Sie sind auf Seite 1von 13

Engineering Structures 24 (2002) 10911103

www.elsevier.com/locate/engstruct

Theory of principal components analysis and applications to


multistory frame buildings responding to seismic excitation
Mark A. Aschheim a,, Edgar F. Black b, Isabel Cuesta c
a
Mid-America Earthquake Center, Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, 205 North
Mathews, Urbana, IL 61801, USA
b
Department of Civil Engineering, University of Notre Dame, 159 Fitzpatrick Hall of Engineering, University of Notre Dame, Notre Dame, IN
46556, USA
c
Los Alamos National Laboratory, P.O. Box 1663, MS C926, LANL ESA-WR, Los Alamos, NM 87545, USA

Received 9 February 2001; received in revised form 3 October 2001; accepted 29 October 2001

Abstract

Described herein is a technique of multivariate statistical analysis applied to the post-processing of dynamic response data. The
data may represent the linear or nonlinear response of structures, and may be obtained from computed simulations or from the
measured response of instrumented structures. When applied to displacement response data, an ordered set of orthonormal mode
shapes is obtained. The principal components analysis (PCA) mode shapes coincide with or are related to the elastic mode shapes
for linear elastic systems, and depart from these shapes as nonlinear response becomes more prominent. The PCA modes provide
an unambiguous and simple description of the predominant mode of structures responding to earthquake ground motions, and
thus improve the theoretical basis of nonlinear static procedures that use equivalent single-degree-of-freedom (SDOF) systems
for representing the response of structures subjected to damaging earthquake ground motions (e.g. the capacity spectrum and dis-
placement coefficient methods). Where greater fidelity is desired, the most efficient representations are obtained by including as
few PCA modes as are needed for the degree of precision desired. This paper presents the theory of PCA and illustrates its
application to a 12-story frame building responding linearly and nonlinearly to earthquake ground motions. Equivalent SDOF
models of the structure are developed based on the PCA mode shapes, and these are applied to estimate the computed displacement
histories. 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Dynamic response; Mode shapes; Nonlinear static procedures; Principal components analysis; Equivalent single-degree-of-freedom sys-
tems; Seismic design

1. Introduction unable to describe with specificity what this mode


shape is or the degree to which response is in this mode.
Linear and nonlinear dynamic response can generate Principal components analysis (PCA) identifies these
large amounts of time series data that may be difficult mode shapes precisely and quantifies the degree to which
to interpret, particularly when structures respond nonlin- response is in each mode.
early. Although various measures such as displacement Principal components analysis determines an
ductility and dissipated hysteretic energy are useful indi- orthonormal set of basis vectors that best represent the
cators of inelastic demand, only qualitative descriptions response data. The ordering is such that the first basis
of the displacement response have been available. vector (hereafter referred to as a PCA mode) is the best
Researchers have referred to the predominant mode of fit to the observed response history. The second PCA
response of a nonlinear system [15], but have been mode is the best fit to the residuals, and so on, for each
mode. When applied to the displacement response of lin-
ear systems that have uniform values of nodal mass, the
Corresponding author. Tel.: +1-217-333-7384; fax: +1-217-265-
mode shapes are observed to coincide with the elastic
8040. mode shapes. For nonuniform mass distributions, the
E-mail address: aschheim@uiuc.edu (M.A. Aschheim). PCA modes and elastic modes are related according to

0141-0296/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 2 ) 0 0 0 3 6 - 6
1092 M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103

relationships that are described herein. The empirically- 2. Theory of PCA


determined mode shapes may deviate from those
obtained for elastic response as response becomes non- Consider a vector of displacement response data v rep-
linear, and depend (indirectly) on properties of both the resenting the displacement relative to the ground at n
structure and the excitation. The PCA mode shapes are discrete locations of a MDOF system at an instant of
eigenvectors computed from the response data; the cor- time. There are t observations of the n1 vector v over
responding eigenvalues provide a quantitative measure time. It is obvious that the displacement response, v, at
of the degree to which the response of the structure is any instant of time represents a linear combination of
represented by each PCA mode shape. Thus, it is poss- the unit basis vectors that provide an orthonormal basis
ible to identify the first mode of a structure responding for the n-dimensional space that contains v. The devi-
nonlinearly and to quantify the degree to which response ation of v from its mean over t observations,v, may be
is represented by this mode. Explicit knowledge of the expressed in terms of a new orthonormal basis, B:
properties of the structure (e.g. the mass and stiffness
matrices) is not needed for determination of the PCA vv Bu (1)
mode shapes; these properties are reflected in the dis- where u represents the displacements relative to their
placement response history. means in terms of the basis vectors contained in B. Since
Since the PCA mode shapes are determined to maxim- B is orthonormal, BTB I. Therefore, BT B1, and
ize variances, they are natural choices to be used in premultiplying (1) by BT gives
establishing parsimonious representations of the
response. Representations of the response that use the u BT(vv) (2)
first or first two PCA modal responses are preferable to with the mean of u being the n1 vector 0.
approximations in which responses based on the elastic Let the covariance of the n1 vector v be represented
mode shapes are superposed, because a larger number by the nn covariance matrix Cv. The (i, j) element of
of elastic mode shapes may be necessary to represent Cv is the covariance between the displacements at the
the displacement profile exhibited by systems ith and jth degrees of freedom, vi and vj, over the t obser-
responding nonlinearly. The PCA mode shapes may be vations, given by
useful for developing equivalentsingle-degree-of-free-

t
dom (SDOF) representations of the response of multi- 1
cov(vi,vj) (v v )(v v ) (3)
degree-of-freedom (MDOF) systems. The computed t k 1 i,k i j,k j
PCA mode shapes portray the behavior of structures
being designed to resist earthquakes; thus, it is possible where vi and vj are the means of vi and vj, respectively,
in design to modify member strengths and stiffnesses over the t observations. Because cov(vi, vj) is computed
iteratively until the PCA mode shapes indicate that unde- using real-valued data, Cv is real, and because
sirable mechanisms do not develop in the structure. PCA cov(vi,vj) cov(vi,vj), Cv is symmetric.
also is useful when modeling the response of instru- Because Cv is real and symmetric, it can be diag-
mented structures, because model properties may be onalized by an orthogonal matrix
constrained by requiring that PCA mode shapes obtained TCv (4)
from the computed data approximately match those
obtained from the recorded data. Examples of this appli- where consists of the orthonormal eigenvectors of Cv
cation are given by Wissawapaisal and Aschheim [21] and is a diagonal matrix containing the eigenvalues
and Inel [13], who studied the seismic response of instru- li of Cv. By convention, the eigenvectors of Cv are
mented bridges. Independently, Gutie rrez and Zaldivar arranged in sequence so that their corresponding eigen-
[10] recently applied PCA to data collected in pseudo- values are in descending order (l1l2ln).
dynamic and shake table tests. By operating with standard identities on Eq. (2), the
The idea of PCA originates in work by Pearson [18] covariance of u, Cu, can be expressed in terms of Cv:
and was developed independently by Hotelling [11].
Cu BTCvB. (5)
PCA finds modern use in statistical treatments of empiri-
cal data [16,17,20]. This paper presents the theoretical The parallel structures of Eqs. (4) and (5) indicate that
basis for the technique using the nomenclature of struc- Cu is a diagonal matrix consisting of the eigenvalues of
tural engineering and applies the technique to the linear Cv if the new basis B is selected to be the set of
and nonlinear responses of a 12-story moment-resistant orthonormal eigenvectors of Cv.
frame building. The effectiveness of the PCA represen- As Cu is diagonal, cov(ui,uj) 0 for i j, leading to
tations of lateral forces and lateral displacements is com- the result that the displacements ui and uj (expressed in
pared, and equivalent SDOF (ESDOF) systems based terms of the orthonormal basis B ) are uncorrelated
on the predominant PCA mode are used to estimate the for i j. Similarly, each displacement ui has variance
roof displacement history. var(ui) cov(ui,ui) li.
M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103 1093

In subsequent sections it will be shown that:



n

var(aTv) var(fTi v) a2i li li. (10)


1
1. the eigenvectors of Cv are, in fact, the principal
components of the displacement response v (in other Similarly, the principal components are uncorrelated
words, f1 is so oriented that it maximizes the vari- because all covariances (between different components)
ance, f2 maximizes the variance with f1 removed, and are identically 0. Because is a basis for the space that
so on); and contains the deviation of v from its mean, eqs. (1) and
2. the variance in the displacement response explained (2) provide that v can be expressed as a linear combi-
by each principal component is proportional to the nation of the eigenvectors of Cv
eigenvalue associated with the principal component. v u v (11)
where
2.1. Proof that the eigenvectors are principal
components u T(vv ). (12)
Eq. (11) represents the displacement response exactly,
To see that the eigenvectors of Cv are the principal and thus the entire set of n eigenvectors represents 100%
components of the displacement history, consider the lin- of the variance in v, which is given by tr(). Because
ear combination aTv that has maximum variance. The the bases are orthonormal, tr() tr(Cv), which indi-
weights a are selected to maximize var(aTv), under the cates that the total variance is a constant for any given
constraint that a has unit length, or aTa 1. Because data set. A subset of principal components may be used
is orthonormal and is a basis for the displacement to obtain efficient representations of v. Because the eig-
space, a can be expressed as a linear combination of the envectors are sequenced according to their eigenvalues,
basis vectors the most efficient representations of v using k eigenvec-
a a. (6) tors will be obtained if the first k eigenvectors are selec-
ted:


By applying a standard identity, k

var(a v) a Cva.
T T
(7) v fiui v . (13)
1
Eqs. (4) and (5) result in The proportion of total variance represented by the k
Cv BCuB .
T T
(8) components used in eq. (13) is (l1 l2
lk) / tr(). Because the principal components are orthog-
Substituting (6) and (8) into (7) leads to onal, the coefficients yi are invariant with changes in k,


n and because they have unit length, their values are
var(aTv) aTTTa aTa a2i li. (9) determined by the simple dot product given in (12). The
1 number of eigenvectors to be used depends on the
desired level of precision. Since each eigenvector
The constraint aTa 1 is equivalent to aTa 1, so

explains a proportion of the variance given by


a2i 1. Then, var(aTv) is maximized by picking a to (li) / tr() inspection of quickly reveals the value
be the unit vector that returns f1 to a (a1 1 and added by additional components. The authors experi-
aj 0 for all j 1). Thus, fT1v maximizes the variance, ence with civil engineering structures responding to
and therefore f1 is the first principal component. Sub- earthquake excitations is that the first PCA mode often
sequent combinations to maximize the variance should is sufficient to represent 90% or more of the variance in
be uncorrelated with those that have already been taken. the displacement response, while the first two PCA
Since the first k principal components (f1, f2,fk) are modes often represent 9599% of the variance in the
uncorrelated, the (k+1)th principal component to maxim- displacement response.
ize var(aTv) must select the largest remaining eigen- Three unusual cases may occur:
value, lk+1, for 1k+1n. Thus, the (k+1)th principal
1. the number of observations, t, is less than the degrees
component is given by fk+1.
of freedom, n;
2. one or more of the displacement degrees of freedom
2.2. The proportion of response explained by each is a linear combination of the other displacement
eigenvector responses; and
3. not all eigenvalues are distinct (that is, one or more
The variance explained by any eigenvector fi is its eigenvalues are repeated).
eigenvalue li, since fi is selected by choosing an appro-
priate unit vector a in Eq. (6), for which Eq. (9) provides In the first two cases, the rank of Cv will be reduced to
1094 M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103

a value less than n, and the number of nonzero eigenval- limit, coincide with the transformed elastic mode shapes
ues will be reduced correspondingly. Meaningful charac- given by LTel.
terization of dynamic response ordinarily will require t Two special cases are of interest:
to be substantially larger than n, so the first case usually
is not encountered. In the second case, a reduced basis 1. If M is diagonal with uniform masses, M mI
consisting of the eigenvectors associated with the non- LLT, requiring that L LT mI. Because this
zero eigenvalues will be sufficient to represent the reduces LTv and LTel to scalar operations on v and
response using Eq. (12). In the third case, the eigenvec- el, respectively, the PCA mode shapes obtained on
tors associated with the repeated roots are not unique, v will coincide with the elastic mode shapes after nor-
yet remain orthogonal to the other eigenvectors. Thus, malization.
they may be chosen arbitrarily to be orthonormal to 2. If M is diagonal with entries mi, M LLT requires
themselves and will be orthonormal to the other eigen- that L LT with the ith diagonal entry of L given
vectors. by mi. For this case, the PCA mode shapes obtained
All data presented in this paper are of the same type on LTv will coincide with the transformed elastic
(i.e. displacements, forces or accelerations). When mixed mode shapes given by LTel. Thus, the elastic mode
data types are present (e.g. rotations and displacements), shapes are given by el (LT)1F L1F , pro-
either the data should be scaled to be of similar numeric vided that sufficient response data v is available and
value, or formulations based on correlations may be response is linear.
preferable to the present covariance formulation.

2.3. Relationship between the PCA and the elastic 3. Principal components in the response of a 12-
mode shapes story building

A subsequent example illustrates that the PCA mode In the following, PCA is applied to the computed
shapes coincide with the elastic mode shapes for linear dynamic response of a 12-story moment-resistant steel
elastic structures having uniform nodal masses, provided frame building subjected to the 1940 NS El Centro earth-
that the data are obtained at small enough time quake ground motion. Linear elastic response is con-
increments over a duration of sufficient length. This sidered as well as nonlinear response involving either
observation is discussed for general distributions of mass beam-hinging or weak-story mechanisms. Estimates of
as follows. the displacement response are presented, based on
The free vibration eigenproblem of elastic dynamics, ESDOF models that are established using the predomi-
given by Kfel lMfel, may be transformed to a stan- nant mode shapes determined by PCA.
dard form [2] by introducing a Cholesky factorization of
the mass matrix given by M LLT. The entries of L, 3.1. Building description and modeling
Lij, are zero for i j. This factorization allows the elastic
dynamics eigenproblem to be restated as K f lf , The 12-story steel moment-resistant frame building
where K L1K(LT)1. The transformed elastic mode (Fig. 1) was designed using the uniform building code
shapes, given by f LTfel, are orthogonal to one [12] distribution of lateral forces for uniform floor
another while satisfying the orthogonality relationships masses equal to 551 kN per floor. The base shear
required of the elastic mode shapes: strength was established to limit drift response; the
j fTel, iLLTfel,j fTel, iMfel,j dij
Tif design is described in more detail as the Flexible-12
f (14)
frame in Black and Aschheim [3]. The frame was
where dij 1 for i j and dij 0 otherwise. designed only for lateral loads in order to validate a
Because the elastic modal responses are independent design methodology. Lateral response was computed
and, in general, are uncorrelated with one another, the using DRAIN-2DX [19]. Flexural response was modeled
displacements u i and u j expressed in terms of the trans- using beam-column elements (Type 02) extending along
formed mode shapes LTel will be uncorrelated, for beam and column centerlines; the post-yield stiffness
closely-spaced elastic response data of sufficient dur- was set equal to 10% of the initial stiffness.
ation. Therefore, the off-diagonal terms in Cu will tend
to zero as t increases. If the principal components are 3.2. Linear elastic response
determined for transformed displacements v LTv,
then Cu is a diagonal matrix provided that the displace- Elastic response was obtained by subjecting the build-
ments u are expressed in terms of the set of orthonormal ing to the 1940 NS El Centro ground motion with accel-
basis vectors that are the eigenvectors of Cv . The PCA eration amplitudes scaled by 0.5. The computed dis-
mode shapes obtained for LTv exhibit the orthogonality placement of the roof relative to the ground is shown in
required of the elastic mode shapes [Eq. (14)], and, in the Fig. 2; the peak of 0.1682 m occurred at 5.61 s. The first
M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103 1095

Fig. 3. Mode shapes obtained from modal and PCAs coincide for
large data sets, for structures responding elastically with uniform
mass distributions.

Table 1
Modal response data for elastic response to El Centro

Mode Viscous Natural Spectral Proportion Cumulative


damping period displacement of variance proportion
(percent of (s) (m) (%) of variance
critical
damping)

1 5.00 2.168 0.1081 96.85 96.85


2 2.77 0.798 0.0483 2.87 99.72
3 2.82 0.470 0.0295 0.26 99.98

summarizes elastic periods and spectral response data for


the first three elastic modes, along with the modal damp-
Fig. 1. Framing configuration and steel member sizes.
ing resulting from the use of mass and stiffness pro-
portional damping in DRAIN-2DX. The USEE 2001
computer program [14] was used to compute the spectral
displacements of Table 1, for the scaled record and
damping present in the dynamic analyses. Because the
PCA mode shapes coincided with the elastic mode
shapes, the proportion of variance represented by these
modes, given by li /tr() is shown on the same lines
of the table. The elastic spectral data characterize peak
quantities while the variance data characterize the degree
to which each mode represents the displacement
response over the duration of response. It can be
observed that 97% of the variance in the displacement
response is represented by the first mode, while the first
two modes represent nearly 100% of the variance in the
Fig. 2. Computed roof displacement history for elastic response to displacement response history.
the El Centro ground motion. Based on the prominence of the first mode in the dis-
placement response and the success that other investi-
three elastic mode shapes are shown by solid lines in gators have had in using ESDOF models, one may
Fig. 3. The PCA modes were computed using 80 s of speculate that an ESDOF system may yield accurate esti-
displacement response data at 0.01-s intervals, and are mates of the peak displacement and the displacement
plotted in the same figure using circles. The PCA mode response history. Using the ESDOF model of ATC-40
shapes computed for this relatively large data set are [1] with the principal component renormalized to unit
seen to coincide with the elastic mode shapes. Table 1 amplitude at the roof results in 1 1.372, where 1 is
1096 M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103

of the 5-s window. The solid line indicates that response


after about 38 s is nearly entirely in the first mode. The
first two modes are sufficient to represent nearly all the
response for 5-s intervals that begin during the first 20
s. The third mode plays a significant role in the response
for 5-s intervals that begin between 20 and 35 s. Three
or fewer modes are seen to be sufficient to represent
nearly 100% of the variance in the displacement
response over the entire history. Since it is possible that
the principal components may rotate (while maintaining
orthonormality) as the moving window advances, the
mode shapes are plotted in Fig. 6 together with the elas-
tic mode shapes. Fig. 6 indicates that the principal
Fig. 4. Comparison of roof displacement response computed for the components deviate little from the elastic mode shapes
MDOF system and estimated with the ESDOF system, for elastic during most intervals. Therefore, the changes in the pre-
response.
dominance of the mode shapes over time in Fig. 5 relate
primarily to the excitation of higher modes rather than
a term that is often described at the participation factor substantial changes in the mode shapes themselves. This
of the first mode [5]. Therefore, the roof displacement is can be expected for elastic response. The increased pro-
estimated to be 1.372 times the displacement of a SDOF minence of the second and third mode contributions at
oscillator having a period of 2.168 s and viscous damp- about 28 s (Fig. 5) causes the roof to oscillate more rap-
ing equal to 5% of critical damping. Fig. 4 compares the idly from about 28 to 37 s (Fig. 2). Since the roof dis-
estimated roof displacement history with values com- placements during this interval are relatively small, it
puted for the MDOF system. The estimated peak is clear that the higher modes do not cause large roof
response can be obtained from Table 1 as 1.372(0.1081 displacements in and of themselves, for this excitation.
m)=0.148 m, a value that is 11.8% less than the com- The PCA technique also may be applied to the
puted peak of 0.1682 m. This difference is attributed to dynamic story shears and to the lateral forces that equi-
the influence of higher modes. librate these story shears in a static analysis. Fig. 7 shows
To better understand the dynamic response of the sys- the first three PCA modes of the lateral force distri-
tem and the role of higher modes on response, PCA may bution, computed in a moving window analysis over 80-
be applied to smaller intervals of the response history. s of response data, along with the lateral force distri-
Advancing the interval over time results in a moving butions that correspond to the elastic mode shapes. The
window analysis. Such an analysis identifies how well lateral force distribution represented by the first PCA
the various modes represent the response as a function mode varies over time, sometimes resembling a first
of time as well as possible changes in the mode shapes mode distribution and other times resembling the second
with time. A 5-s window was advanced in 1-s increments or third elastic modal force distributions. The proportion
over the displacement response data. The proportion of of variance in the lateral forces explained by each PCA
variance explained by each of the three components is mode over 80 s of response data is presented in Table
plotted in Fig. 5, with values plotted at the trailing edge 2. It can be observed that all three PCA modes contribute

Fig. 5. The proportion of variance and cumulative proportion of vari-


ance in the displacement response represented by the first three princi-
pal components, obtained in a moving window analysis of elastic Fig. 6. Principal components of the displacement response obtained
response data. in a moving window analysis of elastic response data.
M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103 1097

computed over 80-s of response data, is given in Table


2. The first PCA mode represents a greater proportion
of the modal shears than is the case for the equilibrating
lateral forces. Table 2 and Figs. 6 8 indicate that the
lateral forces and to a lesser extent the story shears vary
significantly relative to the first mode even though the
displacements are represented well by the first mode.
This may be explained by noting that the lateral forces
associated with mode i are proportional to w2i Sd,ii, where
wi is the modal circular frequency and Sd,i is the spectral
displacement associated with the modal frequency. The
frequency content of the El Centro motion is such that
the spectral displacements associated with the higher
Fig. 7. Principal components of the lateral forces required to stati- modes are small, and thus the first PCA mode is suf-
cally equilibrate the story shears computed by dynamic analysis, ficient to represent nearly all of the displacement
obtained in a moving window analysis of elastic response data. response. However, the squares of the higher mode fre-
quencies elevate the lateral force contributions suf-
ficiently that the first three PCA modal force distri-
Table 2
Efficiency of modal representations of various quantities (elastic butions are needed to represent the variance in the lateral
response) forces with a similar degree of precision.

Quantity Proportion of variance Cumulative proportion 3.3. Nonlinear response


(%) of variance (%)
The original frame was provided with relatively strong
1st 2nd 3rd 1st 2nd 3rd columns such that mechanisms develop that involve
mode mode mode mode mode mode
beam plastic hinging. This section presents two cases of
Lateral forces 37.27 28.20 22.88 37.27 65.47 88.35 nonlinear response to the El Centro ground motion. The
Story shears 80.80 14.62 3.87 80.80 95.42 99.29 first case develops beam-hinging response, consistent
Lateral 96.85 2.87 0.26 96.85 99.72 99.98 with the original design. In the second case, the column
displacements flexural strengths in the first story were artificially weak-
ened to induce a weak-story response to develop. For
both cases the El Centro ground motion was applied with
significantly to the lateral forces required to equilibrate amplitude scaled by a factor of 2.
the story shears computed in the dynamic analysis.
A moving window analysis of the story shears resulted 3.3.1. Beam-hinging response
in the modal shears shown in Fig. 8. These are seen to The computed nonlinear response of the beam-hinging
vary somewhat from the elastic modal shears, parti- frame resulted in the roof displacement response history
cularly for some 5-s intervals. The proportion of vari- shown in Fig. 9. The peak roof displacement of 0.5966
ance in the story shears represented by each PCA mode, m occurred at 5.64 s. Fig. 10 shows the first three elastic
and PCA mode shapes. The response was intermittently
inelastic, and the PCA mode shapes computed over the

Fig. 8. Principal components of the dynamic story shears, obtained


in a moving window analysis of elastic response data. Fig. 9. Roof displacement history of the beam-hinging frame.
1098 M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103

time, and Fig. 12 shows the PCA mode shapes obtained


for each window, together with the elastic mode shapes.
The large departures from the third elastic mode shape
in Fig. 12 occurred at the end of the record at a time
that virtually all of the response was represented by the
first two principal components. Comparison of Figs. 5
and 11 indicates a small reduction in the variance
explained by the first principal component during the
early portion of the response when the largest roof dis-
placements occurred. The variance explained by the first
two principal components is nearly the same for both
cases. Comparison of Figs. 6 and 12 reveals little differ-
ence in the PCA mode shapes, even though moderate
Fig. 10. The elastic (solid line) and PCA mode shapes (dashed line inelastic response developed in the beam-hinging case.
with circles) are nearly the same even for intermittent inelastic In the present case, the elastic mode shapes were con-
response of the beam-hinging frame, when computed over the 80-s sidered adequate approximations of the predominant
duration of response data. mode shapes, based on Fig. 10. Nonlinear static
(pushover) analyses were done by applying lateral forces
in proportion to the first mode shape amplitude and mass
at each floor level. A bilinear curve was fitted to the
capacity curve (Fig. 13) to determine the yield strength
and displacement for response in the first mode. The dis-
placement of the roof at yield is 0.353 m, or 0.72% of
the height of the building, and the base shear coefficient
at yield is 0.173. The post-yield stiffness is 17.5% of the
initial stiffness. The elastic mode shape and uniform
mass distribution result in 1 1.372, following the
ATC-40 [1] formulation. The yield strength coefficient

Fig. 11. The proportion of variance and cumulative proportion of


variance represented by the first three principal components, obtained
in a moving window analysis of the beam-hinging frame response data.

80-s duration of response data are nearly equal to the


elastic mode shapes. Because the elastic and PCA mode
shapes are nearly identical, results for each mode are
tabulated together in Table 3. The spectral displacements
of Table 3 are the peak elastic displacements computed
for the scaled record, using the viscous damping present
in the nonlinear dynamic analysis.
A moving window PCA analysis was done using a 5- Fig. 12. The principal components obtained in a moving window
s window advanced in 1-s increments. Fig. 11 shows the analysis of the displacement response data tend to follow the elastic
proportion of variance represented by each mode over mode shapes, even for beam-hinging responses.

Table 3
Response data for the beam-hinging frame

Mode Viscous damping (percent of Period (s) Spectral Peak displacement Proportion of Cumulative proportion
critical damping) displacement (m) (m) variance (%) of variance (%)

1 5.00 2.168 0.4323 0.4207 96.75 96.75


2 2.77 0.798 0.1930 0.1930 2.84 99.58
3 2.82 0.470 0.1181 0.1181 0.37 99.96
M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103 1099

using the ESDOF system captures the essence of the roof


displacement response computed for the MDOF system.
The peak displacements of the ESDOF systems corre-
sponding to the first three modes are reported in Table
3. The 1st mode ESDOF system reached a peak ductility
of m 1.64 and peak displacement of 0.4207 m; the
estimate of peak roof displacement due to the first mode
is 1.372(0.4207 m)=0.5772 m. The contribution of the
second mode to the peak roof displacement is estimated
as 0.5682(0.1930)=0.1097 m, based on 2 0.5682,
although its timing may differ from the timing of the
first mode peak. An SRSS combination results in a peak
roof displacement estimate of (0.57722 0.10972)0.5
Fig. 13. Capacity curve obtained by applying lateral forces pro- 0.5875m, nearly equal to the first mode estimate. Rela-
portional to the product of the elastic modal amplitude and mass at
each floor in a nonlinear static analysis.
tive to the computed peak displacement of 0.5966 m, the
first mode estimate is 3.3% low and the SRSS estimate
for the first two modes is 1.5% low.
of the ESDOF system, Cy, determined by the ATC-40
method is 0.220. This value may be determined more 3.3.2. Weak-story response
directly by: To introduce a weak-story response, the frame was


artificially modified by reducing the strengths of the low-
2p 2xy est story columns to 25% of their original values. The
Cy (15)
T1 g elastic properties were not changed; hence all building
frames described in this paper have the same elastic
where xy is the yield displacement of the ESDOF system
mode shapes and natural periods. The weak-story model
and T1 is the fundamental period of vibration of the
was subjected to the same ground motion record used
MDOF system. When the ESDOF system is determined
for the beam-hinging frame. The computed roof dis-
with a mode shape that differs from the elastic mode
placement history is shown in Fig. 15. Peak roof dis-
shape, Eq. (15) ensures the initial period of the ESDOF
placements of 0.4011 and 0.3998 m occurred at 4.69
system matches the fundamental period of the MDOF
and 5.69 s, respectively, and a small permanent displace-
system. (The periods obtained with the ATC-40 [1] and
ment remained at the end of the response.
FEMA [7,8] formulations do not, in general, match the
Fig. 16 plots the PCA mode shapes computed for 80
fundamental period of the MDOF system.) Roof dis-
s of displacement response data. Although a weak-story
placement histories were estimated based on the poten-
mechanism was not apparent in the displaced shape at
tially nonlinear response of the analogous SDOF oscil-
the end of the response, the PCA mode shapes clearly
lator. The estimated roof displacement history is
identify the intermittent weak story responses. The vari-
compared with the roof displacement history computed
ance in the displacement response represented by the
in the nonlinear response of the MDOF system in Fig.
first three modes is given in Table 4. While the mode
14. It can be observed that the response history estimated
shapes are clearly different from the elastic and beam-
hinging cases, the first and second modes were able to
represent similar proportions of the variance. The pro-

Fig. 14. Comparison of roof displacement histories computed for the


beam-hinging MDOF system and estimated using an ESDOF system
based on the first mode. Fig. 15. Roof displacement history of the weak-story frame.
1100 M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103

Fig. 16. PCA mode shapes computed over the 80-s response clearly Fig. 18. The first three PCA mode shapes, obtained in a moving win-
indicate weak story behavior. dow analysis of the weak-story frame displacement response data.
Weak story behavior is clearly evident for 5-s intervals that begin in
the first 12 s of response.
Table 4
Response data for the weak-story frame
interval. The mode shapes at 3 s are plotted in Fig. 19,
PCA mode Proportion of Cumulative proportion of and these mode shapes are used subsequently to estimate
variance (%) variance
response based on an ESDOF model.
The data of Fig. 19 indicate that response in the first
1 96.57 96.57
2 2.75 99.33 mode changes from the elastic mode shape at low ampli-
3 0.56 99.89 tudes to become dominated by a weak-story mechanism
as displacements increase. To represent such a response,
the ideal ESDOF system would reflect this change in
portion of variance explained by each mode is plotted predominant mode shape as displacements increase. Ide-
cumulatively as a function of time in Fig. 17, based on ally, the capacity curve would be generated using an
a moving window analysis (a 5-s window advanced in adaptive load pattern [4,6,9] such that the deflected
1-s increments). Comparison to Fig. 11 reveals overall shape of the system corresponds to the predominant
similarities. The corresponding mode shapes obtained in mode of response. The change in deflected shape affects
the moving window analysis are plotted in Fig. 18. Weak the mass that participates in the response; the lateral
story behavior is evident in the first three modes, parti- resistance provided by the structure relative to the partic-
cularly for the 5-s intervals that begin during the first 12 ipating mass ideally would be used to define the load-
s. Peak roof displacements occurred at 4.69 and 5.69 s, deformation response of the ESDOF system. These
which are approximately centered in the 5-s windows theoretical considerations provide a context in which
that begin between 2 and 3 s. The first, second, and third decisions are made about how an ESDOF model should
mode shapes appear not to change significantly in this be developed. Because ESDOF systems necessarily omit
(or oversimplify) interactions with higher modes, they

Fig. 17. The proportion of variance and cumulative proportion of


variance represented by the first three principal components in a mov- Fig. 19. The PCA mode shapes for the 5-s window beginning at
ing window analysis of the weak-story frame response data. 3 s.
M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103 1101

will always be approximate. The benefit of an ESDOF


analysis is the ease with which approximate information
can be generated, and to preserve this utility, analyses
reported here make use of an invariant lateral load pat-
tern throughout the linear and nonlinear portions of
response. The lateral load patterns examined here are
based on the PCA mode shape, a displaced shape asso-
ciated with the development of the weak-story mech-
anism, and the elastic mode shape. The yield strength
coefficient is determined using Eq. (15) to ensure that
the period of the ESDOF system matches the fundamen-
tal period of vibration of the MDOF system.
Because a nonlinear dynamic analysis had been done,
it was possible to determine the capacity curve (Fig. 20) Fig. 21. Estimated roof displacement history using the PCA mode
by applying lateral forces proportional to the first PCA shape in conjunction with an analogous SDOF oscillator, and the
response computed by nonlinear dynamic analysis of the MDOF sys-
mode shape amplitude (Fig. 19) and mass at each floor. tem.
A bilinear curve fit to the capacity curve had a roof dis-
placement at yield of 0.140 m, base shear coefficient at
yield of 7.55%, and post-yield stiffness equal to 52% of 1. applying lateral forces proportional to the floor
the initial stiffness. Using the ATC-40 [1] formulation masses in the pushover analysis; and
with the PCA mode shape results in 1 1.307. There- 2. using elastic modal properties (1 1.372 and
fore, the yield displacement of the ESDOF system is T1 2.17s) to convert between the ESDOF system
0.140 / 1.307 0.107m. Eq. (15) requires the yield and the MDOF system.
strength coefficient of the ESDOF system, Cy, to be
9.14%. The viscous damping, as a percent of critical This results in a roof displacement at yield of 0.118 m,
damping, and post-yield stiffness, as a percent of initial base shear coefficient at yield of 7.58%, and a peak roof
stiffness, associated with response of the MDOF system displacement of 1.372(0.322)=0.442 m. This estimate is
in the first mode were preserved in the ESDOF system. about as good as the one computed using the first PCA
The peak displacement of the ESDOF system is 0.343 mode but was made without first identifying the PCA
m; thus, the estimate of the peak roof displacement is mode shape.
1.307(0.343) 0.448m. This value exceeds the com- If lateral forces are applied proportional to the first
puted peak response of 0.4011 m by 12%. Even so, the elastic mode, the roof displacement at yield is 0.153 m
estimated roof displacement history compares well to the and the base shear coefficient at yield is 7.47%. Using
roof displacement history computed for the MDOF sys- the elastic mode shape and period to obtain the ESDOF
tem, as shown in Fig. 21. system results in a first mode estimate of peak roof dis-
Alternative load patterns may be applied to generate placement of 0.494 m, 23% higher than the computed
the capacity curve. If the weak-story mechanism has peak.
been identified, a blind estimate of the displacement
response may be made by:
4. Conclusions

The theory of PCA was presented using the nomencla-


ture of structural engineering. Examples illustrated the
application of the theory for a 12-story moment-resistant
frame building having uniformly-distributed lumped
masses, subjected to earthquake excitations. Linear
response and nonlinear responses characterized by
beam-hinging or weak-story mechanisms were con-
sidered. Displacement estimates were made using equi-
valent SDOF models, based on the principal compo-
nents identified in the response data. The following
conclusions can be made.

1. The displacement response of MDOF structures sub-


Fig. 20. Capacity curve obtained by applying forces in proportion to jected to base excitation may be characterized by a
the amplitude of the first PCA mode (at 3 s) and mass at each floor. predominant mode. The precise mode shape and the
1102 M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103

degree to which the response is characterized by this mechanisms are not apparent in the final displaced
mode shape may be determined for a given excitation shape of the structure. The PCA technique may be
by the analytical procedure known as PCA. This pro- useful for evaluating when the distribution of member
cedure is applicable to data obtained from structures strengths and stiffnesses is adequate to prevent the
responding linearly or nonlinearly. The PCA mode development of undesirable mechanisms.
shapes identified in different intervals of time may
differ from one another.
2. In the cases investigated, nearly all of the variance in Acknowledgments
the displacement response can be represented by one
or two principal components. The first mode contri- The development of this paper involved a number of
bution to the total variance in the displacement people. Early applications of the method were by gradu-
response was observed to approach 100% towards the ate students Andrew Jackson, Chatdanai Wissawapaisal,
end of the response. Lateral forces and story shears and Mehmet Inel. Keith Hjelmstad and Ping Gu pro-
vary more rapidly than do the displacements and often vided helpful comments to an early manuscript. The
require inclusion of a larger number of principal careful review and comments provided by anonymous
components to achieve similar precision. reviewers are greatly appreciated. The support of the
3. The PCA mode shapes coincide with the elastic mode National Science Foundation via a CAREER award
shapes, when they are determined for the displace- (CMS-9984830) to the first author is gratefully acknowl-
ment response of linear systems having uniform mass edged. This work was supported in part by the Earth-
distributions using data sets of sufficient size. quake Engineering Research Centers Program of the
Relationships between the PCA and elastic mode National Science Foundation under Award No. EEC-
shapes were provided for nonuniform mass distri- 9701785.
butions. The sequencing of the PCA and elastic mode
shapes may differ, depending on the degree to which
response in a given mode is excited by the ground References
motion.
4. The first PCA mode represented 95% of the vari- [1] ATC-40, Seismic evaluation and retrofit of concrete buildings.
ance of the displacement response for each of the 12- Report No. ATC-40. Redwood City (CA): Applied Technology
Council; 1996.
story buildings and excitations studied, when the [2] Bathe KJ. Finite Element Procedures in Engineering Analysis,
analysis was done over an 80-s window. This was the Englewood Cliffs (NJ): Prentice-Hall; 1982, p. 5739.
case despite the potential for significant higher modes [3] Black EF, Aschheim M. Seismic design and evaluation of multi-
that might be expected for a frame building having a story buildings using yield point spectra. CD Release 00-04. Uni-
fundamental period of vibration of 2.17 s. During versity of Illinois (Urbana): Mid-America Earthquake Center;
2000.
smaller intervals of the response, the relative contri- [4] Bracci JB, Kunnath SK, Reinhorn AM. Seismic performance and
bution of the second and third modes was larger, retrofit evaluation of reinforced concrete structures. ASCE Jour-
although this occurred when the roof displacements nal of Structural Engineering, American Society of Civil Engin-
were substantially below their peak values. eers 1997;123:310.
5. The beam-hinging frame resulted in predominant [5] Chopra AK, Goel RK. Capacity-demand-diagram methods based
on inelastic design spectrum. Earthquake Spectra
mode shapes that were nearly equal to the elastic 1999;15(4):63756.
mode shapes. An equivalent SDOF model based on [6] Elnashai AS. Advanced inelastic static (pushover) analysis for
the first mode was useful for estimating the peak dis- seismic design and assessment. In: G. Penelis International Sym-
placement response and roof displacement response posium on Concrete and Masonry Structures; 2000, p. 2334.
history. The predominant mode shapes of the weak- [7] FEMA-273. NERHP Guidelines for the Seismic Rehabilitation of
Buildings. Report FEMA-273. Federal Emergency Management
story frame were very different from the elastic mode Agency; 1997.
shapes and clearly indicated the development of a [8] FEMA-274. NERHP Commentary on the Guidelines for the Seis-
weak-story mechanism. An equivalent SDOF model mic Rehabilitation of Buildings. Report FEMA-274. Federal
based on the PCA mode shape determined in the 5- Emergency Management Agency; 1997.
s interval in which the peak roof displacement [9] Gupta B, Kunnath SK. Spectra-based pushover procedure for
seismic evaluation. Earthquake Spectra 2000;16(2):36791.
occurred was useful for estimating the peak roof dis- [10] Gutie rrez E, Zaldivar JM. The application of Karhunen-Loe ve,
placement of the weak-story frame. For these cases, or principal component analysis method, to study the non-linear
estimates of peak roof displacement based on the first seismic response of structures. Earthquake Engineering and
PCA mode were within 12% of the peak quantities Structural Dynamics 2000;29:126186.
computed in nonlinear dynamic analyses of the [11] Hotelling H. Analysis of a complex of statistical variables into
principal components. Journal of Educational Psychology
MDOF structures. 1933;24:417520.
6. The PCA technique may be used to identify the devel- [12] ICBO. Uniform building code. International Conference of Build-
opment of weak-story mechanisms even when such ing Officials, Whittier, CA; 1997.
M.A. Aschheim et al. / Engineering Structures 24 (2002) 10911103 1103

[13] Inel M. Displacement-based strategies for the performance-based applications in the social sciences, Series No. 07-069. Newbury
seismic design of short bridges considering embankment flexi- Park (CA): Sage, 1989.
bility. Ph.D. thesis, University of Illinois, Urbana (in preparation). [18] Pearson K. On lines and planes of closest fit to systems of points
[14] Inel M, Bretz E, Black E, Aschheim M, Abrams D. USEE 2001 in space. Phil. 1901;2:55972.
utility software for earthquake engineering: program, report, and [19] Prakash V, Powell GH, Campbell S. Drain-2DX base program
users manual. CD Release 01-05. University of Illinois (Urbana): description and user guide, v. 1.10, Report No. UCB/SEMM-
Mid-America Earthquake Center. Available for download from 93/17. Berkeley: Department of Civil and Environmental Engin-
http://mae.ce.uiuc.edu/. eering, University of California; 1993.
[15] Krawinkler H. Pushover analysis: why, how, when, and when not [20] Venables WN, Ripley BD. Modern Applied Statistics with S-
to use it. In: Proceedings of the 65th Annual Convention, Struc- Plus. 2nd ed. Berlin: Springer, 1997
tural Engineers Association of California, Oct. 16, Maui [21] Wissawapaisal C, Aschheim M. Modeling the transverse response
(Hawaii), p. 1736; 1996. of short bridges subjected to earthquakes. CD Release 00-05.
[16] Mardia V K, Kent JT, Bibby JM. Multivariate analysis. London: University of Illinois (Urbana): Mid-America Earthquake
Academic Press, 1979. Center; 2000.
[17] Dunteman GH. Principal components analysis. Quantitative

Das könnte Ihnen auch gefallen