Sunteți pe pagina 1din 12

Engineering Structures 26 (2004) 117–128

www.elsevier.com/locate/engstruct

An upper-bound pushover analysis procedure for estimating the


seismic demands of high-rise buildings
Tysh Shang Jan ∗, Ming Wei Liu, Ying Chieh Kao
Department of Civil Engineering, Feng Chia University, Taichung, Taiwan, ROC

Received 12 December 2002; received in revised form 28 August 2003; accepted 4 September 2003

Abstract

When evaluating the seismic demands of tall buildings, engineers are more likely to adopt simplified non-linear static analytical
procedures, or pushover analyses, instead of the more complicated non-linear response history analysis. Since the conventional
procedure has some drawbacks in predicting the inelastic seismic demands of high-rise buildings, some improved procedures have
recently been investigated. In this paper, a new simplified pushover analysis procedure, which considers higher mode effects, is
proposed. The basic features of the proposed procedure are: the response spectrum-based higher mode displacement contribution
ratios, a new formula for determining the lateral load pattern and the upper-bound (absolute sum) modal combination rule for
determining the target roof displacement. Five different-height buildings are used as examples to illustrate the application of the
proposed procedure. Three other types of analysis are performed for the purposes of comparison. It will be demonstrated that the
proposed procedure can more precisely estimate important response attributes of tall buildings, such as roof displacements, story
drift ratios and plastic hinge rotations.
 2003 Elsevier Ltd. All rights reserved.

Keywords: Seismic demands; Pushover analysis; Target roof displacement

1. Introduction can provide more accurate predictions for floor displace-


ments and story drift ratios than TLP procedure, but they
In nascent triangular-load pushover analysis (TLP), still have certain drawbacks when high-rise structures
the seismic demands are computed by non-linear static are being evaluated [11,13]. First, predictions of the roof
analysis of the structure subjected to lateral forces. These displacements and story drift ratios are neither precise
forces are monotonically increased—with an invariant nor conservative. Secondly, they cannot accurately esti-
linear height-wise distribution—until a predetermined mate the extent of plastic hinge rotation, which is an
target displacement is reached. When structures are gov- important criterion when appraising the structural safety
erned primarily by first mode response, this procedure of buildings. In order to improve the reliability of the
has been shown to provide a relatively good estimate evaluation of seismic demands for tall buildings, and
of the global deformation response [1,2,4,6,7,12,15,16]. ensure they are conservative, this research was under-
However, when higher-mode effects are important, this taken in order to find a new analytical procedure.
procedure is clearly flawed. Recently, several investi- Since the higher mode effects on seismic demand are
gators [3,8,9,10] have proposed new pushover analysis dependent on the frequency contents of ground motion
procedures, which will primarily consider higher mode and the characteristics of structural systems [14], this
effects, and will follow elastic modal combination rules. paper, first, studies the higher mode effects on the dis-
For low and medium-rise structures, these procedures placement responses based on the response history
analysis (RHA), and then proposes a new procedure for
determining the lateral load pattern and target displace-

Corresponding author. Tel.: +8864-24517250 ext 3111; fax: ment. As practical examples, the following four analyti-
+8864-24516982. cal procedures are used to estimate seismic demands: the
E-mail address: psjan@fcu.edu.tw (T.S. Jan). non-linear response history analysis (NL–RHA), TLP,

0141-0296/$ - see front matter  2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2003.09.003
118 T.S. Jan et al. / Engineering Structures 26 (2004) 117–128

the modal pushover analysis (MPA) and the method of


analysis proposed in this paper. The results obtained by
NL–RHA are treated as ‘exact’ solutions and are com-
pared with those obtained by other procedures. The non-
linear analysis program DRAIN-2D+ [17] is used to pro-
cess various non-linear dynamic and static analyses.

2. Higher-mode contributions to elastic


deformation response

Based on the current seismic code in Taiwan, 2, 5,


10, 20 and 30-storey moment resisting frames of strong
column-weak beam system (see Appendix B) were
designed. Optimal sections of the frames were selected
automatically with the aid of SAP2000 [18]. In order to
analyse the higher-mode contribution to the deformation
response, 13 horizontal records were chosen from Chi-
Chi earthquake in Taiwan, and were applied to the
designed frames. In total, 65 sets of structural motions Fig. 1. The contribution ratios of higher modes to the first mode.
were obtained in the elastic response history analysis.
The procedure for this analysis is outlined as follows:
3. Proposed upper-bound pushover analysis
1. By selecting one of the five designed frames and one procedure
of the 13 horizontal ground motion records üg(t), the
governing equation of motion—after static conden- The principle steps of the proposed pushover analysis
sation—can be written as follows [3]: procedure are as follows:
mü ⫹ cu̇ ⫹ ku ⫽ ⫺m␫üg(t) (1)
1. Select a building and find its elastic structural proper-
where u is a vector of N lateral floor displacements ties—the natural frequencies and mode shapes. Also,
relative to the ground, m, c, and k are the mass, normalize mode shape fn such that the roof degree
classical damping, and lateral stiffness matrices of the of freedom frn = 1 for all modes.
system, and ␫ is the influence column vector with each 2. Use the elastic response spectrum of the selected
element equal to unity. earthquake to determine the upper-bound of the con-
2. The displacement response u(t) can be determined by tribution ratio of the 2nd mode, (q2/q1)UB, as given by
solving Eq. (1) numerically, and a constant vector u0 the following expression (see Appendix E):

冉冊 | |
is selected from u(t) at the time instant when the roof
displacement reaches its elastic maximum. q2 ⌫2D2
⫽ (3)
3. The modal co-ordinate qi,0 of u0 can be obtained as q1 UB ⌫1D1
follows [5]:
where ⌫n (n = 1,2) is the modal participation factor
fTimu0
qi,0 ⫽ (2) and Dn (n = 1,2) is the displacement obtained from
fTimfi the elastic displacement response spectrum.
where fi is the ith natural vibration mode of the struc- 3. Determine the distribution vector of the lateral loads
ture. over the height of the building using the following
formula (see Appendix E):

冉冊
From step 1 to 3, the displacement–response contri-
q2
bution of a higher mode compared to the fundamental fs,UB ⫽ w21mf1 ⫹ w22mf2· (4)
mode can be expressed as a ratio qi,0/q1,0. Fig. 1 shows q1 UB

the ratios obtained for the first four higher modes as a where wn (n = 1,2) is the natural frequency for the
function of the fundamental period. From the envelope nth-mode.
curves of the contribution ratios, it can be seen that the 4. Determine the target roof displacement ur,UB as given
first two modes dominate the displacement response and by the following relationship (see Appendix E):

冋 冉冊册
the third or higher modes can be ignored. Hence, only
the first two modes are considered in the following pro- q2
ur,UB ⫽ ur,TLP 1 ⫹ (5)
posed pushover analysis procedure. q1 UB
T.S. Jan et al. / Engineering Structures 26 (2004) 117–128 119

where ur,TLP is the target displacement predicted by 1. For the 2 and 5-story buildings [Figs. 2(a) and 3(a)],
TLP of which the detailed procedure is presented in for which the 2nd mode effects are not significant,
Appendix A. both TLP and MPA procedures are better than the
5. The seismic demands of a given structure are determ- proposed procedure in predicting floor displacements.
ined by a non-linear static analysis of the structure Although, in the 2-story case, the proposed procedure
subjected to the lateral forces which have a distri- is slightly better in predicting the 2nd floor displace-
bution vector in the form fs,UB [Eq. (4)], and are pro- ment, the error is greater at the 1st floor, as compared
gressively increased from zero until the target dis- to those predicted by the TLP and MPA methods. For
placement ur,UB [Eq. (5)] is reached. the 5-story case, the upper-bound procedure overesti-
mates the 2nd mode contribution ratio and thus under-
estimates the displacements up to the 3rd floor, and
4. Numerical examples overestimates that of the fifth floor, while the dis-
placements predicted by the TLP and MPA methods
For comparison purposes, simulations were made for are in good agreement with those calculated by the
the NL–RHA, TLP, MPA and the proposed upper-bound NL–RHA.
pushover analysis procedure. The structures shown in the 2. For the 10-story building [Fig. 4(a)], the TLP and
first column in Table 1 were designed to meet Taiwan’s MPA procedures overestimate floor displacements at
current earthquake code, their configurations and first the lower stories and underestimate those at the upper
three natural periods being outlined in Appendix B. The stories; the proposed procedure, on the other hand,
earthquakes chosen in column 3 are those having more underestimates floor displacements at the lower sto-
striking second-mode effects on the structure specified ries and overestimates those at the upper stories. The
in the same row, more characteristics of these earth- errors made by all methods are of the same order
quakes being found in Appendix C. The earthquake data though they may be in opposite sign.
were scaled in such a way as to show a PGA, as can be 3. For the 20- and 30-story buildings [Figs. 5(a) and
seen in the table, such that one of the plastic hinge 6(a)], for which the 2nd mode effects are more sig-
rotations of the structure reached 0.015 rad in the NL– nificant, the proposed procedure is much better than
RHA procedure. For the MPA procedure in the follow- the TLP and MPA procedures. For the 20-story build-
ing study, the first three modes were analysed and com- ing, it still underestimates the floor displacements, but
bined according to the SRSS rule. the errors are much smaller than those estimated by
The numerical results are shown in Figs. 2–7. As in TLP and MPA procedures. The contribution of the
Fig. 2, for example, Fig. 2(a,b) show the peak values of 2nd mode is obvious if one deducts the contribution
floor displacements and story drift ratios, respectively, of the linear mode from the total deflections. For the
while the ellipses in Fig. 2(c–f) show the locations of 30-story building, the displacements estimated by the
plastic hinge rotations with the size of each ellipse rep- proposed procedure fit the NL-RHA results extremely
resenting the degree of plastic hinge rotation with a ref- well, while errors made by other procedures are great-
erence size of 0.015 rad on the side of each frame. The er.
values (in rad) of the largest plastic hinge rotations are
written on the sides of the hinges and are shown in Table When predicting story drift ratios, the following
2. The first five largest plastic hinge rotations of each characteristics are observed:
building estimated by different methods are also outlined
in Table 2. The results are discussed in the following 1. For the 2-story building [Fig. 2(b)], the TLP and MPA
sections: accurately estimate the story drift ratio at the first
After estimating floor displacement demands, the fol- story but underestimate it at the second story. The
lowing phenomena were found: proposed procedure overestimates the ratio at the
second story and underestimates it at the first story.
2. For the 5-story building, the TLP and MPA pro-
Table 1 cedures are very effective. They achieve greater accu-
Data of the five analytical cases
racy than the proposed procedure [Fig. 3(b)].
No. of T1 (s) Earthquake PGA q2/q1[ Eq.(3)]
3. For the 10-story building, the TLP and MPA pro-
stories stations (g) cedures overestimate ratios at the lower stories and
underestimate those at the upper stories of the build-
2-storey 0.559 TCU076 0.663 2.02% ing; the proposed procedure, on the other hand, under-
5-storey 1.216 TCU076 0.51 6.03% estimates ratios at the lower stories and overestimates
10-storey 2.077 TCU076 0.43 11.13% those at the upper stories [Fig. 4(b)]. However, the
20-storey 3.425 CHY046 0.439 19.87%
30-storey 4.393 TCU067 0.4 29.72%
MPA procedure is excellent in estimating the ratios
at the near-top stories of the building.
120 T.S. Jan et al. / Engineering Structures 26 (2004) 117–128

Fig. 2. Peak values of floor displacements, story drifts and plastic hinge rotations of a 2-story building (TCU076, PGA = 0.663 g).

Fig. 3. Peak values of floor displacements, story drifts and plastic hinge rotations of a 5-story building (TCU076, PGA = 0.51 g).
T.S. Jan et al. / Engineering Structures 26 (2004) 117–128 121

Fig. 4. Peak values of floor displacements, story drifts and plastic hinge rotations of a 10-story building (TCU076, PGA = 0.43 g).

4. For the 20-story building [Fig. 5(b)], the proposed plastic hinge rotations at upper levels and less severe
procedure is better than the TLP and MPA procedures ones at lower levels of the 10-story building.
in predicting story drift ratios. Although the proposed 4. For the 20- and 30-story buildings, only the proposed
procedure still underestimates ratios at the lower sto- procedure can predict both the locations and severities
ries and overestimates them at the upper stories, the of the plastic hinge rotations with reasonable accu-
errors are much smaller than that estimated by the racy, [Figs. 5(f) and6(f)], as compared with the NL–
TLP and MPA procedures. For the 30-story building RHA procedure [Figs. 5(c) and 6 (c)]. Very small
[Fig. 6(b)], the proposed procedure is much more plastic hinge rotations are estimated by the TLP and
effective than the TLP and MPA procedures in MPA procedures for these two buildings [Figs. 5(d,e)
obtaining story drift ratio profiles at all stories. and 6 and Table 2].

In evaluating plastic hinge rotations, some of follow- Using the numerical results of the NL-RHA procedure
ing observations are made: as a reference point, the errors in roof drift and the errors
in the maximum story drift ratio are shown in Fig. 7(a,b)
1. For the 2-story building, as shown in Fig. 2(c–f) and respectively. As can be seen from Fig. 7(a), the errors
Table 2, all procedures can predict plastic hinge made by all methods are acceptably insignificant for
rotations fairly well. buildings no higher than ten stories, but for buildings
2. For the 5-story building, the plastic hinge rotations taller than ten stories—whereas the errors induced by the
predicted by the TLP, MPA and the proposed pro- TLP and MPA procedures are large and underesti-
cedures [Fig. 3(d–f)] are less severe than that by the mated—only the errors arrived at by the proposed pro-
NL–RHA procedure [Fig. 3(c) and Table 2]. cedure are within an acceptable range. As to the
3. Fig. 4(c–f) and Table 2 show that, in comparison with maximum story drift ratios in Fig. 7(b), the errors from
the estimates by NL–RHA procedure, the MPA and the TLP and MPA procedures are insignificant for build-
TLP procedures predict locations of plastic hinge ings up to ten stories high. However, the errors are unac-
rotation reasonably closely, though with less severity, ceptably large for buildings higher than ten stories, while
while the proposed procedure predicts more severe the errors brought about by the proposed procedure are
122 T.S. Jan et al. / Engineering Structures 26 (2004) 117–128

Fig. 5. Peak values of floor displacements, story drifts and plastic hinge rotations of a 20-story building (CHY046, PGA = 0.439 g).

not too serious, around 16%, for such buildings. Again, higher mode contribution ratios are negligible. More
Fig. 7(a,b) suggest that, in the evaluation of seismic numerical simulations were then performed in order to
demands, the TLP and MPA procedures are excellent for evaluate the floor displacements, story drift ratios and
buildings not taller than 10 stories while the proposed hinge plastic rotations of buildings. This was done by
procedure is very effective for buildings taller than 10 applying non-linear response history analysis (NL–
stories. RHA), triangle linear pushover analysis (TLP), modal
pushover analysis (MPA) and the method proposed in
this paper to the designed structures subjected to earth-
5. Conclusions quakes having more striking higher-mode effects in their
characteristics. The results were fully compared and led
A simplified pushover analysis procedure, which takes to the following conclusions:
into account higher mode effects, for the non-linear seis-
mic evaluation of planar building frames, referred to as 1. For low-rise buildings, e.g. 2 and 5-story, the TLP
the upper-bound method, has been proposed. It can be and MPA procedures are better than the proposed pro-
applied to the evaluation of both newly designed and cedure in predicting the non-linear responses, includ-
existing structures. In this study, the elastic displace- ing floor displacements, story drift ratios and hinge
ment-response contribution ratios of the higher modes, plastic rotations.
as compared to the 1st mode, were obtained through 2. For medium high-rise buildings, e.g. 10-story, the
simulations of five elaborately designed buildings (2–30 TLP and MPA procedures slightly overestimate non-
stories) subjected to 13 strong earthquake motions gener- linear responses at the lower levels and underestimate
ated by the history records of the Chi-Chi Earthquake. those at the upper levels of the building, while the
The results showed that the envelope of the 2nd mode proposed procedure, on the contrary, slightly under-
contribution ratio increases in an approximate linear estimates non-linear responses at the lower levels and
fashion with the height of the building and that other overestimates those at the upper levels of the building.
T.S. Jan et al. / Engineering Structures 26 (2004) 117–128 123

Fig. 6. Peak values of floor displacements, story drifts and plastic hinge rotations of a 30-story building (TCU067, PGA = 0.4 g).

3. For high-rise buildings, e.g. 20 and 30-story, with sig- the strong column-weak beam system and were sub-
nificant higher mode effects on dynamic responses, jected to strong ground motions which had more notice-
only the proposed procedure can reasonably predict able higher mode effects on the structures. The con-
the non-linear responses or seismic demands of the clusions were drawn from the initial data. To affirm
buildings. Although the MPA procedure can predict these conclusions more strongly, this new method, of
the story drift ratios very well at the near-top levels course, should be tested for a wide variety of buildings—
of the buildings, it, along with the TLP procedure, is such as reinforced concrete (RC) frames, multi-bay
unable to accurately predict the hinge plastic rotations steel/RC frames and frames with different beam-to-col-
in this situation. umn stiffness ratios—and a variety of ground motion
4. It is important to note that the proposed procedure patterns. Work in this area continues.
has a tendency to overestimate the seismic demands at
the upper stories and underestimate them at the lower
stories, while the TLP procedure has the opposite
tendency. Therefore, for safety reasons, one should
evaluate the seismic demands for high-rise build- Acknowledgements
ings—of medium or greater height-based on the criti-
cal values (absolute maximum values) that are esti-
mated by both the proposed and TLP procedures. This research is funded by the National Science Coun-
cil, Taiwan, ROC, under grant number NSC-90-2211-E-
This paper has focused on developing an upper-bound 035-014. This financial support is gratefully acknowl-
pushover analysis procedure and its initial evaluation in edged. The authors also appreciate the valuable sugges-
predicting the seismic demands on five different steel tions of the reviewers.
buildings with varying heights. These buildings were of
124 T.S. Jan et al. / Engineering Structures 26 (2004) 117–128

Appendix A. : Predicting roof displacements ur,TLP


of MDOF systems using equivalent SDOF systems
[13]

1. Apply a lateral triangle-pattern load over the height


of the building and conduct a pushover analysis.
2. Idealize the Base Shear (V)–Roof Displacement
(ur,TLP) response of the structure into a bilinear
response as shown in Fig. 8(a). The corresponding
force–deformation characteristics of the equivalent
SDOF system is shown in Fig. 8(b). The transform-
ation formulas are as follows:
fTmf
u∗y ⫽ u (A.1)
fTm1 r,y
Qy∗ ⫽ fTQy (A.2)
where Qy is the story force vector at yield; i.e. Vy =
1TQy.
3. Compute the initial period Teq of the equivalent
SDOF system:
uy∗M∗
Fig. 7. Errors in roof drift and the maximum story drift ratio
demands.
Teq ⫽ 2p 冪Q ∗
y
(A.3)

where M∗ = jTm1, denoting the mass of the equival-


ent SDOF system.

Table 2
Magnitudes and locations of the plastic hinge rotationa

Building NL-RHA MPA TLP Proposed

2-story 0.01500 (B-1-R) 0.01342 (B-1-R) 0.01367 (B-1-R) 0.01713 (B-2-R)


0.01467 (B-2-R) 0.01151 (B-2-R) 0.01204 (B-2-R) 0.01328 (B-1-R)
0.00498 (RC-1-b) 0.00451 (RC-1-b) 0.00458 (RC-1-b) 0.00200 (RC-1-b)
0.00493 (LC-1-b) 0.00441 (LC-1-b) 0.00448 (LC-1-b) 0.00182 (LC-1-b)
0.00126 (B-1-L) 0.00000 0.00000 0.00000
5-story 0.01500 (B-2-L) 0.01387 (B-2-R) 0.01391 (B-2-R) 0.01467 (B-3-R)
0.01233 (B-3-R) 0.01191 (B-1-R) 0.01172 (B-1-R) 0.01364 (B-4-R)
0.01216 (B-4-R) 0.01063 (B-3-R) 0.01146 (B-3-R) 0.01348 (B-2-R)
0.01184 (B-1-L) 0.00480 (B-3-R) 0.00660 (B-3-R) 0.00742 (B-1-R)
0.01114 (B-3-L) 0.00267 (RC-1-b) 0.00255 (B-3-L) 0.00434 (B-5-R)
10-story 0.01483 (B-6-R) 0.01419 (B-4-R) 0.01312 (B-4-R) 0.01747 (B-6-R)
0.01418 (B-5-R) 0.01285 (B-3-R) 0.01182 (B-3-R) 0.01656 (B-7-R)
0.01376 (B-4-R) 0.01213 (B-5-R) 0.01135 (B-5-R) 0.01376 (B-8-R)
0.01239 (B-7-R) 0.01023 (B-2-R) 0.00949 (B-2-R) 0.01366 (B-5-R)
0.01108 (B-3-R) 0.00954 (B-6-R) 0.00916 (B-6-R) 0.00946 (B-4-R)
20-story 0.01500 (B-14-R) 0.00145 (B-4-R) 0.00091 (B-4-R) 0.01696 (B-15-R)
0.01442 (B-13-R) 0.00140 (B-5-R) 0.00072 (B-5-R) 0.01680 (B-16-R)
0.01417 (B-15-R) 0.00129 (B-11-R) 0.00050 (B-3-R) 0.01422 (B-14-R)
0.01355 (B-16-R) 0.00120 (B-9-R) 0.00018 (B-6-R) 0.01350 (B-17-R)
0.01150 (B-17-R) 0.00112 (B-10-R) 0.00014 (B-9-R) 0.01099 (B-13-R)
30-story 0.01500 (B-24-R) 0.00000 0.00000 0.01563 (B-24-R)
0.01465 (B-25-R) 0.00000 0.00000 0.01553 (B-25-R)
0.01395 (B-23-R) 0.00000 0.00000 0.01441 (B-23-R)
0.01181 (B-26-R) 0.00000 0.00000 0.01326 (B-26-R)
0.01025 (B-22-R) 0.00000 0.00000 0.01035 (B-22-R)

a
The letters in parentheses are: B: Beam, C: Column, R: Right, L: Left, b: bottom, t: top, and the middle number represents the story of the
building where the plastic hinge rotation takes place.
T.S. Jan et al. / Engineering Structures 26 (2004) 117–128 125

for the fundamental periods. It is assumed that the earth-


quake movements would be aligned in the direction of
the strong-axis of the columns. All buildings are 7 × 7 m
in the horizontal plan, with 4 m heights for the first story,
and 3.2 m for others. The seismic mass at the top level
is 1784 kg, and 2354 kg for others. The structural sys-
tems are composed of Special Moment Resisting Frames
(SMRF). All buildings are designed to satisfy the opti-
mal sections and the strong column-weak beam philo-
sophy. Table 3 presents the details of, and first three
vibration periods for, various frames.

Appendix C. : Earth tremors

Table 4 shows the 13 earth tremor records, which


were chosen from the Chi-Chi Earthquake in Taiwan,
and their elastic response spectra are shown in Fig. 9.
The paper makes references to these earth tremors using
the Record Name.

Appendix D: Response analyses of linear systems


[5, ch. 13]

Response history analysis


Fig. 8. Force–deformation characteristics of MDOF structure and the
equivalent SDOF system. The differential Eq. (1) governing the response of an
MDOF system to earthquake-induced ground motion is
4. Compute the maximum inelastic displacement u∗m of repeated:
the equivalent SDOF system according to the known mü ⫹ cu̇ ⫹ ku ⫽ ⫺m␫üg(t) (D.1)
ground acceleration ẍg(t), and transform um∗ into the
target displacement ur,TLP of the MDOF structure. The The displacement u of an N-DOF system can be
transformation formula is as follows: expressed as a superposition of the modal contributions:


N
fTm1 ∗
ur,TLP ⫽ T u (A.4) u(t) ⫽ fnqn(t) (D.2)
f mf m
n⫽1

where the modal co-ordinate qn(t) is governed by


In order to obtain a satisfactory ur,TLP and its associa- q̈n ⫹ 2znwnq̇n ⫹ w2nqn ⫽ ⫺⌫nüg(t) (D.3)
ted mode shape f, iteration of step 2 to 4 is required.
Initially, we can assume a linear shape vector (f)1, that in which wn, zn and ⌫n are the natural vibration fre-
is the same as the load pattern, to compute the target quency, damping ratio and modal participation factor of
displacement (ur,TLP)1, and a new shape vector (f)2 by the nth mode. And
normalizing the displacement profile of the structure cor-
fTnm␫
responding to the target displacement (ur,TLP)1. Then, use ⌫n ⫽
the shape vector (f)2 to compute a new target displace- fTnmfn
ment (ur,TLP)2. Repeat the above procedure until the tar- is the solution of qn in Eq. (D.3) is given by
get displacement is converged.
qn(t) ⫽ ⌫nDn(t) (D.4)
where Dn(t) is governed by the equation of motion for
Appendix B. : Configurations of steel frames a SDOF system, with natural frequency wn and damping
ratio zn, subjected to üg(t):
Five different single-bay buildings of 2, 5, 10, 20 and
30 stories are designed in order to cover a broad range D̈n ⫹ 2znwnḊn ⫹ w2nDn ⫽ ⫺üg(t) (D.5)
126 T.S. Jan et al. / Engineering Structures 26 (2004) 117–128

Table 3 Table 4
Member details and first three vibration periods for various frames Chi-Chi earthquake records used in this paper

2-storey Column Beam 20-story Column Beam Station PGA (cm/s/s) Station PGA (cm/s/s)

2 W12×22 W12×16 20 W14×22 W12×19 TCU051(S85W) ⫺164.12 TCU065(S57E) 759.95


1 W14×34 W12×22 19 W14×26 W14×22 TCU053(S62E) ⫺215.63 TCU067(S63E) ⫺554.94
5-story 18 W14×30 W14×30 TCU089(N88W) 342.58 TCU068(N43W) ⫺457.74
5 W14×26 W12×19 17 W14×34 W14×30 TCU049(S59E) ⫺240.41 TCU075(S87E) ⫺323.68
4 W14×30 W12×22 16 W14×38 W14×34 CHY029 (S19E) 253.71 TCU076 (S44E) ⫺414.79
3 W14×34 W14×26 15 W14×43 W14×38 CHY046 (N41W) ⫺159.04 TCU102 (S51W) 252.57
2 W14×38 W14×26 14 W14×48 W14×43 TCU052 (N37W) 499.81
1 W14×43 W14×26 13 W14×53 W14×43
10-story 12 W14×61 W14×48
10 W14×26 W14×19 11 W14×68 W14×48
9 W14×30 W14×22 10 W14×74 W14×53
8 W14×34 W14×26 9 W14×82 W14×53
7 W14×38 W14×30 8 W14×90 W14×61
6 W14×43 W14×30 7 W14×99 W14×61
5 W14×48 W14×34 6 W14×109 W14×61
4 W14×53 W14×34 5 W14×120 W14×61
3 W14×61 W14×38 4 W14×132 W14×61
2 W14×68 W14×38 3 W14×145 W14×61
1 W14×74 W14×38 2 W14×159 W14×61
1 W14×176 W14×61
30-story
30 W14×26 W12×19
29 W14×30 W14×26
28 W14×34 W14×30
27 W14×38 W14×34
26 W14×43 W14×38
25 W14×48 W14×43
24 W14×53 W14×48
23 W14×61 W14×48
22 W14×68 W14×53
21 W14×74 W14×61
20 W14×82 W14×61
19 W14×90 W14×61
18 W14×109 W14×68
17 W14×109 W14×68
16 W14×132 W14×68
15 W14×132 W14×68
14 W14×159 W14×74
13 W14×159 W14×74
12 W14×193 W14×82
11 W14×193 W14×82
10 W14×233 W14×82
9 W14×233 W14×82 Fig. 9. Normalized pseudo-acceleration response spectra (5%
8 W14×257 W14×90 damped).
7 W14×257 W14×90
6 W14×283 W14×90
5 W14×283 W14×90 Substituting Eq. (D.4) into Eq. (D.2) gives the floor dis-
4 W14×311 W14×90 placements
3 W14×311 W14×90


2 W14×342 W14×74 N
1 W14×342 W14×61 u(t) ⫽ ⌫nfnDn(t) (D.6)
Frame T1 (s) T2 (s) T3 (s) n⫽1
30-story 4.393 1.594 0.916
20-story 3.425 1.238 0.712 Making use of Eq. (D.2) and the relation of kfn =
10-story 2.077 0.717 0.403 w2nmfn, the equivalent static forces fs(t) = ku(t) can be
5-story 1.216 0.395 0.199 expressed as
2-story 0.559 0.162


N

fs(t) ⫽ w2nmfnqn(t) (D.7)


n⫽1

and at any instant of time t, these forces fs are the exter-


T.S. Jan et al. / Engineering Structures 26 (2004) 117–128 127

nal forces that produce the displacements u at the same get displacement is reached. Both the force distribution
t, in the stiffness component of the structure. and target displacement are based on the assumption that
The roof displacement ur(t) of u(t) can be expressed the response is controlled by the fundamental mode and
in the following form: that the mode shape remains unchanged after the struc-

冉 冘 冊
ture yields.

N N
jrnqn(t) In Section 2, the conclusion is that the first two modes
ur(t) ⫽ jrnqn(t) ⫽ ur1(t)· 1 ⫹ (D.8)
n⫽1
j q (t)
n ⫽ 2 r1 1
provide a reasonably accurate prediction for the elastic
displacement response of structures to earthquakes, and
where ur1(t) = fr1 q 1(t), representing the roof displace- the third or higher mode can be ignored. Similar con-
ment due to the first mode. If fn is normalized such that clusions can also be found in reference [5, ch.18]. Thus,
the degree of freedom of the roof frn = 1, then Eq. (D.8) if we assume that the displacement response is mainly
can be simplified as controlled by the first two modes, Eqs. (D.11)–(D.13)

冉 冘 冊 N can be reduced to
qn(t)
ur(t) ⫽ ur1(t)· 1 ⫹ (D.9)
⌫2D2

where
q (t)
n⫽2 1
q2
q1

| |
⌫1D1
(E.1)

冘 冘
⌫nDn(t)
N N
qn(t)
⫽ (D.10)
q (t) ⌫ D (t) q2
n⫽2 1 n⫽2 1 1 fs ⫽ q1(w21mf1 ⫹ w22mf2 ) (E.2)
q1
which is a summation of the displacement–response con-
tribution ratio of all higher modes to that of the funda-
mental mode.
q2
ur ⫽ ur1(1 ⫹ ) (E.3)
Response spectrum analysis q1
Since fs is a spatial vector and increases monotonically
In structural design applications, the peak response of from zero, Eq. (E.2) can be simply expressed as
a MDOF system is determined by the response spectrum
for a specific ground motion without carrying out a q2
fs ⫽ w21mf1 ⫹ w22mf2 (E.4)
response history analysis. If one chooses the absolute q1
sum (ABSSUM) modal combination rule to determine
the peak response, Eqs. (D.10), (D.7) and (D.9) become In Eq. (E.3), if ur1 reaches ur,TLP—the target roof dis-
placement contributed by only the 1st mode, ur will

冘 冘| |
⌫nDn
N N reach ur,UB—the target roof displacement contributed by
qn
⫽ (D.11) both the 1st and 2nd modes, as long as the deflected
q
n⫽2 1
⌫D
n⫽2 1 1 shape remains unchanged after the structure yields. Eq.


N (E.3) can therefore be expressed as
fs ⫽ w2nmfnqn (D.12) q2
n⫽1 ur,UB ⫽ ur,TLP(1 ⫹ ) (E.5)
q1
ur ⫽ ur1 1 ⫹冉 冘 冊 q
N
qn
n⫽2 1
(D.13)

References
In Eq. (D.11), qn = |⌫nDn| and Dn ⬅ D(Tn, zn)—the
deformation response spectrum ordinate corresponding
[1] Albanesi T, Nuti C, Vanzi I. A simplified procedure to assess
to the natural period Tn and damping ratio zn (n = 1, the seismic response of non-linear structures. Earthquake Spectra
2,..., N). In Eq. (D.13), ur1 is the roof displacement con- 2000;16(4):715–34.
tributed by the 1st mode. [2] Bracci JB, Kunnath SK, Rrinhorn AM. Seismic performance and
retrofit evaluation of reinforced concrete structures. J Struct
Engng 1997;123(1):3–10.
[3] Chopra AK, Goel RK. A modal pushover analysis procedure for
Appendix E. : Explanation of Eqs. (4) ad (5) estimating seismic demands for buildings. Earthquake Engng
Struct Dynam 2002;31:561–82.
In the pushover analysis of FEMA-273, the non-linear [4] Chopra AK, Goel RK. Evaluation of NSP to estimate seismic
deformation: SDF systems. J Struct Engng 2000;126(4):482–90.
static analysis is performed under the condition that the [5] Chopra AK. Dynamic of structures: theory and applications to
spatial distribution of the lateral forces to the structure earthquake engineering. Englewood Cliffs, NJ: Prentice-Hall,
are invariant and monotonically increasing until the tar- 1995.
128 T.S. Jan et al. / Engineering Structures 26 (2004) 117–128

[6] Fajfar P, Gaspersic P. The N2 method for the seismic damage analysis of seismic performance evaluation. Engng Struct
analysis of RC buildings. Earthquake Engng Struct Dynam 1998;20(4-6):452–64.
1996;25:31–46. [13] Liu MW, Kao YC, Jan TS. A new pushover analysis procedure
[7] Gupta A, Krawinkler H. Estimation of seismic drift demands for for high-rise buildings. The 25th Conference on Theoretical and
frame structures. Earthquake Engng Struct Dynam Applied Mechanics, Hu-Wei, Taiwan, ROC; 2002, p. D008.
2000;29:1287–305. [14] Miranda E. Evaluation of site- dependent inelastic design spectra.
[8] Gupta B. Enhanced pushover procedure and inelastic demand J Struct Engng 1993;119:1319–38.
estimation for performance-based seismic evaluation of buildings, [15] Seneviratna GDPK Evaluation of inelastic MDOF effects for seis-
Ph.D. dissertation, University of Centro Florida, Orlando, FL; mic design. Ph.D. dissertation, Dept. of Civil Engineering, Uni-
1999. versity of Stanford; 1995.
[9] Gupta B, Kunnath SK. Adaptive spectra-based pushover pro- [16] Saiidi M, Sozen MA. Simple non- linear seismic analysis of R/C
cedure for seismic evaluation of structures. Earthquake Spectra structures. J Struct Div 1981;107:937–51.
2000;16(2):367–91. [17] Tsai KC, Li JW. DRAIN 2D+, a general purpose computer pro-
[10] Han SW, Wen YK. Method of reliability-based seismic design. gram for static and dynamic analyses of inelastic 2D structures
I: equivalent non-linear system. J Struct Enging supplemented with a graphic processor, VIEW2D. Report no.
1997;123(3):256–65. CEER/R83-03, Center for Earthquake Engineering Research,
[11] Jan TS, Liu MW, Kao YC. Pushover analysis and seismic National Taiwan University; 1994.
demands of structures to near-fault earthquakes, NSC 90-2211- [18] Wilson EL, Habibullah A. SAP—2000 integrated finite element
E- 035-014, Taiwan, ROC; 2002. analysis and design of structures. Computers and Structures
[12] Krawinkler H, Seneviratna GDPK. Pros and cons of a pushover Inc, 1998.

S-ar putea să vă placă și