Sei sulla pagina 1di 8

Sensitivity of Seismic Demand of a Reinforced Concrete Shear-Wall

Building*
Tae-Hyung Lee and Khalid M. Mosalam
Dept. of Civil and Environmental Engineering, University of California, Berkeley, CA, USA

* This article is to be appeared in the Prof. of Ninth International Conference on Application of Statistics and Probability in Civil
Engineering (ICASP 9), San Francisco, California, USA, July 6-9, 2003

Keywords: Nonlinear time-history analysis, Reinforced concrete, Seismic demand, Sensitivity analysis,
Shear-wall building

ABSTRACT: Several sources of uncertainty exist in evaluating the seismic demand of a building. Among
those are intensity measure of ground shaking, ground motion profile, structural strength and stiffness, mass,
and damping. This paper investigates which of these uncertain input parameters are more significant in affect-
ing the building seismic demand. In this study, we adopt a simple deterministic sensitivity analysis where we
assume the probability distribution of each input parameter and select the predetermined lower and upper
bound and best estimate. For a specific demand quantity, the difference between its estimations corresponding
to the lower and upper bounds of an input parameter, a relative importance of the input parameters is deter-
mined. This method is applied to an important reinforced concrete shear-wall laboratory building located on
the campus of the University of California, Berkeley. The result clearly demonstrates the significantly higher
importance of ground motion uncertainties compared with those related to structural properties such as
strength and stiffness. This conclusion suggests the need for a comprehensive study focusing on ground mo-
tion characteristics to reduce their uncertainty.

ables by fixing their values at the best estimate, such


1 INTRODUCTION
as the mean or the median to save computational ef-
fort.
Evaluation of building performance due to seismic
This paper presents a deterministic sensitivity
loading includes various uncertain input parameters.
analysis of a reinforced concrete (RC) shear-wall
Parameter uncertainty causes uncertainty of building
building subjected to seismic loading. It presents a
responses upon which damage and loss analyses are
simple methodology of investigating the relative
based. Accurate estimation of the structural seismic
significance of various variables for determining the
demands is essential for performance-based earth-
seismic demand (referred to as the engineering de-
quake engineering. In that regard, floor acceleration,
mand parameter EDP) of this building. The method-
floor displacement, and inter-story drift ratio are of
ology is illustrated using a university building that
importance for hospital or laboratory buildings
contains heavy equipments and valuable contents.
where heavy and sensitive equipments and contents
exist. The identified sources of input uncertainty that
contribute to uncertainty of these demand measures
2 DETERMINISTIC SENSITIVITY ANALYSIS
include ground motion profile, ground motion inten-
sity, structural strength and stiffness, mass, and
A deterministic sensitivity analysis is performed to
damping, among others.
determine the relative significance of each uncertain
Uncertainty of a variable can be reduced by fur-
variable to EDP uncertainty. The so-called tornado
ther investigation. This can be preceded by a practi-
diagram, commonly used in decision analysis, is
cal investigation to determine which variable is more
employed in this methodology (Porter et. al. 2002).
significant to the structural demand than others.
Subsequently, one conducts more sophisticated in- In this analysis, it is assumed that the output vari-
vestigations to characterize and possibly reduce the able (EDP in this study) is a known deterministic
degree of uncertainty of such significant variable. function (e.g. finite element model (FEM)) of a set
One may ignore uncertainties of relatively insignifi- of input variables whose probability distribution is
cant variables by treating them as deterministic vari- assumed by the analyst. For each input variable, the
best estimate and two extreme values corresponding frequency of Sa for the site of the case-study build-
to upper and lower bounds of its probability distribu- ing in Section 4.
tion are selected. First, the deterministic function is
evaluated to determine the best estimate of the out- 3.2 Uncertainty in Ground Motion Profile
put variable using input variables set to their best es- Different profiles of ground motion with the same
timates. Subsequently, for each input variable, the earthquake IM may produce different outputs of a
function is evaluated twice, using one of the extreme specific EDP of interest. Porter et al. (2002) dis-
values each time while the other input variables are cussed two ways of considering ground motion
set to their best estimates. This process yields two characteristics other than the primary IM, i.e. Sa .
bounding values of the output variable for each input One of these methods is selected in the present study
variable. The absolute difference of these two values, for its simplicity. In this method, one selects a set of
referred to as the swing, is used as an indicator of the ground motion profiles where each of them is scaled
“significance” of the given input variable to the out- according to the target IM selected beforehand. For a
specific EDP of interest, one performs a set of struc-
put variable. One can rank input variables according
tural analyses using a scaled ground motion profile
to their swings. A larger swing implies a more sig-
each time and obtains the set of EDP values. Then
nificant input variable to the uncertainty of the out- the set of ground motion profile can be sorted with
put variable. respect to the magnitude of EDP. Subsequently, one
selects the lower bound ground motion profile such
that EDP from this specific ground motion corre-
3 SELECTED UNCERTAIN VARIABLES sponds to the predetermined lower fractile, e.g. 10th
percentile. The best estimate and upper bound can be
The selection of input variables depends on the out- selected in a similar manner but corresponding to the
put variable of interest. In this study, we select the median and the predetermined upper fractile, respec-
peak absolute (or total) roof acceleration (ARA), tively.
peak absolute roof displacement (ARD), and maxi-
For the best estimate of Sa as the target IM, detailed
mum inter-story drift ratio (IDR) as output variables
among various EDP’s. By absolute, we mean the steps of this process are;
floor response relative to the fixed base plus the 1. Compute the elastic spectral acceleration
ground motion. Accordingly, ground motion profile, Sai at the fundamental period of the struc-
ground motion intensity, structural strength and ture with 5% viscous damping ratio for ith
stiffness, building mass, and viscous damping are se- ground motion profile.
lected as input variables. Other parameters that may 2. Determine the scale factor ( α i ) by
contribute to the EDP uncertainty include soil flexi- α i = target Sa / Sai .
bility, the spatial distribution of the mass, and the 3. Perform a nonlinear time history analysis of
modeling assumptions in the FEM. the structure using the ith scaled ground mo-
tion ( α i times the original ground motion
3.1 Uncertainty in Ground Motion Intensity profile) to obtain EDP corresponding to the
There are various ways of characterizing intensity ith ground motion, EDPi,
measures (IM) of an earthquake. The typical IM’s in 4. Sort ground motion profiles by their EDP’s.
earthquake engineering are measured peak ground 5. Find the median EDP from the set of EDP’s.
motions (acceleration, velocity, and displacement) The corresponding ground motion profile is
and elastic spectral responses (acceleration, velocity, the best estimate.
6. The lower and upper bound ground motion
and displacement). There have been efforts to define
profiles are those corresponding to the
a proper IM of earthquake in relation to structural lower and upper fractile of EDP’s.
performance or damage, where IM is strongly corre- As Porter et al. (2002) discussed, this approach
lated to EDP such that IM becomes an indicator of doesn’t provide any insight why ground motions
structural performance or damage. with equal IM produce different estimates of EDP.
For the purpose of the present study, IM candi-
date must satisfy the following criteria: (1) it is a 3.3 Uncertainty in Strength and Stiffness
proper IM to EDP of interest and (2) its probability Sources of uncertainty in structural strength and
function of occurrence is readily available. Satisfy- stiffness are idealization of material constitutive
ing these criteria, elastic spectral acceleration Sa model, difference of member dimensions between
with 5% viscous damping is selected as an uncertain the structural drawing and as-built members, and
input variable representing earthquake IM. Frankel imperfect knowledge of material properties among
and Leyendecker (2001) provide probabilistic hazard others. Uncertainty in the structural strength and
information in terms of mean annual exceedance stiffness has been of major interest by a number of
probabilistic studies of RC members and systems. A
short review can be found in Lee and Mosalam studies for the performance-based earthquake design
(2003). In most of these studies, researchers consid- methodology of the Pacific Earthquake Engineering
ered uncertainty of material behavior at the stress- Research (PEER) center.
strain level by treating, e.g. concrete compressive 4.1 Building Description
strength f c' , steel yield stress f y , and steel Young’s
The case-study building is a modern RC shear-wall
modulus Es , as random variables. building completed in 1988 to provide high-
Two scenarios are available in examining the ef- technology research laboratories. The building is
fect of uncertainty of strength and stiffness to EDP: 203,800 square feet overall, with 122,000 net use-
(1) treat strength and stiffness as correlated random able square feet of research laboratories, offices and
related support spaces. The building is six stories
variables, and (2) treat them as independent random
plus a basement and is rectangular in plan with over-
variables. The first approach is based on the notion
all dimensions of approximately 306 feet in the lon-
that f c' and the concrete initial Young’s modulus E c gitudinal direction (oriented N16.5W) and 105 feet
are strongly correlated. Mirza et al. (1979) suggested in the transverse direction. The basement is con-
the correlation coefficient of 0.8-0.9 between f c' and tained within the periphery of the building. The
building was designed to meet 1982 Uniform Build-
Ec . On the other hand, f y and Es are known to ing Code (ICBO 1982) and is classified as C2 Build-
have weak correlation where they can be treated as ing Type 9, i.e. RC shear-wall in both directions.
independent random variables without losing gener- The gravity load-carrying system of the building
ality. The second approach aims to investigate the consists of a RC space frame, as shown in Figure 1
individual sensitivity of each of the strength and which includes the global geometry and span dimen-
stiffness variables, regardless of the possible correla- sions. The lateral force-resisting system consists of
tion between them. In the present sensitivity study of coupled shear-walls in the transverse direction (ap-
RC shear-wall building case-study, the authors adopt proximately normal to the Hayward fault) and perfo-
the second approach by treating f c' , Ec , f y , and Es rated shear walls in the longitudinal direction. The
floors consist of waffle slab systems with solid parts
as independent random variables.
acting as integral beams between the columns. The
3.4 Uncertainty in Building Mass waffle slab is composed of a 4.5 inch thick RC slab
Quantification of building mass in structural analysis supported on 20 inch deep joists in each direction.
depends on several factors: materials used in con- The building foundation consists of a 38 inch thick
struction, building dimensions, locations of non- mat foundation. According to the design specifica-
structural elements, and structural model, e.g. nodal tions, the concrete of the shear-walls and the cou-
coordinates. In general, uncertainty in building mass pling beams has nominal 28-day compressive
is an integration of uncertainties in those factors. strength f 'c = 5 ksi . On the other hand, the concrete
Thus, one may subdivide the variable “mass” into
several variables listed above and consider them as for the interior columns and the waffle slab systems
separate uncertain variables. However, since one of has nominal strength f c' = 3 ksi . The reinforcing
the objectives of the approach presented in this study steel is scheduled as ASTM A-615 Grade 40 for #4
is its simplicity, we considered mass uncertainty as a and smaller bars and Grade 60 for #5 and larger
whole. In this context, Ellingwood et al. (1980) sug- bars. The middle frame (labeled 8 in Figure 1) is
gested an adequate model for the probability distri- analyzed in the present study.
bution of dead load as discussed in Section 4. Figure 2 presents the structural elevation view of
3.5 Uncertainty in Viscous Damping frame 8 (as identified in Figure 1) of the building,
A comprehensive discussion of uncertainty in vis- and indicates the story heights and the different la-
cous damping is presented by Porter el at. (2002) in- beling of the building levels. All interior columns of
cluding a summary of several other researchers. the building are square with dimensions 24 inch with
They suggested a reasonable estimate of COV of the transverse reinforcement in the form of #4@8 inch
damping ratio in the range of 0.3 to 0.4. closed ties. The longitudinal reinforcing bars of the
interior columns vary with the levels of the building;
12#11 are used between the foundation level and
4 RC SHEAR-WALL BUILDING CASE-STUDY level 1, 12#10 are used between levels 1 and 3, and
8#8 are used between levels 3 and the roof. The
The presented methodology of deterministic sensi- cross sections of the shear walls are schematically
tivity analysis is applied to a laboratory building lo- shown in Figure 2. All coupling beams between the
cated on the campus of the University of California, shear-walls are 48 inches wide and 24.5 inches deep.
Berkeley. This building represents one of two case-
203'-0"
N
16 o 33' 20'-9" 3@20'=60' 20'-9" 20'-9" 3@20'=60' 20'-9"

19'-9" 22'-10" 19'-4" 22'-10" 19'-9" B

D
104'-6"

3 4 5 6 7 8 9 10 11 12 13

101'-6"

Figure 1: Structural plan view of the building


A B D E G H

Roof
20'-5"

Level 6
13'-6"

Level 5
Figure 3: OpenSees model of frame 8 of the building
13'-6"

Level 4
4.2.2 Modeling Shear-Wall Elements
All shear-wall elements are modeled using one di-
110'-11"

13'-6"

mensional beam elements aligned with the centerline


Level 3
of the shear-wall. For proper idealization of the ge-
ometry, the node at the shear-wall centerline and the
13'-6"

Level 2 node at the boundary of the shear-wall (representing


one end of a coupling beam) are connected by rigid
13'-6"

Level 1 elements, as shown in Figure 4.


18'-6"

19'-9" 14'-10" 8' 19'-4" 8' 14'-10" 19'-9"

104'-6"

Figure 2: Elevation view and sections of frame 8


4.2 Structural Model and Related Uncertainties
The computational model of the building is devel-
oped using the modeling capabilities of the software
framework of OpenSees (McKenna and Fenves
2001). The complete OpenSees model of the build-
ing is illustrated in Figure 3. Figure 4: Modeling of shear-walls
4.2.1 Assumptions and Limitations 4.2.3 Modeling Beam Elements
The current model of OpenSees focuses on modeling All elements in the building model are based on
frame 8 of the building. This model is based on the flexibility formulation of beam elements. Each beam
following assumptions: element has two nodes with two translations and one
1. Two-dimensional idealization is considered. rotation per node. The beam element has four moni-
2. Linear elastic shear force-deformation rela- toring sections with fiber element discretization (Lee
tionship is chosen for all elements relying on and Mosalam 2003). In this discretization, distinc-
the fact that shear failure will not occur prior tion is made between the constitutive model of the
to flexure failure. reinforcing bars, unconfined concrete, and confined
3. Reinforcing bars are assumed fully bonded to concrete. For the reinforcement, a bilinear stress-
the surrounding concrete. strain relationship is used. As for the concrete, the
4. The analyzed frame has a tributary area with
width of 101.5 feet as shown in Figure 1.
modified Kent-Park stress-strain (with zero tensile We adopt a model suggested by Ellingwood et al.
strength) relationships are used (Scott et al. 1982). (1980) for dead load. In this model, a normal distri-
Uncertainties in the structural strength and stiff- bution with the mean as the calculated nominal dead
ness are considered on the level of the constitutive load and COV of 0.1 is assumed. Moreover, all
models. Since the compressive strength of concrete nodal masses are assumed perfectly correlated. In
f c' and the yield stress of the reinforcement f y dic- other words, all nodal masses are increased or de-
creased together with the appropriate ratio.
tate the structural strength, we considered them as
uncertain variables at the same time. It is assumed 4.2.6 Damping Idealization
that f c' has a normal distribution with the mean as The damping characteristic of the building is mod-
the nominal strength and COV of 0.175 according to eled using mass and stiffness proportional damping
with 5% of critical damping for the first two modes
Mirza et al. (1979). On the other hand, f y is as-
of vibration. These modes are estimated from the ei-
sumed to have a lognormal distribution with the gen solution using the initial elastic stiffness matrix.
mean as the nominal yield stress and COV of 0.1 ac- The periods of these two modes are estimated as
cording to Mirza and MacGregor (1979). 0.51 and 0.27 seconds. We assumed a normal distri-
The initial tangent stiffness of the concrete con- bution with the mean as 5% of critical damping and
stitutive model Ec and Young’s modulus of the rein- COV of 0.4 for damping (Porter et al. 2002).
forcement Es control the structural stiffness. They 4.2.7 Solution Strategy
are considered as uncertain variables to represent Newmark β -method is used as the time integrator
uncertainty of the structural stiffness. It is assumed with coefficients γ = 0.50 and β = 0.25 . In general,
that Ec has a normal distribution with the mean as a time step of 1/5 the ground motion time interval is
the nominal initial tangent stiffness and COV of 0.08 used. The modified Newton-Raphson solution algo-
rithm is utilized for solving the nonlinear system of
(Mirza et al. 1979), while Es has a normal distribu- equilibrium equations.
tion with the mean as the nominal Young’s modulus
and COV of 0.033 (Mirza and MacGregor 1979).
4.3 Soil Conditions and Seismic Hazard at the
4.2.4 Boundary Conditions Building Site
Flexible supports are used at the foundation level in The soil conditions at the site of the building are
the vertical direction using truss elements. These summarized by Geomatrix Consultants (2000). The
elements represent soil with a modulus of subgrade building is located at a site consisting of stiff soil of
reaction of 100 lb/in3 obtained by the procedure pre- thickness in the range of 19.7 to 52.5 feet (6 to 16
sented in FEMA273 (1997). The tributary width of meters), with an estimated average of about 39.4 feet
these truss elements included the width of the shear (12 meters) above Franciscan bedrock assumed to be
wall and four times the thickness of the mat founda- not pervasively sheared and to have a shear wave ve-
tion. To simulate the soil behavior, the truss ele-
locity of about 2953 ft/sec (900 m/sec). Older allu-
ments are associated with material having elastic be-
havior in compression and no tensile strength. In this vium overlies the Franciscan rocks at the site. The
study, we do not account for uncertainty in the soil alluvium typically comprises very stiff sandy clay,
flexibility. This type of uncertainty and its effect on with average standard penetration resistance values
the estimated EDP’s will be addressed in the future of 50 or greater and estimated shear wave velocity of
extension of this study. about 1214 ft/sec (370 m/sec). The site is thus classi-
fied as NEHRP category Sc (FEMA273 1997).
4.2.5 Mass Idealization and Gravity Load Frankel and Leyendecker (2001) provide the
The dead load accounted for the self-weight of the mean annual exceedance frequency of Sa for the
waffle slab system and the supporting elements, i.e.
shear-walls and columns. The assumed unit weight latitude and longitude of the site of the building.
of the concrete is 145 pcf. Accordingly, the com- Figure 5 shows the mean site hazard curve for 0.5
puted dead load is 183 psf. Moreover, 25 psf repre- sec period with assumption of B-C soil boundary as
senting building contents are included as a superim- defined by the International Code Council (2000).
posed dead load. The live load of 100 pcf is assumed Since the fundamental period of the demonstration
according to the original design of the building. The building is 0.51 sec and the site class is C, we adopt
mass of the building is modeled using lumped the hazard information shown in Figure 5 without
masses at the nodes. Nodal masses are directly com- modification.
puted from the dead load including the superim-
posed dead load.
Table 1: Ranks of ground motion records according
to different EDP’s
Rank ARA ARD IDR Percentile
1 KB_kobj LP_srtg LP_lex1
2 TO_ttrh02 LP_lgpc KB_kobj 90th
3 LP_lex1 KB_kobj LP_cor
4 LV_mgnp LP_lex1 TO_ttrh02
5 LP_lgpc TO_ttrh02 EZ_erzi
6 MH_clyd EZ_erzi MH_hall
7 LV_fgnr LP_cor LP_srtg
8 LP_cor LP_gilb MH_clyd
9 PF_cs08 MH_clyd LP_lgpc
10 EZ_erzi CL_gil6 LV_mgnp
50th
Figure 5: Probabilistic hazard curve for the site of 11 MH_hall MH_hall CL_gil6
the building with 0.5 sec period and soil type C 12 CL_gil6 LP_gav LV_fgnr
13 PF_temb PF_cs05 PF_cs08
We select three Sa values corresponding to non- 14 TO_ttr007 PF_cs08 LP_gav
15 LP_srtg TO_ttr007 PF_temb
exceedance probabilities of 10%, 50%, and 90% in
16 LP_gav MH_andd CL_clyd
50 years as the lower bound, median, and upper
17 CL_clyd PF_temb LP_gilb
bound, respectively. Since the relationship between
18 LP_gilb LV_mgnp MH_andd
mean annual exceedance frequency and spectral ac-
celeration is not given in a functional form, but in 19 MH_andd CL_clyd PF_cs05 10th
20 PF_cs05 LV_fgnr TO_ttr007
discrete points, Sa values for a given exceedance
frequency is obtained by linear interpolation of two Among the examined 20 records, EZ_erzi,
adjacent data points. From Figure 5 and following MH_hall, CL_gil6, and LV_mgnp produce the me-
the assumption of Poisson arrivals of earthquakes dian EDP’s as shown in Table 1. In general, we must
(Porter et al. 2002), the selected Sa values are 0.15g, select the median ground motion record correspond-
0.40g, and 1.21g for the lower bound, median, and ing to each EDP of interest, separately. Instead, we
upper bound of ground motion IM, respectively. select CL_gil6 as the median ground motion record
Pacific Earthquake Engineering Research Center for all EDP’s for simplicity even though it is not ex-
(PEER) Testbeds Program (http://peertestbeds.net) actly the median record for ARA. However, this re-
provides a set of 20 recorded ground accelerations to cord is ranked next to the median record for ARA
be used for the site of the building. For each ground (MH_hall) and the difference of ARA values pro-
motion, we compute the elastic spectral acceleration duced by the two records is only 1.9%.
Sa with 0.51 sec period and 5% damping ratio to de- 4.4 Results
termine the scale factor corresponding to the median
Table 2 summarizes values of input variables
IM, Sa = 0.40g . Subsequently, we perform a set of used in the sensitivity analyses where the lower and
20 nonlinear time history analyses using these scaled upper bound, and best estimate represent the 10th
ground motion profiles with all uncertain input vari- and 90th percentiles and median of the variable,
ables in structural properties set to its best estimate. respectively. It is noted that different sets of the
Each simulation produces three different EDP’s: lower bound, best estimate, and upper bound inputs
ARA, ARD, and IDR. Upon the completion of the of ground motion profile are selected according to
20 simulations, we obtain 20 ARA, ARD, and IDR different EDP’s. EDP values corresponding to the
sets. Finally, the set of the 20 scaled ground motions set of best estimate input variables are ARA= 0.46g,
is sorted three times in the descending order, each ARD=4.16 inches, and IDR=0.2%.
time with respect to different EDP. Ranks of ground Figure 6, Figure 7, and Figure 8 show the results
motions according to EDP’s are listed in Table 1. of the sensitivity studies for ARA, ARD, and IDR,
Names of the ground motions are as defined in the respectively. These figures present the results in the
ground motion database that can be found in the
www site http://peertestbeds.net.
Table 2: Estimates of input variables in simulations
Lower Best
Variable
bound estimate
the correlation with the selection of ground motion fornia, Berkeley. Opinions and findings presented
and IM. are those of the authors and do not necessarily re-
flect the views of the sponsor or PEER.
5 CONCLUDING REMARKS
REFERENCE
This paper presents a deterministic sensitivity study
for a RC shear-wall laboratory building located on Ellingwood, B., Galambos, T.V., MacGregor, J.G., and Cornell,
the campus of the University of California, Berkeley. C.A. 1980. Development of a probability-based load crite-
rion for American National Standard A58. National Bureau
This study investigates the sensitivity of seismic of Standards, Washington, D.C.: pp222
demand (referred to as the engineering demand pa- Federal Emergency Management Agency (FEMA). 1997.
rameter, EDP) in terms of the peak absolute roof ac- NEHRP Guidelines for the seismic rehabilitation of build-
celeration (ARA), peak absolute roof displacement ings, Developed by the Building Seismic Safety Council for
FEMA (Report No. FEMA 273), Washington, D.C.
(ARD), and maximum inter-story drift ratio (IDR) to Frankel, A. and Leyendecker, E.V. 2001. Uniform hazard re-
a set of uncertain input variables. The uncertain sponse spectra and seismic hazard curves for the United
variables considered in this study are intensity States, US Geological Survey, Menlo Park, CA.
measure (IM) of earthquake, ground motion profile, Geomatrix Consultants. 2000, Geologic hazards investigation,
Central campus, University of California, Berkeley, Cali-
structural strength and stiffness, and building mass fornia, Appendix One, report prepared for The Economic
and damping. We adopt Sa (elastic spectral accel- Benefits of a Disaster Resistant University.
Hayward Fault Paleoearthquake Group (HPEG). 1999. Timing
eration with 0.5 sec period and 5% viscous damping) of Paleoearthquakes on the Northern Hayward Fault–
as ground motion IM. For nonlinear time history Preliminary Evidence in El Cerrito, California, Open-File
analyses, the building is modeled using detailed fi- Report 99-318, Online version 1.0, U.S. Geological Survey,
ber element methodology. Menlo Park, CA.
International Code Council. (2000). International Building
It is observed that EDP’s in this study are more Code 2000, International Conference of Building Officials,
sensitive to uncertainty in ground motion character- Whittier, CA, pp. 756.
istics ( Sa and ground motion profile) than that in International Conference of Building Officials (ICBO) 1982.
Uniform Building Code, Whittier, CA.
structural properties (strength, stiffness, mass and Lee, T.-H. and Mosalam, K.M. 2003. Probabilistic fiber ele-
damping) of the building. This implies that a better ment modeling of reinforced concrete columns. Proc. Ninth
understanding of the characteristic of ground motion Int. Conf. on Applications of Statistics and Probability in
profiles and hazard information will greatly reduce Civil Eng., ICASP9, San Francisco, July 6-9, 2003.
McKenna, F. and Fenves, G.L. 2001. OpenSees Manual. PEER
uncertainty of seismic demand of the studied build- Center, http://opensees.berkeley.edu/OpenSees.
ing. ARA and ARD are primarily sensitive to ground Mirza, S.A. and MacGregor, J.G. 1979. Variability of me-
motion profile given Sa while IDR is more sensitive chanical properties of reinforcing bars. J. of the Structural
Div., ASCE, 105(ST5): 921-37.
to Sa . This suggests that Sa may not be the best IM Mirza, S.A., Hatzinikolas, M., and MacGregor, J.G. 1979. Sta-
for ARA and ARD while it can serve as a good IM tistical description of strength of concrete. J. of the Struc-
tural Div., ASCE, 105(ST6): 1021-37.
for IDR for the studied RC shear-wall building. Porter, K.A., Beck, J. L., and Shaikhutdinov, R.V. 2002. Sensi-
The methodology of deterministic sensitivity tivity of building loss estimates to major uncertain variables.
analysis presented in this paper does not provide any Earthquake Spectra, 18(4): 719-43.
insight or effect of correlation between variables. As Scott, B.D., Park, R., and Priestley, M.J.N. 1982. Stress-strain
a future extension of the current study, a more accu- behavior of concrete confined by overlapping hoops at low
and high strain rates. ACI Journal, 79(1), Jan.-Feb.: 13-27.
rate sensitivity analysis will be pursued for the same
RC shear-wall building to overcome the shortcom-
ings of the presented analysis. The future study will
also include the 3-dimensional modeling of the
building to investigate the effect of torsional mode
of the building and the spatial variability of the mass.

6 ACKNOWLEDGEMENT

This work was supported primarily by the Earth-


quake Engineering Research Centers Program of the
National Science Foundation under Award Number
EEC-9701568 to the Pacific Earthquake Engineering
Research (PEER) Center at the University of Cali-

Potrebbero piacerti anche