You are on page 1of 23

This article was downloaded by: [University of Bucharest ]

On: 17 November 2014, At: 22:40


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Earthquake Engineering


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/ueqe20

Sensitivity Analysis for Soil-Structure


Interaction Phenomenon Using
Stochastic Approach
a

M. Moghaddasi , J. G. Chase , M. Cubrinovski , S. Pampanin &


A. Carr

Department of Civil and Natural Resources Engineering , University


of Canterbury , Christchurch , New Zealand
b

Department of Mechanical Engineering , University of Canterbury ,


Christchurch , New Zealand
Accepted author version posted online: 12 Apr 2012.Published
online: 14 Sep 2012.

To cite this article: M. Moghaddasi , J. G. Chase , M. Cubrinovski , S. Pampanin & A. Carr (2012)
Sensitivity Analysis for Soil-Structure Interaction Phenomenon Using Stochastic Approach, Journal of
Earthquake Engineering, 16:7, 1055-1075, DOI: 10.1080/13632469.2012.677570
To link to this article: http://dx.doi.org/10.1080/13632469.2012.677570

PLEASE SCROLL DOWN FOR ARTICLE


Taylor & Francis makes every effort to ensure the accuracy of all the information (the
Content) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions

Journal of Earthquake Engineering, 16:10551075, 2012


Copyright A. S. Elnashai & N. N. Ambraseys
ISSN: 1363-2469 print / 1559-808X online
DOI: 10.1080/13632469.2012.677570

Sensitivity Analysis for Soil-Structure Interaction


Phenomenon Using Stochastic Approach
M. MOGHADDASI1 , J. G. CHASE2 , M. CUBRINOVSKI1 ,
S. PAMPANIN1 , and A. CARR1

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

Department of Civil and Natural Resources Engineering, University of


Canterbury, Christchurch, New Zealand
2
Department of Mechanical Engineering, University of Canterbury, Christchurch,
New Zealand
This article analyses 1.36 million realistic soil-structure interaction (SSI) scenarios in a systematic
fashion to define the correlation between soil, structural, and system parameters and interaction
effects on the structural response. In the analyses, a soil-shallow foundation-structure model that
satisfies design building code requirements is utilized. It has been identified that soil shear wave
velocity, shear wave velocity degradation ratio, structure-to-soil stiffness ratio, and structural aspect
ratio combined with the system stiffness are the key parameters whose variation significantly affects
variation in structural response. The critical range of variation of these parameters resulting in a
detrimental SSI effects is also defined.
Keywords Soil-Structure Interaction; Sensitivity Analysis; Stochastic Approach; Influential SoilStructure Parameters; Critical Range of Variation

1. Introduction
It has been clearly demonstrated in the accompanying article [Moghaddasi et al., 2012] that
the structural response of a soil-structure system to seismic forces is strongly affected by the
impact of uncertainty in soil and structural parameters accompanied with the inherent randomness of the ground motion. For the single-degree-of-freedom (SDOF) model assumed,
it has also been shown that soil-structure interaction (SSI) effects cannot be always safely
ignored, given the likelihood of having amplification in the structural response due to foundation flexibility. Thus, at least for critical scenarios, SSI effects have to be taken into
account in the seismic design procedure.
However, significant complexities and variation in the structural response make the
identification of the critical scenarios a challenging task. A considerable step towards
identification of these critical scenarios is to: (1) define the correlation between different parameters and the observed variation in response modification factors; and (2) to
comprehensively characterize and quantify the scenarios causing either reduction or amplification in the structural response. Thus, those scenarios causing the greatest likelihood of
exceeding demand can be more precisely defined.

Received 11 July 2011; accepted 14 March 2012.


Address correspondence to M. Moghaddasi, Department of Civil and Natural Resources Engineering,
University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand; E-mail: masoud.moghaddasi@
gmail.com

1055

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

1056

M. Moghaddasi et al.

In this context, Veletsos and Nair [1975] and Bielak [1975] showed that the difference
between seismically induced linear response of a fixed-base and a flexible-base system is
strongly affected by structural aspect ratio, soil Poisons ratio, soil hysteretic damping ratio,
a dimensionless parameter expressing the relative stiffness of foundation and structure,
and a dimensionless parameter representing soil-to-structure mass. Following this study, a
more comprehensive investigation was carried out by Ciampoli and Pinto [1995]. They concluded that structural response of a nonlinear system does not show any systematic dependencies on the parameters regulating SSI phenomena, and it is statistically reduced due to
foundation flexibility. Later on, Stewart et al. [1999a,b] used a comprehensive database of
recorded data, and concluded that ratio of structure-to-soil stiffness has the greatest influence on the structural response of a soil-structure system to seismic forces. In addition,
it has been established that structural aspect ratio, foundation embedment, and foundation
flexibility are the other parameters with significant effect on inertial interaction. Finally,
based on the framework of dimensional analysis, Zhang and Tang [2009] showed that SSI
effects are highly dependent on the structure-to-pulse frequency, foundation-to-structure
stiffness ratio, and foundation damping ratio. They also presented certain limits for these
controlling parameters to distinguish whether or not SSI effects were significant.
Given the existing controversy in the previous findings and the assumed limitations
in the analysis performed, it is believed that the most rational way to identify critical SSI
scenarios is to make use of a probabilistic approach. With this in mind, the results of a
comprehensive probabilistic simulation explained in detail in the accompanying paper are
summarized in this article, and then used to: (1) define the correlation and dependency
between structural response modification factors and model parameters; (2) identify the
key parameters having a significant effect on the structural response; (3) present the trend
of variation of SSI effects due to change of these parameters; and finally (4) quantify the
critical range of variation of the parameters of consequence that causes detrimental SSI
scenarios (i.e., scenarios with amplified structural response).
This probabilistic analysis is a critically important step towards understanding and
reliably characterizing the complex SSI problem. It should also be noted that the presented
outcomes are limited to a SDOF system, and that the scenarios presented do not consider
the extreme conditions, such as those imposed by liquefiable soils or near-fault effects on
the ground motion. However, the cases covered represent a significant majority of typical
design cases and scenarios.

2. Methodology and Monte Carlo Simulation


2.1. Outlines of the Procedure for the Probabilistic Study
A Monte Carlo technique was used to study sensitivity of inelastic seismic structural
response of an established rheological soil-shallow foundation-structure system to different model parameters. Four measures of structural response are considered: (1) structural
distortion us that represents the horizontal displacement of the structure relative to the foundation; (2) structural drift dr that is the summation of foundation rocking and normalized
structural distortion by the effective height; (3) total displacement utot that represents the
displacement measured at the roof level including lateral displacement resulted from foundation response and structural distortion; and (4) structural acceleration as that represents
the total transmitted acceleration (or force) to the structure. A large number of nonlinear time-history simulations were run over models with randomly selected parameters
using a suite of recorded ground motions that were then scaled. Parameters of these systems were systematically defined by a random process carefully ensuring realistic models,

Sensitivity Analysis for Soil-Structure Interaction

1057

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

as well as covering a specific period range in the design spectrum. The period range of
T FB = 0.2, 0.3 . . . 1.8 s was considered to: (1) represent fixed-base (FB) structures with
total height of 330 m and (2) satisfy the period-height relationship adopted in the New
Zealand Standard [NZS1170.5, 2004] for seismic design of structures. For each period considered TFB , 1,000 corresponding soil-foundation-structure (SFS) models were generated
by assembling the randomly defined parameters for the SFS system and using commonly
accepted relationships between various model parameters. The number 1,000 was chosen
with the intention to: (1) give the best fit statistical distribution for the randomly selected
parameters and (2) increase the confidence level of the Monte-Carlo simulation compared
to the exact solution [Fishman, 1996]. All the nonlinear time-history simulations were
carried out using the finite element program Ruaumoko 2D [Carr, 2009].
2.2. Dynamic Soil-Foundation-Structure Model
The soil-foundation-structure (SFS) system investigated in this study is constituted from
a rheological soil-foundation element and a structure (Fig. 1), following the substructure
technique. The structure is modelled as a yielding SDOF system with a stiffness-degrading
force-deformation hysteresis rule as Takeda (bilinear envelope with strain hardening and
stiffness degradation) assuming a 5% post-yield stiffness and unloading and reloading
parameters of = 0.3 and = 0.2, respectively. This SDOF representation is an approximate model of a multi-story building vibrating in its fundamental natural mode. The
considered SDOF structure is assumed to have the same period and viscous damping coefficient as that of the corresponding FB system and is characterized by its mass ms , lateral
stiffness ks , coefficient of relative viscous damping cs , and effective height he .
The soil-foundation element was modelled by a lumped-parameter model representing
a rigid circular footing resting on the soil surface and having a perfect attachment to the soil.
Moreover, the foundation was assumed to have no mass or mass moment of inertia about
the horizontal axis. For evaluating the dynamic soil impedances incorporating soil nonlinearity, the frequency-independent coefficients of a rheological Cone model [Wolf, 1994]
were modified using the conventional equivalent linear method [Seed and Idriss, 1970].

FIGURE 1 Soil-shallow foundation-structure model for horizontal and rocking foundation


motions (color figure available online).

1058

M. Moghaddasi et al.

To avoid more complication in time-domain analysis, soil material damping was assumed
to be viscous instead of hysteretic. In the model presented, the horizontal degree of freedom and the rocking degree of freedom are considered as representatives of translational
and rocking motions of foundation, respectively, and the effects of vertical and torsional
motions are ignored.

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

2.3. Uncertainty in System Parameters and Randomness in Ground Motions


In seismic analysis, there are two recognized principal sources of uncertainty that need to
be addressed: (1) model parameters and (2) input ground motion. In this research, both
types of uncertainties are covered. A brief overview of the stochastic selection process is
presented here, but for more detailed information the interested reader is referred to the
accompanying article.
2.3.1. Selection of Uncertain Model Parameters. All soil parameters defining the soilfoundation element were considered as uncertain parameters. Initial soil shear wave
velocity (Vs )0 , shear wave velocity degradation ratio (Vs )sec /(Vs )0 , where (Vs )sec represents the degraded shear wave velocity, soil mass density and Poissons ratio were
defined as independent parameters. For each of them, a realistic range was defined, and
then 1000 uniformly distributed values were assigned to that range.
Randomly varying structural parameters include: structural effective height he , foundation radius r, and structural mass ms . Depending on these randomly generated parameters,
the values for structural stiffness ks and structural damping cs were then calculated.
To achieve realistic SFS models, the selection of these structural parameters was constrained by commonly accepted relationships for the structure and/or the whole SFS
system. These relationships can be found in Table 1 of Moghaddasi et al. [2012].
2.3.2. Selection of Ground Motions. Forty large-magnitude, small-distance ground
motions recorded on stiff/soft soil, i.e., soil type C with Vs = 180 360 m/s and soil
type D with Vs < 180 m/s to a depth of 30 m based on USGS soil geomatrix classification, were used as input ground motions in the Monte Carlo simulations. The time-histories
were obtained from the PEER and NGA databases available online at the link http://peer.
berkeley.edu/nga.
The number 40 was chosen to reduce the variance in the response due to record-torecord variability and obtain an estimate of median response within a factor of 0.1 with
a 95% probability [Shome et al., 1998]. In addition, these recorded ground motions were
selected to satisfy the constrains of: (1) magnitude in the range of 6.57.5; (2) closest
distance to fault rupture in the range of 1540 km; and (3) peak ground acceleration (PGA)
greater than 0.1 g.
The selected records were then scaled to distribute PGAs within the range of 0.30.8 g,
assuming that a nonlinear behavior of the structure will be caused by those intensity levels.
Given the rigorous scaling criteria and recommendations in NZS 1170.5, all scaling factors
were chosen to be between 0.3 and 3.0.

3. Correlation Between SSI Effects and Model Parameters


To investigate the correlation between SSI induced modification in structural response and
randomly selected model parameters, the variation in the structural response modification
factors, for all considered groups of models categorized based on TFB , was examined as a
function of:

Sensitivity Analysis for Soil-Structure Interaction

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

1059

soil parameters: (1) soil mass density ; (2) Poisons ratio ; (3) initial soil shear
wave velocity (Vs )0 ; and (4) shear wave velocity degradation ratio (Vs )sec /(Vs )0
structural parameters: (1) structural effective height he ; (2) foundation radius r;
(3) structural mass ms ; and (4) structural aspect ratio he /r
soil-structure system parameters: (1) ms /r2 ; (2) ms / r2 he ; (3) = (Vs )sec TFB /he ;
and (4) = (Vs )hsece TFB (he /r)0.25 .

Note that the combined model parameters selected are based on the previous studies in
literature, such as Ciampoli and Pinto [1995], Stewart et al. [1999a], and Wolf [1994],
which introduce the regulating parameters in SSI phenomenon. In this context, he /r represents the structural aspect ratio. This parameter is not a complete parameter in terms
of describing both soil and structural characteristics. However, it is a geometric parameter of immediate engineering and design significance, as well as easily and robustly
determined. Furthermore, ms /r2 and are measures of structure-to-soil mass ratio, and
is a representative of structure-to-soil stiffness ratio. Finally, the combined effect of
and he /r is captured in , a parameter suggested by Ciampoli and Pinto [1995] to
be the best suited in measuring the limit condition where SSI effects are not worth to
consider.
In the following, the possibility of having either a linear or nonlinear correlation is
analyzed and discussed. For presentation purpose, structural response modification factors
are denoted as: (us )SSI /(us )FB , drSSI /drFB , (utot )SSI /(utot )FB , and (as )SSI /(as )FB , representing
modification in structural distortion, structural drift, total displacement, and structural
acceleration due to SSI effects, respectively.

3.1. Linear Correlations


The existing linear dependency of the structural response modification factors on the
considered parameters is presented through Pearson correlation coefficient . Pearson correlation coefficients are obtained by dividing the covariance of two variables by the product
of their standard deviations and represent the level of correlation across a range [Dowdy
and Wearden, 1983]. Assuming X is the calculated structural response modification factor with mean and standard deviation values of X and X , and Y is the model parameter
with mean and standard deviation values of Y and Y , the Pearson correlation coefficient
between these two random variables is defined:
(X, Y) =

E[(X X )(Y Y )]
COV(X, Y)
=
X Y
X Y

(1)

where E represents the expected value and COV means covariance.


The Pearson correlation coefficient is +1 in a perfectly increasing (positive) linear
relationship, and -1 in the case of a perfectly decreasing (negative) linear relationship.
It approaches zero when there is less of a correlation between variables. In all other cases,
it gives values between -1 and +1 indicating the degrees of linear dependence between
the variables. If the variables are independent, the Pearson correlation coefficient is zero.
However, the converse is not always true.
The Pearson correlation coefficients representing the linear dependency between structural response modification factors and the adopted soil, structural, and soil-structure
system parameters are presented in Figs. 2, 3, and 4, respectively. Note that in the spectral

1060

M. Moghaddasi et al.

(Vs)0

(Vs)sec /(Vs)0

[(u ) /(u ) ,Y]

TK

(Vs)0

(Vs)sec /(Vs)0

TK
0.5

FB

s FB

[dr /dr ,Y]

0.5

SSI

s SSI

0
0.5

0.5

1
0.2

0.4

0.6

0.8

(a)

1.2

1.4

1.6

1.8

0.2

(Vs)0

0.8

1
1.2
TFB (s)

(Vs)0

1.4

1.6

1.8

(Vs)sec /(Vs)0

1
TK

[(a ) /(a ) ,Y]

TK
0.5

0.5

s FB

tot FB

0.6

(Vs)sec /(Vs)0

[(u ) /(u ) ,Y]

0.4

(b)

TFB (s)

s SSI

tot SSI

0
0.5

0.5

1
0.2

0.4

0.6

0.8

(c)

1.2

1.4

1.6

1.8

0.2

0.4

0.6

0.8

(d)

TFB (s)

1
1.2
TFB (s)

1.4

1.6

1.8

FIGURE 2 Pearson correlation coefficient spectra representing the correlation between


structural response modification factors and soil parameters.
he

ms

he

he/r

/(u ) ,Y]

TK
/dr ,Y]

0.5

[dr

he/r

0.5

FB

s FB

SSI

s SSI

[(u )

ms

TK

0.5

0.5

1
0.2

0.4

0.6

0.8

(a)

1.2

1.4

1.6

ms

he/r

0.8

1.2

1.4

1.6

1.8

he

ms

he/r

/(a ) ,Y]

0.5

[(a )

TK

0.5

0.5

s FB

tot FB

0.6

TFB (s)

TK

/(u ) ,Y]

0.4

(b)

TFB (s)

he

0.2

1.8

s SSI

tot SSI

[(u )

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

0.5

0.2

(c)

0.4

0.6

0.8

1
TFB (s)

1.2

1.4

1.6

1.8

0.2

(d)

0.4

0.6

0.8

1.2

1.4

1.6

1.8

TFB (s)

FIGURE 3 Pearson correlation coefficient spectra representing the correlation between


structural response modification factors and structural parameters.

Sensitivity Analysis for Soil-Structure Interaction

ms/r^3

0.5

0.5
0

0.5
1

0.2

0.4

0.6

0.8

1.2

1.4

1.6

0.2

1.8

0.4

0.6

0.8

ms/r^3

1.2

ms/r^3

1.4

1.6

1.8

1
TK

TK

[(a ) /(a ) ,Y]

0.5

s FB

0.5
0

s SSI

tot FB

1
TFB (s)

TFB (s)

tot SSI

FB

SSI

s SSI

s FB

[dr /dr ,Y]

0.5

[(u ) /(u ) ,Y]

TK

TK

[(u ) /(u ) ,Y]

ms/r^3
1

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

1061

0.5
1

0
0.5
1

0.2

0.4

0.6

0.8

1
TFB (s)

1.2

1.4

1.6

1.8

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

TFB (s)

FIGURE 4 Pearson correlation coefficient spectra representing the correlation between


structural response modification factors and soil-structure system parameters.
format presented, all considered scenarios are taken into account and categorized based on
values of fundamental structural period TFB .
3.1.1. Correlation between SSI Effects and Soil Parameters. In terms of soil parameters, as
shown in Fig. 2, only (Vs )0 and (Vs )sec /(Vs )0 have a more pronounced linear correlation with
(us )SSI /(us )FB and (as )SSI /(as )FB . The existence of linear correlation for the other different
scenarios can almost always be ignored. In addition, the existing correlations are stronger
for (us )SSI /(us )FB compared to those for (as )SSI /(as )FB . However, note that the correlations
observed are not a very strong linear correlation as the Pearson correlation coefficients are
not very close to 1 ( 0.5). It should also be noted that, the existing linear correlations (or Pearson correlation coefficients) are almost unchanged for all periods considered,
meaning, the dependency of (us )SSI /(us )FB and (as )SSI /(as )FB on (Vs )0 and (Vs )sec /(Vs )0 are
independent from TFB .
3.1.2. Correlation between SSI Effects and Structural Parameters. If the linear correlation
between structural parameters and structural response modification factors are considered,
no significant correlation exist as all Pearson correlation coefficients are small compared to
1. This result is illustrated in Fig.3. Therefore, no linear trend can be defined to correlate
the variation in structural parameters to the modification in structural response due to SSI
effects.
3.1.3. Correlation between SSI Effects and Soil-Structure System Parameters. When soilstructure system parameters are taken into account (Fig. 4), and are the only parameters

1062

M. Moghaddasi et al.

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

that have a more pronounced linear relationship with (us )SSI /(us )FB and (as )SSI /(as )FB .
In addition, the correlation between and (us )SSI /(us )FB or (as )SSI /(as )FB is more likely
to be linear compared to that for , as the corresponding Pearson correlation coefficients are closer to 1. It is also noted that the evaluated Pearson correlation coefficients,
representing the dependency of structural response modification factors to and , are
independent from the periods considered. Parameters and also show a small linear
correlation with drSSI /drFB and (utot )SSI /(utot )FB when stiff structures (TFB 0.6 s) are
considered.

3.1.4. General Comments. Finally, considering the increasing (positive) or decreasing


(negative) linear relationship between structural response modification factors and soil,
structural, and soil-structure system parameters, it is concluded that (us )SSI /(us )FB and
(as )SSI /(as )FB are increased when: (1) (Vs )0 increases; (2) smaller degradation in shear
wave velocity occurs; (3) increases; or (4) decreases. These trends are discussed in
more detail in Sec. 4.

3.2. Nonlinear Correlations


To examine the possibility of having a nonlinear correlation between the response modification factors and the soil, structural, and soil-structure system parameters, the graphs
showing data distribution are considered. In these graphs, the response modification factors for each selected group of models with a specific period are represented based on the
variation of a certain parameter. By examining all possible scenarios, it has been seen that
the measured structural response modification factors only have an obvious relationship
with and . The graphs are noisy for all other cases. Therefore, parameters having no linear correlation with the response modification factors (i.e., , , he , r, ms , he /r, ms / r2 he
and ms /r3 ) also have no distinguishable nonlinear dependency. In other words, SSI
induced modification in structural response does not have a significant linear or nonlinear
correlation with any of these parameters.
To avoid presenting unnecessary information, only the graphs showing the variation of (us )SSI /(us )FB and (utot )SSI /(utot )FB with and are illustrated and discussed
in the following. In these graphs, the Pearson correlation coefficients previously presented are also shown to specify the existing relationship in terms of being a linear or
nonlinear.
Figure 5 illustrates the relationship between (us )SSI /(us )FB and for different TFB values. Clearly, there is a strong directionality in the data presented. In this context, as the
Pearson correlation coefficients are small, the existing relationship can be considered as a
nonlinear correlation. Equally, Fig. 6 shows the relationship between (us )SSI /(us )FB and .
In this case, in addition to the strong directionality, it can be seen that the Pearson correlation coefficients are located in the range of
= 0.5 0.7. Thus, the existing dependency
between (us )SSI /(us )FB and is better to be assumed as a linear correlation rather than a
nonlinear type. These observations are in complete agreement with that has been presented
in Sec. 3.1.3.
If the relationship between (utot )SSI /(utot )FB and is examined (Fig. 7), it can be clearly
concluded that the existing correlation between (utot )SSI /(utot )FB and is nonlinear, respecting the small values of Pearson correlation coefficients observed. In addition, when the
correlation between (utot )SSI /(utot )FB and is considered (Fig. 8), only a nonlinear correlation can be distinguished when < 0.5. Above this value, the dependency vanishes very
quickly.

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

Sensitivity Analysis for Soil-Structure Interaction

1063

FIGURE 5 Correlation and dependency between structural distortion modification factors


and = (Vs )sec TFB /he (color figure available online).

FIGURE 6 Correlation and dependency between structural distortion modification factors


and = (Vs )hsece TFB (he /r)0.25 (color figure available online).
3.3. Linear and Nonlinear Correlations in Summary
In summary, on the basis of the results presented for both linear and nonlinear correlations, it can be concluded that structural response modification factors due to SSI
effects do not show a systematic dependency on the model parameters except for
(Vs )0 , (Vs )sec /(Vs )0 , = (Vs )sec TFB /he and = (Vs )hsece TFB (he /r)0.25 .

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

1064

M. Moghaddasi et al.

FIGURE 7 Correlation and dependency between total displacement modification factors


and = (Vs )sec TFB /he (color figure available online).

FIGURE 8 Correlation and dependency between total displacement modification factors


and = (Vs )hsece TFB (he /r)0.25 (color figure available online).

4. Variation of SSI Effects with Model Parameters


To quantify the variation of structural response modification factors due to change in the
model parameters of consequence, a robust statistical presentation was adopted. In this
regard, all the examined scenarios are considered together, regardless of the initial grouping based on TFB . This approach is acceptable since the observed correlation between

Sensitivity Analysis for Soil-Structure Interaction

1065

4.1. Dependency on Initial Shear Wave Velocity


The dependency of structural response modification factors on (Vs )0 is shown in Fig. 9.
Clearly, if smaller values of (Vs )0 or softer soil conditions are considered, larger variation
in the response modification factors are expected. This variation in terms of dr and utot
could result in either a large reduction or amplification in the response. However, for us and
as , this variation causes the possibility of a large reduction or only very small amplification.
This observation can be explained noting that a large foundation response is expected
at softer soil conditions. As a result, a large rigid body motion accompanied with a large
damping can be imposed to the SFS system that, in turn, is responsible for a significant
amplification in dr and utot , and a significant reduction in us and as . In addition, the scenarios with a decreased dr and utot might be correspondent to the cases where foundation
imposed structural displacement is not as significant as structural distortion, which is also
decreased due to SSI effects.

1.5

SSI

TK
1.5
FB

5th Prct.
50th Prct.
75th Prct.
95th Prct.

dr /dr

s SSI

(u ) /(u )

s FB

TK

0.5

1
0.5

0
100

(a)

200

300

400

100

(b)

(Vs)0 (m/s)

400

300

400

TK
s FB

1.5

(a ) /(a )

s SSI

tot FB
tot SSI

300

0.5
0

1.5
1
0.5
0

100

(c)

200
(Vs)0 (m/s)

TK

(u ) /(u )

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

the response modification factors and (Vs )0 , (Vs )sec /(Vs )0 , and are not changing, but
are almost the same for all considered fundamental structural periods. Thus, the variation of structural response modification factors due to change in the model parameters of
consequence is independent of the TFB values.
To carry out this quantification, the existing dependency of the response modification factors to the parameters considered were presented for the 5th , 50th , 75th , and 95th
percentiles representing different levels of likelihood. The distance between the 5th and
95th percentile boundary lines shows the possible variation in the response. The larger the
distance, the higher the variation. The line assigned to the 50th percentile values shows
the central trend of the response, and the boundary lines assigned to the 75th , and 95th
percentiles are seen as the response trend at the higher levels of probability.

200

(Vs)0 (m/s)

300

400

100

(d)

200

(Vs)0 (m/s)

FIGURE 9 The effects of soil shear wave velocity on structural response modification
factors.

M. Moghaddasi et al.

As (Vs )0 increases, the variation in the response modification factors reduces significantly, such that the response modification factors approaches to 1.0. This trend is because
an increase in (Vs )0 corresponds to the stiffer foundation condition. Consequently, SFS systems are forced to a more similar behavior to that of the corresponding fixed-base systems.
In Fig. 9, it is also shown that, at the median values, only (us )SSI /(us )FB is strongly
affected by the variation of (Vs )0 . In this context, when smaller values of (Vs )0 are
considered, smaller values of (us )SSI /(us )FB or higher reduction in us are expected.
However, if higher levels of probability are considered, different interpretations appear.
For (us )SSI /(us )FB and (as )SSI /(as )FB , the 75th and 95th percentiles do not show a significant
dependency on the change of (Vs )0 . It thus indicates that the maximum expected modification in us and as is independent from (Vs )0 . In other words, the maximum modification
in us and as may occur for any values of (Vs )0 . However, different probabilities have to be
considered.
When modification in dr and utot at the 75th and 95th percentiles is considered, an
increase in (Vs )0 tends to sharply decrease the response modification factors. This observation indicates that foundation imposed structural response is significantly reduced due to
the increase of (Vs )0 and, thus, the likelihood of having a large amplification in dr and utot
decreases sharply.
4.2. Dependency on Shear Wave Velocity Degradation Ratio
The effects of (Vs )sec /(Vs )0 on the structural response modification factors are shown in
Fig. 10. The trends and conclusions are very similar to those indicated for (Vs )0 , noting
smaller degradation ratio corresponds to the scenarios with smaller shear wave velocity
or softer soil conditions. Clearly, wider variation in the structural response modification
factors are expected when higher degradation in shear wave velocity occurs, i.e., when
smaller values of (Vs )sec /(Vs )0 are considered. This trend is justified as higher levels of
2

2
TK

5th Prct.
50th Prct.
75th Prct.
95th Prct.

1.5
1

1.5

drSSI/drFB

(us)SSI/(us)FB

TK

0.5

1
0.5

0
0.1

0.2

0.3

(a)

0.4
0.5
(Vs)sec /(Vs)0

0.6

0.1

0.7

0.3

0.4
0.5
(Vs)sec /(Vs)0

0.6

0.7

0.2

0.3

0.4
0.5
(Vs)sec /(Vs)0

0.6

0.7

2
TK

TK

(as)SSI/(as)FB

1.5
1
0.5

1.5
1
0.5

0
0.1

(c)

0.2

(b)

(utot)SSI/(utot)FB

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

1066

0.2

0.3

0.4
0.5
(Vs)sec /(Vs)0

0.6

0.1

0.7

(d)

FIGURE 10 The effects of shear wave velocity degradation ratio on structural response
modification factors.

Sensitivity Analysis for Soil-Structure Interaction

1067

4.3. Dependency on Parameter


The dependency of structural response modification factors on is discussed next. Note
that, as defined in Sec.3, combines the effects of three parameters including: (1) degraded
soil shear wave velocity (Vs )sec ; (2) fundamental structural period TFB ; and (3) structural
effective height he . Since (Vs )sec is directly related to (Vs )0 and (Vs )sec /(Vs )0 , is then
highly influenced by these two parameters. Therefore, similar trends to those have been
observed for (Vs )0 and (Vs )sec /(Vs )0 might be expected.
The results are presented in Fig.11. Clearly, the variation in the response reduces very
sharply with the increase of , such that the variation can be practically ignored when
> 20. In addition, since the response modification factors approach to 1.0, it can be
concluded that SSI does not have any reduction or amplification effects on the structural
response after > 20.
This observation is in complete agreement with that has been previously presented
for (Vs )0 and (Vs )sec /(Vs )0 . Specifically, because smaller can be correspondent to the
scenarios with the softer soil condition or larger shear wave velocity degradation.

2
TK

5th Prct.
50th Prct.
75th Prct.
95th Prct.

1.5
1

1.5
drSSI/drFB

(us)SSI/(us)FB

TK

0.5

1
0.5

0
0

10

15

20

25

30

35

(a)

10

15

20

25

30

35

20

25

30

35

2
TK

TK
(as)SSI/(as)FB

1.5
1
0.5
0

1.5
1
0.5
0

(c)

(b)

2
(utot)SSI/(utot)FB

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

degradation will result in the more flexible foundations, as well as more added damping to
the system. Consequently, the response of the SFS system gets more dissimilar to that of
the corresponding fixed-base condition.
In this context, similar to that has been presented for (Vs )0 , a large reduction or amplification in dr and utot may occur, while only a large reduction or a very small amplification
is expected for us and as . Obviously, the reduction is due to the large amount of damping added to the system, and the amplification is due to the large foundation motion
imposed. In addition, at the median values, only (us )SSI /(us )FB is sensitive to the variation of
(Vs )sec /(Vs )0 , while at the 75th and 95th percentiles, only drSSI /drFB and (utot )SSI /(utot )FB are
significantly influenced by (Vs )sec /(Vs )0 . It has also been noted that the maximum expected
modification in us and as is independent from (Vs )sec /(Vs )0 .

10

15

20

25

30

35

(d)

10

15

FIGURE 11 The effects of = (Vs )sec TFB /he . on structural response modification factors.

1068

M. Moghaddasi et al.

4.4. Dependency on Parameter


Finally, the dependency of structural response modification factors on is presented in this
section. As defined in Sec. 3, is a parameter combining the effects of: (1) degraded soil
shear wave velocity (Vs )sec ; (2) fundamental structural period TFB ; (3) structural effective
height he ; and (4) structural aspect ratio he /r. Similar to , is also highly subjective to
the effects of (Vs )0 and (Vs )sec /(Vs )0 , but in the reverse order.
The results for the dependency of structural response modification factors on are
illustrated in Fig. 12. Obviously, the degree of variation is less for smaller that, in turn,
corresponds to the scenarios with greater (Vs )sec (stiffer soil conditions), greater TFB and he
(more flexible structures), and greater he /r (greater structural aspect ratio). In addition, the
response modification factors approaches to 1.0 when smaller is considered.
When increases, the response modification factors for (us )SSI /(us )FB and
(as )SSI /(as )FB decrease even for large percentile values. In this regard, at the 95th percentiles, the likelihood of having amplification in us and as is negligible after >
0.5.
2

2
TK

5th Prct.
50th Prct.
75th Prct.
95th Prct.

1.5
1

1.5

drSSI/drFB

(us)SSI/(us)FB

TK

0.5

1
0.5

0
0

0.5

1.5

(a)

1.5

1.5

2
TK

TK

(as)SSI/(as)FB

1.5
1
0.5
0

1.5
1
0.5
0

(c)

0.5

(b)

(utot)SSI/(utot)FB

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

Smaller may also be related to the scenarios with greater structural stiffness or
greater structural height, i.e. smaller TFB or greater he . In both cases, foundation motion
is dominant and, thus, larger variation in SSI effects can be expected. After all, it should be
noted that it is the combined effect of soil stiffness, structural period and structural height
that could result in the trends observed due to the variation of .
For scenarios with < 20, if the median values are considered, only (us )SSI /(us )FB and
(as )SSI /(as )FB are affected by the variation of . As a result of this dependency, smaller values of cause higher reduction in the response modification factors. However, if higher
levels of probability are considered, (us )SSI /(us )FB and (as )SSI /(as )FB increase when
increases and approach to 1.0, while drSSI /drFB and (utot )SSI /(utot )FB reduce sharply to 1.0.

0.5

FIGURE 12 The effects of =


factors.

1.5

(d)
he
(h /r)0.25
(Vs )sec TFB e

0.5

on structural response modification

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

Sensitivity Analysis for Soil-Structure Interaction

1069

In contrast, if the dependency of drSSI /drFB and (utot )SSI /(utot )FB on is considered,
the variation in the response modification factors increases with . This increase results
in a wide degree of variation, with the amount of amplification being greater than that of
reduction. Observing a large amplification in dr and utot , but only a large reduction in us
and as can be due to the significant foundation response occurring as increases. Large
foundation response, consequently, reduces the transmitted displacement (or force) to the
structure, while it increases the structural responses including rigid body motion due to
foundation flexibility.
It should be also noted that scenarios with > 0.5 0.7 are the unusual cases, i.e.
tall and narrow stiff structures located on very soft soils, that are considered as outliers
in the response modification spectrum shown in Moghaddasi et al. [2012]. This fact is
also distinguished in Figs. 7 and 8 as the number of data points presented is significantly
reduced when > 0.5 0.7. It basically means only a small number of models generated
satisfy this condition. Therefore, the results presented for > 0.5 0.7 can be practically
ignored.

5. Risk of Detrimental SSI Effects Based on Model Parameters


To evaluate the risk of detrimental SSI effects, as compared to fixed-base assumptions,
on structural response due to variation in the model parameters of consequence, two main
aspects are analyzed: (1) the probability of having amplification in the response of the SFS
system as compared to the response of the corresponding FB structure, and (2) the level
of increase in the response due to SSI consideration. For this purpose, two amplification
levels (A.L.) were taken into account: 1.1 and 1.2, and the probability of having scenarios
with the response modification factors greater than each level were calculated. For the
considered amplification levels, the corresponding values of median percentage increase in
the response (Med[P.I.]) were also evaluated.
5.1. Amplification Risk due to Variation of Initial Shear Wave Velocity
The risk of detrimental SSI effects on structural response due to variation in (Vs )0 is
presented in Fig. 13. On the left side, the probabilities of amplification in the structural
response are presented, whereas on the right side, the corresponding values of median
percentage increase are shown.
As expected from the results presented in Sec. 4.1, the probability of amplification in
the response is very small for us and is almost negligible for as , whereas the probability of
amplification in dr and utot cannot be simply neglected. It is also noted that the probability
of amplification in the response is reduced as the higher level of amplification is considered.
The probability of amplification in the response decreases when (Vs )0 increases.
It implicitly indicates that when structures on stiffer soil conditions are considered, the
probability of having detrimental SSI effects will be less critical. If (Vs )0 changes from
100350 m/s, the observed probability values at A.L.=1.1 vary from 10%5% for us ;
40%15% for dr; and 50%20% for utot , and are almost 0% for as . Equally, these probabilities at A.L.=1.2 are in the range of: 5%1% for us ; 30%5% for dr; and 35%5%
for utot .
If the values of median percentage increase in the response are taken into account,
a reduction is also expected when = (Vs )sec TFB /he increases. Therefore, the degree of
amplification in the structural response decreases similar to the probability of amplification
when stiffer soil conditions are considered. The corresponding values of median percentage increase are in the range of: 20%15% for us ; 30%15% for dr; 30%15% for utot ;

M. Moghaddasi et al.

100
80

100
TK

A.L.=1.1
A.L.=1.2

60
40
20
0
50

100

150

200
250
(Vs)0 (m/s)

300

350

Med[P.I.] (%)

Pr[(dr)

20
0
50

100

150

350

40
20
100

150

200
250
(Vs)0 (m/s)

TK

300

350

400

dr /dr >A.L.
SSI

FB

60
40
20
0
50

400

100

150

(d)

200
250
(Vs)0 (m/s)

300

350

400

100
TK

80

60
40
20
0
50

100

150

200
250
(Vs)0 (m/s)

300

350

TK

(u ) /(u ) >A.L.
tot SSI

40
20
100

150

(f)

100

tot FB

60

0
50

400

200
250
(Vs)0 (m/s)

300

350

400

100
TK

80

Med[P.I.] (%)

s SSI

s FB

Pr[(a ) /(a ) >A.L.] (%)

300

100

(e)

(g)

200
250
(Vs)0 (m/s)

Med[P.I.] (%)

tot SSI

tot FB

Pr[(u ) /(u ) >A.L.] (%)

(c)

60
40
20
0
50

s FB

60

80

40

80

s SSI

100
TK

60

80

(u ) /(u ) >A.L.

(b)

100
80

TK

0
50

400

SSI

FB

/(dr) >A.L.] (%)

(a)

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

80

Med[P.I.] (%)

s SSI

s FB

Pr[(u ) /(u ) >A.L.] (%)

1070

100

150

200
250
(Vs)0 (m/s)

300

350

(h)

(a ) /(a ) >A.L.
s SSI

s FB

60
40
20
0
50

400

TK

100

150

200
250
(Vs)0 (m/s)

300

350

400

FIGURE 13 Risk of detrimental SSI effects based on variation in initial soil shear wave
velocity: (left) probability of amplification in the response; (right) level of increase in the
response.

and 15%10% for as when A.L.=1.1 is considered. Equally, for A.L.=1.2, the values of
median percentage increase are in the range of: 30%25% for us ; 40%30% for dr; 45%
30% for utot ; and 30%20% for as . It should be noted that the values presented for as are
mostly related to the outliers (uncommon soil-structure scenarios) as the probability of
amplification in as is negligible. This point is also valid for the results followed in case of
(Vs )sec /(Vs )0 , and .

Sensitivity Analysis for Soil-Structure Interaction

1071

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

5.2. Amplification Risk due to Variation of Initial Shear Wave Velocity


Degradation Ratio
The risk of detrimental SSI effects on structural response due to variation in (Vs )sec /(Vs )0 is
shown in Fig. 14. Regardless of the values, the trends and conclusions are similar to those
for (Vs )0 . In this comparison, smaller degradation ratios have to be considered instead of
smaller shear wave velocities.
The probabilities of amplification in the response and the corresponding values of
median percentage decrease when (Vs )sec /(Vs )0 increases or smaller soil shear wave velocity degradation is considered. In this context, when (Vs )sec /(Vs )0 changes from 0.20.7,
the probability of amplification at A.L.=1.1 vary from: 10%5% for us ; 35%20% for dr;
40%20% for utot , and are almost 0% for as . Equally, these probabilities at A.L.=1.2 are
in the range of: 2%1% for us ; 30%15% for dr; and 30%15% for utot .
The corresponding values of median percentage increase are in the range of: 20%15%
for us ; 30%15% for dr; 30%20% for utot ; 15%10% for as at A.L.=1.1 and 30%25%
for us ; 40%30% for dr; 45%30% for utot ; and 25%20% for as at A.L.=1.2.
5.3. Amplification Risk due to Variation of
The risk of detrimental SSI effects on structural response due to variation in is shown
in Fig. 15. Clearly, an increase in yields smaller probability values such that after
> 20 any amplification more that 10%, and after > 15 any amplification more that 20%
in the structural response may be practically ignored. In addition, an increase in is accompanied with a decrease in the values of median percentage increase. These observations
once more highlight that SSI effects are less critical for soil-structure scenarios including
stiffer soil condition, greater structural stiffness, and smaller structural height.
When changes from 220, the probability values at A.L.=1.1 vary from: 10%0%
for us ; 35%0% for dr; 45%0% for utot , and are almost 0% for as . Equally, these probabilities at A.L.=1.2 are in the range of: 5%0% for us ; 35%0% for dr; and 30%15% for utot .
The corresponding values of median percentage increase are in the range of: 20%10% for
us ; 35%10% for dr; 35%10% for utot ; about 10% for as at A.L.=1.1, and 35%25% for
us ; 45%25% for dr; 45%20% for utot ; and about 20% for as at A.L.=1.2.
5.4. Amplification Risk due to Variation of
Finally, Fig.16 illustrates the risk of detrimental SSI effects on the structural response due to
variation in . In this case, two different trends are seen for the probability of amplification
in us and as compared to the probability of amplification in dr and utot . The probability
of amplification in us is very small and decreases as increases, while the probability of
amplification in as is almost negligible. In addition, an amplification in us can be practically
ignored when > 1.0. In contrast, the probability of amplification in dr and utot increases
as increases. The reason behind this increase is previously described in Sec. 4.4.
When changes from 0.11.5, the probability values at A.L.=1.1 vary from: 8%0%
for us ; 20%50% for dr; and 25%60% for utot , and are almost 0% for as . Equally, these
probabilities at A.L.=1.2 are in the range of: 2%0% for us ; 5%45% for dr; and 8%50%
for utot .
In terms of the median percentage increase, an increasing trend exists for us , dr, and
utot when increases, whereas the values of median percentage increase for as are almost
constant for all values. The values of median percentage increase are in the range of:
15%25% for us ; 15%65% for dr; 15%80% for utot ; and about 15% for as at A.L.=1.1,
and 25%35% for us ; 25%90% for dr; 25%95% for utot ; and about 15% for as at
A.L.=1.2.

M. Moghaddasi et al.
100

100
TK

A.L.=1.1
A.L.=1.2

80

80

Med[P.I.] (%)

Pr[(u s)SSI/(us)FB>A.L.] (%)

1072

60
40
20
0
0.1

0.2

0.3

(a)

0.4
0.5
(Vs)sec /(Vs)0

0.6

Med[P.I.] (%)

Pr[dr SSI/drFB>A.L.] (%)

0
0.1

0.2

0.3

0.4
0.5
(Vs)sec /(Vs)0

0.6

20
0.2

0.3

0.4
0.5
(Vs)sec /(Vs)0

TK

0.6

0.7

dr /dr >A.L.
SSI

FB

60
40
20
0
0.1

0.7

0.2

0.3

(d)

0.4
0.5
(Vs)sec /(Vs)0

0.6

0.7

100

100
TK

80

Med[P.I.] (%)

Pr[(u tot)SSI/(utot)FB>A.L.] (%)

(c)

60
40
20
0
0.1

0.2

0.3

(e)

0.4
0.5
(Vs)sec /(Vs)0

0.6

TK

(u ) /(u ) >A.L.
tot SSI

tot FB

60
40
20
0
0.1

0.7

0.2

0.3

(f)

0.4
0.5
(Vs)sec /(Vs)0

0.6

0.7

100

100
TK

80

Med[P.I.] (%)

Pr[(a s)SSI/(as)FB>A.L.] (%)

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

40

80

20

60
40
20
0
0.1

s FB

100
TK

40

80

s SSI

(b)

60

80

(u ) /(u ) >A.L.

60

0
0.1

0.7

100
80

TK

0.2

(g)

0.3

0.4
0.5
(Vs)sec /(Vs)0

0.6

(h)

(a ) /(a ) >A.L.
s SSI

s FB

60
40
20
0
0.1

0.7

TK

0.2

0.3

0.4
0.5
(Vs)sec /(Vs)0

0.6

0.7

FIGURE 14 Risk of detrimental SSI effects based on variation in shear wave velocity
degradation ratio: (left) probability of amplification in the response; (right) level of increase
in the response.

6. Conclusions
A comprehensive statistical evaluation of SSI effects has been conducted using a simplified soil-shallow foundation-structure model with equivalent soil domain and nonlinear
structural behavior following the Takeda type hysteretic model. The key findings can be
summarized as follows:

100
80

A.L.=1.1
A.L.=1.2

40
20
0
10

15

30

60
40
20

35

10

15

20

25

30

35

(b)
100

TK

80

TK

80

Med[P.I.] (%)

Pr[(dr) SSI/(dr) FB>A.L.] (%)

25

100

60
40
20
0

drSSI/drFB>A.L.

60
40
20
0

10

15

20

25

30

35

(c)

10

15

100

20

25

30

35

(d)
100

TK

80

TK

80

Med[P.I.] (%)

Pr[(u tot)SSI/(utot)FB>A.L.] (%)

20

60
40
20
0

(utot)SSI/(utot)FB>A.L.

60
40
20
0

10

15

20

25

30

35

(e)

10

15

100

20

25

30

35

(f)
100

TK

80

TK

80

Med[P.I.] (%)

Pr[(as)SSI/(as)FB>A.L.] (%)

(us)SSI/(us)FB>A.L.

0
5

(a)

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

TK

80

60

60
40
20
0

(as)SSI/(as)FB>A.L.

60
40
20
0

(g)

1073

100
TK

Med[P.I.] (%)

Pr[(u s)SSI/(us)FB>A.L.] (%)

Sensitivity Analysis for Soil-Structure Interaction

10

15

20

25

30

35

(h)

10

15

20

25

30

35

FIGURE 15 Risk of detrimental SSI effects based on variation in = (Vs )sec TFB /he : (left)
probability of amplification in the response; (right) level of increase in the response.
1. From all considered soil, structural, and soil-structure system parameters only
he
(he /r)0.25 have a pro(Vs )0 , (Vs )sec /(Vs )0 , = (Vs )sec TFB /he , and = (Vs )sec
TFB
nounced correlation with structural response modification factors due to SSI
effects.
2. An increase in (Vs )0 , (Vs )sec /(Vs )0 and cause a less variation in the resulted
structural response modification factors. In addition, as these parameters increase,
the response modification factors approach 1.0, indicating the behavior of the
soil-structure systems are more similar to the behavior of the corresponding
fixed-base system.

M. Moghaddasi et al.
100

100
TK
80

40
20
0

0.5

1.5

100
TK

TK

80

drSSI/drFB>A.L.

80
Med[P.I.] (%)

Pr[(dr) SSI/(dr) FB>A.L.] (%)

20

(b)

100

60
40
20
0

60
40
20
0

0.5

1.5

(c)

0.5

100

1.5

(d)

100
TK

TK

80

(utot)SSI/(utot)FB>A.L.

80
Med[P.I.] (%)

Pr[(u tot)SSI/(utot)FB>A.L.] (%)

40

1.5

60
40
20
0

60
40
20
0

0.5

1.5

(e)

0.5

1.5

(f)

100

100
TK

TK

80

(as)SSI/(as)FB>A.L.

80
Med[P.I.] (%)

Pr[(a s)SSI/(as)FB>A.L.] (%)

60

0
1

0.5

(a)

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

(us)SSI/(us)FB>A.L.

80

60

60
40
20
0

60
40
20
0

(g)

TK

A.L.=1.1
A.L.=1.2

Med[P.I.] (%)

Pr[(u s)SSI/(us)FB>A.L.] (%)

1074

0.5

1.5

(h)

0.5

1.5

FIGURE 16 Risk of detrimental SSI effects based on variation in = (Vs )hsece TFB (he /r)0.25 :
(left) probability of amplification in the response; (right) level of increase in the response.
3. Considering the median values (the 50th percentiles), an increase in (Vs )0 ,
(Vs )sec /(Vs )0 and results in an increase in the structural distortion modification
factors, while modification in dr, utot , and as are only weakly sensitive to the
parameters considered.
4. The likely maximum modification in us and as are independent from the variation
of (Vs )0 , (Vs )sec /(Vs )0 , and . However, if dr or utot is considered, a sharp reduction in the maximum modification factors is observed due to an increase in (Vs )0 ,
(Vs )sec /(Vs )0 , and .

Downloaded by [University of Bucharest ] at 22:40 17 November 2014

Sensitivity Analysis for Soil-Structure Interaction

1075

5. When increases, structural distortion modification factors and structural acceleration modification factors reduce very sharply at the median and likely maximum
values. In contrast, the variation in structural drift modification factors and total
displacement modification factors gets more significant as increases. This variation mostly results in amplification in structural drift and total displacement than
reduction.
6. In terms of quantification of the risk, the probability of amplification in the response
is very small for us and is almost negligible for as , whereas the probability of
amplification in dr and utot cannot be simply neglected.
7. An increase in (Vs )0 , (Vs )sec /(Vs )0 , and results in a reduction in the probability of amplification in the structural response modification factors, as well
as a reduction in the values of median percentage increase. Specifically, when
> 20, detrimental SSI effects on the structural response can be practically
ignored.
8. An increase in reduces the probability of amplification in us such that the amplification can be practically ignored when > 1.0. However, the probabilities of
amplification and the corresponding values of median percentage increase for dr or
utot rise when increases.

References
Bielak, J. [1975] Dynamic behaviour of structures with embedded foundations, Earthquake
Engineering and Structural Dynamics 3(3), 259174.
Carr, A. [2009] Ruaumoko 2D, Nonlinear FEM Computer Program, University of Canterbury,
New Zealand.
Ciampoli, M. and Pinto, P. E. [1995] Effects of soil-structure interaction on inelastic seismic
response of bridge piers, Journal of Structural Engineering 121(5), 806814.
Dowdy, S. and Wearden, S. [1983] Statistics for Research, Wiley, New York.
Fishman, G. S. [1996] Monte Carlo: Concepts, Algorithms, and Applications, Springer, New York.
Moghaddasi, M., Cubrinovski, M., Chase, J. G., Pampanin, S., and Carr, A. [2012] Stochastic quantification of soil-shallow foundation-structure interaction, Journal of Earthquake Engineering
16(6), 820850.
NZS1170.5 [2004] Structural design actions, part 5: earthquake actions, Standards New Zealand,
Wellington, New Zealand.
Seed, H. B. and Idriss, I. M. [1970] Soil moduli and damping factors for dynamic response analysis,
Report No. EERC 70-109, Earthquake Engineering Research Centre, University of California,
Berkeley.
Shome, N., Cornell, C. A., Bazzurro, P., and Carballo, J. E. [1998] Earthquakes, records, and
nonlinear responses, Earthquake Spectra 14(3), 469500.
Stewart, J. P., Fenves, G. L., and Seed, R. B. [1999a] Seismic soil-structure interaction in buildings. I:
analytical methods, Journal of Geotechnical and Geoenvironmental Engineering 125(1), 2637.
Stewart, J. P., Seed, R. B., and Fenves, G. L. [1999b] Seismic soil-structure interaction in buildings. II: empirical findings, Journal of Geotechnical and Geoenvironmental Engineering 125(1),
3848.
Veletsos, A. S. and Nair, V. D. [1975] Seismic interaction of structures on hysteretic foundations,
Journal of Structural Engineering 101(1), 109129.
Wolf, J. P. [1994] Foundation Vibration Analysis Using Simple Physical Models, Prentice-Hall,
Englewood Cliffs, New Jersey.
Zhang, J. and Tang, Y. [2009] Dimensional analysis of structures with translating and rocking foundations under near-fault ground motions, Soil Dynamics and Earthquake Engineering 29(10),
13301346.

You might also like