Vous êtes sur la page 1sur 26

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2002; 31:693718 (DOI: 10.1002/eqe.149)

Earthquake-induced 0oor horizontal accelerations in buildings


M. E. Rodriguez1; ; , J. I. Restrepo2 and A. J. Carr 3
1National

University of Mexico; Apartado Postal 70-290; CP 04510; Mexico City; Mexico


of Structural Engineering; University of California; San Diego; 9500 Gilman Drive;
La Jolla; CA 92093-0085; U.S.A.
3 Department of Civil Engineering; University of Canterbury; Private Bag 4800; Christchurch; New Zealand
2 Department

SUMMARY
Floor horizontal accelerations are needed for obtaining forces for the design of diaphragms, for the
design of their connections and for the design of non-structural components and equipment supported
by structures. Large 0oor horizontal accelerations have been recorded in buildings during earthquakes.
Such accelerations have been responsible for inertia forces causing damage to services and are a major
reason for structural damage and even building collapse.
This paper describes an analytical investigation into earthquake-induced 0oor horizontal accelerations
that arise in regular buildings built with rigid diaphragms. The paper also describes several methods
prescribed by design standards and proposes a new method. The method is based on modal superposition
modiAed to account for the inelastic response of the buildings lateral force resisting system. Results
obtained from time-history inelastic analysis are compared with the proposed method. Copyright ? 2001
John Wiley & Sons, Ltd.
KEY WORDS:

0oor accelerations; diaphragms; non-structural components; non-linear analysis; seismic


demands; building codes

1. INTRODUCTION
The impact and cost of the consequences of damage caused by earthquakes worldwide during
the past 12 years has raised the question of whether current building seismic design procedures
are satisfying the needs of modern society. Most seismic design standards are based on a lifeprevention approach where structural damage is accepted providing that collapse is avoided.
No other economic parameters, such as the cost of damage to equipment and stored goods and
the cost associated with loss of operation following a moderate=strong earthquake, are currently

Correspondence to: M. E. Rodriguez, National University of Mexico, Apartado Postal 70-290, CP 04510, Mexico
City, Mexico.
E-mail: mrod@servidor.unam.mx
Contract=grant sponsor: New Zealand Foundation for Research Science and Technology; contract=grant number:
UOC 808.

Copyright ? 2001 John Wiley & Sons, Ltd.

Received 15 January 2001


Revised 31 August 2001
Accepted 31 August 2001

694

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 1. Floor acceleration magniAcation obtained from records in instrumented buildings during the
Northridge earthquake (Adapted from Hall, 1995).

accounted for in the design process. The philosophy behind the newly proposed performancebased seismic design intends to encompass, directly or indirectly, those parameters within a
set of objectives aimed at ensuring predictable behaviour of the entire building envelope [1].
Within the framework of performance-based seismic design, signiAcant eLort has been made
in recent years at developing new methodologies [2; 3] and new structural systems in a way
that the design objectives can be achieved [4]. The design methodologies tend to give more
emphasis on obtaining better estimates of lateral deformations. Nevertheless, there are areas
relevant to performance-based seismic design that have been the subject of limited research
work. For example, the determination of 0oor horizontal accelerations in buildings with
emphasis on practical applications is one of them [58]. Floor accelerations are needed for
obtaining in-plane forces for the design of diaphragms and their connections to the primary
lateral force resisting system. Of particular importance are those diaphragms that incorporate
signiAcant openings and are built using precast concrete 0oor systems. Furthermore, 0oor
horizontal accelerations are needed for determining forces for the design of non-structural
components and equipment supported on the 0oors.
It has been reported that damage to diaphragms and their connections was a major cause
of poor building behaviour, and even collapse, during the 1988 Armenia [9], the 1994
Northridge [10; 8] and the 1999 central Colombia [11] earthquakes. It has also been reported
that damage to services caused business interruption in several buildings in the Northridge
earthquake [12]. In fact, records obtained during the Northridge earthquake in multistorey
buildings, other than base-isolated, showed that 0oor peak horizontal accelerations were
generally greater than those recorded at the ground level. Figure 1 plots the maxima 0oor
acceleration magniAcation versus peak ground acceleration (PGA) reported by Hall [12] for
25 multistorey buildings. The maxima 0oor acceleration magniAcation was obtained as the
ratio of the maxima 0oor horizontal acceleration to the PGA. Floor acceleration magniAcations ranged between 1.1 and 4.6. These results are in agreement with those found by Soong
et al. [6] for data also obtained in the 1994 Northridge earthquake. Large 0oor accelerations
have been also observed in experimental work. For example in the pseudo-dynamic test of a
large-scale Ave-storey precast concrete building reported by Priestley et al. [3] 0oor acceleraCopyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

695

tions computed from the loading algorithm led to in-plane 0oor forces that were in excess of
those estimated with the Uniform Building Code [13]. These researchers suggested that the
augmented forces were caused by higher modes. Moreover, other researchers have reported
evidence of the importance of higher mode eLects on in-plane 0oor forces. Eberhard and
Sozen [14], examining the base-shear response of structures in terms of incremental stiLness
and using modal analysis, found that after a mechanism formed, further changes in force
distribution occurred as a result of the 0uctuation of the higher modes.
The main objective of the paper is to improve the current understanding on how 0oor horizontal accelerations arise during earthquakes in buildings. Emphasis is given to the interaction
between a building non-linear response and the magnitude of the 0oor accelerations. Using an
approach based on the concepts of linear structural dynamics, the authors propose a simple
method for determining 0oor horizontal accelerations and design forces.

2. REVIEW OF CURRENT DESIGN PROVISIONS


A review of seismic design provisions revealed that several approaches are used in the
evaluation of the horizontal forces needed for the design of diaphragms, non-structural
components and equipment in buildings. For example, and as described later in this section, the
UBC [13] requires for the design of diaphragms horizontal forces greater than those used in the
design of the lateral force resisting system. The UBC also contains provisions for the design
of non-structural components and equipment supported at the 0oors. These provisions are
diLerent from those for the design of diaphragms. In Mexico the design forces for the lateral
force resisting system and for the diaphragms are generally considered equal. In New Zealand,
the Loadings Standard [16], gives provisions for the design of building parts. According
to this standard, diaphragms can be designed as a rigid part, which is a non-structural
component or equipment having a short-period of vibration such that the response ampliAcation under earthquake loading is negligible. The approach given by the Loadings Standard
accounts for the eLects of overstrength in lateral force resisting systems and leads to forces
that are greater than the design forces required for the primary lateral force resisting system. It
is interesting to note that in New Zealand, diaphragms can also be designed for forces derived
from the Concrete Structures Standard [17]. The procedure for the design of diaphragms in this
standard is based on the horizontal forces used for the design of the primary lateral force
resisting system multiplied by an overstrength factor. Design provisions of New Zealand, UBC
and Mexico are described below in more detail.
2.1. New Zealand design standards
In New Zealand, 0oor horizontal accelerations can be derived from the Loadings Standard [16]
or from the Concrete Structures Standard [17]. The Loadings Standard [16] gives recommendations for the design of rigid parts in buildings. Its commentary implies that in-plane forces
in diaphragms can be evaluated using provisions for rigid parts. The 0oor horizontal force,
Fph , inferred from the standard is,
Fph = 2Wp Cfi Rp
Copyright ? 2001 John Wiley & Sons, Ltd.

(1)

Earthquake Engng Struct. Dyn. 2002; 31:693718

696

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

where Fph is the design horizontal force, Wp is the weight of the part, Rp is the risk factor
for the part, which varies from 1 to 1.1 and Cfi is a basic horizontal seismic coePcient for
a part located at level i.
At ground level, the coePcient Cfi becomes Cfo , which for the ultimate limit state is
deAned as:
Cfo = 0:25RZ

(2)

where R is the building risk factor, which, for buildings other than of secondary nature, varies
between 1 and 1.3, and Z is the seismic zone factor, which varies between 0.6 and 1.2.
The 0oor horizontal seismic coePcient for the uppermost level of a building, Cfn , is deAned
as,
Cfn =

Ch (T1 ;  o ) Fn
Ch (T1 ; ) Wn

(3)

where Fn is the horizontal force evaluated for the primary lateral force resisting system at the
uppermost level of the building, Wn is the seismic weight at that level. Parameter Ch (T1 ; )
is the seismic hazard acceleration coePcient obtained from the design spectrum for the fundamental period of the building, T1 , and for the design structural ductility factor ; Ch (T1 ;  o )
is the seismic coePcient obtained from the design spectrum for the period T1 and ductility
factor  o associated with overstrength. The Loadings Standard recommends the use of  o = 1
unless capacity design is applied to the structure to justify a larger value. The standard does
not give explicit guidelines for calculating  o . This lack of guidance has usually meant that
design engineers generally select the default value of  =  o = 1. When using unity values
for  and  o , Equation (3) simpliAes to Cfn = Fn =Wn , which is the seismic coePcient at the
uppermost principal seismic weight associated with the buildings elastic response.
CoePcient Cfi for any level i below the uppermost level of a building can be found
using two diLerent methods depending on whether the equivalent static method or the modal
response spectrum method is being used in the structural analysis.
Where the modal response spectrum method of analysis is used, coePcient Cfi is given by,
Cfi =

Ch (T1 ;  o ) Fi
Ch (T1 ; ) Wi

(4)

where Fi is the horizontal force found for the primary lateral force resisting system for level
i and Wi is the seismic weight at this level.
It is interesting to note that 0oor accelerations can also be derived in New Zealand from
the provisions for the seismic design of diaphragms given by the Concrete Structures Standard
[17]. The importance of the structural role of a diaphragm has always been recognized by the
standard to the extent that design recommendations are contained in a separate chapter. The
procedure outlined in the current standard is based on the principles of capacity design, where
design diaphragm forces are equal to the design lateral forces multiplied by an overstrength
factor. The 0oor horizontal acceleration inferred from this standard can be obtained by dividing
the diaphragms design lateral force by its mass.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

697

2.2. Uniform Building Code [13]


The 0oor horizontal force deAned by the UBC for non-structural components and equipment
is,


ap Ca Ip
hx
Fp =
1+3
Wp
(5)
Rp
hn
where ap is the structure ampliAcation factor for the part, Rp is the response modiAcation
factor for the part, and hx and hn are the part elevation measured from the ground and the
uppermost level, respectively. CoePcients ap and Rp vary from 1.0 to 2.5 and from 1.0 and
4.0, respectively. When using Equation (5) Fp shall be no less than 0:7Ca Ip Wp and need not
be more than 4:0Ca Ip Wp . The horizontal force in elastic parts inferred from Equation (5) can
be determined using ap = 1:0 and Rp = 1:0.
The UBC requires diaphragms to be designed to resist the force Fpx given by:

Ft + ni=x Fi
Fpx = n
wpx
(6)
i=x wi
where the horizontal force Ft is the portion of the design base shear force, V; concentrated at
the top level of the structure in addition to the horizontal force Fn . Fx is the design horizontal
force applied to level x, wi and wx are the portion of the total seismic dead weight, W; located
or assigned to level i or x, respectively; and wpx is the weight of the diaphragm at level x.
Forces Fn and Fx are derived from the equivalent static method.
Force Fpx shall not be less than 0:5Ca Iwpx , and nor greater than 1:0Ca Iwpx , where I is the
building importance factor. This factor varies between 1.0 and 1.5.
The design base shear force V in this code is a fraction of the force required for elastic
response. The ratio of the force required for elastic response and the design force is deAned as
the R factor. This factor varies between 2.2 and 8.5. According to the deAnition given by the
UBC, R is a numerical coePcient representing the inherent ductility capacity and overstrength
of the lateral force resisting system.
2.3. Mexico city building code [15]
As noted above, designers in Mexico use for the design of diaphragms the horizontal forces
derived for the primary lateral force resisting system. These forces are obtained by dividing
the elastic forces by a factor that recognizes the ductility and overstrength inherent in the
structural system.
2.4. Application of current design provisions
To illustrate the application of the approaches described above, forces for the design of
diaphragms and non-structural components were obtained for a regular twelve-storey building
whose primary lateral force resisting system consisted on ductile reinforced concrete structural
walls. Floor accelerations were determined by dividing the design forces by the mass of the
diaphragm or partition in consideration. Floor accelerations were normalized by the peak
ground acceleration obtained from the design spectrum. These ratios are referred to as 0oor
acceleration magniAcations, which will be used extensively in the sections below.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

698

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 2. Floor acceleration magniAcations derived for a twelve-storey wall building.

Figure 2(a) compares the 0oor acceleration magniAcations found from the two methods
given by the New Zealand Standards. Floor acceleration magniAcations were calculated using
the modal response spectrum method of the Loading Standard [16] for  = 5 with the default
value of  o = 1. The design response spectrum for intermediate soils was used in this example. Floor acceleration magniAcations were also determined from the procedure given in the
Concrete Structures Standard [17] assuming an overstrength factor,  = 1:8. Factor  is deAned as the ratio of the maximum moment at the base of walls obtained from the non-linear
analysis to the design bending moment. These methods reveal contrasting diLerences. For
example the magniAcation found from the modal response spectrum method of the Loadings
Standard gives a top 0oor magniAcation of 4.9, while a value of 0.5 is obtained from the
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

699

Concrete Structures Standard method. In fact, the values obtained from the latter standard
suggest that there is de-ampliAcation of the ground acceleration at all levels, which is contrary
to the data plotted in Figure 1.
Figure 2(b) shows the 0oor acceleration magniAcations found from the UBC [13] design
recommendations for the design of diaphragms. The base shear force for this building was
found using the equivalent static method with R = 4:5 and I = 1:0. The peak ground acceleration to estimate the 0oor magniAcation was assumed equal to Ca I . Two observations can be
made from the results obtained: (i) the values obtained from the analyses are always within
0.5 and 1.0, which indicates 0oor acceleration de-ampliAcation rather than magniAcation, and,
(ii) the values depend on the seismic coePcient Ca . Also plotted in Figure 2(b) is the 0oor
acceleration magniAcation derived from the recommendation for the design of parts. In this
case the peak ground acceleration was estimated to be equal to Ca Ip . CoePcients ap and Rp
for this example were made equal to 1.0. As can be seen in Figure 2(b), 0oor acceleration
magniAcations using the design procedure for parts are signiAcantly greater than those found
using the design provisions for diaphragms.
For diaphragms designed in accordance with the Mexico City Building code [15] 0oor
acceleration magniAcations are expected to be lower than those derived using procedures
from the New Zealand Concrete Structures Standard [17] and from the UBC [13]. The reason
for this is the later codes deAne 0oor horizontal forces that are larger than those derived for
the primary lateral force resisting system.
In summary, 0oor acceleration magniAcations derived from the diLerent approaches diLer
signiAcantly in their results. In some cases the magniAcations obtained are too high when
compared to the results obtained during the 1994 Northridge earthquake. In other cases the
results actually indicate de-ampliAcation at the acceleration at the 0oors. The consequence of
such diLerences can be of signiAcance when performance-based is implemented in the design
of diaphragms or parts.
3. ALTERNATIVE METHOD FOR EVALUATING FLOOR HORIZONTAL
DESIGN FORCES
3.1. E3ect of non-linear response on the magnitude of the 4oor acceleration magni5cation
Parametric non-linear time history analyses were performed on three, six and twelve-storey
high buildings to observe the eLects of diLerent variables on the development and magnitude
of the 0oor acceleration. In these analyses the diaphragms were considered rigid in their plane.
The plan view of a typical 0oor on the three- and twelve-storey high buildings is shown in
Figure 3. As can be seen there, four 4:7 m long by 250 mm thick cantilever walls provide
the lateral force resistance in the short direction of the three-storey high building, whereas
the lateral force resistance in the short direction of the twelve-storey high building is provided by eight 7 m long by 250 mm thick cantilever walls. The six-storey high building, not
shown here, had a distribution, number and thickness of structural walls similar to those of
the three-storey building but with lengths equal to 7 m [18]. The 0oor system was assumed
to consist of one-way precast prestressed 0oor units with 80 mm of cast-in-place concrete
topping. The response of the buildings to earthquake input ground motion was investigated
for the short direction only. The seismic weight per 0oor was assumed equal to 4700 kN. It
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

700

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 3. Plan view of typical 0oor of the prototype three- and twelve-storey buildings investigated.

was assumed that the walls would provide the entire primary lateral force resistance in this
direction.
The storey heights for the twelve-storey high building were equal to 4 m for the Arst two
levels and 3 m for the remaining levels. For the three-storey high building, the height of the
Arst level was 4 m, while the remaining levels were 3 m high. These buildings were analyzed
and designed in accordance with the New Zealand Loadings Standard [16] and the New
Zealand Concrete Structures Standard [17]. The initial stiLness of these walls was determined
using an eLective moment of inertia equal to 0:25Ig , where Ig is the moment of inertia of
gross concrete section. Designs were carried out for limited ductility response,  = 3, and for
fully ductile response,  = 5. Further details are given in Rodriguez et al. [18].
The analyses were carried out for two earthquake records and two types of non-linear
response. A synthetic record, named SIM1, with a PGA equal to 0:42g, was generated to
match the design spectra for intermediate soil conditions for the highest seismic region in New
Zealand. SpeciAc details of this record can be found elsewhere [18]. The other earthquake
record was SYLM949, with a PGA equal to 0:80g, obtained during the 1994 Northridge
earthquake. The acceleration amplitude of these records was scaled to observe the gradual
eLect that inelastic response has on the 0oor acceleration. The non-linear response at the
base of the walls was represented either by a Takeda or origin-centered hysteresis rules. The
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

701

Figure 4. Elastic dynamic characteristics of the three- and twelve-storey buildings investigated.

Takeda hysteresis rule is generally used to characterize behaviour of cast-in-place reinforced


concrete walls or the behaviour of precast concrete walls designed to respond as if monolithic. The origin-centered hysteresis rule, termed modiAed origin-centered rule (MOC) in
the computer program Ruaumoko [19], was used to characterize the non-linear response of
self-centering precast concrete hybrid walls, which are post-tensioned with partially unbonded
tendons and incorporate energy dissipation devices. The behaviour of such walls is described
elsewhere [7; 20]. The Takeda and MOC hysteresis rules used in the parametric analysis had
identical initial and post-elastic backbone moment-rotation behaviour. The non-linear time
history dynamic response of the building was performed using the computer program Ruaumoko [19].
The structural model incorporated a Rayleigh damping formulation proportional to the mass
and initial stiLness matrices [19]. The Arst mode of vibration was given a 5 per cent damping
ratio. Care was taken to ensure that damping ratios would be similar for the Arst modes of
all buildings and to ensure that the higher translational modes would not be highly damped.
Figure 4 shows the mode shapes, periods of vibration, damping ratios and participation factors for the three- and twelve-storey buildings. Owing to the Rayleigh damping formulation,
the second and third modes were signiAcantly under-damped. While under-damping of these
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

702

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 5. Top 0oor acceleration magniAcation obtained for the twelve-storey wall buildings for diLerent
scaled SYLM949 input ground motions.

modes could be expected to have a small eLect on the peak displacement response of the
buildings, it would be expected to have a greater eLect on the magnitude of the 0oor accelerations. The Newmark constant average acceleration method was used to integrate the equation
of motion in the time-history analyses [19]. The integration time-step was equal to 1=1000 sec
to ensure satisfactory acceleration results. Data from the response was stored at 1=100 sec
intervals. Further details of the model are described in Reference [18].
The trends obtained for the 0oor horizontal accelerations were generally similar for the
response of the buildings to the SYLM949 and SIM1 records. Figure 5 plots the top 0oor
acceleration magniAcation against the scale factor [SF] used for the SYLM949 record for the
two types of hysteresis rules investigated. There are three distinct regions in this Agure. In
the Arst region, from SF0 to SF = 0:08, the magniAcation is constant because the building
responds elastically. In the second region, from SF = 0:08, where the building reaches the
elastic limit, to about SF = 0:2, the magniAcation decreases very rapidly. In the third region,
from SF = 0:2 onwards there is little dependency between the magniAcation and the SF.
Figure 6 plots envelopes for the 0oor acceleration magniAcation for the twelve-storey high
building, for the SYLM949 record scaled by 0.08, 0.2, 0.4 and 1, and for the Takeda hysteresis
rule. Each dot corresponds to the magniAcation in a level of the building. It is clear from
Figure 6 that the maximum 0oor acceleration magniAcation occurs at the top of the building.
The following additional observations can be inferred from this Agure: (i) the maximum
0oor acceleration magniAcation occurs when the building responds elastically, and, (ii) the
0oor acceleration magniAcation tends to diminish as the SF, and hence the ductility demand,
increase.
Figure 7 is similar to Figure 6 except that the non-linear response of the building was
modelled with the MOC hysteresis rule. In contrast with the trend observed for the response
obtained for the Takeda hysteresis rule, the maximum magniAcation does not always occur
at the top of the building. This plot does not show any clear trend, which also contrasts
with the response observed for the Takeda hysteresis rule. This is caused by the shock eLect
resulting from the sudden change in stiLness at the origin of the non-linear response as will
be discussed later on.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

703

Figure 6. Floor acceleration magniAcations for each level of the twelve-storey wall building responding
to four scaled SYLM949 input ground motions with a Takeda hysteresis rule.

Figure 7. Floor acceleration magniAcation for each level of the twelve-storey wall building responding
to four scaled SYLM949 input ground motions with an origin-centered hysteresis rule.

The response plotted in Figures 57 suggests that the 0oor acceleration magniAcation, and
therefore the peak 0oor acceleration, is somewhat related to the ductility demand but is not
inversely proportional to it, as is inferred from recommendations given for the design of
diaphragms in the Mexico City Building Code [15] and in the New Zealand Standards [16; 17].
In the case of the Uniform Building Code [13] the in-plane 0oor forces are often controlled by
the limits 0.5 and 1:0(CaIWpx ), which means that in this case they are not related to ductility
demand.
Floor pseudo-acceleration response spectra derived from the response of elastic one-degreeof-freedom oscillators, responding to the uppermost 0oor acceleration time-histories, will
be used here to provide an understanding of the main variables that aLect 0oor acceleraCopyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

704

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 8. Top 0oor pseudo-acceleration response spectra (5 per cent damping ratio) normalized by the peak ground acceleration obtained from scaled SYLM949 input ground motions for
the twelve-storey wall building.

tions. Uppermost 0oor-response spectra for the twelve-storey building, for 5 per cent damping ratio, were derived from the response of the building to the SYLM949 input ground
motion for SF equal to 0.08, 0.4 and 1.0 for the two types of hysteresis rules investigated. A value of SF = 0:08 demarcates the limit of the buildings elastic response while the
value of SF = 0:4 and SF = 1:0 result in moderate and high displacement ductility demands,
respectively. Figure 8 depicts the 0oor-response spectra. In this Agure the spectral ordinates
have been normalized by the peak ground acceleration obtained from the scaled input ground
motion. The solid-bold line in Figure 8 represents response spectrum corresponding to the
buildings elastic response. The three of peaks in the response are related to the translational
modes with periods of free vibration T1 , T2 and T3 .
It is interesting to note that inelastic behaviour in the primary lateral force resisting system,
shown by the shaded and solid-thin lines in Figure 8, results in a reduction of the response
around the natural periods of free vibration. This reduction is greatest for the response around
T1 . The response associated with the Takeda hysteresis rule, see Figure 8(a), shows a diLerent
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

705

Figure 9. Momentcurvature response at the base of a wall in the twelve-storey building for the Arst 7
sec of the SYLM949 input ground motion for the two types of hysteresis rule.

trend to that observed for the MOC rule, see Figure 8(b). In the case of the response with a
Takeda hysteresis rule, an increase in the ductility demand in the walls results in further, but
not signiAcant, reduction of the acceleration in the period band around T1 while the response
around the higher mode periods is largely unaLected. When the response is obtained using
the MOC hysteresis rule, see Figure 8(b), a large ductility demand also results in further,
but not signiAcant reduction of the response around T1 . However, the peaks in the spectral
ordinates associated with the higher modes periods have an amplitude much greater than those
associated with relative moderate ductility response, that is, with the response for SF = 0:4.
The diLerent trends observed between the response at high ductility demands for the two
hysteresis rules can be explained with the aid of the momentcurvature response at the critical
region at the base of the walls obtained from the analyses, see Figure 9. The characteristic
self-centering response of the MOC hysteresis rule produces a sudden change of stiLness at the
origin, see Figure 9(a). The sudden change from low to high stiLness is felt in the structure as
a shock that feeds energy into the higher modes. The shock eLect is more pronounced when
the ratio between the reloading stiLness and the unloading stiLness is high. For example, for
the SYLM 949 record such ratio was 4.5 when SF = 0:4 and 12 when SF = 1:0 [18]. It should
be noted, however, in an actual building built with a self-centering system that the change
in stiLness at the origin should not be as abrupt as that given by the MOC hysteresis rule.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

706

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

For this reason the shock eLect should be expected to be less pronounced than that obtained
theoretically.
3.2. Proposed First Mode Reduced method for determining 4oor magni5cations
The 0oor acceleration Anq corresponding to mode q at the uppermost level of the building,
can be expressed by
Anq = Tq qn

Sa (Tq ; q )
Rq

(7)

where Tq is the participation factor for mode q; qn is the amplitude of mode q at level n; Sa
is the spectral acceleration, Tq and q are the period of free vibration and damping ratio,
respectively, associated with mode q, and R q is a reduction factor to account for the eLect of
ductility on the primary lateral force resisting system.
Equation (7) is based on linear-elastic theory and is adapted here for evaluating the response
of non-linear systems. In linear-elastic systems, the natural modes of vibration can clearly
be deAned by ensuring orthogonality between modes with respect to the stiLness and mass
matrices. As soon as the structure exceeds the elastic limit, the modes are no longer orthogonal
to each other with respect to the stiLness matrix. Notwithstanding this, the modes still provide
a set of independent vectors that can conveniently be used in the analysis of non-linear systems
[21; 14; 22]. Noteworthy is the fact that when a structure is deformed beyond the elastic limit
the modal characteristics change instantaneously every time the stiLness changes.
For hysteresis rules characterized by a relatively smooth non-linear response, such as the
Takeda rule employed here, it appears from the results of the parametric analysis discussed
previously that R q R q+1 and R q 1. For rules showing self-centering characteristics, such
as the MOC rule utilized in this study R1 1. However, the reduction factors for the higher
modes, R2 ; R3 : : : R r , seem to be larger than one for low ductility demands but could be less
than one for high ductility demands.
Modal accelerations can be combined to obtain an approximation to the 0oor acceleration.
This can be achieved using a suitable modal combination technique. The square-root-of-thesum-of-the-squares technique [SRSS] is chosen in this investigation because of its simplicity.
The limitation of the SRSS technique when combining the response of modes with close
periods is not considered to signiAcantly aLect the results of this study. When combining the
absolute 0oor accelerations due to the translational modes in the direction begin considered,
the SRSS results in the following 0oor horizontal acceleration, An , for the uppermost principal
seismic mass,



2

r
q Sa (Tq ; q )

Tq n
An =
q=1

Rq

(8)

The observations obtained from the non-linear analysis suggest that Equation (8) can be
simpliAed by assuming that Arst mode is the only mode aLected by the ductility. This
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

simpliAcation results in R2 = R3 = = R r = 1. Thus, Equation (8) becomes,






r
Sa (Tq ; q ) 2 
+
[Tq qn Sa (Tq ; q )]2
An =  T1 1n
R1
q=2

707

(9)

where R1 = OMe =OMo ; OMe is the maximum overturning moment required for elastic response
and OMo is the maximum overturning moment found from the non-linear time-history analysis
considering also the eLects of overstrength.
Results obtained from the non-linear time history dynamic analyses, to be presented in
the following section, indicate that the 0oor accelerations in the lower 0oors is strongly
in0uenced by the horizontal ground excitation as well as by the shape of the hysteresis rule.
The following interpolation function is proposed to obtain A i at any 0oor in the building:
A i = Ui Ao

(10)

where Ao is the peak ground acceleration and factor Ui is the 0oor acceleration magniAcation
factor. This factor is given by,

or

Ui = An =Ao for 0oors located between 0:2hi =hn 61


 

hi
An
Ui = 5
1 + 1 for 0oors located between 06hi =hn 60:2
hn
Ao

(11)

where hi is the height of the 0oor in consideration, and hn is the height of the uppermost
level of the building both measured from the base.

4. COMPARISON OF FLOOR ACCELERATION MAGNIFICATIONS OBTAINED


FROM THE FIRST MODE REDUCED METHOD AND FROM
NON-LINEAR TIME-HISTORY DYNAMIC ANALYSES
In this paper results of the parametric analysis are only presented for the three- and twelvestorey buildings designed for  = 5 and subjected to the full-scale SYLM949 and SIM1
records. Complete results of the analyses are reported elsewhere [18]. Although only few
examples of the First Mode Reduced method are given in this paper, the previously discussed Andings of Eberhard and Sozen [14] on higher mode eLects suggest that the method
might work for other structures with diLerent parameters.
Figures 10 and 11 plot the distribution in height of the 0oor acceleration magniAcations
for the three- and twelve-storey buildings when subjected to the SYLM949 record. These
Agures plot the magniAcations obtained from the Takeda and MOC hysteresis rules. The
line associated with black-circle markers in Figures 10 and 11 corresponds to those results
obtained from the non-linear analysis. The line associated with triangular markers corresponds to the 0oor magniAcation predicted using the First Mode Reduced method described
above. The reduction factors for the Arst mode, R1 , for the three- and twelve-storey buildings,
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

708

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 10. Floor acceleration magniAcation for all levels for the three-storey building
subjected to SYLM949 ground motion.

were evaluated according to the deAnition given in Section 3, and are shown in the legend
box in both Agures.
Figure 10 shows the worst case where the First Mode Reduced methods under-predicts
the 0oor acceleration magniAcations. Much better agreement was found for the 0oor acceleration magniAcations for higher rise buildings and for all buildings when subjected to the
synthetic record [18]. For example, Figure 11 compares the magniAcation predicted by the
First Mode Reduced method with the magniAcation derived from the non-linear analysis
for the two hysteresis rules investigated. In this case 0oor acceleration magniAcations predicted with the First Mode Reduced method envelopes the magniAcations derived from the
non-linear analyses, except in the Arst 0oor of the building responding with the MOC hysteresis rule. It is interesting to note that, as shown in Figures 10 and 11, 0oor acceleration
magniAcations associated with the origin-centered hysteresis rule were generally greater than
those associated with the Takeda rule, particularly in the lower 0oors of the buildings. The
diLerence was particularly large in low-rise buildings responding to a near fault input ground
motion [18].
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

709

Figure 11. Floor acceleration magniAcation for all levels for the twelve-storey
building subjected to SYLM949 motion.

5. SIMPLIFICATION OF THE FIRST MODE REDUCED METHOD


FOR USE IN DESIGN
The First Mode Reduced method described in Section 3 is suitable for use in conjunction
with a response spectrum modal analysis but is rather cumbersome for use in routine design.
A simpliAcation of this method, for use in compliance with a generic design spectrum, such
as that prescribed by the New Zealand Loadings Standard [16], will be derived in this section.
5.1. Step 1
The Arst step is to assume that the damping ratio for all modes is equal to 5 per cent as
is normally assumed in most seismic design standards. This assumption enables the spectral acceleration terms in Equations (9)(11) to be obtained from a single spectrum. The
seismic coePcient, Ch (Tq ; 1), for fully elastic response corresponding to period Tq is the spectral acceleration for 5 per cent damping, Sa (Tq ; 0:05), divided by the acceleration of gravity.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

710

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 12. DeAnition of parameters for deriving a simple method compatible with a design spectrum.

That is:
Ch (T1 ; 1) =

Sa (Tq ; 0:05)
g

(12)

The seismic coePcient, Cpn , for a part or diaphragm at the uppermost level is equal to,
Cpn =

An
g

(13)

Substituting Equations (12) and (13) into Equation (9)




2

r

Ch (T1 ; 1)
Cpn =  T1 1n
+
[Tq qn Ch (Tq ; 1)]2
R1
q=2

(14)

where Ch (T1 ; 1) is the seismic coePcient derived for elastic response from the design spectrum
for the fundamental period in the direction being considered, see Figure 12.
In the simpliAed method described in this section 0oor acceleration magniAcations are
deAned by the ratio Cpi =Cho , where coePcients Cpi and Cho are deAned as:
Cpi = Ai =g

and

Cho = A o =g

(15)

For example, in the current New Zealand Loadings Standard [16] coePcient Cho is equal to
0.4, 0.42 and 0.42 for hard, intermediate and soft subsoil categories, respectively.
The relationship between Cpi and Cho is obtained by substituting Equation (15) into Equation (10). Thus:
Cpi = Ui Cho

(16)

where the magniAcation factor Ui , deAned in Equation (11) in terms of acceleration, can
be written in terms of the seismic coePcients. This is achieved by substituting Equations
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

711

(13) and (15) into Equation (11),

or

Ui = Cpn =Cho for 0oors located between 0:2hi =hn 61


 

Cpn
hi
Ui = 5
1 + 1 for 0oors located between 06hi =hn 60:2
hn
Cho

(17)

5.2. Step 2
The second step assumes that all the natural periods of free vibration corresponding to the
higher modes in the considered direction are in the period band that coincides with the maximum spectral ordinate, see Figure 12. It is also assumed that for these periods the corresponding elastic seismic coePcient is taken conservatively as Ch; max . Thus, this implies that
Equation (14) can be written as:


2
$1
Ch (T1 ; 1) + $h2 (!Cho )2
(18)
Cpn =
R1
where coePcient ! = Ch; max =Cho , see Figure 12, $1 is Arst mode contribution coePcient and
$h is a coePcient that accounts for the contribution of the higher modes of response. These
coePcients are given by,

r

1
(Tq qn ) 2
(19)
$1 = T1n and $h =
q=2

Upper and lower bounds for $1 and $h can be obtained from modal analyses of 0exural and
shear cantilevers representing buildings with regular mass and stiLness distribution. Flexural
beams characterize the behaviour of slender walls where shear deformations are negligible. In
contrast, shear beams characterize the behaviour of squat walls or moment resisting frames
with stiL beams and inextensible columns. Figure 13 plots the variation of $1 and $h with the
number of levels in a building. CoePcients $1 and $h obtained from modal analyses for the
three buildings investigated are also shown in Figure 13. It is evident in Figure 13 that, when
the number of levels is greater than two, 0exural beams have the largest values of coePcients
$1 and $h . An upper bound solution applicable to all types of buildings is obtained by deAning
coePcients $1 and $h as:
$1 = 1 for single-storey buildings and $1 = 1:5 for multi-storey buildings

(20a)

$h = 0:53 ln(n)

(20b)

and

where n is number of the levels in the building.


5.3. Step 3
The deAnition of Arst mode reduction factor R1 given in Section 3 is appropriate for use in
conjunction with non-linear time-history dynamic analyses of multi-storey buildings. Consequently, a simpliAcation for use in practical design seems warranted. It should be expected
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

712

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 13. Variation of coePcients $1 and $h2 with the number of levels in a building.

that the displacement ductility demand in a building be less than that chosen for design due to
the eLects of overstrength. Buildings are expected to have values of R1 ranging between =
and 1, whichever is greater. CoePcient  depends on the overstrength of the overall structure
and on the relationship between the ordinates in the elastic and inelastic response spectra. A
value of  = 2, proposed in this paper, seems appropriate for buildings designed in accordance
with the principles of capacity design [23]. Hence:

(21)
R1 = or 1; whichever is greater
2
The following expression is obtained for coePcient Cpn by substituting Equation (20b) into
Equation (18) and by making ! = 2:5,


2
$1
2
Ch (T1 ; 1) + 1:75 ln(n)Cho
(22)
Cpn =
R1
The horizontal design force, Fph , for a rigid part or diaphragm at level i between, and including,
the base and the top level is,
Fph = Sp Rp ZCpi Wp

(23)

where Sp is a structural performance factor, Rp is the risk factor for the part or diaphragm,
Z is the seismic zone factor and Wp is the weight of the part or diaphragm in
consideration.
Figure 14 compares the 0oor acceleration magniAcations determined from the simpliAed
method and from the non-linear analyses of the three- and twelve-storey buildings. Floor accelCopyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

713

Figure 14. Floor acceleration magniAcation for the three- and twelve-storey buildings when subjected to the synthetic SIM1 input ground motion and the simpliAed method.

eration magniAcations determined from the simpliAed approach are in general good agreement
with the results obtained from the non-linear analysis, particularly for those cases associated
with the Takeda hysteresis rule.
Figure 15 shows the top 0oor acceleration magniAcation against the displacement ductility demand for four buildings. It is evident in this Agure that 0oor acceleration magniAcation becomes rather insensitive to the ductility demand as the number of levels increase. The only case where de-ampliAcation is found, that is when the 0oor acceleration
is less that the peak ground acceleration, is in the one-storey building. This is because in
a single-degree-of-freedom system, the 0oor acceleration is directly controlled by the capacity of the critical mechanism. Such capacity is a function of ductility factor, , selected
in design. Note also that the top 0oor acceleration magniAcations found using the simpliAed method are in the range of values observed during the Northridge earthquake, see
Figure 1.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

714

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

Figure 15. Top 0oor acceleration magniAcation obtained from the simple design approach vs
displacement ductility and number of levels.

6. CONCLUSIONS
This paper summarized an investigation into earthquake induced 0oor horizontal accelerations
that develop in regular buildings. The main Andings and conclusions of this investigation
are:
1. A parametric non-linear time-history dynamic analysis of cantilever wall buildings was
conducted in this investigation. Three-, six-, and twelve-storey buildings, designed for limited
ductility response and for fully ductile response, were subjected to two diLerent
input ground excitations. One excitation was a synthetic record designed to match one of the
design response spectra given by the New Zealand Loadings Standard [16]. The other excitation was the near-fault record SYLM949 obtained during the 1994 Northridge earthquake.
The response of the buildings was investigated using a Takeda hysteresis rule representing
behaviour of cast-in-place walls and an origin-centered hysteresis rule representing the
behaviour of self-centering precast concrete walls post-tensioned with partially unbonded tendons and incorporating mechanical energy dissipation devices.
2. The investigation found that the maxima 0oor acceleration magniAcations nearly always
occur at the uppermost 0oor of a building. It was also found that the response immediately beyond the elastic limit signiAcantly reduces the 0oor accelerations at the uppermost
0oor.
3. Floor accelerations associated with the origin-centered hysteresis rule were generally
greater than those associated with the Takeda rule, particularly in the lower 0oors of the
buildings investigated. The diLerence was particularly large in low-rise buildings responding
to a near fault input ground motion.
4. A procedure for deriving the design horizontal forces was proposed in the paper. The
procedure assumes that ductility only aLects 0oor accelerations associated with the Arst mode
of the response. A simple version of the method, for use in routine design, was also proposed
and compared, with generally good agreement, with the results from non-linear time history
analyses.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

715

ACKNOWLEDGEMENTS

The Anancial support of the Public Good Science Fund administered by the New Zealand Foundation
for Research Science and Technology, Contract UOC 808 for the 1998=2000 grant cycle, is gratefully
acknowledged.
Thanks are due to Prof. J.D. Jaramillo, from University EaAt in Medellin, Colombia, to Dr B. Deam,
Leicester Steven Lecturer at the University of Canterbury, and to Prof. A Rutenberg from Technion
Israel Institute of Technology for their useful comments. Thanks are also due to Mr J.J. BlandVon, Ph.D.
candidate from the National University of Mexico who helped with several analyses. Thanks are also
given to the anonymous reviewers for their critical reading of the manuscript and useful suggestions.

APPENDIX: NOMENCLATURE
= horizontal force factor according to UBC 1997
= absolute acceleration of the seismic weight at level n due to mode q; mm= sec2
= absolute acceleration at 0oor i; mm= sec2
= peak ground acceleration, mm= sec2
= seismic coePcient (deAned for six seismic zone factors and six soils proAle types)
= 0oor horizontal seismic coePcient at level i (NZS 4203:1992)
= uppermost level horizontal seismic coePcient (NZS 4203:1992)
= 0oor horizontal seismic coePcient applicable at or below the base of a building
(NZS 4203:1992)
Ch (Tq ; 1) = seismic coePcient derived for elastic response from the basic hazard acceleration
design spectra for the q translational mode of natural vibration in the direction being
considered
= maximum spectral acceleration coePcient
Ch max
= peak ground acceleration and acceleration of gravity ratio
Cho
= basic horizontal seismic coePcient for a part at level i (NZS 4203:1992)
Cpi
= basic horizontal coePcient for a part or diaphragm
Cpn
= horizontal force on primary lateral force resisting system at level i, kN (NZS
Fi
4203:1992)N
= is the horizontal force found for the primary lateral force resisting system at the
Fn
uppermost level of the building, kN (NZS 4203:1992)
= horizontal force acting on a part according to NZS 4203:1992
Fph
= design horizontal force for diaphragms, kN (UBC 1997)
Fpx
= portion of the base shear, V , considered concentrated at the top level of the
Ft
structure in addition to Fn , kN (UBC 1997)
= seismic design force applied to level x, kN (UBC 1997)
Fx
= height, measured from the principal seismic weight at level i, m
hi
= height, measured from the uppermost principal seismic weight, m
hn
i
= index deAning a level or storey in a building. Note level i is immediately above
storey i
I
= building importance factor (UBC 1997)
= importance factor for a part (UBC 1997)
Ip
ap
Aqn
Ai
Ao
Ca
Cfi
Cfn
Cfo

Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

716

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

= reloading 0exural stiLness, kN m2


= unloading 0exural stiLness, kN m2
= number of levels in building
= maximum overturning moment for elastic response, kN m
= overturning moment in a structure at the development of overstrength, kN m
= node number
= highest mode of vibration considered to in0uence the absolute 0oor acceleration
= risk factor for a structure (NZS 4203:1992) also
= base shear force reduction (UBC 1997)
= risk factor applicable to a part (NZS 4203:1992), also
Rp
= component response modiAcation factor for the part deAned by UBC 1997
= reduction factor in the spectral acceleration Sa to account for the eLect of strucRq
tural ductility on mode q
= spectral acceleration, mm= sec2
Sa
Sp
= structural performance factor (NZS 4203:1992)
= natural period for a buildings Arst mode of translational vibration in the direction
T1
being considered, sec
T2 : : : Tq = natural period of the buildings higher translational modes of vibration in the
direction being considered, sec
= natural period of vibration of the part in the direction being considered, sec
Tp
(NZS 4203:1992)
= natural of free vibration associated with mode of vibration q, sec
Tq
V
= base shear force, kN (UBC 1997)
W
= total seismic dead load, kN (UBC 1997)
wi and wx = portion of the total seismic dead load, W, located or assigned to level i or x,
kN (UBC 1997)
= seismic weight at level i, kN
Wi
= seismic weight at level n, kN
Wn
= weight of a part in NZS 4203:1992, kN
Wp
= weight of the diaphragm at level x according to UBC 1997, kN
wpx
x
= index referring to a level in a building
kr
ku
n
OMe
OMo
q
r
R

Greek letters

Tq

q
qn
$1
$h

o
Ui
!

=
=
=
=
=
=
=
=
=
=
=

overstrength factor
participation factor for mode q
damping ratio for a one-degree-of-freedom system
damping ratio associated with mode q
amplitude of mode q at level n
Arst mode contribution coePcient
higher mode contribution coePcient
displacement ductility factor (NZS 4203:1992)
displacement ductility factor including overstrength required by NZS 4203:1992
0oor magniAcation factor at level i
Ch max =C ho

Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

FLOOR HORIZONTAL ACCELERATIONS

717

Abbreviations
DF
=
NZS =
MCBC =
MOC =
PGA =
PSA =
SF
=
SRSS =
UBC =

displacement ductility factor


New Zealand Standard
Mexico City Building Code
modiAed origin-centered hysteresis rule
peak horizontal ground acceleration
pseudo-spectral acceleration
input ground motion scale factor
square-root-of-the-sum-of-the-squares modal combination technique
Uniform Building Code
REFERENCES

1. SEAOC, Vision 2000A Framework for Performance-based Design, vols. IIII. Structural Engineers of
California: Sacramento, 1995.
2. Moehle JP. Displacement-based design of RC structures subjected to earthquakes. Earthquake Spectra 1992;
8(3):403 428.
3. Priestley MJN. Performance-based seismic design. State-of-the-Art Report, 12th World Conference on
Earthquake Engineering, Auckland, New Zealand. Paper 2831, 2000.
4. Priestley MJN, Sritharan S, Conley J, Pampanin S. Preliminary results and conclusions from the PRESS fourstorey precast concrete test building. PCI Journal 1999; 44(6):42 67.
5. Lam N, Wilson J, Doherty K, GriPths M. Horizontal seismic forces on rigid components within multi-storey
buildings. Proceedings of the Australasian Structural Engineering Conference, Auckland, New Zealand, 1998.
6. Soong TT, Bachman RE. Drake RM. Implications of the 1994 Northridge earthquake on design guidelines
for nonstructural components. Proceedings of the NEHRP Conference and Workshop on Research on the
Northridge California Earthquake of January 17, 1994, California Universities for Research in Earthquake
Engineering, vol. III, 1988; 441 448.
7. Kurama Y, Pessiki S, Sause R, Lu L. Seismic behaviour and design of unbonded post-tensioned precast concrete
walls. PCI Journal 1999; 44(3):72 89.
8. Wood S, Stanton J, Hawkins N. New seismic design provisions for diaphragms in precast concrete parking
structures. PCI Journal 2000; 45(1):50 65.
9. Wyllie Jr LA, Filson JR (eds), Armenia Earthquake Reconnaissance Report. Earthquake Spectra Publication
89-01, Special Supplement, August 1989; 175.
10. Fleischman S, Pessiki R, Rhodes A. Seismic behaviour of precast parking structure diaphragms. PCI Journal
1998; 43(1):38 53.
11. Restrepo JI, Cowan H. E1 Eje Cafetero Earthquake, Colombia of January 25 1999. Bulletin of the New Zealand
Society for Earthquake Engineering 2000; 33(1).
12. Hall JF (ed.). Northridge Earthquake of January 17, 1994, Reconnaissance Report, vol. 1. Earthquake Spectra,
Supplement C to vol. 11, Publication 95 - 03, 1995; 523.
13. UBC 1997. Uniform Building Code. International Conference of Building O?cials, Whittier, CA, vol. 2, 1997.
14. Eberhard M, Sozen M. Behavior-based method to determine design shear in earthquake-resistant walls. Journal
of Structural Engineering, ASCE 1993; 119(2):619 640.
15. Mexico City Building Code. Reglamento de Construcciones para el Distrito Federal, Diario OAcial. Mexico
City, 1993.
16. NZS 4203:1992. Code of practice for general structural design and design loadings for buildings. Standards
New Zealand, 1992.
17. NZS 3101:1995. Concrete Structures Standard. Standards New Zealand, 1995.
18. Rodriguez M, Restrepo JI, Carr AJ. Earthquake resistant precast concrete buildings: 0oor accelerations in
buildings. Research Report 2000-6, Department of Civil Engineering, University of Canterbury, Christchurch,
2000.
19. Carr AJ. Ruaumoko, Computer Program Library, Department of Civil Engineering, University of Canterbury,
1998; 226.
20. Rahman A, Restrepo JI. Earthquake resistant precast concrete buildings: seismic performance of cantilever walls
prestressed using unbonded tendons. Research Report 2000-5, Department of Civil Engineering, University of
Canterbury, Christchurch, 2000.
Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

718

M. E. RODRIGUEZ, J. I. RESTREPO AND A. J. CARR

21. Skinner RI, Robinson WH, McVerry GH. An Introduction to Seismic Isolation. Wiley: New York, 1993.
22. Reinhorn AM. Inelastic analysis techniques in seismic evaluations. In Seismic design methodologies for the next
generation of codes, Workshop, Bled, Slovenia, Krawinkler H, Fajfar P (eds). Balkema Publishers: Rotterdam,
1997.
23. Paulay T, Priestley MJN. Seismic Design of Reinforced Concrete and Masonry Buildings. Wiley: New York,
1992.

Copyright ? 2001 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2002; 31:693718

Vous aimerez peut-être aussi