Class of Irregularity

Engineering Structures 30 (2008) 12721291
www.elsevier.com/locate/engstruct
Analysis for preliminary design of a class of torsionally coupled buildings
with horizontal setbacks
Dhiman Basu
, N. Gopalakrishnan
Structural Dynamics Lab, Structural Engineering Research Center, Chennai 600 113, India
Received 24 July 2006; received in revised form 1 July 2007; accepted 3 July 2007
Available online 4 September 2007
Abstract
Simplied method of analysis of a special class of torsionally coupled buildings with horizontal setbacks is developed that can be executed
with a plane frame analysis by means of a personal computer. Since most buildings may not exactly satisfy all the classication criteria, it is shown
that an averaging technique may be used in such cases up to a certain limit. Perturbation analysis is carried out in determining such a limit, and
numerical examples are presented to validate this. As a whole, the proposed simplied analysis may be used as a convenient and offhand tool at
the preliminary stage of design.
c 2007 Elsevier Ltd. All rights reserved.
Keywords: Special class of buildings; Torsional coupling; Seismic analysis; Proportional stiffness; Horizontal setbacks; Dynamic analysis; Preliminary design
1.0. Introduction
Buildings are, hardly ever, truly symmetric. Consequently,
lateral vibration of buildings during seismic excitation is
always coupled with torsional vibration. A number of studies
have been reported in the literature addressing the issue
of lateraltorsional coupling ([3,68,1013,15,17] and many
more). Further, most seismic codes often suggest an equivalent
static analysis against a specied lateral load prole for regular
and nominally irregular buildings taking into account the
torsional effects. Therefore, behavior of asymmetric buildings
under monotonic or equivalent static loading has also received
adequate attention ([1,2,4,5] and many more). On the other
hand, when irregularity exceeds certain nominal limit, for
example, building with horizontal setbacks, complete dynamic
analysis is a must according to most seismic codes.
Instead of complete dynamic analysis, simplied dynamic
analysis is often preferred, especially, at the stage of
preliminary design. In this connection, a special class of
torsionally coupled buildings has been reported in the literature
[7,8,11,12] wherein the lateral stiffness and mass are distributed
throughout the building in a specic way. Kan and Chopra [11,
12] reported that a multi-story building belongs to the category
Corresponding author. Tel.: +91 44 2254 9147; fax: +91 44 2254 1508.
E-mail address: dhiman basu@rediffmail.com (D. Basu).
of a special class if (i) the center of mass (CM) of all the oors
lie on one vertical line and radius of gyration about the vertical
axis passing through the CM is the same for all the oors,
(ii) principal planes of all the lateral load-resisting elements
form an orthogonal grid system and (iii) lateral stiffness
matrices of all the lateral load-resisting elements oriented along
either of the two orthogonal directions are proportional to
a characteristic lateral stiffness matrix along that direction;
however, these two characteristic lateral stiffness matrices may
not be identical. Further, it has been shown [11,12] that the
response behavior of such a shear building may be obtained
through appropriate combination of the results calculated from
the analysis of two smaller systems, namely, (a) corresponding
torsionally uncoupled building and (b) equivalent one-story
torsionally coupled building. Hejal and Chopra [7,8] extended
the concept of this simplied analysis to buildings comprising
moment-resisting frames (MRFs) with the imposition of
an additional constraint, which states identical characteristic
lateral stiffness matrices along both the orthogonal directions.
In other words, lateral stiffness matrices of all the lateral load-
resisting elements are proportional.
However, conditions of the existence of the special class of
buildings are hardly, ever, truly satised in practice, e.g., (i)
lateral stiffness matrices of the constituting frames may not
be exactly proportional and (ii) CM of all the oors may not
0141-0296/$ - see front matter c 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2007.07.013
D. Basu, N. Gopalakrishnan / Engineering Structures 30 (2008) 12721291 1273
be located exactly on a single vertical line. None of the two
previous studies [11,12,7,8] has investigated the applicability
of the simplied procedure in such cases. In the rst study [11,
12], shear building is assumed and the existence of which is
in itself questionable. In the second study [7,8], numerically
proportional lateral stiffness matrices for the MRFs are directly
assumed. This is because MRFs do not, in general, yield to
exactly proportional lateral stiffness matrices. Furthermore,
in order to proportion the MRFs so as to yield proportional
lateral stiffness matrices, Hejal and Chopra [7,8] made an
attempt by introducing a factor, called the joint rotation index,
which is based on the assumption of uniform story height and
uniform bay width, but these conditions are hardly, if ever,
met in practice. Surprisingly, in both the previous studies,
orthogonality of the building was assumed and that seems to
be superuous.
A building is said to be with horizontal setbacks if there
exists at least two points within its plan area when joined
through a straight line, the line runs out of the plan area.
In the absence of a stiff shear wall, the lateral stiffness
matrices of the constituting frames of a building with horizontal
setbacks closely satisfy the desired proportionality criterion
of the special class of buildings. Consequently, any building
without stiff shear walls may also be treated as satisfying all
the classication criteria if the CM of all the oors lie on a
single vertical line and the radius of gyration about this vertical
line is the same for all the oors. The objective of this paper is
two-fold: rst, to explore the concept of this simplied analysis
to the special class of buildings with horizontal setbacks
and second, to investigate the applicability of the procedure
where mass proportionality criteria of the special class are not
exactly satised. In order to meet these objectives, rst, the
formulation presented in Hejal and Chopra [7,8] is extended
to the special class of buildings with horizontal setbacks,
as reported herein followed by a numerical example on a
ten-storied C-shape building with MRFs. Second, a rigorous
perturbation analysis is carried out in order to restrict the
limit of the scattering of the oor CMs (from the vertical line
passing through the average CM of the building) up to which
the simplied analysis can be applied with an acceptable error
for all practical purposes. Numerical example on the same C-
shape-MRFs building, but with vertically-nonaligned CMs, is
then presented to substantiate the results of the perturbation
analysis. Finally, the same C-shape-MRFs building, but with
another two sets of vertically-nonaligned CMs, are analyzed so
as to assess the conservativeness in the limiting criteria derived
from the perturbation analysis. In all the numerical examples
presented, results of the simplied method are compared with
that calculated using SAP2000 [14].
The methodology developed in this paper does not impose
any constraint over the shape of the diaphragm and the
orientation of the lateral load-resisting elements. Therefore, the
proposed approach is equally applicable to buildings of V-, Y-
etc. shape provided that the necessary conditions are satised.
However, without losing the generality, C-shaped building
is chosen for illustration. Further, 3D FEM analysis of the
special class of torsionally coupled buildings with horizontal
setbacks is no longer impossible in the present era of computer
advancement. Consequently, one may argue against the utility
of this simplied analysis in the present scenario. Nevertheless,
such a simplied analysis may always be preferred to assess
the design force resultants at preliminary stage of design;
3D FEM analysis may be carried out at the nal stage of
verication. Therefore, the proposed simplied analysis can
be applied as a convenient and offhand tool with sufcient
accuracy at the preliminary stage and requires only plane frame
analysis.
2.0. Development of the methodology
To formulate the methodology presented in this paper, an
arbitrarily shaped diaphragm of a typical oor of an N-story
building comprising different wings is considered. CM of all
the oors is assumed to be lying on the same vertical line; the
radius of gyration about the vertical axis passing through the
CM is assumed to be the same for all the oors. Dening the
degrees of freedom at the CM of respective oors, mass matrix
of the building can be expressed as the direct product of two
smaller matrices as follows:
3N3N
[M] =
33
[CM]
B

NN
[m] where, (1)
[CM]
B
=
_
1 1 r
2
_
diag
. (2)
Here, [m] is a diagonal matrix with elements as the lumped
mass at the respective oor levels and may be considered
as the mass matrix of the characteristic frame; r the radius
of gyration of a typical oor about the vertical axis passing
through the CM; and [CM]
B
may be considered as the mass
matrix of an equivalent one-story coupled building. Similarly,
considering the proportionality of the lateral stiffness matrices
of the constituting frames and assuming (i) i th frame of the qth
wing with stiffness proportionality constant as C
i q
is located
at a distance d
i q
from the geometric center of gravity of the
wing and oriented at an angle
i q
in counter clockwise direction
with respect to the longitudinal axis of the wing, (ii) qth wing
of the building, comprising of NEQ number of frames, is
located from the CM of the building at distances d
Lq
and d
Sq
along the longitudinal and transverse directions of the wings,
respectively, and oriented at an angle
q
in counter clockwise
direction with respect to the global X-axis and (iii) [K
] is the
stiffness matrix of the characteristic frame while NW is the
total number of wings, stiffness matrix of the building may also
be expressed as the direct product of two smaller matrices as
follows:
3N3N
[K
B
] =
33
[CL]
B

NN
[K
] where, (3)
[CL]
B
=
NW
q=1
_
[CL]
q B
_
;
[CL]
q B
= [T]
T
q
[CL]
q
[T]
q
;
[CL]
q
=
NEQ
i =1
_
[CL]
i q
_
;
(4a)
1274 D. Basu, N. Gopalakrishnan / Engineering Structures 30 (2008) 12721291
[T]
q
=
_
_
cos
q
sin
q
d
Sq
sin
q
cos
q
d
Lq
0 0 1
_
_
;
[CL]
i q
= [T
el
]
T
i q
C
i q
[T
el
]
i q
;
(4b)
[T
el
]
i q
=
_
cos
i q
sin
i q
d
i q
_
. (4c)
Here, [CL]
B
may be considered as the lateral stiffness
matrix of an equivalent one-story coupled building. For such
a proportioned building, it may be shown that, similar to CMs,
center of rigidity (CR) of the oors are also located on another
vertical line irrespective of the lateral load prole. Eigenvalue
equation for the building is now dened as
[K
B
] {} =
2
[M] {} (5)
where , {} are the frequency and mode shape, respectively,
of the building. Also, dene two smaller eigen equations, one
each for characteristic frame and one-story coupled system as
follows:
_
K
_
{} =
2
[m] {} (6)
[CL]
B
{} =
2
[CM]
B
{} (7)
where, , {} and , {} are the frequencies, mode shape vectors
of the characteristic frame and equivalent one-story coupled
building, respectively. Next, assuming
{} = {} {} (8)
and utilizing Eqs. (1), (3) and (8) in Eq. (5), thereafter,
substituting Eqs. (6) and (7) and carrying out direct product,
it may be shown that
2
i j
=
2
i

2
j
or
i j
=
i

j
where i = 1, . . . 3;
and j = 1, 2, . . . , N. (9)
Therefore, the frequency and mode shape vector of the entire
building in i j th mode are given by the direct product of
those of equivalent one-story coupled building in i th mode
and characteristic frame in j th mode. Such decoupling is not
only valid for frequency and mode shapes, but also holds good
for any other modal response. For example, shear force shared
by the sth frame in qth wing, with stiffness proportionality
constant C
sq
may be expressed as
{V}
sq
i j
= SD(P
i
C
sq
13
[T
el
]
sq
33
[T]
q
31
{}
i
)
(P
j
1N
{
f
}
T
NN
[K
]
N1
{}
j
) (10)
where, P
i
and P
j
are the participation factor (PF) of the one-
story coupled system in i th mode, and the characteristic frame
in j th mode, respectively, {
f
} the inuence vector of the
characteristic frame and SD the spectral displacement. In Eq.
(10), expressions within the rst braces may be identied as
contributions from one-story coupled system in i th mode and
characteristic frame in j th mode.
Nonproportionality of lateral stiffness matrices of constitut-
ing MRFs affects the estimation of mode shapes and hence, the
frame shear force, in a relatively adverse way than that of fre-
quency. Thus, only the expression of frame shear force Eq. (10)
needs to be modied as follows before applying the simplied
method to buildings with MRFs.
{V}
sq
i j
= SD(P
i
13
[T
el
]
sq
33
[T]
q
31
{}
i
)
(P
j
1N
{
f
}
T
C
sq
NN
[K
]
N1
{}
j
)
= SD(P
i
13
[T
el
]
sq
33
[T]
q
31
{}
i
)
(P
j
1N
{
f
}
T
NN
[K
sq
]
N1
{}
j
) (11)
where, [K
sq
] is the lateral stiffness matrix of the frame under
consideration. This is to minimize the error in the story-wise
distribution of the frame shear force, only.
3.0. Numerical example-1
A ten-storied C-shaped building (Fig. 1) is considered
for this example-problem. Column and beam dimensions are
shown in Fig. 1. Story height is considered 3.5 m. Modulus of
elasticity of concrete is adopted as 2.5510
7
kN/m
2
while the
Poissons ratio is taken as 0.2. The building is considered to be
located in zone-V of IS 1893-2002 [9] and ratio of importance
factor to response reduction factor is taken as unity. Soil
condition is assumed as medium and acceleration spectrum
is chosen in compliance with IS 1893-2002 [9]. Intensity of
uniform mass distribution is taken as 1.0 t/m
2
, 0.9 t/m
2
,
0.8 t/m
2
and 0.7 t/m
2
for the oors 13, 46, 79 and roof
level, respectively, which leads to lumped mass of 2400 t,
2160 t, 1920 t and 1680 t, respectively. CM is calculated for
a typical oor and X-, Y-coordinates of which are 32.656 m,
18.594 m, respectively, with respect to point H (Fig. 1) as
origin. Radius of gyration about the vertical axis passing
through the CM is calculated for a typical oor as 24.639 m.
Considering frame A1 as the characteristic frame, stiffness
proportionality constants for the frames A1A6, A7A14,
B1B6, B7B10 and B11B16 are evaluated, using plane
frame analysis, as 1.0, 0.2975492, 0.610739, 0.2032498 and
0.4907527, respectively. The building is then analyzed using
the proposed method. Frequency and modal mass participation
ratio (MMPR) for the excitation along X- and Y-directions,
for the rst 15 modes are presented in Table 1 (columns (2),
(5) and (6), respectively). Similar responses are also presented
(Table 1, columns (3) (7) and (8) respectively) when frame B1 is
used as the characteristic frame. Shear force induced in frames
A1, A11, A14, B1 and B16 are presented in Table 2 (columns
(2), (5), (8), (11) and (17), respectively) when excitation
is considered along X-direction and frame A1 is taken as
the characteristic frame. Moreover, for frames B1 and B16,
shear forces are also presented (Table 2, columns (14) and
(20), respectively) when the excitation is considered along Y-
direction and characteristic frame is chosen as B1. In order to
compare these responses, the example-building is then analyzed
using SAP2000 and the responses of interest are tabulated
(Tables 1 and 2).
4.0. Range of applicability of the proposed method using
perturbation analysis
It may be intuitively judged that, when CM of all the oors
do not lie on a single vertical line but the scattering in their
Table 1
Comparison of modal responses calculated using the proposed method and SAP2000 for example-1
Mode
number
Frequency (rad/s) Modal mass participation ratio (%)
Proposed SAP2000 Proposed SAP2000
Characteristic frame
A1
Characteristic frame
B1
Characteristic frame A1 Characteristic frame B1 X-
direction
Y-direction
X-
direction
Y-
direction
X-
direction
Y-
direction
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
1 4.82 4.83 4.84 0.3 79.1 0.3 78.9 0.5 78.3
2 4.92 4.93 4.95 42.1 0.9 42.0 0.9 45.9 1.4
3 5.34 5.35 5.37 37.6 0.1 37.5 0.1 33.7 0.1
4 14.30 14.35 14.38 0.0 10.3 0.0 10.4 0.2 10.2
5 14.61 14.66 14.65 5.5 0.1 5.5 0.1 6.1 0.4
6 15.85 15.91 15.94 4.9 0.0 4.9 0.0 4.3 0.0
7 24.79 24.94 24.98 0.0 4.0 0.0 4.0 0.2 3.5
8 25.31 25.47 25.36 2.1 0.1 2.1 0.1 2.2 0.5
9 27.47 27.64 27.65 1.9 0.0 1.9 0.0 1.5 0.0
10 36.81 37.10 37.04 0.00 2.1 0.0 2.1 0.4 1.4
11 37.59 37.88 37.53 1.1 0.0 1.1 0.0 0.9 0.7
12 40.80 41.12 41.14 1.0 0.0 1.0 0.0 0.8 0.0
13 49.52 50.00 49.62 0.0 1.4 0.0 1.4 0.5 0.5
14 50.56 51.10 50.42 0.8 0.0 0.8 0.0 0.4 0.9
15 54.88 55.42 55.46 0.7 0.0 0.7 0.0 0.5 0.0
Fig. 1. Typical oor plan of the example-building.
locations is small, the proposed method may still be applied
by using the average location of CM and the average radius
of gyration, provided the stiffness proportionality criterion of
the constituting frames are closely satised. A perturbation
analysis will now be carried out in order to ascertain the limit
of such scattering; up to this limit, the proposed method may be
applied with an acceptable margin of error at the preliminary
stage of design. Since the building considered in this paper does
not comprise shear wall, stiffness proportionality criterion of
the constituting frames are closely satised and hence, while
Table 2
Comparison of frame shear force calculated using the proposed method and SAP2000 for example-1
Floor Frame-A1 Frame-A11 Frame-A14
X-direction excitation X-direction excitation X-direction excitation
Proposed SAP %error Proposed SAP %error Proposed SAP %error
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
10 607.9 598.5 1.6 240.3 262.8 8.6 196.3 215.7 9.0
9 1154.1 1112.0 3.8 460.2 488.6 5.8 376.1 401.4 6.3
8 1595.6 1535.9 3.9 643.9 679.7 5.3 526.3 557.8 5.6
7 1952.8 1877.2 4.0 792.3 830.6 4.6 647.7 681.2 4.9
6 2288.5 2197.7 4.1 929.1 968.6 4.1 759.6 794.1 4.3
5 2582.5 2476.9 4.3 1051.6 1090.9 3.6 859.9 894.0 3.8
4 2846.0 2725.2 4.4 1165.3 1203.0 3.1 952.9 985.7 3.3
3 3106.8 2968.8 4.7 1275.8 1312.8 2.8 1043.2 1075.9 3.0
2 3312.3 3159.2 4.8 1353.0 1395.4 3.0 1106.3 1144.6 3.4
1 3417.8 3272.5 4.4 1204.4 1284.3 6.2 984.8 1061.1 7.2
Frame-B1 Frame-B16
X-direction excitation Y-direction excitation X-direction excitation Y-direction excitation
Proposed SAP2000 %error Proposed SAP2000 %error Proposed SAP2000 %error Proposed SAP2000 %error
(11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
382.0 401.8 4.9 561.3 577.5 2.8 403.2 441.6 8.7 338.2 327.5 3.2
747.4 729.7 2.4 1050.8 1093.3 3.9 784.5 831.0 5.6 656.5 610.9 7.5
1030.5 1007.1 2.3 1436.4 1483.1 3.2 1081.4 1130.9 4.4 898.1 846.2 6.1
1258.7 1230.4 2.3 1738.5 1783.5 2.5 1320.6 1367.3 3.4 1087.5 1033.6 5.2
1472.8 1438.7 2.4 2016.8 2059.8 2.1 1546.5 1588.0 2.6 1263.4 1206.6 4.7
1660.2 1619.1 2.5 2261.2 2302.1 1.8 1743.9 1780.4 2.1 1417.6 1357.4 4.4
1830.5 1780.1 2.8 2486.1 2525.5 1.6 1924.2 1954.4 1.5 1560.0 1494.8 4.4
2004.4 1940.7 3.3 2718.1 2760.5 1.5 2113.2 2133.8 0.9 1710.2 1634.9 4.6
2139.9 2063.2 3.7 2908.1 2963.4 1.9 2268.1 2285.6 0.8 1839.9 1744.4 5.5
carrying out the perturbation analysis, perturbation is given
only in the mass matrix. Moreover, in perturbation analysis,
eigenvalue problem is formulated with respect to the CR of the
respective oors as the reference point.
4.1. Perturbation equation from the rst principle
Eigenvalue problems of the unperturbed and perturbed
systems are given by
0
K = M and (12a)
K = (M +B) (12b)
where, K is the stiffness matrix; M, (M +B) the mass
matrices,
0
, the eigenvalues, and , the eigenvectors
of the unperturbed and perturbed systems, respectively. Here,
perturbation in the mass matrix may be dened as E = B.
Now, using the fundamental principle of perturbation [16],
expressions for the perturbed eigenvalues and eigenvectors may
be arrived at. Resulting expressions from the second order
perturbation are as follows:
i
=
0i
+
0i
_
T
0i
E
0i
_ n
j =1
j =i
0i
0 j
0i

0 j
_
T
0i
E
0 j
_
2
(13)
i
() =
0i
+
n
j =1
j =i
__

0 j
0i

0 j
_
_
T
0 j
E
0i
_
_
1 +
_
1
0i

0 j
_
_
0 j
_
T
0 j
E
0 j
_

0i
_
T
0i
E
0i
__
__
0 j
. (14)
4.2. Error estimation by perturbation analysis
Let m
i
be the lumped mass at the CM of the i th oor and r
i
be its radius of gyration about the vertical axis passing through
the CM. Therefore, the average location of the CM of the entire
building may be calculated as
X
CM
=
m
i
x
i
m
i
and (15a)
Y
CM
=
m
i
y
i
m
i
. (15b)
Here, x
i
and y
i
are measured with respect to an arbitrarily
chosen origin. Radius of gyration of the i th oor about the
vertical axis passing through the average CM is then given
by,
r
2
i
= r
i
2
+b
2
x
i
+b
2
y
i
(16)
where b
x
i
and b
y
i
are the distances between the actual and av-
erage CMs at the i th oor level along the X- and Y-directions,
respectively. Therefore, average radius of gyration for the entire
building is given by
r
2
=
m
i
r
i
m
i
=
m
i
_
r
2
i
+b
2
x
i
+b
2
y
i
_
m
i
. (17)
This is the radius of gyration assumed for all the oors in the
simplied analysis because (i) b
x
i
and b
y
i
are small enough
when compared to the dimensions of the building and (ii) r
2
i
does not differ much from oor-to-oor.
Stiffness matrix of the building with respect to the CR can
be expressed as
[K] =
_
_
K
xx
K
xy
0
K
yx
K
yy
0
0 0 K
r
_
_
. (18)
Since, CRs and CMs of all the oors are lying on two different
vertical lines for the model assumed in the simplied analysis
(average CM model), denoting the distance between these two
vertical lines along the X- and Y-directions as e
x
and e
y
,
respectively, mass matrix of the actual building with respect to
the CR may be expressed in Box I
However, in the simplied analysis, the stiffness matrix is
assumed to be the same as that of the actual building while the
mass matrix is assumed to be
[M]
B
=
_
_
_
[m
i
] 0 [m
i
e
y
]
0 [m
i
] [m
i
e
x
]
_
m
i
e
y
_
[m
i
e
x
] [m
i
_
r
2
+e
2
x
+e
2
y
_
]
_
_. (19)
Let us now denote the actual building as system-A and
that assumed in the simplied analysis as system-B. For
convenience in the perturbation analysis, in addition, system-C
is assumed wherein the stiffness matrix is considered the same
as before while the mass matrix is assumed to be,
[M]
C
=
_
_
_
[m
i
] 0 0
0 [m
i
] 0
0 0
_
m
i
_
r
2
+e
2
x
+e
2
y
__
_
_. (20)
The system-C, hereafter, may be referred to as the unperturbed
system. Further, comparing the i th diagonal entry of the M
sub-matrices of system-A and system-B (or C), the difference

may be noted as
m
i
d
2
i
= m
i
_
b
x
i
_
b
x
i
+2e
x
_
+b
y
i
_
b
y
i
+2e
y
__
[as r
2
i
r
2
].
(21)
Moreover, denoting L and S to be the maximum dimensions
of the plan area of the building along the X- and Y-directions,
respectively, and letting
e
x
L
=
x
;
e
y
L
=
y
;
b
x
i
L
=
x
i
;
b
y
i
L
=
y
i
and r
2
=
L
L
2
=
S
S
2
, it may be shown that
d
2
i
r
2
=
_
_
2
x
i
+2
x
L
+

2
y
i
+2
y
y
i
S
_
_
. (22)
Typically,
x
i
and
y
i
are of the order 10
1
(or even 10
2
);
x
,
y
of the order 10
1
in case of highly eccentric system; and
L
,
S
of the order 10
1
. Therefore, it may be shown that
d
2
i
r
2
is of the order 10
1
to 10
3
; this result will be used in the
subsequent stages of the perturbation analysis.
Here, the objective is to relate the order of error in the
estimated frame shear forces when using system-B in lieu
of system-A with the order of the scattering in location of
the oor CMs. This is done in a step-by-step fashion: First,
using the general expressions for the perturbed eigenvalues
and eigenvectors Eqs. (13) and (14), error equations for
the eigenvalues and eigenvectors are formulated. Second,
these error equations are greatly simplied using certain
pertinent approximations. Third, likelihood order of the error
in eigenvalues and eigenvectors are related with the order of the
scattering in location of the oor CMs. Finally, expected order
of the error in the estimation of the frame shear forces is related
with the order of the scattering in location of the oor CMs.
The complete derivation has been presented in Appendix. This
is because, (i) readers of this paper have to deal with only the
end results of the perturbation analysis and (ii) entire derivation
process is extremely rigorous and likely to distract the readers
from the main stream.
In Appendix, it has been shown that the error in frame shear
forces, estimated using the simplied procedure, on account
of the vertically-nonaligned oor CMs might be restricted to
10% (which might be considered reasonable at the stage
of preliminary design) if the following two conditions are
satised:
|
x
| 0.002 and (23a)
= 0.002. (23b)
Here,
x
= {u}
T
1
[
b
x
i
m
i
r
]
diag
{u}
1
,
y
= {u}
T
1
[
b
y
i
m
i
r
]
diag
{u}
1
are the indices for assessing the applicability of the simplied
analysis and {u}
1
the mass normalized fundamental mode shape
of the characteristic frame. If
x
= C
1
10
and
Y
=
C
2
10
with |C
1
|, |C
2
| close to unity, e.g., 1.0 to 2.0, it has
been further shown that the maximum expected difference for
the eigenvalues ( ) while assuming system-B in lieu of system-
A is of the order 10
and that for the eigenvectors is of the

order 10
+1
{u}
1
.
5.0. Numerical example-2: Validation of the proposed
method in building with scattered CM
The building considered for this example is the same as that
considered in example-1 except the location of CMs. Offsets
in the location of CMs from that considered in example-1 are
shown in Table 3 (columns (2), (3)). These locations of the
CMs are chosen in such a way that location of the average
CM coincides with that considered in example-1. Also, average
of square of radius of gyration about the vertical axis passing
through the average CM is calculated as 609.9 m
2
, which is
nearly the same as that considered in example-1. Hence, the
building considered in example-1 nearly represents the building
idealized in the proposed method (average CM model) when
applied to this example-problem. Therefore, response of the
average CM model associated with the building considered
in this example may be considered the same as that obtained
in example-1. Next, considering frame A1 as the characteristic
frame, using Eqs. (23a) and (23b),
x
and
y
are evaluated
for this building as 0.00076 and 0.00129, respectively. Clearly,
Eqs. (23a) and (23b) are satised for this example-building. The
[M]
A
=
_
_
_
[m
i
] 0
__
e
y
+b
y
i
_
m
i
_
0 [m
i
]
__
e
x
+b
x
i
_
m
i
_
__
e
y
+b
yi
_
m
i
_
[(e
x
+b
xi
) m
i
]
_
m
i
_
r
2
i
+
_
b
x
i
+e
x
_
2
+
_
b
y
i
+e
y
_
2
__
_
_.
Box I.
Table 3
Offsets in CM for example-problems-2, -3 and -4 with respect to example-problem-1
Floor Offset in CM for example-2 Offset in CM for example-3a Offset in CM for example-3b
X-direction (m) Y-direction (m) X-direction (m) Y-direction (m) X-direction (m) Y-direction (m)
(1) (2) (3) (4) (5) (6) (7)
1 1.0 0.5 2.0 0.5 6.0 3.5
2 1.5 1.0 3.0 1.0 1.5 1.0
3 2.5 0.5 5.0 0.5 4.5 4.5
4 0.5 1.0 1.0 1.0 3.0 2.5
5 1.5 1.5 3.0 1.5 1.5 2.5
6 1.0 0.5 2.0 0.5 4.5 5.0
7 1.5 1.5 3.0 1.5 5.0 4.0
8 0.5 0.5 1.0 0.5 3.0 2.0
9 2.0 1.0 4.0 1.0 2.0 6.0
10 0.0 0.0 0.0 0.0 0.0 0.0
Table 4
Comparison of frame shear force of the perturbed and the average CM model calculated using SAP2000 for example-2
Perturbed Average CM %error Perturbed Average CM %error Perturbed Average CM %error
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
10 602.4 598.5 0.7 257.8 262.8 1.9 211.7 215.7 1.9
9 1140.6 1112.0 2.5 474.7 488.6 2.9 389.9 401.4 2.9
8 1576.1 1535.9 2.6 661.3 679.7 2.8 542.8 557.8 2.8
7 1900.8 1877.2 1.2 819.9 830.6 1.3 672.1 681.2 1.4
6 2229.3 2197.7 1.4 956.7 968.6 1.3 783.9 794.1 1.3
5 2494.3 2476.9 0.7 1085.4 1090.9 0.5 888.7 894.0 0.6
4 2753.6 2725.2 1.0 1194.2 1203.0 0.7 977.7 985.7 0.8
3 2995.8 2968.8 0.9 1304.3 1312.8 0.7 1068.1 1075.9 0.7
2 3196.0 3159.2 1.2 1382.3 1395.4 1.0 1133.1 1144.6 1.0
1 3306.5 3272.5 1.0 1272.0 1284.3 1.0 1050.3 1061.1 1.0
Frame-B1 Frame-B16
Perturbed Average CM %error Perturbed Average CM %error Perturbed Average CM %error Perturbed Average CM %error
(11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
400.2 401.8 0.4 583.4 577.5 1.0 440.6 441.6 0.2 323.4 327.5 1.3
703.9 729.7 3.7 1057.3 1093.3 3.4 854.2 831.0 2.7 636.2 610.9 4.0
985.7 1007.1 2.2 1461.7 1483.1 1.5 1145.9 1130.9 1.3 860.6 846.2 1.7
1231.7 1230.4 0.1 1785.3 1783.5 0.1 1371.1 1367.3 0.3 1032.7 1033.6 0.1
1439.1 1438.7 0.0 2042.1 2059.8 0.9 1592.4 1588.0 0.3 1226.5 1206.6 1.6
1638.8 1619.1 1.2 2295.8 2302.1 0.3 1776.5 1780.4 0.2 1368.5 1357.4 0.8
1798.4 1780.1 1.0 2507.7 2525.5 0.7 1950.2 1954.4 0.2 1516.4 1494.8 1.4
1974.9 1940.7 1.7 2774.2 2760.5 0.5 2117.7 2133.8 0.8 1631.3 1634.9 0.2
2088.0 2063.2 1.2 2965.6 2963.4 0.1 2276.9 2285.6 0.4 1747.2 1744.4 0.2
example-building, also referred to as the perturbed model, is
then analyzed using SAP2000. Resulting shear forces shared by
the frames A1, A11, A14, B1 and B16 due to excitation along
X-direction are presented in Table 4 (columns (2), (5), (8),
(11)and (17), respectively). For frames B1 and B16, induced
shear forces due to Y-direction excitation are also presented in
Table 4 (columns (14) and (20), respectively). Similar responses
for the average CM model calculated using SAP2000 are
again presented in Table 4 for ready reference and comparison.
Finally, frame shear forces calculated for the average CM
model using the proposed method are compared with that
calculated for the perturbed model using SAP2000 and the
resulting relative errors are presented in Table 5 for the selected
frames.
Table 5
Error in the calculation of frame shear force when the proposed method is used for the building with perturbed CM (example-2) (compared with the results of
SAP2000)
Floor Frame-A1 Frame-A11 Frame-A14 Frame-B1 Frame-B16
X-direction excitation X-direction excitation Y-direction excitation X-direction excitation Y-direction excitation
(1) (2) (3) (4) (5) (6) (7) (8)
10 0.9 6.8 7.3 4.6 3.8 8.5 4.6
9 1.2 3.1 3.5 6.2 0.6 8.2 3.2
8 1.2 2.6 3.0 4.6 1.7 5.6 4.4
7 2.7 3.4 3.6 2.2 2.6 3.7 5.3
6 2.7 2.9 3.1 2.3 1.2 2.9 3.0
5 3.5 3.1 3.2 1.3 1.5 1.8 3.6
4 3.4 2.4 2.5 1.8 0.9 1.3 2.9
3 3.7 2.2 2.3 1.5 2.0 0.2 4.8
2 3.6 2.1 2.4 2.5 1.9 0.4 5.3
1 3.4 5.3 6.2 2.3 4.1 1.3 5.4
6.0. Numerical example-3: Validation of the range of
applicability
Two example-problems are furnished for this purpose,
namely, examples-3a and 3b. Both the example-buildings differ
from that considered in example-1 only in their locations of the
oor CMs. Offsets in location of CMs from that considered in
example-1 are shown in Table 3 (columns 47). Moreover, the
location of the CMs for both the examples are chosen in such a
way that example-1 nearly represents the average CM model
in either case. Considering frame A1 as the characteristic frame,
x
and
y
are then calculated (Eqs. (23a) and (23b)) as 1.527
10
3
and 1.288 10
3
, respectively, for the example-3a and
2.35710
2
and 2.42310
2
, respectively, for the example-
3b. Clearly, the simplied method should be acceptable for the
example-3a but not for the example-3b. This will be further
veried by the analysis using SAP2000.
A. Example-3a: Scattering in location of the CMS within the
acceptable limit
When the excitation is considered along the X-direction,
shear forces shared by the frames A1, A11, A14, B1 and
B16 are presented in Table 6 (columns (2), (5), (8), (11) and
(17), respectively). Further, when the excitation is considered
along the Y-direction, induced shear forces for the frames
B1 and B16 are also presented in Table 6 (columns (14)
and (20), respectively). Moreover, rst three mass normalized
mode shapes are presented in Table 7 (columns 5, 9 and 13,
respectively).
B. Example-3b: Scattering in location of the CMS beyond
the acceptable limit
Frequency and, MMPR associated with the excitation along
the X- and Y-directions, are presented in Table 8 (columns
(2), (4) and (5), respectively) for the rst 15 modes. When
the excitation is considered along the X-direction, shear forces
shared by the frames A1, A11, A14, B1 and B16 are presented
in Table 9 (columns (2), (5), (8), (11) and (17), respectively).
For the frames B1 and B16, induced shear force due to the
excitation along the Y-direction are also presented in Table 9
(columns (14) and (20), respectively). Moreover, rst three
mass normalized mode shapes are presented in Table 10
(columns 5, 9 and 13, respectively).
For the purpose of direct comparison, results of the average
CM model (example-1) are again presented in Tables 610.
7.0. Results and discussions
In the case of building with vertically-aligned CMs
(example-1), close agreement in frequencies and MMPRs
between the simplied analysis and SAP2000 may be seen
(Table 1). It may also be noted that the simplied analysis
yields consistent modal response irrespective of the selection
of characteristic frame as A1 or B1 (Table 1). However, the
stiffest frame may always be preferred for the selection of the
characteristic frame. Further, comparing the simplied analysis
with SAP2000 for the frame shear forces, maximum error is
noted as 9%. More precisely, as the design force resultant
often governs at the ground-story level, the error at that level
may be a better option for such comparison, which is noted as
7%.
In the case of example-2, scattering in location of CMs is
such that
x
= 7.6 10
4
and
y
= 1.29 10
3
. Modal
responses calculated for the perturbed and average CM model
using SAP2000 are found well in agreement with each other;
these results are, however, not presented for brevity. When the
frame shear forces are compared, maximum error of 4 % is
noted for the upper-story levels while only 1% is observed
at the base level (Table 4). Since, SAP2000 is used for both
the models, these comparisons truly reect the effect of the
scattering of CMs irrespective of the proportionality of the
lateral stiffness matrices of the constituting MRFs. Moreover,
while applying the simplied analysis to the perturbed model
overall error in the estimation of frame shear force is noted
as 7% for the upper-story level and 6% for the base level
when compared to the response of the perturbed model using
SAP2000 (Table 5). It may be noted that, this error includes the
effect of nonproportionality of lateral stiffness matrices of the
constituting MRFs also.
Validation of the limit predicted by the perturbation
formulation is substantiated through examples-3a and 3b.
Example-3a is considered to furnish the case wherein scattering
in location of the CMs remains within the prescribed limit; on
the other hand, example-3b represents the case wherein the
Table 6
Comparison of frame shear force of the perturbed and the average CM model calculated using SAP2000-examples-3a and -1
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
10 618.3 598.5 3.2 247.1 262.8 6.3 203 215.7 6.3
9 1172 1112 5.1 457.2 488.6 6.9 375 401.4 7
8 1618.5 1535.9 5.1 638 679.7 6.5 523.3 557.8 6.6
7 1952.4 1877.2 3.9 794.1 830.6 4.6 650.6 681.2 4.7
6 2288.3 2197.7 4 929.8 968.6 4.2 761.4 794.1 4.3
5 2560.7 2476.9 3.3 1058.2 1090.8 3.1 865.8 894 3.3
4 2825.6 2725.2 3.6 1164.7 1203 3.3 953 985.7 3.4
3 3076.3 2968.8 3.5 1271.9 1312.8 3.2 1040.9 1075.9 3.4
2 3281.8 3159.2 3.7 1345.9 1395.4 3.7 1102.5 1144.6 3.8
1 3390.4 3272.5 3.5 1237.5 1284.3 3.8 1021.6 1061.1 3.9
Frame-B1 Frame-B16
(11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
393.9 401.8 2 588.8 577.5 1.9 436.8 441.6 1.1 324.4 327.5 1
679 729.7 7.5 1024.3 1093.3 6.7 885.1 831 6.1 657.9 610.9 7.1
963.8 1007.1 4.5 1446.1 1483.1 2.6 1169.7 1130.9 3.3 870.1 846.2 2.7
1226.5 1230.4 0.3 1796.1 1783.5 0.7 1380.6 1367.2 1 1023.2 1033.6 1
1432.5 1438.7 0.4 2037.5 2059.8 1.1 1603.8 1588 1 1234.9 1206.6 2.3
1649.6 1619.1 1.8 2302 2302.1 0 1776.3 1780.4 0.2 1366 1357.3 0.6
1807 1780.1 1.5 2506.1 2525.5 0.8 1952.6 1954.4 0.1 1519 1494.8 1.6
2000.3 1940.7 3 2805.4 2760.5 1.6 2110.3 2133.8 1.1 1608.6 1634.9 1.6
prescribed limit is exceeded. Clearly, example-2 would also
sufce the purpose of example-3a. Nevertheless, example-3a is
considered, as the
x
in example-2 is one order less than the
prescribed limit.
In the case of example-3a,
x
= 1.527 10
3
and
y
=
1.288 10
3
, and hence, = 3. Therefore, maximum
expected difference in eigenvalues (reciprocal of the square
of frequency in this paper) between the perturbed (example-
3a) and average CM (example-1) models should be of the
order of 10
3
, especially, in the rst three modes. Similarly,
maximum expected difference in the mass normalized mode
shapes, especially, in the rst three modes should be of the order
of 10
2
times the mass normalized rst characteristic mode
shape. Further, maximum expected error in the estimation of
frame shear force (as derived in Appendix) should be of the
order of 10
0
, i.e., 0%10%.
Comparison of modal responses, e.g., frequency, MMPR
etc., shows close agreement between the perturbed and
average CM model; these results are not presented for
brevity. However, rst three mass normalized mode shapes
are compared numerically and graphically. First, comparing
columns 4, 8, and 12 with columns 5, 9 and 13, respectively,
of Table 7, it may be seen that rst three mass normalized
mode shapes do match well in perturbed and average CM
models. Moreover, order of their difference (columns 7, 11 and
15, Table 7) is also well below the predicted maximum limit
except at a few places indicated in italic font. On the other
hand, Figs. 24 show a graphical comparison for the rst three
mass normalized mode shapes, respectively. It may be noted,
as evident from the modal mass participation ratio in example-
1, rst mode shape is predominantly along Y-direction while
second and third mode shapes are dominant in X-direction.
Therefore, the small jaggedness observed in the Y-component
of the second and third mode shapes (Figs. 3(b) and 4(b)),
respectively) does not contribute much to the overall response.
Next, columns (4), (7), (10), (13), (16), (19) and (22) of Table 6
show that the errors in the estimation of the frame shear forces
for the selected frames are also well within the maximum
predicted limit of 10%; based on ground-story level, maximum
error in the estimation of frame shear force may be considered
as around 4%.
In order to assess the conservativeness in the limit of
scattering predicted by the perturbation formulation (Eqs.
(23a) and (23b)), example-3b is considered. Here,
x
=
2.357 10
2
,
y
= 2.423 10
2
and hence, =
2. Clearly, the proposed method should not be applied to
this building. Nevertheless, if applied (i) maximum expected
difference in eigenvalues between the perturbed (example-
3b) and average CM (example-1) models should be of the
order 10
2
, especially, in the rst three modes, (ii) maximum
expected difference in the mass normalized mode shapes,
especially, in the rst three modes should be of the order 10
1
times the mass normalized rst characteristic mode shape and
(iii) maximum expected error in the estimation of the frame
shear force should be of the order 10
1
, i.e., 10%100%.
From columns (4), (7), (10), (13), (16), (19) and (22) of
Table 9, maximum error in the estimation of the frame shear
force may be noted to be as much as about 50%. This is in
compliance with what is predicted by the perturbation analysis.
Clearly, the proposed method cannot be applied to example-3b.
T
a
b
l
e
7
E
r
r
o
r
i
n
m
o
d
e
s
h
a
p
e
s
o
n
u
s
i
n
g
a
v
e
r
a
g
e
c
e
n
t
e
r
o
f
m
a
s
s
m
o
d
e
l
a
n
d
p
e
r
t
u
r
b
e
d
m
o
d
e
l
e
x
a
m
p
l
e
s
-
3
a
a
n
d
-
1
D
i
r
e
c
t
i
o
n
F
l
o
o
r
l
e
v
e
l
C
h
a
r
a
c
t
e
r
i
s
t
i
c
M
o
d
e
-
1
M
o
d
e
-
2
M
o
d
e
-
3
A
v
e
r
a
g
e
P
e
r
t
u
r
b
e
d
E
r
r
o
r
o
r
d
e
r
A
v
e
r
a
g
e
P
e
r
t
u
r
b
e
d
E
r
r
o
r
o
r
d
e
r
A
v
e
r
a
g
e
P
e
r
t
u
r
b
e
d
E
r
r
o
r
o
r
d
e
r
1
2
3
4
5
6
=
(
5
4
)
/
3
7
8
9
1
0
1
1
1
2
1
3
1
4
1
5
X
-
d
i
r
e
c
t
i
o
n
1
9
.
2
5
E
0
4
7
.
9
7
E
0
5
7
.
7
9
E
0
5
1
.
9
5
E
0
3
7
.
2
6
E
0
4
7
.
1
5
E
0
4
1
.
1
9
E
0
2
6
.
0
7
E
0
4
6
.
2
1
E
0
4
1
.
5
1
E
0
2
2
2
2
.
4
1
E
0
3
2
.
0
0
E
0
4
1
.
7
9
E
0
4
8
.
7
3
E
0
3
1
.
8
5
E
0
3
1
.
7
2
E
0
3
5
.
2
4
E
0
2
1
.
5
7
E
0
3
1
.
7
1
E
0
3
5
.
8
2
E
0
2
2
3
3
.
9
2
E
0
3
3
.
2
2
E
0
4
3
.
1
5
E
0
4
1
.
7
9
E
0
3
2
.
9
8
E
0
3
2
.
9
4
E
0
3
1
.
1
2
E
0
2
2
.
5
5
E
0
3
2
.
6
0
E
0
3
1
.
3
5
E
0
2
2
4
5
.
3
4
E
0
3
4
.
3
5
E
0
4
3
.
8
8
E
0
4
8
.
8
1
E
0
3
4
.
0
4
E
0
3
3
.
7
6
E
0
3
5
.
1
7
E
0
2
3
.
4
7
E
0
3
3
.
7
7
E
0
3
5
.
7
5
E
0
2
2
5
6
.
6
1
E
0
3
5
.
3
8
E
0
4
5
.
5
2
E
0
4
2
.
1
2
E
0
3
5
.
0
0
E
0
3
5
.
1
1
E
0
3
1
.
6
2
E
0
2
4
.
2
9
E
0
3
4
.
1
9
E
0
3
1
.
6
3
E
0
2
2
6
7
.
7
2
E
0
3
6
.
2
6
E
0
4
5
.
7
3
E
0
4
6
.
8
7
E
0
3
5
.
8
3
E
0
3
5
.
5
3
E
0
3
3
.
8
4
E
0
2
5
.
0
1
E
0
3
5
.
3
4
E
0
3
4
.
3
0
E
0
2
2
7
8
.
6
2
E
0
3
6
.
9
9
E
0
4
7
.
1
2
E
0
4
1
.
5
1
E
0
3
6
.
5
1
E
0
3
6
.
6
4
E
0
3
1
.
5
5
E
0
2
5
.
5
9
E
0
3
5
.
4
6
E
0
3
1
.
5
7
E
0
2
2
8
9
.
3
1
E
0
3
7
.
5
6
E
0
4
6
.
8
8
E
0
4
7
.
3
0
E
0
3
7
.
0
4
E
0
3
6
.
6
7
E
0
3
3
.
9
5
E
0
2
6
.
0
4
E
0
3
6
.
4
6
E
0
3
4
.
4
4
E
0
2
2
9
9
.
7
9
E
0
3
7
.
9
4
E
0
4
7
.
0
1
E
0
4
9
.
5
0
E
0
3
7
.
4
0
E
0
3
6
.
8
8
E
0
3
5
.
3
6
E
0
2
6
.
3
5
E
0
3
6
.
9
4
E
0
3
5
.
9
9
E
0
2
2
1
0
1
.
0
0
E
0
2
8
.
1
6
E
0
4
7
.
6
0
E
0
4
5
.
5
7
E
0
3
7
.
6
1
E
0
3
7
.
3
4
E
0
3
2
.
6
6
E
0
2
6
.
5
2
E
0
3
6
.
8
3
E
0
3
3
.
0
5
E
0
2
2
Y
-
d
i
r
e
c
t
i
o
n
1
9
.
2
5
E
0
4
9
.
0
3
E
0
4
8
.
9
5
E
0
4
8
.
6
5
E
0
3
3
1
.
1
2
E
0
4
1
.
5
1
E
0
4
4
.
2
2
E
0
2
1
.
8
6
E
0
5
5
.
9
0
E
0
5
4
.
3
7
E
0
2
2
2
2
.
4
1
E
0
3
2
.
3
6
E
0
3
2
.
3
3
E
0
3
1
.
2
5
E
0
2
2
3
.
0
4
E
0
4
4
.
6
7
E
0
4
6
.
7
7
E
0
2
5
.
9
0
E
0
5
2
.
3
4
E
0
4
7
.
2
7
E
0
2
2
3
3
.
9
2
E
0
3
3
.
8
5
E
0
3
3
.
9
5
E
0
3
2
.
5
3
E
0
2
2
5
.
0
4
E
0
4
8
.
1
0
E
0
5
1
.
4
9
E
0
1
1
.
0
5
E
0
4
5
.
3
2
E
0
4
1
.
6
3
E
0
1
1
4
5
.
3
4
E
0
3
5
.
2
6
E
0
3
5
.
2
4
E
0
3
2
.
8
1
E
0
3
3
6
.
9
4
E
0
4
7
.
7
2
E
0
4
1
.
4
6
E
0
2
1
.
5
0
E
0
4
2
.
3
4
E
0
4
1
.
5
7
E
0
2
2
5
6
.
6
1
E
0
3
6
.
5
3
E
0
3
6
.
6
3
E
0
3
1
.
4
8
E
0
2
2
8
.
6
7
E
0
4
2
.
5
6
E
0
4
9
.
2
4
E
0
2
1
.
9
2
E
0
4
4
.
7
3
E
0
4
1
.
0
1
E
0
1
1
6
7
.
7
2
E
0
3
7
.
6
3
E
0
3
7
.
5
7
E
0
3
7
.
3
9
E
0
3
3
1
.
0
2
E
0
3
1
.
3
6
E
0
3
4
.
4
7
E
0
2
2
.
3
0
E
0
4
6
.
1
0
E
0
4
4
.
9
2
E
0
2
2
7
8
.
6
2
E
0
3
8
.
5
4
E
0
3
8
.
6
5
E
0
3
1
.
3
7
E
0
2
2
1
.
1
4
E
0
3
3
.
7
2
E
0
4
8
.
9
3
E
0
2
2
.
6
1
E
0
4
5
.
7
3
E
0
4
9
.
6
8
E
0
2
2
8
9
.
3
1
E
0
3
9
.
2
4
E
0
3
9
.
2
8
E
0
3
4
.
7
3
E
0
3
3
1
.
2
4
E
0
3
9
.
3
1
E
0
4
3
.
3
1
E
0
2
2
.
8
7
E
0
4
3
.
9
4
E
0
5
3
.
5
1
E
0
2
2
9
9
.
7
9
E
0
3
9
.
7
2
E
0
3
9
.
5
7
E
0
3
1
.
5
5
E
0
2
2
1
.
3
1
E
0
3
2
.
3
3
E
0
3
1
.
0
4
E
0
1
3
.
0
6
E
0
4
1
.
4
4
E
0
3
1
.
1
6
E
0
1
1
1
0
1
.
0
0
E
0
2
1
.
0
0
E
0
2
1
.
0
0
E
0
2
2
.
9
9
E
0
4
4
1
.
3
5
E
0
3
1
.
3
1
E
0
3
3
.
4
8
E
0
3
3
.
1
9
E
0
4
2
.
9
7
E
0
4
2
.
1
9
E
0
3
-
d
i
r
e
c
t
i
o
n
1
9
.
2
5
E
0
4
4
.
5
2
E
0
6
4
.
5
8
E
0
6
6
.
4
9
E
0
5
5
2
.
4
3
E
0
5
2
.
5
3
E
0
5
1
.
0
8
E
0
3
2
.
8
0
E
0
5
2
.
6
8
E
0
5
1
.
3
0
E
0
3
3
2
2
.
4
1
E
0
3
1
.
1
2
E
0
5
1
.
1
4
E
0
5
8
.
3
1
E
0
5
5
6
.
2
7
E
0
5
6
.
5
3
E
0
5
1
.
0
8
E
0
3
7
.
3
5
E
0
5
7
.
0
4
E
0
5
1
.
2
9
E
0
3
3
3
3
.
9
2
E
0
3
1
.
7
9
E
0
5
1
.
8
3
E
0
5
1
.
0
2
E
0
4
4
1
.
0
2
E
0
4
1
.
0
6
E
0
4
1
.
0
2
E
0
3
1
.
2
0
E
0
4
1
.
1
5
E
0
4
1
.
2
8
E
0
3
3
4
5
.
3
4
E
0
3
2
.
4
0
E
0
5
2
.
4
2
E
0
5
3
.
7
5
E
0
5
5
1
.
3
8
E
0
4
1
.
4
4
E
0
4
1
.
1
2
E
0
3
1
.
6
4
E
0
4
1
.
5
7
E
0
4
1
.
3
1
E
0
3
3
5
6
.
6
1
E
0
3
2
.
9
5
E
0
5
2
.
9
3
E
0
5
3
.
0
2
E
0
5
5
1
.
7
2
E
0
4
1
.
7
9
E
0
4
1
.
0
6
E
0
3
2
.
0
4
E
0
4
1
.
9
5
E
0
4
1
.
3
6
E
0
3
3
6
7
.
7
2
E
0
3
3
.
4
2
E
0
5
3
.
3
2
E
0
5
1
.
3
0
E
0
4
4
2
.
0
0
E
0
4
2
.
0
9
E
0
4
1
.
1
7
E
0
3
2
.
3
8
E
0
4
2
.
2
8
E
0
4
1
.
3
0
E
0
3
3
7
8
.
6
2
E
0
3
3
.
8
0
E
0
5
3
.
6
4
E
0
5
1
.
8
6
E
0
4
4
2
.
2
4
E
0
4
2
.
3
3
E
0
4
1
.
0
4
E
0
3
2
.
6
6
E
0
4
2
.
5
5
E
0
4
1
.
2
8
E
0
3
3
8
9
.
3
1
E
0
3
4
.
0
9
E
0
5
3
.
8
1
E
0
5
3
.
0
1
E
0
4
4
2
.
4
2
E
0
4
2
.
5
2
E
0
4
1
.
0
7
E
0
3
2
.
8
8
E
0
4
2
.
7
6
E
0
4
1
.
2
9
E
0
3
3
9
9
.
7
9
E
0
3
4
.
2
9
E
0
5
3
.
8
9
E
0
5
4
.
0
9
E
0
4
4
2
.
5
5
E
0
4
2
.
6
5
E
0
4
1
.
0
2
E
0
3
3
.
0
2
E
0
4
2
.
9
0
E
0
4
1
.
2
3
E
0
3
3
1
0
1
.
0
0
E
0
2
4
.
3
9
E
0
5
3
.
9
6
E
0
5
4
.
2
8
E
0
4
4
2
.
6
2
E
0
4
2
.
7
3
E
0
4
1
.
0
9
E
0
3
3
.
1
0
E
0
4
2
.
9
8
E
0
4
1
.
1
9
E
0
3
3
Table 8
Comparison of modal responses of perturbed and average CM model calculated using SAP2000examples-3b and -1
Mode number Frequency (rad/s) Modal mass participation ratio (%)
Perturbed model Average CM model Perturbed model Average CM model
X-direction Y-direction X-direction Y-direction
(1) (2) (3) (4) (5) (6) (7)
1 4.782 4.838 14.6 41.7 0.5 78.3
2 4.894 4.946 24.5 37.3 45.9 1.4
3 5.38 5.365 41 0.7 33.7 0.1
4 14.137 14.376 2.5 4.6 0.2 10.2
5 14.539 14.653 3 5.8 6.1 0.4
6 16.071 15.937 5 0.1 4.3 0.0
7 24.313 24.977 0.9 1.2 0.2 3.5
8 25.29 25.362 1 2.7 2.2 0.5
9 27.836 27.654 2.1 0.2 1.5 0.0
10 37.077 37.043 0 2.2 0.4 1.3
11 38.731 37.533 1.8 0 0.9 0.7
12 39.387 41.137 0.3 0 0.8 0.0
13 49.184 49.618 0.4 0.3 0.5 0.5
14 50.599 50.419 0.4 1 0.4 0.9
15 55.188 55.458 0.6 0.1 0.5 0.0
Table 9
Comparison of frame shear force of the perturbed and the average CM model calculated using SAP2000examples-3b and -1
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
10 662.9 598.5 9.7 187 262.8 40.5 156 215.7 38.3
9 1102 1112 0.9 358.8 488.6 36.2 297 401.4 35.1
8 1607.8 1535.9 4.5 467.9 679.7 45.3 389.3 557.8 43.3
7 2060.4 1877.2 8.9 555.8 830.6 49.4 464.4 681.2 46.7
6 2357.8 2197.7 6.8 656 968.6 47.7 547.8 794.1 45
5 2698.5 2476.9 8.2 725.8 1090.8 50.3 607.8 894 47.1
4 3005.2 2725.2 9.3 793.8 1203. 51.5 666.6 985.7 47.9
3 3227.6 2968.8 8 875.3 1312.8 50 734.4 1075.9 46.5
2 3437.9 3159.2 8.1 926.7 1395.4 50.6 779.2 1144.6 46.9
1 3557.1 3272.5 8 852.4 1284.3 50.7 725.3 1061.1 46.3
Frame-B1 Frame-B16
(11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
403.7 401.8 0.5 477.9 577.5 20.8 418.1 441.6 5.6 413.7 327.5 20.8
772.3 729.7 5.5 993 1093.3 10.1 778.7 831 6.7 794.7 610.9 23.1
1082.4 1007.1 7 1375.9 1483.1 7.8 1050.6 1130.9 7.6 1005.9 846.2 15.9
1266.5 1230.4 2.9 1579.9 1783.5 12.9 1355.6 1367.2 0.9 1237.1 1033.6 16.5
1486.2 1438.7 3.2 1912.1 2059.8 7.7 1545.4 1588. 2.8 1417.8 1206.6 14.9
1650.5 1619.1 1.9 2113.6 2302.1 8.9 1757.7 1780.4 1.3 1576.8 1357.3 13.9
1789.4 1780.1 0.5 2278.2 2525.5 10.9 1961.1 1954.4 0.3 1740 1494.8 14.1
1977.6 1940.7 1.9 2538.8 2760.5 8.7 2113.1 2133.8 1 1892.9 1634.9 13.6
However, frequencies of the perturbed and average CM
models are still in close agreement with each other (columns
(2) and (3) of Table 8). On the other hand, MMPRs in both the
models are signicantly different (columns (4)(7) of Table 8).
Moreover, maximum difference in mode shapes for the rst
three modes is noted to be of the order of 10
1
times the mass
normalized rst characteristic mode shape (columns 7, 11 and
15 of Table 10), which is also predicted by the perturbation
analysis. Such a large variation in rst three mass normalized
mode shapes is also evident from the graphical comparison
presented in Figs. 57.
Therefore, these results show that upper bound of the order
of expected error in the estimation of the eigenvalues, mode
shapes and frame shear forces predicted by the perturbation
analysis is accurate enough. Thus, the proposed conditions
(Eqs. (23a) and (23b)) may be used to assess the applicability of
the simplied analysis wherein the conditions of special class
are not truly met. However, it may be noted that the perturbed
T
a
b
l
e
1
0
E
r
r
o
r
i
n
m
o
d
e
s
h
a
p
e
s
o
n
u
s
i
n
g
a
v
e
r
a
g
e
c
e
n
t
e
r
o
f
m
a
s
s
m
o
d
e
l
a
n
d
p
e
r
t
u
r
b
e
d
m
o
d
e
l
e
x
a
m
p
l
e
s
-
3
b
a
n
d
-
1
D
i
r
e
c
t
i
o
n
F
l
o
o
r
l
e
v
e
l
C
h
a
r
a
c
t
e
r
i
s
t
i
c
M
o
d
e
-
1
M
o
d
e
-
2
M
o
d
e
-
3
A
v
e
r
a
g
e
P
e
r
t
u
r
b
e
d
E
r
r
o
r
o
r
d
e
r
A
v
e
r
a
g
e
P
e
r
t
u
r
b
e
d
E
r
r
o
r
o
r
d
e
r
A
v
e
r
a
g
e
P
e
r
t
u
r
b
e
d
E
r
r
o
r
o
r
d
e
r
1
2
3
4
5
6
=
(
5
4
)
/
3
7
8
9
1
0
1
1
1
2
1
3
1
4
1
5
X
-
d
i
r
e
c
t
i
o
n
1
9
.
2
5
E
0
4
7
.
9
7
E
0
5
3
.
3
0
E
0
4
2
.
7
1
E
0
1
7
.
2
6
E
0
4
4
.
5
9
E
0
4
2
.
8
9
E
0
1
6
.
0
7
E
0
4
7
.
7
4
E
0
4
1
.
8
1
E
0
1
1
2
2
.
4
1
E
0
3
2
.
0
0
E
0
4
9
.
6
5
E
0
4
3
.
1
8
E
0
1
1
.
8
5
E
0
3
1
.
2
8
E
0
3
2
.
3
6
E
0
1
1
.
5
7
E
0
3
1
.
8
4
E
0
3
1
.
1
1
E
0
1
1
3
3
.
9
2
E
0
3
3
.
2
2
E
0
4
2
.
0
2
E
0
3
4
.
3
2
E
0
1
2
.
9
8
E
0
3
2
.
4
7
E
0
3
1
.
3
1
E
0
1
2
.
5
5
E
0
3
2
.
3
8
E
0
3
4
.
2
9
E
0
2
2
4
5
.
3
4
E
0
3
4
.
3
5
E
0
4
1
.
9
3
E
0
3
2
.
8
1
E
0
1
4
.
0
4
E
0
3
2
.
6
5
E
0
3
2
.
6
1
E
0
1
3
.
4
7
E
0
3
4
.
2
8
E
0
3
1
.
5
2
E
0
1
1
5
6
.
6
1
E
0
3
5
.
3
8
E
0
4
2
.
3
9
E
0
3
2
.
8
0
E
0
1
5
.
0
0
E
0
3
3
.
2
8
E
0
3
2
.
6
1
E
0
1
4
.
2
9
E
0
3
5
.
2
9
E
0
3
1
.
5
1
E
0
1
1
6
7
.
7
2
E
0
3
6
.
2
6
E
0
4
4
.
0
3
E
0
3
4
.
4
1
E
0
1
5
.
8
3
E
0
3
4
.
9
2
E
0
3
1
.
1
8
E
0
1
5
.
0
1
E
0
3
4
.
5
5
E
0
3
5
.
9
3
E
0
2
2
7
8
.
6
2
E
0
3
6
.
9
9
E
0
4
2
.
8
5
E
0
3
2
.
4
9
E
0
1
6
.
5
1
E
0
3
4
.
0
3
E
0
3
2
.
8
8
E
0
1
5
.
5
9
E
0
3
7
.
2
4
E
0
3
1
.
9
1
E
0
1
1
8
9
.
3
1
E
0
3
7
.
5
6
E
0
4
3
.
4
9
E
0
3
2
.
9
3
E
0
1
7
.
0
4
E
0
3
4
.
7
2
E
0
3
2
.
4
9
E
0
1
6
.
0
4
E
0
3
7
.
2
8
E
0
3
1
.
3
3
E
0
1
1
9
9
.
7
9
E
0
3
7
.
9
4
E
0
4
5
.
3
7
E
0
3
4
.
6
7
E
0
1
7
.
4
0
E
0
3
6
.
4
6
E
0
3
9
.
6
3
E
0
2
6
.
3
5
E
0
3
5
.
4
7
E
0
3
8
.
9
9
E
0
2
2
1
0
1
.
0
0
E
0
2
8
.
1
6
E
0
4
4
.
2
2
E
0
3
3
.
3
9
E
0
1
7
.
6
1
E
0
3
5
.
5
0
E
0
3
2
.
1
0
E
0
1
6
.
5
2
E
0
3
7
.
2
8
E
0
3
7
.
5
5
E
0
2
2
Y
-
d
i
r
e
c
t
i
o
n
1
9
.
2
5
E
0
4
9
.
0
3
E
0
4
5
.
3
3
E
0
4
4
.
0
0
E
0
1
1
1
.
1
2
E
0
4
7
.
3
0
E
0
4
6
.
6
8
E
0
1
1
.
8
6
E
0
5
2
.
4
6
E
0
4
2
.
4
6
E
0
1
1
2
2
.
4
1
E
0
3
2
.
3
6
E
0
3
1
.
7
8
E
0
3
2
.
4
1
E
0
1
1
3
.
0
4
E
0
4
1
.
5
8
E
0
3
5
.
2
9
E
0
1
5
.
9
0
E
0
5
1
.
5
5
E
0
4
3
.
9
9
E
0
2
2
3
3
.
9
2
E
0
3
3
.
8
5
E
0
3
3
.
1
5
E
0
3
1
.
7
9
E
0
1
1
5
.
0
4
E
0
4
2
.
3
6
E
0
3
4
.
7
4
E
0
1
1
.
0
5
E
0
4
6
.
3
3
E
0
5
4
.
3
0
E
0
2
2
4
5
.
3
4
E
0
3
5
.
2
6
E
0
3
3
.
4
4
E
0
3
3
.
4
1
E
0
1
1
6
.
9
4
E
0
4
3
.
9
8
E
0
3
6
.
1
6
E
0
1
1
.
5
0
E
0
4
1
.
0
4
E
0
3
1
.
6
7
E
0
1
1
5
6
.
6
1
E
0
3
6
.
5
3
E
0
3
4
.
4
8
E
0
3
3
.
1
0
E
0
1
1
8
.
6
7
E
0
4
4
.
7
6
E
0
3
5
.
8
8
E
0
1
1
.
9
2
E
0
4
1
.
0
1
E
0
3
1
.
2
4
E
0
1
1
6
7
.
7
2
E
0
3
7
.
6
3
E
0
3
6
.
2
2
E
0
3
1
.
8
2
E
0
1
1
1
.
0
2
E
0
3
4
.
6
9
E
0
3
4
.
7
6
E
0
1
2
.
3
0
E
0
4
1
.
0
8
E
0
4
4
.
3
8
E
0
2
2
7
8
.
6
2
E
0
3
8
.
5
4
E
0
3
5
.
2
1
E
0
3
3
.
8
6
E
0
1
1
1
.
1
4
E
0
3
6
.
7
9
E
0
3
6
.
5
5
E
0
1
2
.
6
1
E
0
4
2
.
1
6
E
0
3
2
.
2
0
E
0
1
1
8
9
.
3
1
E
0
3
9
.
2
4
E
0
3
7
.
2
4
E
0
3
2
.
1
4
E
0
1
1
1
.
2
4
E
0
3
5
.
9
4
E
0
3
5
.
0
4
E
0
1
2
.
8
7
E
0
4
2
.
5
5
E
0
4
3
.
4
4
E
0
3
3
9
9
.
7
9
E
0
3
9
.
7
2
E
0
3
7
.
4
1
E
0
3
2
.
3
6
E
0
1
1
1
.
3
1
E
0
3
6
.
4
3
E
0
3
5
.
2
3
E
0
1
3
.
0
6
E
0
4
5
.
3
9
E
0
4
2
.
3
8
E
0
2
2
1
0
1
.
0
0
E
0
2
1
.
0
0
E
0
2
7
.
1
9
E
0
3
2
.
7
9
E
0
1
1
1
.
3
5
E
0
3
7
.
0
0
E
0
3
5
.
6
2
E
0
1
3
.
1
9
E
0
4
1
.
1
2
E
0
3
7
.
9
2
E
0
2
-
d
i
r
e
c
t
i
o
n
1
9
.
2
5
E
0
4
4
.
5
2
E
0
6
2
.
0
1
E
0
5
1
.
6
8
E
0
2
2
2
.
4
3
E
0
5
1
.
7
3
E
0
5
7
.
5
7
E
0
3
2
.
8
0
E
0
5
2
.
5
2
E
0
5
3
.
0
3
E
0
3
3
2
2
.
4
1
E
0
3
1
.
1
2
E
0
5
5
.
1
7
E
0
5
1
.
6
8
E
0
2
2
6
.
2
7
E
0
5
4
.
5
0
E
0
5
7
.
3
6
E
0
3
7
.
3
5
E
0
5
6
.
6
3
E
0
5
2
.
9
9
E
0
3
3
3
3
.
9
2
E
0
3
1
.
7
9
E
0
5
8
.
3
8
E
0
5
1
.
6
8
E
0
2
2
1
.
0
2
E
0
4
7
.
3
4
E
0
5
7
.
3
0
E
0
3
1
.
2
0
E
0
4
1
.
0
9
E
0
4
2
.
8
1
E
0
3
3
4
5
.
3
4
E
0
3
2
.
4
0
E
0
5
1
.
1
4
E
0
4
1
.
6
9
E
0
2
2
1
.
3
8
E
0
4
1
.
0
0
E
0
4
7
.
1
2
E
0
3
1
.
6
4
E
0
4
1
.
4
8
E
0
4
3
.
0
0
E
0
3
3
5
6
.
6
1
E
0
3
2
.
9
5
E
0
5
1
.
4
1
E
0
4
1
.
6
9
E
0
2
2
1
.
7
2
E
0
4
1
.
2
4
E
0
4
7
.
2
6
E
0
3
2
.
0
4
E
0
4
1
.
8
4
E
0
4
3
.
0
2
E
0
3
3
6
7
.
7
2
E
0
3
3
.
4
2
E
0
5
1
.
6
5
E
0
4
1
.
6
9
E
0
2
2
2
.
0
0
E
0
4
1
.
4
5
E
0
4
7
.
1
3
E
0
3
2
.
3
8
E
0
4
2
.
1
5
E
0
4
2
.
9
8
E
0
3
3
7
8
.
6
2
E
0
3
3
.
8
0
E
0
5
1
.
8
4
E
0
4
1
.
6
9
E
0
2
2
2
.
2
4
E
0
4
1
.
6
3
E
0
4
7
.
0
8
E
0
3
2
.
6
6
E
0
4
2
.
4
0
E
0
4
3
.
0
2
E
0
3
3
8
9
.
3
1
E
0
3
4
.
0
9
E
0
5
2
.
0
0
E
0
4
1
.
7
1
E
0
2
2
2
.
4
2
E
0
4
1
.
7
6
E
0
4
7
.
0
9
E
0
3
2
.
8
8
E
0
4
2
.
5
8
E
0
4
3
.
2
2
E
0
3
3
9
9
.
7
9
E
0
3
4
.
2
9
E
0
5
2
.
1
0
E
0
4
1
.
7
1
E
0
2
2
2
.
5
5
E
0
4
1
.
8
6
E
0
4
7
.
0
5
E
0
3
3
.
0
2
E
0
4
2
.
7
0
E
0
4
3
.
2
7
E
0
3
3
1
0
1
.
0
0
E
0
2
4
.
3
9
E
0
5
2
.
1
6
E
0
4
1
.
7
1
E
0
2
2
2
.
6
2
E
0
4
1
.
9
1
E
0
4
7
.
0
7
E
0
3
3
.
1
0
E
0
4
2
.
7
7
E
0
4
3
.
2
8
E
0
3
3
Fig. 2. Comparison of rst mode shape on using the average center of mass
model and the perturbed modelexamples-3a and -1; (a) X-direction, (b) Y-
direction and (c) theta-direction.
models are used as the basis of estimation of error in frame
shear forces in the example-problems (Tables 6 and 9) while
the average CM model is used as the basis for the same in the
formulation of perturbation analysis.
It may be noted that the shear wall is not included in
the example-buildings. This is because the simplied analysis
assumes the stiffness proportionality of the constituting MRFs.
This implies that if one frame is predominantly deecting
in a exural mode, other frames will also follow the similar
pattern. This does not happen when only a few selected frames
contain squat shear walls, which force such frames to deect
in a shear type pattern and frameshear wall interaction comes
into the picture. In such cases, simplied procedure does not
work. However, if the constituting shear walls are of slender
type such that, even in their presence, constituting frames
continue to deect in a exural pattern, similar to the frames not
containing shear walls, simplied procedure and the subsequent
perturbation analysis are still applicable. This has been veried
but not included in this paper. However, inclusion of such
shear walls requires some more studies and thus, may be
excluded from the scope of this paper, and will be reported
separately.
Moreover, one may argue about the randomly scattered CMs
selected at different oors in a proportionally framed structural
arrangement without shear walls, as evident from the example-
problems. However, it may be noted that the scattering assumed
in the example-problems is to illustrate the limit of applicability
Fig. 3. Comparison of second mode shape on using the average center of mass
of the proposed procedure and does not necessarily reect the
true scattering available in the actual buildings. Further, when
the degree of scattering is considered, absence of shear walls
may not play a signicant role as the presence of inll masonry
walls nearly compensates it. Also, it is a common practice,
unlike the mass, not to rely upon the stiffness provided by the
inll masonry walls, which may not be available after one or
two cycles of seismic excitation.
Further, this paper deals with the elastic response spectrum
analysis of the MRF-buildings with horizontal setbacks. In
contrast, the MRF-buildings should be designed and detailed
for their ductile behavior because of the possible inelastic
excursion under strong shaking. Consequently, one may argue
about the use of elastic analysis. However, it may be noted that
elastic analysis has its own importance. First, MRF-building
should be designed such that it can sustain minor shaking
without any damage of even the nonstructural components.
Hence, an elastic analysis is required to ensure the necessary
elastic limit of the building. On the other hand, ductile design of
the MRF-buildings following the most seismic codes requires
an elastic (static or dynamic, depending upon the order of mass
and stiffness irregularity) analysis based on the gross cross-
sectional properties of the elements. Member force resultant
thus calculated is then reduced by suitable response reduction
factor to ascertain the design member force resultant. Capacity
design principle is then adopted to suppress the undesirable
shear failure modes and to detail the constituting members
Fig. 4. Comparison of third mode shape on using the average center of mass
so as to meet the seismic demand. Therefore, following most
seismic codes, it is seen that only elastic analysis is required
for the preliminary design of MRF-buildings. Further, as long
as the dynamic analysis is limited within the elastic range (as
ductility is usually incorporated by virtue of code specied
response reduction factor), elastic time history analysis
may be performed with modal decomposition. Therefore, the
proposed simplied method is equally applicable to modal time
history analysis, which however, has not been illustrated. Since,
formulation for perturbation analysis is based on mode-by-
mode decomposition, nal results are equally applicable to time
history analysis also.
It may also be noted that the effect of higher modes
are neglected in perturbation analysis for its simplicity but
not in the simplied analysis. Further, simplied method
does not work in the presence of vertical setbacks. This, in
some sense, may also limit the applicability of the proposed
simplied method. However, in practice, buildings with only
horizontal setbacks are also seen in quite large numbers.
Finally, detailed nonlinear 3D FEM analysis of special class
of torsionally coupled buildings with horizontal offsets is no
longer impossible in the present era of computer advancement.
Consequently, one may argue against the use of the proposed
simplied method. However, a simplied method is always
preferred to assess the design force resultants at the preliminary
stage of design; detailed nonlinear analysis may be carried out
Fig. 5. Comparison of rst mode shape on using the average center of mass
model and the perturbed modelexamples-3b and -1; (a) X-direction, (b) Y-
at the nal stage of verication. Therefore, the proposed method
can be applied as a convenient and offhand tool with sufcient
accuracy at the preliminary stage of design and requires only
the execution of plane frame analysis with a personal computer.
8.0. Summary and conclusions
Simplied method of analysis is proposed for the special
class of torsionally coupled buildings with horizontal setbacks.
The method can also be applied to the cases wherein all the
classication criteria for the special class of buildings are
not strictly satised. Range of applicability of the simplied
analysis with an acceptable margin of error is also proposed.
Acknowledgement
This paper is published with the approval of the Director,
Structural Engineering Research Center, Chennai, India.
Appendix. Limit of scattering in location of the oor CMS
using perturbation analysis
A.1. Formulation of error equations for eigenvalues and
eigenvectors
Here, the objective is to ascertain the error in computation
of eigenvalues and eigenvectors if system-B is used in lieu
Fig. 6. Comparison of second mode shape on using the average center of mass
of system-A. For this purpose, system-C is considered as the
unperturbed system and the error in using system-C in lieu of
system-A is evaluated. Next, the error in using system-C in lieu
of system-B is evaluated. Difference between these two errors
may be considered as the error in using system-B instead of
using system-A.
A.1.1. Case-1:System-A and system-C
Eigen equations of system-C and system-A may be
expressed as follows:
0
[K] {} = [M]
C
{} (A.1)
A
[K] {
A
} = [M]
A
{
A
} (A.2)
where, [K] is the stiffness matrix, [M]
C
, [M]
A
the mass
matrices,
0
,
A
the eigenvalues, and ,
A
the eigenvectors of
system-C and system-A, respectively. Perturbation in the mass
matrix is given by
[E]
A
=
_
_
_
0 0
__
b
y
i
+e
y
_
m
i
_
0 0
__
b
x
i
+e
x
_
m
i
_
__
b
y
i
+e
y
_
m
i
_ __
b
x
i
+e
x
_
m
i
_
[m
i
d
2
i
]
_
_.
(A.3)
It may be noted that the unperturbed system has two distinct
types of eigenvectors, e.g., pure translational (uncoupled from
Fig. 7. Comparison of third mode shape on using the average center of mass
torsional rotation but two translations along any two orthogonal
lateral directions may be coupled) and pure torsional; 2N
number of pure translational and N number of pure torsional
eigenvectors constitute all together 3N number of eigenvectors.
Let
0i
be the i th eigenvector of the unperturbed system. If
0i
is pure translational
{
0i
} = {
T L
0i
}
T
= {
x
T
0i

y
T
0i
0
T
}. (A.4)
On the other hand, if
0i
is a pure torsional eigenvector
{
0i
} = {
T R
0i
}
T
=
_
0
T
0
T

T
0i
_
. (A.5)
Let us now evaluate the expression
T
0i
E
A
0i
. In the case of a
pure translational eigenvector, utilizing Eqs. (A.3) and (A.4), it
may be shown that
T L
T
0i
E
A
T L
0i
= 0. (A.6)
Similarly, in the case of pure torsional eigenvector, utilizing
Eqs. (A.3) and (A.5), in conjunction with mass normalization
criteria, and noting that (d
i
/ r)
2
is of the order 10
1
to 10
3
(Eq. (22) of the main text), it may be shown that
T R
T
0i
E
A
T R
0i
is of the order 10
1
to 10
2
and hence, may be approximated as
zero. Therefore, it is seen that, irrespective of pure translational
and pure torsional eigenvectors,
T
0i
E
A
0i
= 0. (A.7)
Now using Eq. (A.7) in the expressions of eigenvalue and
eigenvector (Eqs. (13) and (14) in the main text), the following
simplied formulae can be derived:
Eigenvalues:
i
=
0i
+
n
j =1
j =i
_

0i
0 j
0i

0 j
_
_
T
0i
E
A
0 j
_
2
. (A.8)
Eigenvectors:
i
=
0i
+
n
j =1
j =i
_

0 j
0i

0 j
_
_
T
0 j
E
A
0i
_
0 j
. (A.9)
Let us now evaluate the expression
T
0i
E
A
0 j
as follows:
(i) When
0i
and
0 j
are two different pure translational
eigenvectors: It may be shown that
T L
T
0i
E
A
T L
0 j
= 0. (A.10)
(ii) When
0i
is a pure translational eigenvector and
0 j
is
a pure torsional eigenvector:
T L
T
0i
E
A
T R
0 j
=

x
T
0i
E
A
13

0 j
+

y
T
0i
E
A
23

0 j
. (A.11)
(iii) When
0i
is a pure torsional eigenvector and
0 j
is a pure
translational eigenvector:
T R
T
0i
E
A
T L
0 j
=

T
0i
E
A
13

x
0 j
+

T
0i
E
A
23

y
0 j
. (A.12)
(iv) When
0i
and
0 j
are two different pure torsional
eigenvectors:
T R
T
0i
E
A
T R
0 j
=

T
0i
E
A
33

0 j
. (A.13)
Further, using the orthogonality criteria and noting that (d
i
/ r)
2
is of the order 10
1
to 10
3
(Eq. (22) of the main text), it may
be shown that
T R
T
0i
E
A
T R
0 j
0. (A.14)
Estimation of eigenvalue of the perturbed system (
i
) by
expanding Eq. (A.8)
(i) Expanding Eq. (A.8) and thereafter utilizing Eqs.
(A.10) and (A.11),
A
i
associated with any pure translational
eigenvector of the unperturbed system is given by
A
i
=
T L
0i
+
N
j =1
_

T L
0i

T R
0 j
T L
0i

T R
0 j
_
x
T
0i
E
A
13

0 j
+

y
T
0i
E
A
23

0 j
_
2
. (A.15)
(ii) Expanding Eq. (A.8) and thereafter utilizing Eqs. (A.12)
and (A.14),
A
i
associated with any pure torsional eigenvector
of the unperturbed system is given by
A
i
=
T R
0i
+
2N
j =1
_

T R
0i

T L
0 j
T R
0i

T L
0 j
_
T
0i
E
A
13

x
0 j
+

T
0i
E
A
23

y
0 j
_
2
. (A.16)
Estimation of the eigenvector of the perturbed system (
i
)by
expanding Eq. (A.9)
(i) Expanding Eq. (A.9) and thereafter utilizing Eqs.
(A.10) and (A.12),
A
i
eigenvector of the unperturbed system is given by
A
i
=
T L
0i
+
N
j =1
_

T R
0 j
T L
0i

T R
0 j
_
T
0 j
E
A
13

x
0i
+

T
0 j
E
A
23

y
0i
_
T R
0 j
. (A.17)
(ii) Expanding Eq. (A.8) and thereafter utilizing Eqs. (A.11)
and (A.14),
A
i
of the unperturbed system is given by
A
i
=
T R
0i
+
2N
j =1
_

T L
0 j
T R
0i

T L
0 j
_
x
T
0 j
E
A
13

0i
+

y
T
0 j
E
A
23

0i
_
T L
0 j
. (A.18)
A.1.2. Case-2: System-B and system-C
Let us now consider the unperturbed system as system-C
as before but the perturbed system is system-B. Therefore,
perturbation in the mass matrix can be expressed as
[E
B
] =
_
_
[0] [0] [e
y
m
i
]
[0] [0] [e
x
m
i
]
[e
y
m
i
] [e
x
m
i
] [0]
_
_
. (A.19)
Noting the similarity between [E
B
] and [E
A
], results for
system-B can be directly expressed by the formulae already
derived for system-A by replacing [E
A
] by [E
B
]. These results
are, however, not presented for brevity.
A.1.3. Case-3: System-A and System-B
Now, from the results of the perturbation analysis of
systems-A and -B with respect to the unperturbed system-C,
one may write
For eigenvalues (i)
_
A
i

B
i
_
associated with any pure
translational eigenvector of the unperturbed system is
A
i

B
i
=
N
j =1
_

T L
0i

T R
0 j
T L
0i

T R
0 j
_
x
T
0i
_
E
A
13
+ E
B
13
_

0 j
+

y
T
0i
_
E
A
23
+ E
B
23
_

0 j
_
x
T
0i
_
E
A
13
E
B
13
_

0 j
+

y
T
0i
_
E
A
23
E
B
23
_

0 j
_
. (A.20)
(ii)
_
A
i

B
i
_
of the unperturbed system is
A
i

B
i
=
T R
0i
+
2N
j =1
_

T R
0i

T L
0 j
T R
0i

T L
0 j
_
T
0i
_
E
A
13
+ E
B
13
_

x
0 j
+

T
0i
_
E
A
23
+ E
B
23
_

y
0 j
_
T
0i
_
E
A
13
E
B
13
_

x
0 j
+

T
0i
_
E
A
23
E
B
23
_

y
0 j
_
.
(A.21)
For eigenvectors
(i)
_
A
i

B
i
_
eigenvector of the unperturbed system is
A
i

B
i
=
T L
0i
+
N
j =1
_

T R
0 j
T L
0i

T R
0 j
_
T
0 j
_
E
A
13
E
B
13
_

x
0i
+

T
0 j
_
E
A
23
E
B
23
_

y
0i
_
T R
0 j
.
(A.22)
(ii)
_
A
i

B
i
_
associated with any pure torsional eigenvec-
tor of the unperturbed system is
A
i

B
i
=
T R
0i
+
2N
j =1
_

T L
0 j
T R
0i

T L
0 j
_
x
T
0 j
_
E
A
13
E
B
13
_

0i
+

y
T
0 j
_
E
A
23
E
B
23
_

0i
_
T L
0 j
.
(A.23)
A.2. Simplication of error equations for the eigenvalues and
eigenvectors
The contribution from the higher modes may be neglected
in the perturbation analysis and therefore, it may be sufcient
to account for the effects of the rst three modes. These
three modes show predominantly translation along X-direction,
predominantly translation along Y-direction and predominantly
torsional behavior. However, in each of these three modes, any
typical constituting frame vibrates in its rst mode pattern,
i.e., without any point of contraexure. Therefore, it may be
sufcient to take i = 1 and 2 in Eq. (A.20) and i = 1 in
Eq. (A.21) for the eigenvalue perturbation. Similarly for the
eigenvectors, one may consider i = 1 and 2 in Eq. (A.22) and
i = 1 in Eq. (A.23). Further, it may be noted that the factor
in terms of within the rst brace in Eqs. (A.20)(A.23) dies
out rapidly if j differs from i . Therefore, it may be sufcient to
consider j = 1 in Eqs. (A.20)(A.22) and j = 1 and 2 in Eqs.
(A.22) and (A.23) for all practical purposes.
Utilizing these simplications to the eigenvalue perturba-
tion, one may write from Eq. (A.20)
A
1

B
1
=
_

T L
01

T R
01
T L
01

T R
01
_
x
T
01
_
E
A
13
+ E
B
13
_

01
+

y
T
01
_
E
A
23
+ E
B
23
_

01
_
x
T
01
_
E
A
13
E
B
13
_

01
+

y
T
01
_
E
A
23
E
B
23
_

01
_
(A.24)
A
2

B
2
=
_

T L
02

T R
01
T L
02

T R
01
_
x
T
02
_
E
A
13
+ E
B
13
_

01
+

y
T
02
_
E
A
23
+ E
B
23
_

01
_
x
T
02
_
E
A
13
E
B
13
_

01
+

y
T
02
_
E
A
23
E
B
23
_

01
_
(A.25)
and from Eq. (A.21),
A
1

B
1
=
_

T R
01

T L
01
T R
01

T L
01
_
T
01
_
E
A
13
+ E
B
13
_

x
01
+

T
01
_
E
A
23
+ E
B
23
_

y
01
_
T
01
_
E
A
13
E
B
13
_

x
01
+

T
01
_
E
A
23
E
B
23
_

y
01
_
+
_

T R
01

T L
02
T R
01

T L
02
_
T
01
_
E
A
13
+ E
B
13
_

x
02
+

T
01
_
E
A
23
+ E
B
23
_

y
02
_
T
01
_
E
A
13
E
B
13
_

x
02
+

T
01
_
E
A
23
E
B
23
_

y
02
_
. (A.26)
Similarly, for the eigenvectors, one may write from Eq. (A.22)
A
1

B
1
=
_

T R
01
T L
01

T R
01
_
T
01
_
E
A
13
E
B
13
_

x
01
+

T
01
_
E
A
23
E
B
23
_

y
01
_
T R
01
(A.27)
A
2

B
2
=
_

T R
01
T L
02

T R
01
_
T
01
_
E
A
13
E
B
13
_

x
02
+

T
01
_
E
A
23
E
B
23
_

y
02
_
T R
01
(A.28)
and from Eq. (A.23),
A
1

B
1
=
_

T L
01
T R
01

T L
01
_
x
T
01
_
E
A
13
E
B
13
_

01
+

y
T
01
_
E
A
23
E
B
23
_

01
_
T L
01
+
_

T L
02
T R
01

T L
02
_
_

x
T
02
_
E
A
13
E
B
13
_

01
+

y
T
02
_
E
A
23
E
B
23
_

01
_
T L
02
. (A.29)
At this stage, it may be noted that
E
A
13
+ E
B
13
=
__
b
y
i
+2e
y
_
m
i
_
diag
; (A.30a)
E
A
13
E
B
13
=
_
b
y
i
m
i
_
diag
; (A.30b)
E
A
23
+ E
B
23
=
__
b
x
i
+2e
x
_
m
i
_
diag
; (A.31a)
E
A
23
E
B
23
=
_
b
x
i
m
i
_
diag
. (A.31b)
It is now required to investigate the rst three eigenvectors
of the unperturbed system-C. In doing so, eigenvalue equation
of the unperturbed system may be expressed as
0
_
_
K
xx
K
xy
0
K
xy
K
yy
0
0 0 K
_
_
_
_
_
x
0
y
0
0
_
_
_
=
_
_
[m
i
] 0 0
0 [m
i
] 0
0 0 [m
i
( r
2
+e
2
x
+e
2
y
)]
_
_
_
_
_
x
0
y
0
0
_
_
_
. (A.32)
Stiffness proportionality of the constituting frames and
uncoupled nature of the mass matrix enable us to decouple
the eigenvalue problem into two smaller eigenvalue problems,
namely, characteristic frame problem and equivalent one-story
torsionally uncoupled problem. Letting , {u} be the eigenvalue
and eigenvector of the characteristic frame; , =
_
_
T
the eigenvalue and eigenvector of the equivalent one-story
system; and
0
,
0
=
_

x
0

y
0

0
_
T
the eigenvalue and
eigenvector of the entire building, it may be shown that
0
= (A.33)
0
=
_

x
0

y
0

0
_
T
=
_
_
T
{u} . (A.34)
Further, mass normalization of the i j th eigenvector leads to
_
2
xi
+
2
yi
+
2
i
_
r
2
+e
2
x
+e
2
y
__
{u}
T
j
[m
i
] {u}
j
= 1.
(A.35)
Let us nownormalize the eigenvector of the characteristic frame
and that of one-story system as
2
xi
+
2
yi
+
2
i
_
r
2
+e
2
x
+e
2
y
_
= 1 (A.36)
and
{u}
T
j
[m
i
] {u}
j
= 1. (A.37)
Therefore, if i is a pure translational mode,
i
= 0 and
consequently,
2
xi
+
2
yi
= 1. (A.38)
Otherwise, if i is a pure torsional mode,
xi
=
yi
= 0, and
consequently,
2
i
= ( r
2
+e
2
x
+e
2
y
)
1
. (A.39)
It may be noted that, for the rst three modes of the building,
eigenvector of the characteristic frame is essentially its rst
eigenvector {u}
1
. In other words, oor-wise variation of the
translations along X- and Y-directions and torsional rotation is
the same in the rst three modes of the unperturbed system.
Utilizing this simplication along with Eqs. (A.30a),
(A.30b), (A.31a) and (A.31b), the following simplications
may be carried out for Eqs. (A.24)(A.29).
For eigenvalues
(i)
_
A
1

B
1
_
associated with the rst pure translational
eigenvector of the unperturbed system is given by (from Eq.
(A.24))
A
1

B
1
=
_
T R
01

T L
01
_
1 _
x1
1
Y +2e
y
x1
1
+
y1
1
X +2e
x
y1
1
_
x1
1
Y +
y1
1
X
_
. (A.40)
(ii)
_
A
2

B
2
_
associated with the second pure translational
(A.25))
A
2

B
2
=
_
T R
01

T L
02
_
1 _
x2
1
Y +2e
y
x2
1
+
y2
1
X +2e
x
y2
1
_
x2
1
Y +
y2
1
X
_
. (A.41)
(iii)
_
A
1

B
1
_
associated with the rst pure torsional
(A.26))
A
1

B
1
=
_
T L
01

T R
01
_
1
x1
Y +2e
y
x1
+
y1
X +2e
x
y1
_
x1
Y +
1
y1
X
_
+
_
T L
02

T R
01
_
1
x2
Y +2e
y
x2
+
1
y2
X +2e
x
y2
_
x2
Y +
1
y2
X
_
. (A.42)
For eigenvectors
(i)
_
A
1

B
1
_
associated with the rst pure translational
(A.27))
_
A
1

B
1
_
=
_
T R
01
T L
01
1
_
1
x1
Y +
1
y1
X
_
T R
01
. (A.43)
(ii)
_
A
2

B
2
_
associated with the second pure translational
(A.28))
_
A
2

B
2
_
=
_
T R
01
T L
02
1
_
1
x2
Y +
1
y2
X
_
T R
01
. (A.44)
(iii)
_
A
1

B
1
_
associated with the rst pure torsional
(A.29))
_
A
1

B
1
_
=
_
T L
01
T R
01
1
_
1
_
x1
1
Y +
y1
1
X
_
T L
01
+
_
T L
02
T R
01
1
_
1
_
x2
1
Y +
y2
1
X
_
T L
02
. (A.45)
Here, in Eqs. (A.40)(A.45), it is assumed that
0
=
1
0
; (A.46a)
Y = {u}
T
1
_
b
y
i
m
i
_
diag
{u}
1
; (A.46b)
X = {u}
T
1
_
b
x
i
m
i
_
diag
{u}
1
. (A.46c)
A.3. Estimation of error in eigenvalues and eigenvectors
From Eqs. (A.40)(A.45), it may be seen that the difference
in eigenvalues and in eigenvectors between system-A and
system-B is zero if X = Y = 0. Moreover, for a given order
of X and Y, it may also be possible to ascertain the order of
the differences in eigenvalues and in eigenvectors between the
systems -A and -B. For this purpose, let us dene,
Y =
y
r and (A.47a)
X =
x
r (A.47b)
where,
y
= {u}
T
1
_
b
y
i
r
m
i
_
diag
{u}
1
and (A.48a)
x
= {u}
T
1
_
b
x
i
r
m
i
_
diag
{u}
1
. (A.48b)
Further, letting
x
= C
1
10
and
Y
= C
2
10
with |C
1
|,
|C
2
| close to unity, e.g., 1.02.0, the order of the difference
in eigenvalues and eigenvectors may be evaluated from Eqs.
(A.40)(A.45) as follows:
Utilizing Eqs. (A.47) and (A.48) in Eq. (A.40), it may be
shown that
A
1

B
1
=
_
T R
01

T L
01
_
1
_
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
1
_
x1
y
+2
e
y
r

x1
+
y1
x
+2
e
x
r

y1
_
x1
y
+
y1
x
_
. (A.49)
It may be noted that
2
x1
,
2
y1
are of the order 10
1
as evident
from Eq. (A.38). Thus,
x1
,
y1
are also of the order 10
1
but
more close to unity. Therefore, last term in Eq. (A.49) can be
of the order maximum 10
. Further,
e
x
r
,
e
y
r
are of the order
10
1
and can be even close to 0.5 for a highly torsionally
coupled building. Therefore, if 1, third term in Eq. (A.49)
then may be of the order maximum 10
1
but close to unity or
even slightly more than unity. Similarly, second term in Eq.
(A.49) is also of the order 10
1
but may not be very close to
unity. Finally, in a torsionally coupled building, the rst three
frequencies are very close to each other and thus, rst term
in Eq. (A.49) is of the order 10
1
. Therefore, it is seen that
A
1

B
1
may be of the order maximum 10
(+1)
. Similarly,
Eq. (A.41) may be expressed as
A
2

B
2
=
_
1
T R
01

T L
02
_
_
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
2
x2
y
+2
e
y
r

x2
+
y2
x
+2
e
x
r

y2
_
x2
y
+
y2
x
_
. (A.50)
Following the same reasoning as before, it may be shown that
A
2

B
2
(+1)
. Similarly
from Eq. (A.42) one may write,
A
1

B
1
=
_
T L
01

T R
01
_
1
_
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
1
x1
y
+
2e
y
r

x1
+
y1
x
+
2e
x
r

y1
_
_
x1
y
+
y1
x
_
+
_
T L
02

T R
01
_
1
_
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
1
x2
y
+
2e
y
r

x2
+
y2
x
+
2e
x
r

y2
_
x2
y
+
y2
x
_
. (A.51)
Following the same reasoning as before, it may be shown that
A
1

B
1
.
Similarly, from Eq. (A.43) it may be shown for the
eigenvector that
_
A
1

B
1
_
=
_
T R
01
T L
01
1
_
1
_
x1
y
+
y1
x
_
_
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
1
2
T R
01
. (A.52)
Note that,
T R
01
=
_
0 0
1
_
{u}
1
and
1
is of the order
10
2
or at least 10
1
, as evident from Eq. (A.39), the maximum
order of
_
A
1

B
1
_
may be calculated as 10
{u}
1
. Similarly,
Eq. (A.44) may be written as
_
A
2

B
2
_
=
_
T R
01
T L
02
1
_
1
_
x2
y
+
y2
x
_
_
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
1
2 _
0 0
1
_
T
{u}
1
(A.53)
and the maximum order of
_
A
2

B
2
_
may be shown as
10
{u}
1
. Similarly, from Eq. (A.45)
_
A
1

B
1
_
=
_
T L
01
T R
01
1
_
1 _
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
1
2
x2
y
+
y1
x
_ _
x1

y1
0
_
T
{u}
1
T L
02
T R
01
1
_
1 _
1 +
_
e
x
r
_
2
+
_
e
y
r
_
2
_
1
2
x2
y
+
y2
x
_ _
x2

y2
0
_
T
{u}
1
(A.54)
and the maximum order of
_
A
1

B
1
_
can be calculated as
10
+1
{u}
1
.
Therefore, it is seen that, if system-B is used in lieu of
system-A, difference in the eigenvalues ( ) is of the order
maximum 10
and that for the eigenvectors is of the order

maximum10
+1
{u}
1
where {u}
1
is the mass normalized rst
mode shape of the characteristic frame.
A.4. Estimation of error in frame shear forces
Let the value of be chosen such that error in eigenvalue
(or frequency) estimation is small enough to make a signicant
change in the estimation of the spectral ordinate. Further,
resulting change in participation factor is also assumed small
enough for all practical purposes. In view of these, error in
the estimation of the frame shear force in any of the rst three
modes of the building is attributed only from the error in the
estimation of the eigenvectors. Let =
_
i x

i y

i
_
T
be the eigenvector in any (say, i th) of the rst three modes
of the average CM model, i.e., system-B. Similarly, let
/
be
the associated eigenvector of the actual (perturbed) system-A.
Therefore, error in the estimation of this mode shape is given
by
_
_
=
_
S
i x
S
i y
S
i
_
T
_
10
+1
_
{u}
1
(A.55)
where S
i x
, S
i y
and S
i
are of the order maximum 10
0
. Let
be the inclination, measured in the counter clockwise direction,
of any frame with respect to the X-axis. Therefore, error in
the estimation of the modal shear force for the frame may be
expressed as
_
F
/
F
_
= S
di
P
i
_
S
i x
cos S
i y
sin DS
i
_
_
10
+1
_
[K]
j
{u}
1
(A.56)
where D is the distance of the frame from the CR, S
di
the
spectral displacement, P
i
the participation factor, and [K]
j
the
stiffness matrix of the frame under consideration.
In the case of average CM model, i.e., system-B, since CMs
of all the oors are located on one vertical line and all the CRs
lie on another vertical line, it may be shown that
T
= { h
x
{u}
T
1
h
y
{u}
T
1
h
{u}
T
1
} (A.57)
where, h
x
, h
y
and h
are scalar quantities. Moreover, mass

normalization criteria lead to
h
2
x
+h
2
y
+h
2
( r
2
+e
2
x
+e
2
y
)
2e
y
h
x
h
+2e
x
h
y
h
= 1. (A.58)
Therefore, if the mode shows translation predominantly along
X-direction, h
x
1, h
y
0, and h
0. Similarly, if the
mode is predominantly translational along Y-direction, h
x

0, h
y
1, and h
0; and, for a predominantly torsional

mode, h
x
0, h
y
0, and h
( r
2
+e
2
x
+e
2
y
)
1
.
Finally, considering system-B as the basis, percentage error
in the estimation of the modal frame shear force may be shown
as
__
F
/
F
_
/F
_
100 = C
0
10
+3
(A.59)
where
C
0
=
_
S
i x
cos S
i y
sin DS
i
_
/
_
h
x
cos h
y
sin Dh
_
. (A.60)
Since, S
i x
, S
i y
are of the order 10
0
, DS
i
usually of the order
10
1
or even less, (h
x
, h
y
) of the order 10
1
, Dh
of the order
10
2
or even less, and most importantly, signs of the respective
terms in the numerator and the denominator are consistent, it
may be easily seen that C
0
is of the order of 10
0
. For example,
if is taken as 3, maximum error in the estimation of the frame
shear force is of the order 10
0
, i.e., 0%10%; for equals to 2,
the maximum error is increased to 10
1
order, i.e., 10%100%.
The formulation is also valid for the remaining two of the rst
three modes of the building and the resultant upper bound of
the error is valid even after applying modal combination too.
Therefore, the applicability conditions of the simplied method
to a building with vertically-nonaligned CMs may be stated as
|
x
| 0.002 and (A.61a)
= 0.002. (A.61b)
Here,
x
= {u}
T
1
_
b
x
i
m
i
r
_
diag
{u}
1
,
y
= {u}
T
1
_
b
y
i
m
i
r
_
diag
{u}
1
are the indices for assessing the applicability of the simpli-
ed procedure and {u}
1
the mass normalized fundamental mode
shape of the characteristic frame.
References
[1] Balendra T. A simplied model for lateral load analysis of asymmetrical
buildings. Engineering Structures 1983;5(3):15462.
[2] Basu D, Jain SK. Alternative method to locate centre of rigidity in
asymmetric buildings. Earthquake Engineering and Structural Dynamics
2007;36(7):96573.
[3] Chandler AM, Hutchinson GL. Torsional coupling effects in the
earthquake response of asymmetric buildings. Engineering Structures
1986;8(4):22236.
[4] Ferhi A, Truman KZ. Behavior of asymmetric building systems under a
monotonic loadI. Engineering Structures 1996;18(2):13341.
[5] Ferhi A, Truman KZ. Behavior of asymmetric building systems under a
monotonic loadII. Engineering Structures 1996;18(2):14253.
[6] Gupta RK, Hutchinson GL. Horizontal and vertical seismic response of
torsionally coupled buildings. Engineering Structures 1994;16(1):1124.
[7] Hejal R, Chopra AK. Earthquake response of torsionally-coupled build-
ings. Report no. UCB/EERC-87/20. Berkeley: Earthquake Engineering
Research Center, Univ. of Calif. December 1987.
[8] Hejal R, Chopra AK. Earthquake response of torsionally-coupled farmed
buildings. Journal of Structural Engineering ASCE 1989;115(4):83451.
[9] IS: 1893 (Part 1)-2002, Criteria for earthquake resistant design of
structures: General provisions and buildings. New Delhi: Bureau of Indian
Standards. 2002.
[10] Jiang W, Hutchinson GL, Wilson JL. Inelastic torsional coupling of
building models. Engineering Structures 1996;18(4):288300.
[11] Kan CL, Chopra AK. Coupled lateral torsional response of buildings
to ground shaking. Report no. UCB/EERC-76/13. Berkeley: Earthquake
Engineering Research Center, Univ. of Calif. May 1976.
[12] Kan CL, Chopra AK. Torsional coupling and earthquake response of
simple elastic and inelastic systems. Journal of Structural Engineering-
ASCE 1981;107(ST8):156988.
[13] Nelson TK, Wilson JL, Hutchinson GL. Review of the torsional coupling
of asymmetrical wall-frame buildings. Engineering Structures 1997;
19(3):23346.
[14] SAP2000. Structural analysis program. Berkeley (CA): Computer and
Structures, Inc; 1995.
[15] Thambiratnam DP, Corderoy JB. Effects of asymmetry on the response of
multistorey buildings to earthquakes. Engineering Structures 1994;16(3):
21021.
[16] Wilkinson JH. The algebraic eigenvalue problem. Clarendon Press; 1965.
[17] Yoon YS, Smith BS. Estimating period ratio for predicting torsional
coupling. Engineering Structures 1995;17(1):5262.

Class of Irregularity

Transféré par

Informations du document

Description originale:

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Class of Irregularity

Transféré par

Droits d'auteur :

Formats disponibles

Engineering Structures 30 (2008) 12721291

sub-matrices of system-A and system-B (or C), the difference

and that for the eigenvectors is of the

and that for the eigenvectors is of the order

are scalar quantities. Moreover, mass

0; and, for a predominantly torsional

Vous aimerez peut-être aussi