Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Departamento de Ingenieria Estructural y Geotecnica, Pontificia Universidad Catolica de Chile, Av. Vicuna Mackenna 4860, Macul, Santiago 782-0436, Chile
Department of Civil, Structural & Environmental Engineering, University at Buffalo, 212 Ketter Hall, Buffalo, NY 14260, USA
article
info
Article history:
Received 2 July 2007
Received in revised form
28 May 2008
Accepted 12 June 2008
Available online 18 June 2008
Keywords:
Seismic pounding
Separation distance
a b s t r a c t
This study examines the accuracy of the Double Difference Combination (DDC) rule (also known simply
as the CQC rule) in predicting the separation necessary to prevent seismic pounding between linear
structural systems. Seismic excitations were modeled as modulated and filtered modulated Gaussian
white noise random processes, and adjacent structures were modeled as 5%-damped SDOF systems having
a wide range of values of natural periods. Results obtained through Monte Carlo simulations indicate that
the accuracy of the DDC rule depends not only on the ratio of the natural periods of the structures, but
also on the relationship between the values of the natural periods and the value of the period associated
with the main frequency of the excitation.
2008 Elsevier Ltd. All rights reserved.
1. Introduction
Seismic pounding occurs when the separation between adjacent buildings is not large enough to accommodate the displacement response of the structures relative to each other during
earthquake events. Depending on the characteristics of the colliding buildings [1], pounding might cause severe structural damage in some cases [2], and even collapse is possible in some
extreme situations [3]. Further, even in those cases where it does
not result in significant structural damage, pounding always induces higher floor accelerations in the form of large magnitude,
short duration pulses [4], which in turn cause greater damage to
building contents [2]. For these reasons, it is widely accepted that
pounding is an undesirable phenomenon that should be prevented
or mitigated. This is recognized in seismic design codes and regulations worldwide, which typically specify minimum separations to
be provided between adjacent buildings. For instance, according
to the 2003 edition of the International Building Code, minimum
separations are given by:
S = XA + XB
(adjacent buildings separated by a property line)
S=
XA2 + XB2
(1)
Corresponding author. Tel.: +56 2 354 7684; fax: +56 2 354 4243.
E-mail address: dlg@ing.puc.cl (D. Lopez-Garcia).
0266-8920/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.probengmech.2008.06.002
(2)
S=
XA2 + XB2 2 XA XB
(3)
T
TA
8 A B A + B TA
TB
B
=
(4)
2 2
2
T A 2
T
T
TA
2
2
+ 4A B 1 + TA
+
4
1 TA
A
B
T
T
B
211
(5)
(6)
where:
b
a=
TD
b=
1 + [ln ( 1)]
c=
(7)
ln ()
TD
(8)
(9)
UREL (t ) = UA (t ) UB (t ) .
(10)
212
Fig. 3. Sample realization of a modulated white noise excitation process (TD = 15 s).
Estimates of XREL obtained using the DDC rule (i.e., calculated using
Eqs. (3) and (4)) are denoted by S.
Monte Carlo simulations (500 samples) are performed as
follows. Realizations u g (t) of the excitation process U g (t) are
generated according to standard simulation techniques [19,20].
Realizations uA (t) and uB (t) of response processes UA (t) and UB (t)
are obtained by numerically solving the corresponding equations
of motion, i.e.:
u A (t ) +
u B (t ) +
4 A
TA
4 B
TB
u A (t ) +
u B (t ) +
4 2
TA2
4 2
TB2
uA (t ) = u g (t )
(11)
uB (t ) = u g (t )
(12)
(13)
also been shown [23] that the effects of traveling seismic waves are
also negligible when the distance between the adjacent structures
is relatively small. For these reasons, it was assumed that both
adjacent structures experience the same excitation at the same
time.
Finally, it must be noted that the extreme values of response
processes UA , UB and UREL calculated with Eqs. (14)(16) are
double-sided extreme values, while the values needed to correctly
estimate the separation necessary to avoid pounding are actually
one-sided extreme values [10,11]. However, double-sided extreme
values are considered because, in practice, the quantities XA and
XB in Eq. (3) are always estimates of double-sided extreme values,
and then, for consistency, values of S obtained using doublesided extreme values of processes UA and UB (i.e., XA and XB )
are compared with double-sided extreme values of process UREL
(i.e., XREL ). Monte Carlo simulations performed considering all the
excitation processes and all the combinations of natural periods TA
and TB that will be described later indicate that the ratio of doublesided extreme values to one-sided extreme values of the response
process UREL ranges from 1.01 to 1.10, i.e., double-sided extreme
values of UREL are only slightly conservative (by 10% at most).
(14)
(15)
(16)
213
Fig. 4. Sample realization uA (t ), uB (t ) and uREL (t ) of displacement response processes UA (t ), UB (t ), UREL (t ). Excitation process: modulated white noise (TD = 15 s).
(t), UB (t) and UREL (t) become more similar to those corresponding
to the stationary conditions under which the equation of the DDC
rule was derived.
4. Response to modulated filtered white noise
More realistic seismic excitation processes can be obtained
by filtering, in the frequency domain, a process having whitenoise characteristics. A widely used filtered white noise excitation
process is that defined by the modified KanaiTajimi equation,
sometimes also referred to as the CloughPenzien equation, which
is given by [19]:
Sg () = HCP () HKT () S0
(17)
where:
HKT () =
g4 + 4g2 g2 2
2
g2 2 + 4g2 g2 2
(18)
214
Fig. 5. Values of the S /XREL ratio: modulated white noise excitation process.
4
2
f2 2 + 4f2 f2 2
(19)
r
1 +
m =
1 + 8g2
2g
(20)
215
Fig. 6. Values of the S /XREL ratio: modulated white noise excitation process.
DDC rule is now always conservative when TA /TB > 0.80 rather
than when TA /TB > 0.75. Secondly, when TA /TB < 0.80, the S /XREL
ratio is now a function not only of the period ratio but also, up
to some degree, of the values of TA and TB as well (Fig. 11 shows
that, the greater the values of TA and TB , the more unconservative
the DDC rule, up to 13% in this case). Finally, the DDC rule is
now not always unconservative when TA /TB < 0.80, i.e., Fig. 10
shows that there are a few cases for which S /XREL > 1 when
TA /TB < 0.80. It was found that periods TA and TB are, in all of
these cases, relatively small. More insight into this last observation
is provided by Fig. 12, which was obtained by considering several
pairs of adjacent structures A and B, where TA = 0.050 s,
0.055 s, . . . , 0.750 s and TB = 0.050 s, 0.055 s, . . . , 0.750 s.
When the excitation is a modulated white noise (Fig. 12, left),
the contour lines separating the conservative (S /XREL > 1) and
unconservative (S /XREL < 1) regions essentially coincide with
the lines analytically expressed by TB = 0.75TA and TB = 1.33TA ,
an observation that is entirely consistent with what was observed
216
Fig. 7. Modified KanaiTajimi modulated filtered white noise excitation process: power spectral density function Sg () (g = 12.50 rad/s, g = 0.60, f = 2.00 rad/s,
f = 0.70 and S0 = 200 cm2 /s3 ).
Fig. 8. Modified KanaiTajimi modulated filtered white noise excitation process (g = 12.50 rad/s, g = 0.60, f = 2.00 rad/s, f = 0.70 and S0 = 200 cm2 /s3 ): sample
realization (TD = 30 s).
Fig. 9. Modified KanaiTajimi modulated filtered white noise excitation process (g = 12.50 rad/s, g = 0.60, f = 2.00 rad/s, f = 0.70 and S0 = 200 cm2 /s3 and
TD = 30 s): mean pseudo-acceleration response spectrum (damping ratio = 0.05).
217
Fig. 10. Values of the S /XREL ratio: modified KanaiTajimi modulated filtered white noise excitation process (g = 12.50 rad/s, g = 0.60, f = 2.00 rad/s, f = 0.70,
S0 = 200 cm2 /s3 and TD = 30 s).
Fig. 11. Values of the S /XREL ratio: modified KanaiTajimi modulated filtered white noise excitation process (g = 12.50 rad/s, g = 0.60, f = 2.00 rad/s, f = 0.70,
S0 = 200 cm2 /s3 and TD = 30 s).
Fig. 12. Contour lines of values of the S /XREL ratio: [left] modulated white noise (TD = 30 s); [right] modified KanaiTajimi modulated filtered white noise excitation process
(g = 12.50 rad/s, g = 0.60, f = 2.00 rad/s, f = 0.70, S0 = 200 cm2 /s3 and TD = 30 s).
218
Fig. 13. Modified KanaiTajimi modulated filtered white noise excitation process (g = 7.50 rad/s, g = 0.30, f = 2.00 rad/s, f = 0.70, S0 = 200 cm2 /s3 and
TD = 30 s): [top] sample realization; [bottom] mean pseudo-acceleration response spectrum (damping ratio = 0.05).
219
Fig. 14. Values of the S /XREL ratio: modified KanaiTajimi modulated filtered white noise excitation process (g = 7.50 rad/s, g = 0.30, f = 2.00 rad/s, f = 0.70,
S0 = 200 cm2 /s3 and TD = 30 s).
Fig. 15. Contour lines of values of the S /XREL ratio: modified KanaiTajimi
modulated filtered white noise excitation process (g = 7.50 rad/s, g = 0.30,
f = 2.00 rad/s, f = 0.70, S0 = 200 cm2 /s3 and TD = 30 s).
Fig. 18 confirms once again that the accuracy of the DDC rule may
be characterized in
q terms of Tm : as before, in the region roughly
defined by TB <
220
Fig. 16. Modified KanaiTajimi modulated filtered white noise excitation process (g = 2.50 rad/s, g = 0.10, f = 0.30 rad/s, f = 0.70, S0 = 200 cm2 /s3 and
TD = 30 s): [top] sample realization; [bottom] mean pseudo-acceleration response spectrum (damping ratio = 0.05).
Fig. 17. Values of the S /XREL ratio: modified KanaiTajimi modulated filtered white noise excitation process (g = 2.50 rad/s, g = 0.10, f = 0.30 rad/s, f = 0.70,
S0 = 200 cm2 /s3 and TD = 30 s).
221
6. Conclusions
Fig. 18. Contour lines of values of the S /XREL ratio: modified KanaiTajimi
modulated filtered white noise excitation process (g = 2.50 rad/s, g = 0.10,
f = 0.30 rad/s, f = 0.70, S0 = 200 cm2 /s3 and TD = 30 s).
ratios are shown in Fig. 22. The latter figure indicates again that
the accuracy of the DDC rule can be characterized, for practical
purposes, in terms of Tm : in the region roughly defined by TB <
Tm TA , the DDC rule is always conservative regardless of the value
of the period ratio.
222
Fig. 20. Modulated filtered white noise excitation processes compatible with design response spectra (TD = 30 s): power spectral density functions Sg ().
Fig. 21. Modulated filtered white noise excitation processes compatible with design response spectra (TD = 30 s): sample realizations.
when the value of the TA /TB ratio is less than 0.75, the DDC
rule is always unconservative, at most by 13% in the case
of seismic excitations expected at most sites, and up to 20%
in the extreme case of excitations having markedly narrowband characteristics.
(3) When the seismic excitation is characterized in terms of a design response spectrum, the period Tm associated with the main
frequency m is the period defining the boundary between
223
Fig. 22. Contour lines of values of the S /XREL ratio: modulated filtered white noise excitation process compatible with design response spectra (TD = 30 s).
[8] Grigoriu M. Mean-square structural response to stationary ground acceleration. ASCE Journal of the Engineering Mechanics Division 1981;107(5):
96986.
[9] Penzien J. Evaluation of building separation distance required to prevent
pounding during strong earthquakes. Earthquake Engineering & Structural
Dynamics 1997;26(8):84958.
[10] Hong HP, Wang SS, Hong P. Critical building separation distance in reducing
pounding risk under earthquake excitation. Structural Safety 2003;25(3):
287303.
[11] Wang SS, Hong HP. Quantiles of critical separation distance for nonstationary
seismic excitations. Engineering Structures 2006;28(7):98591.
[12] Lin JH. Separation distance to avoid seismic pounding of adjacent buildings.
Earthquake Engineering & Structural Dynamics 1997;26(3):395403.
[13] Lin JH, Weng CC. Spectral analysis on pounding probability of adjacent
buildings. Engineering Structures 2001;23(7):76878.
[14] Filiatrault A, Cervantes M, Folz B, Prion H. Pounding of buildings during
earthquakes: A Canadian perspective. Canadian Journal of Civil Engineering
1994;21(2):25165.
[15] Filiatrault A, Cervantes M. Separation between buildings to avoid pounding
during earthquakes. Canadian Journal of Civil Engineering 1995;22(1):16479.
[16] Valles RE, Reinhorn AM. 1997, Evaluation, prevention and mitigation of
pounding effects in buildings structures. Technical report no. NCEER-97-0001,
National Center for Earthquake Engineering Research, Buffalo, USA.
[17] Saragoni GR, Hart GC. Simulation of artificial earthquakes. Earthquake
Engineering & Structural Dynamics 1974;2(3):24967.
[18] Boore DM. Stochastic simulation of high-frequency ground motions based on
seismological models of the radiated spectra. Bulletin of the Seismological
Society of America 1983;73(6A):186594.
[19] Clough RW, Penzien J. Dynamics of structures. 2nd ed. New York (USA):
McGraw-Hill; 1993.
[20] Soong TT, Grigoriu M. Random vibration of mechanical and structural systems.
Englewood Cliffs (USA): Prentice-Hall; 1993.
[21] Hao H, Liu X. Estimation of required separations between adjacent structures
under spatial ground motions. Journal of Earthquake Engineering 1998;2(2):
197215.
[22] Hao H, Zhang SR. Spatial ground motion effect on relative displacement of
adjacent building structures. Earthquake Engineering & Structural Dynamics
1999;28(4):33349.
[23] Jeng V, Kasai K. Spectral relative motion of two structures due to seismic travel
waves. ASCE Journal of Structural Engineering 1996;122(10):112835.
[24] ASCE. Minimum design loads for buildings and other structures (ASCE 7-05).
Reston (USA): American Society of Civil Engineers; 2005.
[25] Der Kiureghian A, Neuenhofer A. Response spectrum method for multisupport
seismic excitations. Earthquake Engineering & Structural Dynamics 1992;
21(8):71340.
[26] Park YJ. New conversion method from response spectrum to PSD functions.
ASCE Journal of Engineering Mechanics 1995;121(12):13912.
[27] Fischer T, Alvarez M, De la Llera JC, Riddell R. An integrated model for
earthquake risk assessment of buildings. Engineering Structures 2002;24(7):
97998.