D = −u ( t )
D (4.7)
n n n n n n g
N N
{ F( t )} = [ k ]{u( t )} = ∑ [ k ]{φn } qn ( t ) = ∑ ( 2π T ) [ m]{φn } qn ( t )
2
n (4.8)
n =1 n =1
Since the periods in the building structure are widely spaced, the SRSS modal
combination rule is used to determine peak response and the equivalent static force at jth
2 2
F j = ⎡( 2π T1 ) m1, j φ1, j q1 ⎤ + ⎡( 2π T2 ) m2, j φ2, j q2 ⎤ + ...
2 2
⎣ ⎦ ⎣ ⎦ (4.9)
Since { F } is a spatial vector and increases monotonically from zero, Eq. 4.9 can simply
be expressed as:
2 2 2
⎛m φ ⎞ ⎛m φ q ⎞ ⎛m φ q ⎞
F j = ⎜ 1, j 2 1, j ⎟ + ⎜ 2, j 2 2, j 2 ⎟ + ⎜ 2, j 2 2, j 3 ⎟ + ... (4.10)
⎝ T1 ⎠ ⎝ T2 q1 ⎠ ⎝ T3 q1 ⎠
92
FN
.
.
.
.
.
.
.
.
.
Fj
.
.
.
F2
F1
The load distribution presented in Eq. 4.10 is the combination of responses from all the
associated modes. A study has been carried out to check how many modes have
contribution, a set of 20 earthquake ground motions were chosen and applied to the 23
designed stepped frames. The details of the selected ground motions design frames have
As shown in Eq. 4.10, the contribution of a higher mode in comparison with the
fundamental mode can be expressed as a ratio of qi q1 . Figs 4.6 to 4.8 show the ratios
obtained for 2nd, 3rd and 4th modes as a function of the fundamental time period.
93
40
R
S1
30
S2
S3
q2 /q1 (%)
20
10
0
0.5 1.0 1.5 2.0
Fundamental Period, T 1 (s)
40
R
30
S1
S2
q3 /q1 (%)
20 S3
10
0
0.5 1.0 1.5 2.0
Fundamental Period, T 1 (s)
94
40
R
30
S1
S2
q4 /q1 (%)
20 S3
10
0
0.5 1.0 1.5 2.0
Fundamental Period, T 1 (s)
From these figures, it can be seen that in addition to the first mode the 2nd mode
contribution is significant and to some extent the 3rd mode also contributes the structural
response. The 2nd mode contribution is 10% to 40% of the fundamental mode while the
3rd mode contribution is 1% to 10% of the fundamental mode. The contribution from 4th
and higher modes to the structural response is very little and can be ignored. It is also
clearly seen that higher mode contribution increase in significance when the stepped
irregularity increases (maximum in the case of S3 and least in the case of R). The plot of
cumulative mass participation for regular (R) frames and stepped (S3) frames are shown
in Fig. 4.9. Modal analysis results for 23 designed stepped frames with varying
irregularity and height show that 90% of the total mass participates in first three modes
and considering these three modes alone can be sufficient as given in Indian Standard
95
Cum. Mass Participation (%)...
100
90
80
70
0 1 2 3 4
No. of modes
100
90
80
70
0 1 2 3 4
No. of modes
A plot of the lateral load distribution for a 15 storey stepped frame (S4-15) as per
Eq. 4.10 considering only first mode along with combinations of first two, three and four
modes is presented in Fig. 4.10. This figure shows that after third mode, the load shape
96
is almost stationary and there is no significant effect on the load profile from fourth and
higher modes.
15
12
9
Storey
6
First four modes
From the study presented in the previous section (Section 4.3.1) it is clear that the
contribution of the first three modes need to be accounted for in the elastic response of
the stepped buildings. Hence, these three modes are considered to calculate the proposed
lateral load distribution for pushover analysis of stepped building. Considering the first
2 2 2
⎛m φ ⎞ ⎛m φ q ⎞ ⎛m φ q ⎞
F j = ⎜ 1, j 2 1, j ⎟ + ⎜ 2, j 2 2, j 2 ⎟ + ⎜ 2, j 2 2, j 3 ⎟ (4.11)
⎝ T1 ⎠ ⎝ T2 q1 ⎠ ⎝ T3 q1 ⎠
97
Linear time history analyses were carried out for all the 23 designed stepped frames
subjected to 20 earthquake ground motions and the distribution of mean of the storey
shear was calculated. Figs 4.11 to 4.13 present the comparison of the proposed load
profile with the mean story shear profile obtained from linear time history analysis for
three typical stepped frames. The figures show that the proposed lateral load profile
15
12
Storey level
Storey Shear
6 (elastic THA)
Proposed
3 Profile
0
0 0.025 0.05 0.075 0.1
Fraction of base shear
Fig. 4.11: Comparison of the proposed load profile with the mean storey shear profile for
Frame S1-15
98
15
12
Storey level
9
Storey Shear
6 (elastic THA)
Proposed
Profile
3
0
0 0.025 0.05 0.075 0.1
Fraction of base shear
Fig. 4.12: Comparison of the proposed load profile with the mean storey shear profile for
Frame S2-15
15
12
Storey level
9
Storey Shear
6 (elastic THA)
Proposed
Profile
3
0
0 0.025 0.05 0.075 0.1
Fraction of base shear
Fig. 4.13: Comparison of the proposed load profile with the mean storey shear profile for
Frame S3-15
99
Figs 4.14 to 4.16 present the comparison of the proposed lateral load profiles with some
of the existing load profiles available for pushover analysis. FEMA 356 recommends the
adoption of two load profiles: (i) uniform distribution or adaptive distribution and (ii)
distribution for equivalent static analysis. In the present study, the following load
analysis (parabolic)
15
12
9
Storey
6 UNI
TRI
Mode-1
3
IS-1893
Proposed
0
0 0.05 0.1 0.15 0.2
Load Ratio
Fig. 4.14: Comparison of the proposed lateral load profile with other existing load
profiles for pushover analysis of Frame S1-15
100
15
12
9
Storey
UNI
6
TRI
Mode-1
3 IS-1893
Proposed
0
0 0.05 0.1 0.15 0.2
Load Ratio
Fig. 4.15: Comparison of the proposed lateral load profile with other existing load
profiles for pushover analysis of Frame S2-15
15
12
9
Storey
UNI
6 TRI
Mode-1
3 IS-1893
Proposed
0
0 0.05 0.1 0.15 0.2
Load Ratio
Fig. 4.16: Comparison of the proposed lateral load profile with other existing load
profiles for pushover analysis of Frame S3-15
101
These three figures show that, compared to the existing lateral load profiles the proposed
profile, adds more loads in the lower storeys and reduces load in the upper stories. The
mass and stiffness of the stepped buildings get reduced at the upper floors compared to
the lower floors. The reduced mass and reduced stiffness are both responsible for
attracting less seismic forces at the upper floors compared to similar regular buildings
without steps. This is reflecting in proposed lateral load profile. Studies carried out on
other frames with different heights and bays shown similar trends
Estimating seismic demands at low performance levels, such as life safety and collapse
many investigations show that simplified equations based on pushover analysis have led
to good estimates of displacement demands for buildings at the roof level (target
restricted to low- and medium-rise regular buildings. A comparative evaluation of all the
methods available for estimating target displacement from pushover analysis shows that
all of these procedures (except capacity spectrum method) perform well for regular
building frames when compared with the results of nonlinear time history analysis.
However, except for UBPA, none of these procedures works for buildings with stepped
irregularity.
method of FEMA 356 for its application to the stepped buildings as this method is the
102
most popular in practice. Recalling Eq. 2.2 (Chapter 2) the expected maximum roof
Teq2
δ t = C 0 C1C 2 C 3 S d = C 0 C1C 2 C 3 Sa (4.12)
4π 2
system to the exact roof displacement of a building system (MDOF), C1 relates the
represents the effect of pinched hysteretic shape, stiffness degradation and strength
From the above definitions of the coefficients, it is clear that the change in building
geometry will affect C0 significantly whereas it is likely to have very little influence on
the other factors. As per FEMA 356, the values of C0 factor for shear buildings depend
only on the number of storeys and the lateral load pattern used in the pushover analysis.
Table 4.1 presents the values of C0 provided by the FEMA 356 for shear buildings.
Table 4.1: Values of C0 factor for shear building as per FEMA 356
103
Generally, stepped buildings have 5 or more storeys in practice and the C0 factor, as per
FEMA 356, is constant for buildings with 5 or more storeys (Table 4.1). Fig. 4.17 shows
1.4
1.2
C0
0.8
5 10 15 20
Number of Storeys
To assess the validity of the values of C0 factor given in FEMA 356, linear time history
analysis of 78 stepped frames has been carried out for 20 earthquake ground motions,
scaled for PGA = 0.36g. The mean value of the maximum roof displacement of each
frame and the mean value of spectral displacement of corresponding equivalent SDOF
system for all the 20 earthquakes are calculated. The equivalent period (Teq) can be
generated from the base shear versus roof displacement curve (pushover curve) obtained
using the proposed load profile (Eq. 4.11). The elastic spectral displacement
104
representing the seismic ground motion under consideration for a specified damping ratio
(5%). Details of the selected ground motion are presented in Sections 3.5.1 and 3.5.2 (in
Chapter 3). Figs 4.18 to 4.21 show the results obtained for different stepped frames: R
height categories in each of these four types of frames are shown here. Figs 4.18 to 4.21
also show two lines representing the C0 values given by FEMA 356 for triangular load
1.00
15-Storey
Spectral displ. (m) eq. SDOF model...
12-Storey
10-Storey
0.75
8-Storey
6-Storey
C0 = 1.2
0.50
C0 = 1.3
0.25
0.00
0.00 0.25 0.50 0.75 1.00
Roof displ. (m) exact MDOF model
105
1.00
15-Storey
0.25
0.00
0.00 0.25 0.50 0.75 1.00
Roof displ. (m) exact MDOF model
Fig. 4.19: Correlation of time history response for S1 frames (0.80 < η < 0.93)
1.00
15-Storey
Spectral displ. (m) eq. SDOF model...
12-Storey
10-Storey
0.75
8-Storey
6-Storey
C0 = 1.2
0.50
C0 = 1.3
0.25
0.00
0.00 0.25 0.50 0.75 1.00
Roof displ. (m) exact MDOF model
Fig. 4.20: Correlation of time history response for S2 frames (0.66 < η < 0.85)
106
1.00
0.25
0.00
0.00 0.25 0.50 0.75 1.00
Roof displ. (m) exact MDOF model
Fig. 4.21: Correlation of time history response for S3 frames (0.61 < η < 0.76)
It is clear from these figures that, for most of the cases, C0 does not match with the
FEMA prescribed values. For regular frame (R), the deviation is less but as the
irregularity increases the deviation tends to increase. Also, for the lower storey frames the
deviation from the FEMA values is less compared to the higher storey frames. This
indicates that the ratio of elastic roof displacement for an exact MDOF frame to the
elastic spectral displacement for equivalent SDOF system increases with the increase in
the number of storeys (building height) and with decrease in regularity index (i.e.,
increase of irregularity).
These results clearly show that FEMA specified values for C0 may work for low-rise
regular buildings but for high-rise regular buildings and all stepped buildings the FEMA
107
A study was taken up to check the effect of number of bays on the C0 value. Fig. 4.22
presents the variation of C0 value with respect to number of bays when height and
regularity index is constant. This figure shows that C0 factor has hardly any dependence
2.2
2.0
C0 factor..
1.8
It can be observed from Figs 4.18 to 4.21 that C0 factor is actually a function of regularity
index (η) as well as building height (h) and the FEMA approach for constant C0 values
for all the high-rise building (with 5 or more storeys) is not correct. Similar conclusions
can be drawn from Figs 4.23 and 4.24 that present the variation of C0 value with respect
108
2.2
1.9
C0 Factor
1.6
1.3
R-Type S1-Type
S2-Type S3-Type
1.0
15 30 45 60
Building Height (m)
2.2
2.0
1.8
C0 Factor
1.6
6-Storey 8-storey
1.4 10-storey 12-storey
15-storey 18-storey
1.2
0.6 0.7 0.8 0.9 1.0
Regularity Index
109
4.4.1 Proposed C0 factor for Stepped Building
It is seen from the previous discussion that the C0 factor for stepped building is a function
of building height and irregularity index. Based on the time history analysis results for 78
stepped frames, a nonlinear regression analysis has been carried out and the following
empirical equation has been arrived at for calculating the value of C0 factor.
⎛ h ⎞
C0 = 1.5 + 0.5η ( 1 − η ) ⎜ − 0.4 ⎟ (4.13)
⎝ 10 ⎠
Fig. 4.25 presents the correlation between the C0 factor calculated using Eq. 4.13 and the
actual C0 factor obtained from the time history analysis. The average ratio of predicted to
actual C0 factor for 78 samples is 1.001 with a correlation coefficient between actual and
2.2
Predicted C 0 from proposed equation..
2.0
1.8
1.6
1.4
1.4 1.6 1.8 2.0 2.2
Actual value of C 0 factor
110
Figs 4.26 and 4.27 present three dimensional plots of C0 factor as a function of as a
regularity index and building height as obtained from time history analysis and the
C0
Re g
u la r i
t y In ht ( m)
de x He ig
Fig. 4.26: Variation of actual C0 factor with of building height and regularity index
C0
Re
g ula
r it
yI
nd t ( m)
e x He igh
Fig. 4.27: Variation of predicted C0 factor with of building height and regularity index
111
4.5 PROPOSED PUSHOVER ANALYSIS PROCEDURE
Summarised below are a series of steps to be followed for the proposed pushover analysis
i. Perform an eigenvalue analysis and find out the natural periods and mode shapes
of the structure.
ii. Use the elastic response spectrum of the selected earthquake to determine the 2nd
qn Γ n Dn
=
q1 Γ1 D1
iii. Determine the later load distribution (height-wise) for pushover analysis using
2 2 2
⎛m φ ⎞ ⎛m φ q ⎞ ⎛m φ q ⎞
F j = ⎜ 1, j 2 1, j ⎟ + ⎜ 2, j 2 2, j 2 ⎟ + ⎜ 2, j 2 2, j 3 ⎟
⎝ T1 ⎠ ⎝ T2 q1 ⎠ ⎝ T3 q1 ⎠
iv. Determine the target roof displacement as given by the following relationship:
Teq2
δ t = C 0 C1C 2 C 3 S d = C 0 C1C 2 C 3 Sa
4π 2
112
v. The seismic demands of a given structure are determined by pushover analysis
with a lateral load profile {Fj } , and the forces are monotonically increased until
All the 23 designed frames with varying irregularity and height were analysed using the
literature (refer Sections 2.5 and 2.7, Chapter 2). The pushover analysis results are then
compared with the nonlinear time history (NLTHA) analysis results for 20 selected
earthquake ground motions. The details of the time history analysis parameters, including
the input ground accelerations, are explained in Section 3.5 (Chapter 3). All of these
records are scaled for various PGA levels ranging from 0.18g to 0.72g. The main purpose
of increasing the PGA is to force the structural collapse and to get an envelope for the
Figs 4.28 to 4.38 present the comparison of typical pushover curves using the proposed
lateral load profile and some of the existing load profiles available for pushover analysis,
i.e., uniform (UNI), fundamental mode shape (Mode 1) and load profile for upper bound
pushover analysis (UBPA). Figs 4.28 – 4.31 pertain to 15-storey frames of types R, S1,
S2 and S3. Similarly, Figs 4.32 – 4.35 pertain to 10-storey frames (R, S1, S2 and S3),
while Figs 4.36 – 4.38 correspond to 6-storey frames (R, S1 and S2).
113
750
600
Base Shear (kN)
450
300
UNI Mode 1
150
UBPA Proposed
NLTHA TRI
0
0 0.3 0.6 0.9 1.2 1.5
Roof Displacement (m)
750
600
Base Shear (kN)
450
300
UNI Mode 1
150 UBPA Proposed
NLTHA TRI
0
0 0.3 0.6 0.9 1.2 1.5
Roof Displacement (m)
Fig. 4.29: Pushover curve for 15 storey stepped frame (S1-15, η = 0.96)
114
750
600
Base Shear (kN)
450
300
UNI Mode 1
150 UBPA Proposed
NLTHA Series6
0
0 0.3 0.6 0.9 1.2 1.5
Roof Displacement (m)
Fig. 4.30: Pushover curve for 15 storey stepped frame (S2-15, η = 0.89)
750
600
Base Shear (kN)
450
300
Fig. 4.31: Pushover curve for 15 storey stepped frame (S3-15, η = 0.78)
115
600
450
UNI Mode 1
150
UBPA Proposed
NLTHA TRI
0
0.0 0.2 0.4 0.6 0.8 1.0
Roof Displacement (m)
600
450
Base Shear (kN)
300
Fig. 4.33: Pushover curve for 10 storey stepped frame (S1-10, η = 0.91)
116
600
450
Base Shear (kN)
300
Fig. 4.34: Pushover curve for 10 storey stepped frame (S2-10, η = 0.77)
600
450
Base Shear (kN)
300
150
UNI Mode 1
UBPA Proposed
NLTHA TRI
0
0 0.2 0.4 0.6 0.8
Roof Displacement (m)
Fig. 4.35: Pushover curve for 10 storey stepped frame (S3-10, η = 0.69)
117
400
300
Base Shear (kN)
200
0
0 0.1 0.2 0.3 0.4 0.5
Roof Displacement (m)
400
300
Base Shear (kN)
200
0
0 0.1 0.2 0.3 0.4 0.5
Roof Displacement (m)
Fig. 4.37: Pushover curve for 6 storey stepped frame (S1-6, η = 0.84)
118
400
300
Base Shear (kN)
200
100
UNI Mode 1
UBPA Proposed
NLTHA TRI
0
0 0.1 0.2 0.3 0.4 0.5
Roof Displacement (m)
Fig. 4.38: Pushover curve for 6 storey stepped frame (S2-6, η = 0.69)
Pushover curves for triangular load profile are found to be similar to that corresponding
to the Mode 1 load pattern. The figures also present the base shear versus roof
As discussed in Section 2.5.2 (Chapter 2), FEMA 356 recommends the use of two lateral
shape or storey shear distribution from response spectrum analysis) with an intention to
bound the range that may occur during actual dynamic response. The above figures
(Figs 4.28 - 4.30, 4.32 - 4.33 and 4.36) show that for regular building (η = 1.0) and less
irregular stepped buildings (η ≥ 0.9) the two load patterns from FEMA 356 (uniform
distribution and fundamental mode shape) indeed represent the upper bound and the
119
lower bound of the nonlinear time history analysis results. But this assumption does not
hold good for stepped frames with less regularity index (i.e., highly irregular frames).
The results shown in Figs 4.31, 4.34, 4.35, 4.37 and 4.38 reveal that the two load patterns
recommended by FEMA 356 cannot bound the solution for the stepped building with less
Time history envelops show greater strength and stiffness than that predicted by the
pushover analysis using uniform load pattern for stepped building with less regularity
index. This is due to the fact that the earthquake forces at the upper storeys are less
compared to the lower storeys due to the relatively lesser mass and lesser stiffness at that
level. The proposed load distribution, which has more load intensity at the lower part of
the frame compared to the upper part, reflect the time history analysis results most
closely. It can be seen through Figs 4.28 to 4.38 that pushover analysis with the proposed
lateral load profile can predict the results that match best with the NLTHA results.
between the pushover analysis and nonlinear time history analysis in terms of base shear
versus roof displacement relation. It is numerically simple and very efficient to define the
difference between the ordinates of a pushover curve and the base shear versus roof
displacement envelop obtained from the nonlinear time history analysis for the same
structural model and based on a similar concept due to standard error of displacement
120
Consider the pushover curve S0 – S4 and the set of envelop points obtained from
nonlinear time history analysis D1 – D5 in Fig. 4.39. The coordinates of the vertical
projection of each time history analysis point on the pushover curve are calculated by
80
D4
60
S3 d4
Y Values
D1 S2 d3 D5
40
d1 d2 D3 S4
D2
20
S1 YD2
S0
0
0 0.02 0.04 0.06 0.08 0.1
X Values
Points with no projections (like D5 in Fig. 4.39) are ignored. The pushover curve error
2
1 N
⎛d ⎞
EPC = ∑1 ⎜ Y i ⎟ (4.14)
N ⎝ Di ⎠
where di is the vertical projection of the ith time history analysis point on the pushover
curve, YDi is the Y-coordinate of ith time history analysis point and N is the total number
121
of points considered. A value of pushover curve error index approaching to zero implies
high accuracy in the pushover analysis results (proximity to the time history analysis
results). Table 4.2 presents the pushover curve error index for different frames for various
load patterns used in pushover analysis. The table shows that proposed profile predicts
results with more accuracy compared to the other existing lateral load profiles.
IS-1893
FRAME UNI TRI MODE-1 UBPA Proposed
distribution
R-15 0.18 0.23 0.20 0.22 0.20 0.08
S1-15 0.16 0.25 0.13 0.24 0.17 0.10
S2-15 0.15 0.23 0.11 0.23 0.19 0.11
S3-15 0.11 0.22 0.22 0.19 0.20 0.08
R-10 0.21 0.19 0.20 0.31 0.18 0.08
S1-10 0.20 0.23 0.22 0.32 0.23 0.09
S2-10 0.18 0.20 0.22 0.30 0.22 0.10
S3-10 0.10 0.23 0.23 0.28 0.23 0.11
R-6 0.23 0.18 0.19 0.31 0.17 0.09
S1-6 0.22 0.21 0.20 0.33 0.21 0.11
S2-6 0.16 0.19 0.20 0.32 0.20 0.08
Mean Error 0.17 0.21 0.19 0.28 0.20 0.09
from pushover analysis with different load pattern and compared with the mean value of
the maximum displacements undergone by the same structure in nonlinear time history
analysis for 20 different earthquake ground motions. Figs 4.40 to 4.51 show the
122
differences between time history analysis and the pushover analysis in terms of maximum
S3-15
UNI S2-15
S1-15
R-15
-40 -20 0 20 40
Difference (%)
Fig. 4.40: Difference in maximum roof displacement between nonlinear time history
analyses and pushover analysis with uniform load pattern for 15 storey frames
S3-15
S2-15
Mode 1
S1-15
R-15
Fig. 4.41: Difference between nonlinear time history analyses and pushover analysis
with fundamental mode shape as load pattern for 15 storey frames
123
S3-15
UBPA S2-15
S1-15
R-15
-40 -20 0 20 40
Difference (%)
Fig. 4.42: Difference between nonlinear time history analyses and pushover analysis
with UBPA load pattern for 15 storey frames
S3-15
S2-15
Proposed
S1-15
R-15
-40 -20 0 20 40
Difference (%)
Fig. 4.43: Difference between nonlinear time history analyses and pushover analysis
with proposed load pattern for 15 storey frames
124
S3-10
UNI S2-10
S1-10
R-10
-40 -20 0 20 40
Difference (%)
Fig. 4.44: Difference between nonlinear time history analyses and pushover analysis
with uniform load pattern for 10 storey frames
S3-10
Mode 1 S2-10
S1-10
R-10
-40 -20 0 20 40
Difference (%)
Fig. 4.45: Difference between nonlinear time history analyses and pushover analysis
with fundamental mode shape as load pattern for 10 storey frames
125
S3-10
S2-10
UBPA
S1-10
R-10
-40 -20 0 20 40
Difference (%)
Fig. 4.46: Difference between nonlinear time history analyses and pushover analysis
with UBPA load pattern for 10 storey frames
S3-10
S2-10
Proposed
S1-10
R-10
-40 -20 0 20 40
Difference (%)
Fig. 4.47: Difference between nonlinear time history analyses and pushover analysis
with proposed load pattern for 10 storey frames
126
S2-6
UNI
S1-6
R-6
-40 -20 0 20 40
Difference (%)
Fig. 4.48: Difference between nonlinear time history analyses and pushover analysis
with uniform load pattern for 6 storey frames
S2-6
Mode 1
S1-6
R-6
-40 -20 0 20 40
Difference (%)
Fig. 4.49: Difference between nonlinear time history analyses and pushover analysis
with fundamental mode shape as load pattern for 6 storey frames
127
S2-6
UBPA S1-6
R-6
Fig. 4.50: Difference between nonlinear time history analyses and pushover analysis
with UBPA load pattern for 6 storey frames
S2-6
Proposed S1-6
R-6
-40 -20 0 20 40
Difference (%)
Fig. 4.51: Difference between nonlinear time history analyses and pushover analysis
with proposed load pattern for 6 storey frames
These figures show that uniform (UNI) and UBPA load patterns almost always under-
estimate the maximum displacement the structure undergoes before collapse. However,
for tall stepped frames (S3-15), both of these load patterns overestimate the maximum
128
displacement capacity. The load pattern corresponding to the fundamental mode shape
(Mode-1) and uniform (UNI) load pattern result in a good estimation of maximum
displacement for frames with less irregularity or less height (difference within 20% for R,
S1 and S2 type frames), while the UBPA load pattern shows a very high variation from
the mean time history analysis results for almost all the cases (difference is up to 50%).
The figures show that the proposed load pattern estimates the maximum roof
displacement of stepped frames for any height category with less than 10% error.
Similarly, the base shear capacity values of the frame estimated by pushover analysis
with different load patterns are compared with the mean value of base shear capacity of
the same structure as obtained from nonlinear time history analysis. Figs 4.52 to 4.63
show the differences between time history analysis and the pushover analysis in terms of
S3-15
UNI
S2-15
S1-15
R-15
-50 -25 0 25 50
Difference (%)
Fig. 4.52: Difference between nonlinear time history analyses and pushover analysis
with uniform load pattern for 15 storey frames
129
S3-15
S2-15
Mode 1
S1-15
R-15
-50 -25 0 25 50
Difference (%)
Fig. 4.53: Difference between nonlinear time history analyses and pushover analysis
with fundamental mode shape as load pattern for 15 storey frames
S3-15
S2-15
UBPA S1-15
R-15
-50 0 50
Difference (%)
Fig. 4.54: Difference between nonlinear time history analyses and pushover analysis
with UBPA load pattern for 15 storey frames
130
S3-15
S2-15
Proposed
S1-15
R-15
-50 -25 0 25 50
Difference (%)
Fig. 4.55: Difference between nonlinear time history analyses and pushover analysis
with proposed load pattern for 15 storey frames
S3-10
S2-10
UNI
S1-10
R-10
-50 -25 0 25 50
Difference (%)
Fig. 4.56: Difference between nonlinear time history analyses and pushover analysis
with uniform load pattern for 10 storey frames
131
S3-10
S2-10
Mode 1
S1-10
R-10
-50 -25 0 25 50
Difference (%)
Fig. 4.57: Difference between nonlinear time history analyses and pushover analysis
with fundamental mode shape as load pattern for 10 storey frames
S3-10
S2-10
UBPA S1-10
R-10
-50 0 50
Difference (%)
Fig. 4.58: Difference between nonlinear time history analyses and pushover analysis
with UBPA load pattern for 10 storey frames
132
S3-10
S2-10
Proposed
S1-10
R-10
-50 -25 0 25 50
Difference (%)
Fig. 4.59: Difference between nonlinear time history analyses and pushover analysis
with proposed load pattern for 10 storey frames
S2-6
UNI S1-6
R-6
-50 -25 0 25 50
Difference (%)
Fig. 4.60: Difference between nonlinear time history analyses and pushover analysis
with uniform load pattern for 6 storey frames
133
S2-6
Mode 1 S1-6
R-6
-50 -25 0 25 50
Difference (%)
Fig. 4.61: Difference between nonlinear time history analyses and pushover analysis
with fundamental mode shape as load pattern for 6 storey frames
S2-6
S1-6
UBPA
R-6
-50 0 50
Difference (%)
Fig. 4.62: Difference between nonlinear time history analyses and pushover analysis
with UBPA load pattern for 6 storey frames
134
S2-6
Proposed
S1-6
R-6
-50 -25 0 25 50
Difference (%)
Fig. 4.63: Difference between nonlinear time history analyses and pushover analysis
with proposed load pattern for 6 storey frames
The figures show that the estimation of base shear capacity using UBPA load pattern is
the poorest among all others. It highly over-estimates the base shear capacity (up to
30%). It can be noticed from Figs 4.53, 4.57 and 4.61 that the pushover analysis with
load profile corresponding to the fundamental mode shape always underestimates the
base shear capacity as expected by FEMA 356. Similarly, as per FEMA 356, pushover
analysis with uniform load profile (UNI) should give the upper bound of the base shear
capacity. Figs 4.52, 4.56 and 4.60 show that pushover analysis with uniform load pattern
is unable to predict the upper bound of the base shear capacity for frames with more
irregularity (S2 and S3 types). However, uniform load pattern represents the upper bound
estimation of the base shear capacity for the frames with less irregularity (R and S1
types). The proposed load profile performs better than other lateral load profiles studied
here for all the building frames. It slightly underestimates the base shear capacity (with a
135
4.6.5 Estimation of Target Displacement
Target displacement was calculated for all the building frames studied here using
displacement coefficient method (FEMA 356), capacity spectrum method (ATC 40),
modal pushover analysis (MPA), modified modal pushover analysis (MMPA), upper
bound pushover analysis (UBPA) and proposed method for 0.36g design spectrum and
compared with the mean displacement demand obtained from nonlinear time history
analysis for 20 earthquake ground motion with normalised PGA = 0.36g. FEMA 356 and
ATC 40 estimates are based on the triangular load pattern. Different methods of target
presents the estimated values of target displacement (mm) for each frame and Table 4.4
shows the ratio of ‘estimated (pushover analysis)’ and ‘exact (nonlinear time history
136
Table 4.4: Ratio of ‘estimated’ and ‘exact’ value of target displacement
It can be seen from Table 4.3 and Table 4.4 that the displacement coefficient method
(FEMA-356), modal pushover analysis (MPA), and modified modal pushover analysis
method (ATC-40) overestimates the target displacement for all the cases studied here.
The capacity spectrum method (ATC-40) estimates target displacement that has the
largest deviation from the nonlinear time history analysis. Estimation of target
displacement using upper bound pushover analysis (UBPA) is closest among all the
existing methods. The proposed method estimates target displacement very close to the
time history analysis result (average ratio =1.04) with a coefficient of variation of 3%. An
interesting observation here is that all the methodologies studied here (except capacity
137
spectrum method of ATC 40) result in reasonably accurate estimation of target
displacement for regular frames. However, for the stepped frame, all of these methods
reveals the possible failure mechanism for the structure. The pattern of hinge formation is
solely dependant to lateral load profile used in the pushover analysis. Fig. 4.64 presents
the hinge distribution in a typical stepped frame (S2-15) as obtained from pushover
analysis with conventional triangular load pattern and that with the proposed lateral load
pattern. This figure also presents the hinge distribution for the same frame obtained from
time history analysis. The figure clearly shows that, in pushover analysis, when triangular
load pattern is used, all of the column hinges are concentrated at the upper floors near the
steps, while the beam hinges are well distributed. It cannot predict the possible failure
mechanism at the ground floor columns, which is indicated in the time history analysis
results. Since triangular load pattern has more lateral force intensity at the upper storey
level, the flexural demand in the upper floor columns reach their capacity much before
the ground floor columns. However, the proposed load pattern reveals the possibility of
hinges forming in the ground floor columns also, as proved by the time history analysis.
The same can be observed from displacement and drift profile of stepped frames at
collapse. Figs 4.65 to 4.70 present the displacement and drift profile at collapse for three
15 storey stepped frames (S1-15, S2-15 and S3-15) as obtained from pushover analysis
with different lateral load pattern along with mean displacement and drift profile obtained
138
(a) Pushover analysis using Triangular (b) Pushover analysis using proposed
load pattern load pattern
(c) Time history analysis for Elcentro (d) Time history analysis for Loma
ground motion Prieta ground motion
Fig. 4.64: Distribution of hinges at collapse for 15 storey stepped frame (S2-15)
139
15
12
Storey Level
9
6 Mode-1
TRI
IS-1893
3 Proposed
NLTHA
0
0 0.3 0.6 0.9 1.2
Storey Displacement (m)
Fig. 4.65: Displacement profile at collapse for 15 storey stepped frame (S1-15)
15
12
Mode-1
Storey Level
9 TRI
IS-1893
Proposed
6 NLTHA
0
0 1 2 3 4
Inter-storey Drift (%)
Fig. 4.66: Inter storey drift profile at collapse for 15 storey stepped frame (S1-15)
140
15
12
Storey Level
9
6 Mode-1
TRI
IS-1893
3
Proposed
NLTHA
0
0 0.25 0.5 0.75 1
Storey Displacemenet (m)
Fig. 4.67: Displacement profile at collapse for 15 storey stepped frame (S2-15)
15
12
Mode-1
Storey Level
9 TRI
IS-1893
Proposed
6
NLTHA
0
0 1 2 3 4 5
Inter-storey Drift (%)
Fig. 4.68: Inter storey drift profile at collapse for 15 storey stepped frame (S2-15)
141
15
12
Storey Level
9
6 Mode-1
TRI
IS-1893
3
Proposed
NLTHA
0
0.00 0.25 0.50 0.75 1.00
Storey Displacement (m)
Fig. 4.69: Displacement profile at collapse for 15 storey stepped frame (S3-15)
15
12
Mode-1
Storey Level
9 TRI
IS-1893
6 Proposed
NLTHA
0
0 1 2 3 4 5
Inter-storey Drift (%)
Fig. 4.70: Inter storey drift profile at collapse for 15 storey stepped frame (S3-15)
142
These figures clearly show that inter storey drifts for the upper storeys are comparatively
less for stepped frames. This is due to the fact that the upper storeys, with less relative
stiffness and mass, attract relatively less force during the earthquake ground motion. The
This Chapter begins by proposing an empirical formula (modification of the existing code
formula for regular RC framed building) to calculate fundamental time period of stepped
equation estimates fundamental periods of stepped frames that match fairly close to the
This Chapter then presents an evaluation of existing lateral load profiles for pushover
analysis of stepped frames. The results show that the lateral load patterns recommended
in FEMA-356 cannot always represent the upper and lower bound of the pushover curve
for the stepped building. These load patterns also fail to predict the hinge distribution,
inter-storey drift profile for stepped building. A lateral load profile, appropriate for
stepped building frames, is proposed for use in pushover analysis. This load profile
includes the properties of first three elastic modes and performs better than the existing
Evaluation of the target displacement estimation procedures shows that the displacement
compared to nonlinear time history analysis whereas capacity spectrum method (ATC-
40) always over-estimates. Other alternative methods are mostly complicated and time-
143
consuming and these methods do not perform satisfactorily for stepped buildings. An
stepped building frames based on nonlinear regression analysis. This formula generates
values of the target displacement that are in close agreement with the nonlinear time
The proposed pushover analysis procedure for RC stepped frames has been carried out
for 23 frames with different irregularities and heights. The results have been validated by
nonlinear time history analysis, using twenty earthquake ground motions. The results
obtained show consistently good performance in comparison with the existing methods of
pushover analysis.
144