Escolar Documentos
Profissional Documentos
Cultura Documentos
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/245545818
CITATIONS
READS
17
17
3 authors, including:
Andrei M Reinhorn
Avigdor Rutenberg
SEE PROFILE
SEE PROFILE
\..,/
8203
J. GLUCK,
DSc
A. REINHORN,
DSc
A. RUTENBERG,
DSc
The Paper shows how the existing approaches to the dynamic analysis of asymmetric
tall buildings, using a single storey torsional coupling analogy, can be applied to a
wide class of irregular structures and under what conditions. A step by step procedure
using the response spectrum technique is outlined and illustrated by a numerical
example. A three-dimensional rather than two-dimensional formulation is adopted to
emphasize the generality of the approach. The diserete and distributed parameter
formulations are followed throughout the mathematical exposition.
Introduction
Multi-storey building structures are seldom symmetric in layout, and when
subjected to earthq'u'ake ground motion they respond in coupled lateral' and
torsional vibrations. The destructive effects of torsional vibrations l - 3 have
stimulated investigators to look for simple models to predict' the earthquake
response of torsionally coupled structures. However, the large number of
degrees of freedom involved in the standard modelling of tall buildfngs tends to
mask the basic principles governing torsional coupling effects.
2. It is wen recognized that the coupling of lateral and torsional vibrations
depends on the relation between the layout of the lateral framing system and the
mass distribution. 3 - 7 However, until recently the regularity of the geometrical
and structural layout throughout the height, so typical in tall buildings, was not
taken advantage of in the mass and stiffness formulation of standard dynamic
analysis techniques. A regular structural layout may be defined as one having a
small number of framing systems, each comprising several vertical planar
assemblages with similar stiffness properties (e.g. frames and flexural walls), and a
common variation thereof along the height of the building. For such structures
~the stiffness properties of each system can be represented by a lateral and a tor
sional cantileverS (RC in Fig. 1) at the system's axis of rigidity. The inertia
properties can be likewise represented by masses and mass moments of inertia
along the mass axis of the buildingS (MC in Fig. 1). The distance between the
mass centre and the centre of rigidity is defined as the static eccentricity of the
floor; it plays a key role in torsional coupling analysis.
1.
GLUCK,
REINHORN
AND
RUTENBERG
,./'
Notation
D
d
Young's modulus
E
damping ratio
~
w
natural circular frequencies
diagonal matrix of non-coupled circular frequencies
Subscripts
n
x,y,8
Superscripts
k
ey
IN
. -a
TALL
BUILDINGS
IMe
,--tt--.
RC
ex
Lx
(b)
v
Fig. 1. Principal axes in building
structure; (a) irregular wall-frame
structure, (b) typical floor plan
.j
413
,/
applied to a wider class of structure, and under what conditions. The analysi: \
is presented in the framework of the response spectrum technique,2.15 which i~
a weB-established seismic response procedure. A step by step solution is out
lined and illustrated by means of a numerical example. A three-dimensional
rather than a two-dimensional formulation is adopted (i.e. no symmetry rather
than single' symmetry) in order to emphasize the generality of the approach. For
the same reason, both discrete and distributed parameter formulations are
followed throughout the mathematical exposition.
..
MoD+CD+KD
=F
(1 a)
-eyM
M(r2+e x 2 +ey2 )
KXX
+ [ sym.
in which M is a diagonal mass matrix, C is the damping matrix assumed for
simplicity to be of the' proportional type,15 K is a stiffness matrix, D is a displace
ment vector, F represents the lateral and torsional forcing functions (Le. the
earthquake force vector) and r is the mass radius of gyration. The eccentricities
ex and ey of the mass axis from the reference axis are shown in Fig. L For
regular structures it is assumed that
s
2:
K plJ
p=1
"
where s is the number of different structural systems in the building.
8. In the distributed parameter formulation the vector D is replaced by a
continuous vector function d having three components (x, y, 8), so that the
coefficient matrix represents given spatial distributions of the mass, damping and
stiffness, namely
=
or
kXX
+ [ sym.
414
, ",.,
IN TALL
BUILDINGS
k(z)d(z, t)
= 2:
k,CtI(d)
1=1
9. The evaluation of the overall stiffness matrices in equations (la) and (2a)
from the properties of the individual assemblages follows well-known tech
. niques. 4 8 15 ,18,19 The submatrices and the element with mixed indices repre
sent the coupling between the co-ordinates x, yand 8, and it is evident that sub
stantial savings in computational effort would result if some transformations of
co-ordinates could be found to equate these off-diagonal terms to zero.
10. Consider the case where s= 1, Le. there is only one framing system in
the building-say, uniform flexural walls. If the origin of the co-ordinate
system is at the rigidity centre, transformation to the principal axes gives
..,/
Ku
=0
ktJ
=0
the assumption of geometrical regularity, it folIo~ that K xx , Kyy and Keo are
proportional. The off-diagonal submatrices also vanish in the case of structural
symmetry; however, proportionality between the diagonal submatrices does not
necessarily follow. When the structure consists of two framing systems (s= 2)
and the rigidity axes of the two are not co-linear, it is possible to equate Ku to
zero (i#- j) by means of the eigenvalues p2 and eigenvectors 1/ of the problem 8 19
(k 1 - p 2 k 2 )1/
=0
(3)
(4)
) and in general Pl #- P2 #- Pa. When s> 2, three or more matrices are involved
and, as is known from linear algebra, K cannot in general be' uncoupled into
main diagonal blocks KJJ
II. Consider now the free vibration problem associated with equations (I a)
and (2a)
MoD*+KD*
=0
(5)
(6)
where
=0
(7a)
For a flexural beam system, for example, a= a4 1az 4 and for combined walls and frames
a =kl(84 j8z 4 ) - k2(a 2 j8z 2 ).
415
GLUCK,
REINHORN
AND
RUTENBERG
or explicitly
y
e 2M
-w
r
o
ex 2M
--w
QJo
= 0
(7b)
sym.
Kyy - w 2M
QJy
00]
QJX}
QJe = ['V0x 'Ve 0 {ax}
ae = 'Va
f/Jy
0 0 'Vy ay
Multiplying equation (7a) by 'VT , where T 'denotes the transpose, and
{
14.
f
\
sym.
(10)(
i
-w 2
=0
reX ['VeTM'Vy- 1]
0
ay
\
in which OJ = diag (Wjn), (j = x, y, 8, 11 = 1, ... N) and I is a unit matrix.
I
15. In equation (10) the first matrix is diagonal in each submatrix or block,
and thus can be decomposed into N sets of three coupled homogeneous equations \
t
(triplets) in w 2 , or
sym.
ey
W nt - r
2
2
Wen -W n 1
sym.
[1 + (e; + (e; fJ
o
2
-Wn 1
e;
Wyn -Wnl
aen
2
a yn
Gr
r[
['VxTM'Ve-I] = ['VeTM'V1I-1] = 0 .
416
(12)
I
I
IN TALL BUILDINGS
(13)
or
= -mru'g
(15)
.' where ilg is the ground acceleration in any given direction and r is an orientation
vector given by
= {I, 0, O}T
rx
'9 =
{eyl, I, - exI}T
(16)
= MlJlyn r ynSa,yn
(17)
in which Sayn is the acceleration spectral value of ug in mode n, lJIyl!. is the corres
ponding mode shape and r yn is the modal participation factor given b y15
Fyn
ryn
lJIynTMI
(18)
21. It can be easily shown from equations (9) and (11) that in torsionally
417
GLUCK,
REINHORN
AND
RUTENBERG
coupled systems the modal forces are linear combinations of the non-couPI,C (
planar modes x, y and 8 or
r
Fnl == Mo'llnanlan?'IIn orSani .
(19)
22. To obtain the ratios of the coupled forces and the non-coupled ones at ~ r
diITerent levels of the structure it is necessary to divide the terms in equation (19)
by the corresponding ones in equation (17), i.e.
.
(20a)
F*ntk -- FntkiFynk
or
F*nt k = EAnkantan?BnEr(Sant/Sa.yn), (i = 1,2,3)
(20b)
r;!
where E denotes the partitioned non-dimensional coupling matrix given by
_ ey
f~'
i'
o
1
r
k
l~:
l+e:f+(;f e;
E=
(21)j,'j~
;~!
sym.
1
An k is the matrix of the mode shape ratios
Ank = diag (,pxnkl,pyn k, ,pOnkl,pynk , 1)
and
;Ii'
I'
(22)
(23)
where r xn and rOn are the participation factors of the non-coupled planar rnode ',',
in the x and 8 directions, computed from equation (18) with the appropriate I
index changes.
23. The multipliers in equations (20) reduce to those of a single storey struc- "
ture when An~ and Bn are diagonal unit matrices. The effect of the modal
participation factor ratios in equations (20) is relatively small, and when it is
assumed that Bn = I, the errors are unlikely to exceed one per cent. s
24. The maximum system response is usually obtained by combining the(
modal responses by means of the root square of the sum of squares formula. "
This yields satisfactory results when the natural frequencies are sufficiently l~
separated,3 but is not likely to in torsional coupling problems. In such cases
the modified root square of the sum of squares formula is used 5
F*n
J[.il
(F*nl)2+2
filk~l F*ntF*nk/(l+EJk 2
)]
(24)
.
."
t~k
IWI-Wk/
v(1-g2)/(Wt+Wk)~
1aoo
( 25
When IWI - wkl is large, the double sum term in equation (24) becomes smalle~\
and equation (24) reduces to the standard root square of the sum of squares \
formula.
Solution procedure
25. It has been shown that the single storey torsional coupling analogy holds
rigorously only when the mode shapes in the three principal directions are
identical. However, good agreement with exact results may still be expected
418
.Jmary of the computational procedure is thus also applicable to the more general
case,
(0) Formulate the problem in the three principal co-ordinates of the struc
1 ,..:.
...J
-.it
CD
02
I-
I]]
2671
@J
r-
\.0
Pd
x
I
SC [
\.0
-it
-r-
rn
II
\
\\'
0 fc\J
..., :-
J-
lb'
t'
02
-'I-
...-
>.,
RC
-- ro
o
,..:.
rnJ.
8x40 = 32.0
is F-
-'--
- CO
c\J
c\J
- f-
'-
./
419
!
,
RUTENBERG
(h) Use equation (20a) to compute the resuhant forces, i.e. Fnt=F*nl kF1; nkQi
(;) Distribute the forces acting on each framing system using the appropriate !J~
transformations 19 and then superpose the effects of the n lower triplets
I:
by means of the standard root square of the sum of squares formula.
'r~
Numerical example
26. The floor plan shown in Fig. 2 is that of a 16 storey building, 48 rh high,
which was first analysed by Gluck l8 and later by Rutenberg and Heidebrecht. l9
The structural properties of the walls and the frames are given in Table 1. The
structure has two vertical axes of rigidity: one for the system of flexural walls and
the other for the frames. If the single storey analogy is to be used, only an
approximate solution may be expected. As the structure is uniform along the
height the continuum method is obviously preferred. With the given parameters
the equations of motion read
(26a)
or
~('
:{
'<'f
' '.'.'.'.
".'
,}>'
i'"
2.24 x 10- 7 [
1
-0,8137
-0'8137]{~y}
0:4853
d9
ltli
+{d y}
d""9
RC
_
RC
{/Y}
_ 8.79 X 10-7 [1
-0'46]{d"y} =
(26b)
- 046
O' 36 d" 8 RC
Ie RC
where the primes and dots denote space and time derivatives, respectively.
27. .Following the solution algorithm, the equivalent modal forces are com
puted first.
28. Static uncoupling (equation (3 by means of the transformation l9
- 0'887]
d
(27)
0.462
In the uncoupled co-ordinate system equation (26b) becomes
I,
r:
0-462
Il=Td= [
s
0'887
is carried out.
7
2'32x 10-
[~'106
+
]{ }
::::
OZ4 -
::
48-0
OZ2
OZ4 -
(O~497)2
48-0
if'
U.
OZ2
U8
f..
(28)
/u9
The second matrix in equation (28) represents the statically uncouPled. system in
U y and U9 co-ordinates.
The mass matrix includes a coupling parameter e!r=
0106.
29. The two non-coupled equations of motion are obtained by putting e/r= O.
Thus
uItlt1l-(1'582!48'0)2U"y+2'32 x 107 uy = 0
}
d"'9-(0'497/48-0)2 U"9+2'32 x 107 x O'6492u9 = O '
(29)
The non-coupled properties are given in Table 2.
30. For the given acceleration spectrum (Fig. 3) the storey shears at each
420
r
>~
C'\'\
!
IN TALL BUILDINGS
= 70 N s21m2,
-*
Princi
pal
moment
of
inertia,
m4
Wall
1
2
3
4
5
6
5'7166
2-0834
2-0834
57166
12348
1-2348
Position
co-ordinates
Yi
-18-671
5329
9329
13329
7329
7'329
0
0
0
0
-4-5
+45
0-15
s=
a
26579
26-579
26-579
26'579
26'579
1
2
3
4
5
1-0
53158
53'158
53'158
53'158
53158
Beam
18-985
18985
18985
18-985
18985
- 14-671
- 10-671
-6,671
- 2-671
J329
(}05
(E = 0-05)
0-5
Position
co-ordin
ate Xc
Moment of inertia
m 4 x 104
Frame
Exterior Centre
column column
Xi
'\
2-0
1-5
2-5
T:s
Refer
ence
level
k
\
,
Mode II
Mode III
156'74
70-47
-10-11
-76-79
-121-66
-139-36
-129-87
-98-60
-55-93
-17-06
113-68
-6-36
-91-27
- 112-29
-65-27
21-29
99-57
129-45
99-29
36-79
.. Natural
fre
quency
w, rad s
3-96
Partici
pation
factor
2580
kN
10
9
8
7
6
5
4
3
2
1
(J
ifi
0-496
0-223
-0-032
-0'243
-0-385
-0-441
-0-411
-0-312
-0-177
-0-054
0-411
-0-023
-0-330
-0-406
-0236
0-077
0-360
0-468
0-359
0-133
0-576
0-496
0-417
0-339
0-265
0-195
0-132
0-078
0-037
0-009
0-486
0-209
-0-047
-0-257
-0-395
-0-446
-0-41 I
-0'309
-0-174
-0-053
OA07 205-13
-0-027 ] 79-21
-0-332 152-94
-0-404 126-66
-0-230 100-38
0-083
75-55
0-365
52-19
0-470
31-75
0-359
14-96
0-132
4-02
1949
52-17
4-64
2822
78-81
- 1-330
0'813
2553
-1-383
0-789
421
RUTENBERG
'Or
reference level for the first three modes (j= 3) are computed using equation (1
and are given in Table 2 . . ' /
31. The coupling coefficients are now computed from equation (II) and the
approximate coupled normal modes obtained from equation (9) (Table 3).
32. Results of the computations for stages (e)-(g) in 25 are given in Table 4,
together with the computed equivalent forces.
33. Using the transformations given in reference 13, the forces and moments
acting on the flexural wall system and the frame system are evaluated for each
pair of coupled modes, from which the internal forces in the individual walls and
frames are obtained. The total response may then be evaluated by means of the
standard root square of the sum of squares formula. For comparison, the
numerical results obtained using a standard eigenvalue computer procedure are
also presented in Tables 3 and 4. As agreement to three significant figures was
obtained for the first two pairs of modes, no comparative results are presented
I(
in Table 3.
Conclusion
;
34. It has been shown that the single storey torsional coupling analogy for.' .
multi-storey asymmetric building structures holds when the following two
conditions are satisfied
(a) the stiffness matrix can be uncoupled in plan into three principal di
For most tall buildings these conditions do not hold rigidly, i.e. there are devia
tions from verticality of the mass and stiffness axes 16 or non-identity of the nor
mal modes in .the principal directions. In such cases the single storey analogy
Table 3. Dynamic properties for third pair of coupled modes (n
shapes 4>
Reference
level
k
First mode (i = 1)
Approximate
Exact
0408
-0-022
-0-328
-0-403
, -0'234
0-076
0-358
0-464
0-356
0'132
0050
-0,003
-0041
-0,050
-0-028
0-010
0-045
0-058
0-044
0,016
0'408
-0'022
-0-328
-0-403
-0,234
0-075
0-357
0-464
0-356
0-132
51-576
Participation
factor
0-7693
422
Natural
frequency w,
rad s
modal
Ii
10
9
8
7
6
5
4
3
2
1
= 3):
..
Approximate
0048
-0'004
-0-041
-0-049
-0-027
0-011
0-046
0-058
0-045
0-016
-0113
0-066
0090
0-111
0-064
-0-021
-0-099
-0-128
-0-098
-0-036
51-576
0-7699
0391
-0026
"':'0-319
-0,388
-0221
0-080
0-351
0-452
0-345
0-127
Exact
y
-0'111
0066
0-090
0-111
0-064
-0-021
-0-099
-0-129
-0-099
-0-037
0391
-0-026
-0-320
-0-388
-0-221
0-080
0-351
0-452
0-345
0-127
79-85
79-85
0-1599
0-1569
--
'
_....-____ .. _ _
'~---."':'"
.'t...___",~
~,!
..
",.
1. - /'~.
,:,'-::--~
...~
'''-''
-<
z
l>
()
::D
Reference
level
k
Approximate
n=1
FlIl
10
9
8
7
6
5
4
3
2
I
0995
0-993
0'993
0'992
0-993
0-989
0-987
0982
0996
0-996
FOl
n=2
F!J2
Fu
Exact
n=3
F1I3
F03
n=1
FlIl
Fill
n=2
F1I2
F02
1-012
1-016
1-013
1-012
1-013
1-013
1-012
1012
1-017
0-367
0-352
0-548
0-396
0385
0-381
0-378
0-376
0-374
0-372
1-'012
1-016
1'013
1-012
1-013
1-013
1-012
1-012
1'017
0-367
0'507
0-397
0389
0-385
0-383
0-381
0-379
0-377
1'012
1-001
1-013
1'011
1-010
1-008
1-013
1-012
1-012
1-009
0'368
0418
0-373
0'370
0-365
0-392
0-374
0'372
0'371
0370
0"995
0-993
0-993
0-992
0-993
0989
0-987
0-982
0-996
0996
0'523
0519
0-513
0-508
0-502
0-496
0-489
0-483
0'477
0-470
n=3
F Y3
1'012
1-001
1013
I-Oil
1-010
1-008
1013
1-012
1-012
1-029
F03
FlI!
(J)
Approximate
o
z
n=1
0-363 204-10
0437 177-96
0-370 15l-87
0-366 12565
0-359 99-68
74-72
0-401
0-374 51-51
31-18
0'371
14-90
0-369
4-00
0'368
Final forces, N
F81
107'28
9301
78'61
64'34
50-39
37-40
25-52
15)4
7'12
1-89
n=3
n=2
r
FlI2
Fo'J.
158-62
71-32
-10,27
-77,79
- 123'12
-141-17
-131-56
-99-78
-56-60
- 17-35
5909
25-86
-5'13
-30-40
-47'33
-53-65
-49-74
- 37-54
- 21-20
-643
F1I3
F03
41-83
11504
-267
-637
-9246 -3404
- 13-53 -40-99
-65-92 -23-82
21-46
8'35
37,23
100-86
131-00
48'16
100-48
36-84
37' 12 1 1361
()
o
C
-0
z
G)
z
~
l>
r
OJ
o
z
.j::::.
I',,)
G)
(J)
GLUCK,
REINHORN AND
RUTENBERG
\...-t::
is still a powerful tool, albeit. approximate, for the dynamic analysis of asymmetric buildings. Due to its simplicity, the single storey model may be useful i n '
formulating anti-seismic codes of practice for asymmetric tall buildings.
35. The solution sequence outlined casts the analysis in the standard form
of the response spectrum technique. It is therefore hoped that the Paper will
contribute to the extension of the technique for asymmetric multistorey buildings.
References
oscillations. Proc. 2nd Wid Can/. Earthquake Engng, Tokyo, 1960,2, 879-894.
164.
6. GLUCK J. and REINHORN A. Dynamic eccentricity of static equivalent earthquake
8. REINHORN A. Static and dynamic torsional coupling in tall buildings. (In Hebrew.)
buildings. J. Struct. Div. Am. Soc. Civ. Engrs, 1977, 103, ST 4, Apr., 805-819.
coupled buildings. J. Strucl. Div. Am. Soc. Civ. Engrs, 1977, 103, ST 4, Apr.,
821-838.
12. F'S:NZIEN J. Earthquake response of irregularly shaped buildings. Proc. 4th WId f
Con/. Earthquake Engng, Chile, 1969,2, A3.
-,."
13. ROSENBLUETH E. and ELORDUY E. Response of linear systems to certain transient
disturbances. Proc. 4th WId Con/. Earthquake Engng, Chile, 1969. I, A-I, 185
196.
14. SKINNER R. I. et al. Unbalanced buildings with light-towers under earthquake
forces. Proc. 3rd WId Con/. Earthquake Engng, N.Z., 1965, 2, 586-602.
J5. CLOUGH R. W. and PENZIEN J. Dynamics 0/ structures. MCGraw-Hili. New York,
1975.
16. HOERNER B. J. Modal coupling and earthquake response 0/ tall buildings. California
Institute of Technology. Pasadena, 1971, Report EERL 71-07.
17. RUTENBERG A. et al. On the dynamic properties of asymmetric wall frame struc
18. GLUCK J. A general method/or lateral load distribution between stiffening elements
.)
"}
.1
424