Sie sind auf Seite 1von 11

Engineering Structures 39 (2012) 199–209

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Evaluation of response modification factor (R) of elevated concrete tanks


Mostafa Masoudi a,b,⇑, Sassan Eshghi b, Mohsen Ghafory-Ashtiany b
a
European School for Advanced Studies in Reduction of Seismic Risk (ROSE School), IUSS Pavia, Via Ferrata 1, 27100 Pavia, Italy
b
International Institute of Earthquake Engineering & Seismology (IIEES), No. 26, Arghavan Street, Tehran 19395-3913, Iran

a r t i c l e i n f o a b s t r a c t

Article history: This paper discusses the failure mechanism of elevated concrete tanks with shaft and frame staging (sup-
Received 26 April 2008 porting system) along with seismic behavior of these construction types. In order to modify the current
Revised 8 February 2012 code-based seismic design methodology, computer models have been established to determine the
Accepted 9 February 2012
response modification factors, R, of the shaft and frame staging elevated tanks. The computational models
Available online 22 March 2012
have been subjected to an ensemble of earthquake ground motions. The effects of multi-component
earthquakes, fluid–structure interaction and the P–D effects on the inelastic response of elevated tanks
Keywords:
have been studied by conducting linear and nonlinear response history analyses. According to results
Elevated tank
Earthquake response
of analyses and observed inelastic behavior during past earthquakes, the R factors for shaft and frame
Nonlinear analysis staging elevated concrete tanks have been evaluated regarding the seismicity of the site. Moreover, the
Fluid–structure interaction shortcomings associated to the current simplified seismic analysis and design procedure have been
Response modification factor addressed and a more rational modeling has been suggested especially for the shaft staging systems to
enhance distribution of the ductility demand.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction and frame supported elevated concrete tanks have been evaluated
regarding the seismicity of the site.
Elevated water tanks are important structures in water supply
networks. They should remain operational after earthquakes for vi- 2. Failure mechanism of elevated concrete tanks
tal needs. The poor seismic performance of these inverted pendu-
lum-type constructions has been reported frequently during The supporting system of elevated tanks mainly divided into
major earthquakes [1–12]. Among several reasons to describe fail- the two categories, i.e. shaft staging and frame staging. The differ-
ures and damages in past earthquakes, improper design in struc- ent structural properties of these two supporting systems suggest
tural elements and joints, low overall strength and ductility that the damage patterns and failure mechanisms of these struc-
capacity [8,11–13], torsional amplification of seismic response tures may not be identical as it was confirmed in the past earth-
[2,10,11,14,15] and the effect of soil–structure interaction [16– quake surveys [5,7,9–11]. Therefore, it is necessary to evaluate
20] have been highlighted in the previous studies. The purpose of the behavior and seismic performance of frame and shaft sup-
this paper is to explore a methodology for evaluation of the re- ported tanks separately.
sponse modification factor, R, of reinforced concrete shaft and
frame staging (supporting system) elevated tanks, which behave
2.1. Frame staging
in a ductile manner under seismic loads and expectedly fail in flex-
ural mode instead of shear mode. For this purpose, a shaft staging
Concrete frame staging is widely used as a supporting system
and also a frame staging elevated concrete tank have been ana-
for concrete elevated tanks. For these elevated tanks if the strong
lyzed by linear and nonlinear response history methods for an
column–weak beam design strategy is adapted, where the plastic
ensemble of earthquake ground motions. The analysis was carried
hinges do not occur in the columns, the damage pattern will be
out with and without the effects of fluid–structure interaction. The
as shown in Fig. 1a–c. Note the failure mechanisms shown in the
models were subjected to one, two and three correlated compo-
pictures are non-ductile flexural-shear failure of beams; however,
nents of earthquakes. In this procedure the R factor of the shaft
the columns did not suffer major damages.
If the frame staging of an elevated tank has been designed for
⇑ Corresponding author at: European School for Advanced Studies in Reduction of seismic forces, they would be able to represent a better seismic
Seismic Risk (ROSE School), IUSS Pavia, Via Ferrata 1, 27100 Pavia, Italy. Tel.: +39 performance than other concrete supporting systems. Large redun-
329 3359509. dancy and monolithic beam–column connections result in a proper
E-mail address: mmasoudi@roseschool.it (M. Masoudi). performance under seismic forces [5,7]. However, some previous

0141-0296/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2012.02.015
200 M. Masoudi et al. / Engineering Structures 39 (2012) 199–209

Fig. 1. (a), (b) and (c) Flexural-shear failure is the repeated damage pattern of beams in almost all levels of these three frame stagings, Chile earthquake, 1960 (courtesy of
EERC, UC Berkeley; Steinbrugge collection), and (d) joint failure, Bhuj earthquake, 2001 (figure reprinted from Rai [5]).

field observations [5,7,12] and the results of this study coming in


the following sections suggest that in frame stagings the level of
redundancy should be evaluated with caution because they do
not behave like building frames. In designing the supporting sys-
tems, formation of hinges in columns should be prevented and
the location of hinges should be directed to the beam ends. Impro-
per seismic design may result in typical failure in frame stagings
which have been shown in Fig. 1.

2.2. Shaft staging

Shaft staging is common for high volume elevated tanks. In


these structures the only lateral force resisting system is the con-
crete shaft. In comparison with the frame-type staging which has
a larger redundancy and higher capacity to absorb seismic energy
through inelastic actions, shaft staging is a determinate structure
and lacks redundancy, additional strength and damping to dissi-
pate seismic energy by forming multiple plastic hinges. When
damage occurs in any location along the shaft height, the only path
of transmitting the lateral force is lost and the structure collapses
as has happened in the past earthquakes [1,3,5,7,12].
Three types of crack have been observed in shaft stagings due to
earthquakes [3,5,7,12]. The most common one is circumferential
crack, which is due to flexural action. These cracks form in the sec-
tions closer to the foundation where the bending moment is the
largest. Generally, diagonal flexure-shear cracks appear in the
materials with low tensile strength due to a combination of bend-
ing and shear stresses. In shaft stagings the construction joints are
usually points of weakness that might result in weak sections
against flexure cracks. The second type of cracks is diagonal-cir-
cumferential cracks which are formed due to torsion, bending mo-
ment and shear or the combination of those. The third type of crack Fig. 2. (a) Horizontal and diagonal circumferential cracks, and (b) vertical cracks,
is those which have vertical directions because of high compres- Bhuj earthquake, 2001 (figures reprinted from Rai [5]).
sion stress imposed on the shaft during an earthquake. When com-
pression stress in the shaft because of vertical component of on the outer face, subjected to combined axial and lateral loads,
earthquake and overturning moment, is greater than the ultimate the position of the neutral axis in the cross-section has a major
strength of the section, the Poisson effects causes the diameter to influence on the available curvature ductility of the column [21].
expand which result in vertical cracks in the shaft. This behavior The closer the neutral axis to the internal face, the lower is the lon-
is more often seen in the shaft stagings with poor detailing and gitudinal strain in the unconfined internal face and the column has
insufficient transverse reinforcement. Examples of horizontal and a more ductile behavior. Generally, hollow circular columns with
vertical cracks are shown in Fig. 2. low axial load, moderate longitudinal reinforcement and a reason-
In shaft staging, ductility is more important than strength. In ably thick wall, with a thickness greater than 15% of the outside
hollow circular reinforced concrete columns with reinforcement diameter, behave in a ductile manner [21].
M. Masoudi et al. / Engineering Structures 39 (2012) 199–209 201

3. Fluid–structure interaction The period of convective motion is very long and it is affected by
ground displacement while the period of impulsive motion is rela-
In simplified modeling of elevated tanks [22–25], structural sys- tively short and sensitive to ground acceleration [25]. It should be
tem is usually represented by two separate lumped mass. One noted that if there is no sufficient freeboard available, the whole
mass includes impulsive mass of the fluid, tank self-weight and a mass tends to act in the impulsive mode. ACI 371R-08 [23] recom-
fraction of the self-weight of the supporting system. ACI 371R-08 mends considering a single lumped-mass model for elevated tanks
[23] and ACI 350.3-06 [22] recommend to use the generalized mass if the water weight exceeds 80% of the whole mass (including
as the equivalent mass of the supporting system which results in water) of the system above the base regardless the free board.
the two-thirds of the supporting structure mass while Priestley ASCE/SEI 7-10 [26] allows to consider the fluid–structure interac-
et al. [25] suggested the whole mass of the supporting system to tion only if the sloshing period is more than three times of the per-
be included in this mass. The other mass is convective mass of iod of the staging system assuming the whole liquid as a rigid mass
the fluid. In order to consider the effect of fluid–structure interac- at its volumetric gravity center. This condition is satisfied when ra-
tion in horizontal direction, the Housner’s [24] two-mass mechan- tio of the liquid height and the tank diameter is small. Otherwise,
ical analogy which includes the first mode of the impulsive and the whole mass of the stored liquid should be considered as the
convective motion of the fluid in cylindrical tanks has been used. impulsive mass. In this paper both single mass and two-mass rep-
To calculate the parameters of the model shown in Fig. 3 for cylin- resentation for fluid have been implemented. They have been de-
drical tanks, the following relationships are used: scribed in Section 5 as interacting and non-interacting models.
It is possible to model the fluid–tank interaction by the added
pffiffiffi 
mass approach [28] and the finite element (FE) method. It is impor-
tanh 3 Rh
MI ¼ pffiffiffi M; h < 1:5R ð1Þ tant to note that in this study the aspect ratios and the thick rein-
3 Rh forced concrete wall of the tank structures allow using the rigid
 
R tank model [25]. To study the state of stresses in the tank structure
M I ¼ 1  0:436 M; h > 1:5R ð2Þ the added mass approach is very valuable. However, to evaluate
h
( pffiffiffi !) the global seismic demands of the tank transmitted to the support-
3 4 3 Rh ing system the added mass approach yields to almost identical re-
hI ¼ 1þ pffiffiffi  1 h; h < 1:5R ð3Þ
8 3 tanhð 3 RÞ sults compared to the two mass representation of fluid [29]. An
h
 
R extensive study on reinforced concrete elevated tanks incorporat-
hI ¼ 0:5 þ 0:12 h; h > 1:5R ð4Þ ing fluid–soil–structure interaction [15] shows negligible differ-
h
rffiffiffiffiffiffi ! ences in seismic response of the lumped-mass mechanical
R 27 h analogy and the added mass approach. Also in a case study [13]
M C ¼ 0:455 tanh M ð5Þ
h 8 R on industrial steel tanks supported by relatively short reinforced
rffiffiffiffiffiffi rffiffiffiffiffiffi !
27 g 27 h concrete columns, the lumped-mass approach overestimated the
K C ¼ MC tanh ð6Þ response compared with the detailed FE modeling. The height to
8 R 8 R
8 qffiffiffiffi  9 diameter ratio of these tanks is larger than what normally used
>
< cosh 27 h
 31 >
= for elevated water tanks. The two concentric stainless steel shell
8 R 16
hC ¼ h 1  qffiffiffiffi qffiffiffiffi  ð7Þ in these tanks responded relatively rigid during the 1999 Kocaeli,
>
: 27 h
sinh 27 h >; Turkey earthquake [13]. Generally, steel tanks should be treated
8 R 8 R
as flexible structures and the two-mass mechanical analogy is
not accurate enough to predict their earthquake response [30].
where h and R are the height of the fluid and the internal radius of
However, even in this case, a research [31] on seismic analysis of
the tank. MI, MC, KC, KS, hC and hI are impulsive mass, convective
conventional and isolated LNG tanks shows a very good agreement
mass, convective motion stiffness and stiffness of supporting sys-
between two-mass mechanical analogy and detailed FE modeling
tem and corresponding height of impulsive and convective mass
in predicting global demands. These LNG tanks have a flexible in-
considering the effect of hydrodynamic pressure on the bottom of
ner steel tank with no dynamic interaction with the outer post-
the tank, respectively.
tensioned concrete tank. Another recent study [32] shows almost
Well-known codes [22,23,26,27] prescribe 0.5% damping for the
identical results in the full FE modeling and the simplified two-
convective mode. For reinforced concrete tanks they recommend
mass model of a conical elevated steel tank mounted on a concrete
5% damping for the impulsive mode. In this study damping of
shaft and confirms the codified simple seismic analysis procedure.
the convective motion, CC, and the damping of the supporting
The detailed FEM modeling of the structural system and the
systems, CS, are considered 0.5% and 5% of the critical damping,
fluid requires a significant time for running each analysis case
respectively.
while these previous studies [13,19,32] show a negligible advan-
Kc Kc tage of the full FE modeling compared to its complexity. Further-
/2 /2 more, in this study it is very important to find the earthquake
Mc scale factor leading to a failure mechanism. Thus, due to the inelas-
tic nature of the structural response, it is needed to run the com-
CC puter program for several times to find such a scale factor for
KS each earthquake motion and for each model. Therefore, it seems
MI impractical to use a detailed FE modeling in this study. Further-
more, there is no guarantee that the detailed FE modeling can ex-
actly predict the real behavior of the fluid–structure interaction
MS
CS problem and it needs further development. For example, the ob-
served discrepancies between results of the FE modeling and
experiment [33] reveal that the FE method for fluid–structure
interaction may deviate considerably from reality. Therefore,
assuming the FE analysis as the reference solution is still question-
Fig. 3. Mechanical model for elevated tanks. able and open to debates. Considering the aforementioned
202 M. Masoudi et al. / Engineering Structures 39 (2012) 199–209

researches supporting the idea of using simplified modeling, the The selected model is a concrete frame supported elevated
lumped-mass mechanical analogy seems to be acceptable whereas water tank with a volume of 700 m3. The geometrical configuration
this study focuses on the nonlinear response of the supporting sys- is chosen similar to an existing elevated water tank [37]. It consists
tems rather than the tank structure. of beams and columns on a lateral surface of an incomplete cone.
Vertical motions can generate significant hydrodynamic pres- The staging radius is 4.9 m at the bottom and 2.1 m at the top.
sure on the tank wall in the horizontal direction [34,35]. This effect The horizontal section of the staging is a regular octagon with a
usually alters the hoop stress in the tank wall rather than the over- column in each corner. The staging has six stories with a 5 m
all horizontal response, i.e. base shear and overturning moment. height in all stories. In each story level, columns are connected
Sloshing waves can be developed under vertical excitations; how- with circumferential beams. The dimension of the column sections
ever, it has been observed experimentally that no significant slosh- is 0.80  0.50 m and the dimensions of the beam sections are
ing occurs due to transient vertical excitations [34]. Therefore, it 0.85  0.45 m at each story. Fig. 4 shows two horizontal cross-sec-
has been assumed that the whole mass of the fluid acts as an tions of the frame staging. The tank structure is a cylinder with a
impulsive mass in the vertical direction. radius of 5.5 m, a height of 7 m and a thickness of 0.20 m. The cone
shape roof of the tank has a 5.5 m radius, 0.60 m height and 0.15 m
thickness. The base of the tank is made of a concrete slab with a
4. Model description 0.20 m thickness. The dimensions of the slab radial beams are
0.50  0.30 m.
Two elevated concrete tanks have been selected for numerical This elevated tank has been designed for a dead load in the full
modeling. One of them is supported by a frame staging. The other condition and a lateral force which is determined for a peak ground
one has a shaft supporting system. In the finite element (FE) mod- acceleration of 0.35 g and stiff soil conditions. According to ASCE/
eling of these structures, numerous frame elements have been SEI 7-10 [26] the response modification factor, R, has been consid-
used to model both frame and shaft staging. The Self-weight of ered equal to 3. The seismic provisions of ACI 371R-08 [23] are
each element has been allocated to the element nodes. In this case, consistent with ASCE/SEI 7-10 [26].
the staging mass has a uniform discrete distribution. Therefore, de- To prevent structural instability, it is assumed that the columns
spite what has been recommended for simplified modeling [23,25], do not suffer plastic deformation and the plastic hinges are only
in this way there is no need to consider a fraction of the self-weight formed in the beams. Thus, an elasto-plastic hinge is placed at
of the supporting system into the lumped mass of the tank at top of the both ends of each beam element. Plastic rotation will occur
the staging. in the hinges if the moment reaches yield moment in that cross-
For both staging models, frame elements with lumped plastic section.
hinges at the end of the element have been used. To model the ri- To obtain the idealized elasto-plastic moment-rotation diagram
gid tank structures, rigid frame elements have been used. The mass of hinges, the stress–strain diagram of steel and concrete is assumed
of the tank was distributed in the nodes of elements to represent as in Fig. 5. The yield and ultimate strengths of steel are 392 MPa
the whole mass of the tank structure in translational and rotational and 490 MPa, respectively. The ultimate compressive strength of
directions. Also rigid links connect the impulsive mass of water to concrete is 29 MPa. The horizontal part of Fig. 5b is because of the
the tank wall. The convective mass of water has been connected to partial confinement of concrete along the plastic zones due to trans-
the tank structure by elastic links having a resultant stiffness equal verse reinforcement. It is not reasonable to consider a full confine-
to the stiffness of the convective motion, KC. ment and consequently a very ductile behavior for concrete fibers
while the amount of tensile reinforcement and stirrups spacing
can significantly alter the ductility capacity due to buckling of com-
4.1. Frame staging elevated tank pressive reinforcement [38]. Fig. 6 shows the schematic shape of the
actual and idealized elasto-plastic moment–curvature diagram of
The common types of frame stagings with basic configuration the beam hinges. For interpreting curvature in terms of rotation in
usually have a torsional-to-lateral time period ratio close to one plastic hinges, the length of the plastic hinge is assumed equal to
[14]. In this case, it has been shown that the staging experiences the half-height of the beam cross section [28].
large torsional response under seismic excitations [14,15]. To calculate the effective stiffness of the beam cross sections it
Although it is possible to reduce torsional vulnerability by alter- is reasonable to consider Ec Ie ¼ M 0y =/0y [39] where, M 0y is the first
nate tank staging configurations [36] this study intends to address yield moment corresponding to a curvature, /0y , at which either a
the nonlinear response of common types of frame stagings regard- longitudinal reinforcement reaches its yield strain or the extreme
less accidental torsion.

Columns

Radial Beams
Circumferential
Beams

(a) (b)
Fig. 4. (a) Structural members of the frame staging which are connected to the bottom slab of the tank, and (b) varying cross-section of the frame staging.
M. Masoudi et al. / Engineering Structures 39 (2012) 199–209 203

Fig. 5. Stress–strain diagram of (a) steel, and (b) concrete.

Mi

M'y

φ
φ y φy
'
Fig. 7. Details of the annular section of the shaft staging (figure reprinted from
Möller and Rubinstein [40]).
Fig. 6. Actual and idealized moment–curvature diagram for the hinge regions.

staging is done such that no shear failure would occur before flex-
ural failure. Based on Fig. 7, double U16 hoops were used as cir-
compression concrete fiber exceeds a strain of 0.002. In this case cumferential reinforcement. The vertical spacing between two
study M 0y and /0y are equal to 950.6 kN m and 5.28  103 1/m, sets of hoops is 15 cm.
respectively. The nominal moment capacity, Mi, and curvature To prevent brittle flexural failure and to avoid unwanted fail-
capacity of the beam cross section are equal to 1176.8 kN m and ure modes such as shear failure and large vertical cracks due to
58  103 1/m, respectively. Thus, /y ¼ /0y M i =M 0y is equal to expansion of concrete core, the cross-section for the staging
6.54  103 1/m. Therefore, the expected curvature ductility must be ductile and well-confined. This is possible by proper
capacity is about 9 for beam cross sections. detailing of the section, providing a confined concrete core, like
The effective moment of inertia of the columns is as follows Fig. 7. The radial transverse bars are designed in such a way that
[39]: the circumferential confining stress is equal to the radial confin-
Ie ¼ 0:7Ig ð8Þ ing stress.
In the computer model, the length of the shaft staging was di-
It should be noted that this equation considers the effect of reduced vided into ten equal segments and a hinge was placed at the begin-
moment of inertia under forces exceeding flexural crack strength of ning of each segment. Plastic rotation will occur in the hinges if the
the columns and is not related to the nonlinear effects which con- moment reaches the yield moment in the hinge. To study the effect
sider steel yielding or concrete crushing in columns. As mentioned of the staging mass on the earthquake response, the mass was dis-
before, it is assumed that columns do not enter the inelastic range tributed between the nodes of elements. Fig. 8 shows a general
of behavior. view of the finite element model of this elevated tank and illus-
trates how the rigid tank structure and the interacting water mass
4.2. Shaft staging elevated tank considered in the model.
The elasto-plastic model shown in Fig. 6 has been used for non-
The model chosen is an elevated concrete shaft staging water linear hinges. To calculate the moment–curvature diagram of a sec-
tank. The volume of the tank is 2500 m3 and the height of the stag- tion the standard method of analyzing reinforced concrete sections
ing is 30 m from the top of the foundation to the bottom slab of the with confined core is used. In this research the method of Möller
tank. The external and internal diameters of the staging are 8 and and Rubinstein [40] has been employed. Considering the details
7 m, respectively. The tank structure is similar to an existing water of the staging section shown in Fig. 7, the moment–curvature dia-
tank [4,12]. gram can be calculated. To calculate the rotation capacity of plastic
To design the shaft staging, seismic forces are taken based on a hinges from the curvature capacity of the sections, the equivalent
peak ground acceleration of 0.35 g, stiff soil conditions and a re- plastic hinge length was assumed equal to 0.2D, where D is average
sponse modification factor, R, equal to 3 [26]. The ultimate com- diameter of the shaft [40]. In Fig. 6, ðM 0y ; /0y Þ corresponds to a cur-
pressive strength of concrete, fc0 , is taken at 29 MPa and the vature at which either a longitudinal reinforcement reaches its
longitudinal reinforcement was determined for the critical combi- yield strain or the extreme compression concrete fiber exceeds a
nation of axial force and bending moment. The shear design of the strain of 0.002.
204 M. Masoudi et al. / Engineering Structures 39 (2012) 199–209

Impulsive mass

Convective mass

Rigid frame elements as


rigid tank structure

Elastic bars for modeling


stiffness of the convective
mass

Distributed mass of the


tank structure and the
impulsive water mass in
the vertical direction

Tank weight = 2350 tons


Staging weight = 28.5 tons/m
Water weight = 2500 tons
Seismic weight = 5705 tons
h/R 0.8
h 8 m, hI 8 m, hC 7 m
MI=1125 tons, MC=1285 tons

Fig. 8. Finite element model in the full condition, with sufficient free board.

5. Special considerations for the analysis It should be noted that the program considers the interaction of
axial force and bending moment of structural elements. Should
The nonlinear earthquake response of structures can be calcu- there be a change in the axial force due to the vertical component
lated by means of nonlinear-static (pushover) analysis or nonlinear of earthquake and overturning moment, its effect on the moment-
dynamic analysis. In many cases, pushover analysis may fail to pre- rotation diagram of the hinge would be considered. This has been
dict the real dynamic behavior of the structure regardless the cho- made possible by calculating the moment capacity for correspond-
sen pushover method [41,42]. Therefore, the nonlinear response ing axial force at each instant of time.
history procedure has been considered in this study to reflect the Recent improvements in joint shear strength prediction [44]
dynamic nature of the earthquake response. and joint shear flexibility [45] facilitate estimation of RC joint
Two computer models of elevated tanks have been taken into behavior during earthquake response. However, inclusion of the
account for both supporting systems. In the first model, called joint shear behavior into the dynamic response is a delicate proce-
the non-interacting model, the whole mass of water was consid- dure and needs further clarifications. It may decrease the ductility
ered as impulsive mass and there is no interaction between the demand due to introduction of a new energy dissipation mecha-
water and the rigid wall of the tank. In the second model the nism [46]. Therefore, in this study the beam–column joints of the
fluid–structure interaction has been taken into account. frame staging have been considered to be rigid.
The analyses of these structures were carried out by PERFORM-
3D program [43]. In this program the definition of moment-rota-
tion diagram for the hinge is required. Instead of using complex 6. The method of analysis
material models for strength deterioration and stiffness degrada-
tion in the hysteresis loops, a model in which the area of the hys- The computer models have been subjected to one, two and
teresis loop decreases by a coefficient smaller than 1 has been used three correlated components of nine earthquakes. At first, each
[43]. This coefficient decreases with increasing the hinge rotation. acceleration record was scaled to the extent that led the structure
This process is shown in Fig. 9. to reach the first yield at one of the plastic hinges. The PGA of the

Fig. 9. A method for decreasing the hysteresis loops (figure reprinted from PERFORM-3D manual [43]).
M. Masoudi et al. / Engineering Structures 39 (2012) 199–209 205

scaled record is called ae. Again each record is scaled to the extent confirmed in the next section that the formation of only one
that the ultimate rotation capacity is reached at one of the plastic yielded hinge and failure of solely one hinge in the computational
hinge. Then the PGA of the scaled record is called af. In another model represent the yield state and the failure state of the struc-
word, ae and af represent the PGA of the design basis earthquake ture, respectively. Considering several parameters of ground mo-
and maximum considered earthquake, respectively. It will be tion which affect the seismic response, it is obvious that the ae
and af are not unique for all earthquakes. This method is used for
the interacting and non-interacting models with and without the
10 P–D effect. The af/ae ratio provides a valuable insight for assessing
af /ae uniaxial
8 biaxial the performance of the structure within the inelastic range. It is ex-
triaxial pected to see a smaller af/ae ratio for the cases in which the P–D
6
effects has been included into the analysis. Some results of these
4 analyses are presented in Figs. 10–13.
2

0 7. Results of analysis
Manjil-Roodbar Naghan Tabas El Centro Imperial Valley
7.1. Frame staging elevated tank
Fig. 10. af/ae in the non-interacting model of the frame staging.

The results of the analysis show that in some of the earth-


quakes, the beams of all stories enter the inelastic range. An exam-
14
af /ae uniaxial ple of this behavior is shown in Fig. 14. The real sample of this
12 biaxial behavior is shown in Fig. 1a–c which clearly shows that the
triaxial hypothesis of the plastic hinge formation only in the lower stories
10
is incorrect. Therefore, the design of these staging types requires
8
proper reinforcement details in the beams of all stories.
6 Usually, capacity design philosophy leads too very strong col-
4 umns. They behave relatively rigid compared to the beams. There-
2
fore, formation of only one plastic hinge at the end of one beam or
column base, theoretically causes formation of plastic hinges at
0
both ends of all beams and developing a full mechanism in the
Bam Manjil- Naghan Tabas El Gazli Loma Chi-Chi Imperial
Roodbar Centro Prieta Valley frame staging. This behavior is shown schematically in Fig. 15. This
means that in terms of redundancy, frame stagings supporting a
Fig. 11. af/ae in the interacting model of the frame staging. large mass at top behaves completely different than moment
resisting frames typically used as lateral load bearing systems in
buildings because in the building structures the total mass of the
14
af /ae system is distributed in all story levels. Despite frame stagings hav-
uniaxial
12 ing almost uniform ductility demand distribution in members,
biaxial
triaxial height-wise distribution of mass in building frames causes a lateral
10
displacement pattern in which the system ductility demand is not
8 necessarily distributed uniformly in all beams.
6

2
Rigid elements
0
Bam Manjil- Naghan Tabas El Centro Imperial
Roodbar Valley

Fig. 12. af/ae in the non-interacting model of the shaft staging. Beams entered
inelastic range
18
af /ae uniaxial
16 biaxial
triaxial
14

12

10

0
Bam Manjil- Naghan Tabas El Centro Imperial
Roodbar Valley Fig. 14. The elements of the frame staging entered inelastic range due to the three
components of Imperial Valley earthquake scaled to its af are shown with dashed
Fig. 13. af/ae in the interacting model of the shaft staging. lines.
206 M. Masoudi et al. / Engineering Structures 39 (2012) 199–209

fact is equivalent to overestimating the curvature ductility demand


at base and avoiding this demand along the shaft height.

8. Explanation of the results

The minimum ratio of af/ae obtained from different earthquakes


is the basis of calculation of R factor. Large values of af/ae lead to a
large R factor. According to the next section, omission of the earth-
quakes whose af/ae are greater than 6 will be a logical action. Each
earthquake with a af/ae ratio larger than 6 has been determined
and omitted in the later analyses.
In the nonlinear analyses, the parameter that has a great influ-
ence on af is the ultimate curvature capacity /u of the section. It
should be noted that even if the parameters required for the non-
linear analysis are obtained through testing, it may not be quite
Fig. 15. Relatively rigid columns cause simultaneous formation of plastic hinges at
the beam ends.
reliable because in the testing process a structural member is sub-
jected to a few displacement histories and the parameters of the
nonlinear behavior are obtained for those types of loading which
may differ significantly from a real earthquake loading. Therefore,
7.2. Shaft staging elevated tank if in a specific earthquake the curvature demand is very sensitive to
the scale factor of the record, the results of this earthquake should
The results of analysis show that at some instant of time lateral not be taken into account.
displacement of the staging is not in phase with the first mode It is seen in the results of the analyses that in some earthquakes
shape. The effect of higher modes in the earthquake response of the af/ae ratio has increased in biaxial or triaxial excitations due to
the structure is apparent. In addition, the distribution of inelastic a decrease in ae while the value of af has not experienced any sig-
behavior takes place in a larger length of the staging which is nificant change.
shown in Fig. 16. At least 20% of the staging height has entered In any case if the fluid–structure interaction plays an important
the inelastic range of response and this ratio has reached 80% in role in the earthquake response, the af/ae ratio decreases in com-
some earthquakes. In most practical cases determination of the parison with the non-interacting model. The period of convective
curvature or rotation ductility demand of the shaft structure is motion is very large and it is not in the range of the earthquake
done based on the formation of one plastic hinge at the bottom dominant period. However, if the spectral ordinate in the range
of the staging. Considering the results of analysis showing the for- of the convective motion time period is relatively significant the
mation of multi plastic hinges along the staging height, the ductil- af/ae ratio decreases.
ity demand at base can increase fictitiously because it ignores the In most cases, considering the effect of P–D will decrease the ra-
distribution of the inelastic behavior in other portions of the stag- tio of af/ae but in some earthquakes the P–D effects increases af/ae.
ing. Also this reality necessitates the need for ductile detailing The reason is considerable decrease in ae which leads to an in-
throughout the height of the shaft staging. Therefore, ignoring this crease in af/ae. However, in some cases despite the usual expecta-
tion from the P–D effect, the value of af increases. This is because
the staging displacement is not in phase with the first mode.
Fig. 17 indicates that the P–D effect has caused a decrease in the
bending moment due to lateral force at the bottom of the staging.
Furthermore, in higher mode shapes the P–D effects help the dis-
tribution of plasticity throughout the staging length. For example
in the second mode shape similar to Fig. 17 the moment above
the mid-height is amplified due to the P–D effect.
Inelastic Range

Translational and
Inelastic Range

rotational masses
Inelastic Range

Supporting system
with distributed
mass

(a) (b) (c)


Fig. 16. The length of the shaft staging entered inelastic range due to the records Fig. 17. Decrease in bending moment in sections which are close to the foundation
that scaled to af. (a) Uniaxial Tabas earthquake, (b) uniaxial Imperial Valley due to the P–D effect in displacement states which are not in phase with the first
earthquake, and (c) triaxial Tabas earthquake. mode shape of the structure.
M. Masoudi et al. / Engineering Structures 39 (2012) 199–209 207

Table 1
The value of R factor for the frame staging.

Type of model No. of components af/ae Seismicity parameter (C)


1.50 1.75 2.00 2.50 3.00
Interacting model 1D 4.0 3.2 2.7 2.4 1.9 1.6
2D 3.7 3.0 2.5 2.2 1.8 1.5
3D 3.4 2.7 2.3 2.0 1.6 1.4
Non-interacting model 1D 3.7 3.0 2.5 2.2 1.8 1.5
2D 3.5 2.8 2.4 2.1 1.7 1.4
3D 3.2 2.6 2.2 1.9 1.5 1.3

Table 2
The value of R factor for the shaft staging.

Type of model No. of components af/ae Seismicity parameter (C)


1.50 1.75 2.00 2.50 3.00
Interacting model 1D 4.1 3.3 2.8 2.5 2.0 1.3
2D 3.9 3.1 2.7 2.3 1.9 1.6
3D 3.6 2.9 2.5 2.2 1.7 1.4
Non-interacting model 1D 3.6 2.9 2.5 2.2 1.7 1.4
2D 3.4 2.7 2.3 2.0 1.6 1.4
3D 3.1 2.5 2.1 1.9 1.5 1.2

9. Calculation of (R) factor a tuned liquid damper mounted at top which are used to enhance
the wind and earthquake response of structures [50].
The following approach for selection of R factor is used [47,48]: The code specifications in ASCE/SEI 7-10 [26], ACI 371R-08 [23]
  and FEMA 450 [47] recommend the R value for the considered sta-
IF gings up to 3. In FEMA P-750 [51], the successor of FEMA 450, these
ae ¼ a475 ð9Þ
R provisions remains without any change. However, in Tables 1 and
where IF is the importance factor and a475 is the PGA of the design 2 the calculated R factor for the elevated tanks is less than 3 if the
basis earthquake with a return period of 475 years (10% probability value of C is considered larger than 1.75. This means that for most
of exceedance in 50 years). This relation can be rewritten in the fol- of the seismic regions these codes underestimate the seismic de-
lowing form. sign forces. The current simplified seismic design procedure allows
performing a unidirectional earthquake analysis in many cases.
a475 af However, it is important to consider the detrimental effects of
R ¼ IF  ð10Þ
af ae other components of earthquake into this procedure. Therefore, it
If af is considered as the PGA of the earthquake with a return is reasonable to relay on the response modification factor obtained
period of 2475 years (2% probability of exceedance in 50 years) R by considering triaxial seismic analyses. In this case, the R factor
will be: for any level of seismicity is less than 3. However, despite the US
  codes, Eurocode 8 [27] specifies q (equivalent to R) up to 1.5 for
IF af all elevated tanks except for simple support structures with low
R¼ ð11Þ
C ae seismic risk for which q can be increased up to 2. This seems to
be a more realistic expectation from tank supporting structures.
where C = a2475/a475 and it depends on the seismicity of the site.
Considering the inherent overstrength of structural systems,
The ATC-3-06 provisions [49] suggest that C varies from 1.5 to
q = 1.5 is equivalent to the elastic design forces. It is worth men-
2.0 for high and low seismic regions, respectively. The seismic haz-
tioning that this level of behavior factor is appropriate for historical
ard maps published in the FEMA 450 [47] provisions suggest sim-
masonry towers as well [42].
ilar values for the high seismicity regions and C = 2–3 for the low
This study shows that the response modification factor for
seismicity regions. However, in some moderate seismic regions,
both frame and shaft stagings are very close to each other. This
near large intraplate faults which rupture infrequently the C value
is contrary to the common expectation from frame supporting
may increase to C = 4–6 [47]. Thus, for calculating the R factor it is
systems. The reason is the unique lateral displacement pattern
reasonable to consider a range of C between 1.5 and 3.
of structural frames carrying a large mass at top which has been
To calculate R, it is necessary to know the minimum value of af/
discussed in Section 7.1. Furthermore, well known codes allow to
ae in each model. In Tables 1 and 2 the minimum af/ae ratios for the
use the same response modification factor for both frame and
interacting and non-interacting models of frame and shaft support-
shaft stagings.
ing systems are summarized for uniaxial, biaxial and triaxial
excitations.
It is noteworthy that the minimum value of af/ae decreases with 10. Conclusions
increasing the number of earthquake components because the
structure is overloaded compared to the uniaxial earthquake exci- Despite the US seismic design codes, i.e. ASCE/SEI 7-10 [26], ACI
tation. This shows that the ability of behaving in the inelastic range 371R-08 [23] and FEMA 450 [47] which recommend the R value up
decreases if the other components of earthquakes are taken into to 3, the calculated R factor for the elevated tanks in this research is
account. Furthermore, the values of af/ae in the non-interacting less than 3. However, Eurocode 8 [27] specifies the response mod-
models are less than the interacting models because the sloshing ification factor mainly up to 1.5 and for simple supported elevated
motion of water inside the tank is very similar to the behavior of tanks with low seismic risk up to 2 which results in essentially
208 M. Masoudi et al. / Engineering Structures 39 (2012) 199–209

elastic design forces if the overstrength factor is considered. This [12] Tavakoli S, Haeri SM, Ramazi HR, Eshghi S, Moghaddam H. Manjil-Roodbar
earthquake. Tehran, Iran: International Institute of Earthquake Engineering &
seems to be a more realistic expectation from tank supporting
Seismology (IIEES); 1991 [in Persian].
structures. However, to come up with a solid recommendation [13] Sezen H, Livaoglu R, Dogangün A. Dynamic analysis and seismic performance
for response modification factor it is recommended to consider a evaluation of above-ground liquid-containing tanks. Eng Struct 2008;30:
larger ensemble of earthquakes, more computational models and 794–803.
[14] Dutta SC, Jain SK, Murty CVR. Assessing the seismic torsional vulnerability of
different subsoil categories. elevated tanks with RC frame-type staging. Soil Dynam Earthquake Eng
Capacity design of frame stagings leads to strong columns. Thus, 2000;19:183–97.
for a frame carrying a large mass at its top, it is more likely to have [15] Dutta SC, Jain SK, Murty CVR. Inelastic seismic torsional behaviour of elevated
tanks. J Sound Vib 2001;242:151–67.
simultaneous plastic hinges at the ends of all beams due to its spe- [16] Dutta S, Mandal A, Dutta SC. Soil–structure interaction in dynamic behaviour
cial displacement pattern shown in Fig. 15 rather than uneven and of elevated tanks with alternate frame staging configurations. J Sound Vib
delayed formation of plastic hinges common in building frames. It 2004;277:825–53.
[17] Dutta SC, Dutta S, Roy R. Dynamic behavior of R/C elevated tanks with soil–
can be concluded that frame stagings of elevated tanks are not very structure interaction. Eng Struct 2009;31:2617–29.
redundant lateral load resisting systems and they have limited [18] Haroun MA, Temraz MK. Effects of soil–structure interaction on seismic
capability to redistribute seismic induced moments. This research response of elevated tanks. Soil Dynam Earthquake Eng 1992;11:73–86.
[19] Livaoglu R, Dogangün A. Simplified seismic analysis procedures for elevated
supports the idea of having the same response modification factor tanks considering fluid–structure–soil interaction. J Fluid Struct 2006;22:
for both shaft and frame stagings. 421–39.
The result of nonlinear analyses shown in Fig. 16 indicates that [20] Livaoglu R, Dogangün A. Effect of foundation embedment on seismic behavior
of elevated tanks considering fluid–structure–soil interaction. Soil Dynam
a large portion of the staging length may enter inelastic range and
Earthquake Eng 2007;27:855–63.
multi plastic zones may be formed along the staging height. This [21] Zahn FA, Park R, Priestley MJN. Flexural strength and ductility of circular
behavior has a significant influence on the ductility demands dur- hollow reinforced concrete columns without confinement on inside face. ACI
ing an earthquake. Therefore, assigning just one potential plastic Struct J 1990;87:156–66.
[22] ACI 350.3-06. Seismic design of liquid-containing concrete structures and
hinge at the bottom of the shaft staging and considering hinges commentary. ACI Committee 350. Farmington Hills, MI: American Concrete
only on the lower levels of the frame supporting system overesti- Institute; 2006.
mates the curvature ductility demand at lower levels of the staging [23] ACI 371R-08. Guide for the analysis, design, and construction of elevated
concrete and composite steel-concrete water storage tanks. ACI Committee
while ignores the distribution of ductility demand throughout the 371. Farmington Hills: American Concrete Institute; 2008. p. 41.
staging height. [24] Housner GW. The dynamic behavior of water tanks. Bull Seismol Soc Am
Fluid–structure interaction can lead to a strong or weak perfor- 1963;53:381–7.
[25] Priestley MJN, Davidson BJ, Honey GD, Hopkins DC, Martin RJ, Ramsey G, et al.
mance of the structure in the inelastic range depending on the fre- Seismic design of storage tanks. Recommendation of a study group of the New
quency content of the earthquake acceleration. Because of Zealand National Society for Earthquake Engineering. Wellington, New
significant differences between the results of the linear and nonlin- Zealand: New Zealand Society for Earthquake Engineering; 1986. p. 180.
[26] ASCE/SEI 7-10. Minimum design loads for buildings and other structures.
ear analyses in the interacting and non-interacting models, it is Reston, VA: Structural Engineering Institute, American Society of Civil
suggested to consider the effect of fluid–structure interaction in Engineers; 2010. p. 650.
the analysis and design. [27] European Committee for Standardization (ECS). Eurocode 8: design of
structures for earthquake resistance. Part 4: Silos, tanks and pipelines, prEN
In the nonlinear analysis of elevated tanks, the P–D effect may
1998–4. Brussels, Belgium; 2003.
be important. This effect causes the amplification of seismic in- [28] Westergaard HM. Water pressures on dams during earthquakes. Trans Am Soc
duced moment throughout the staging height in the higher modes. Civil Eng 1933;98:418–33.
[29] Housner GW. Dynamic pressures on accelerated fluid containers. Bull Seismol
Soc Am 1957;47:15–35.
[30] Haroun MA. Dynamic analyses of liquid storage tanks. Pasadena,
Acknowledgments Calif.: Earthquake Engineering Research Laboratory, California Institute of
Technology; 1980.
[31] Christovasilis IP, Whittaker AS. Seismic analysis of conventional and isolated
The authors wish to thank the anonymous reviewers, Dr. M. LNG tanks using mechanical analogs. Earthquake Spectra 2008;24:599–616.
Mehrain from URS Corporation and Prof. F. Arbabi from IIEES for [32] Moslemi M, Kianoush MR, Pogorzelski W. Seismic response of liquid-filled
reviewing this paper and for their valuable comments. elevated tanks. Eng Struct 2011;33:2074–84.
[33] Ozdemir Z, Souli M, Fahjan YM. Application of nonlinear fluid–structure
interaction methods to seismic analysis of anchored and unanchored tanks.
Eng Struct 2010;32:409–23.
References [34] Haroun MA, Tayel MA. Response of tanks to vertical seismic excitations.
Earthquake Eng Struct Dynam 1985;13:583–95.
[35] Kianoush MR, Chen JZ. Effect of vertical acceleration on response of concrete
[1] Astaneh A, Ghafory-Ashtiany M. The Manjil, Iran, earthquake of June 1990.
rectangular liquid storage tanks. Eng Struct 2006;28:704–15.
EERI Special Earthquake Rep, EERI Newslett 1990;24:5–13.
[36] Dutta SC, Jain SK, Murty CVR. Alternate tank staging configurations with
[2] Jain SK, Murty CVR, Chandak N, Seeber L, Jain NK. The September 29, 1993, M6.
reduced torsional vulnerability. Soil Dynam Earthquake Eng 2000;19:
4 Killari, Maharashtra, earthquake in central India. EERI Special Earthquake
199–215.
Rep, EERI Newslett 1994;28:1–8.
[37] Masoudi M, Ghafory-Ashtiany M, Eshghi S. Seismic behavior, analysis and
[3] Mehrain M. Reconnaissance report on the Northern Iran Earthquake of June 21,
modeling of concrete elevated tanks. Res Bull Seismol Earthquake Eng
1990. NCEER. Buffalo, New York: State University of New York at Buffalo; 1990.
2004;23:1–26 [in Persian].
[4] Memari AM, Ahmadi MM, Rezai B. A study of the failure of the water tower in
[38] Lopes AV, Lopes SMR, do Carmo RNF. Effects of the compressive
Rasht. International Institute of Earthquake Engineering & Seismology (IIEES);
reinforcement buckling on the ductility of RC beams in bending. Eng Struct
1994 [in Persian].
2012;37:14–23.
[5] Rai DC. Elevated tanks. Earthquake Spectra 2002;18:279–95.
[39] Paulay T, Priestley MJN. Seismic design of reinforced concrete and masonry
[6] Rai DC. Seismic retrofitting of R/C shaft support of elevated tanks. Earthquake
buildings. New York: John Wiley & Sons; 1992.
Spectra 2002;18:745–60.
[40] Möller O, Rubinstein M. Reliability-based design of R/C water tank structures
[7] Rai DC. Performance of elevated tanks in Mw 7.7 Bhuj earthquake of January
under seismic action. Earthquake Eng Struct Dynam 1992;21:665–78.
26th, 2001. Proc Indian Acad Sci – Earth Planet Sci 2003;112:421–9.
[41] Peña F, Lourenco PB, Mendes N, Oliveira DV. Numerical models for the seismic
[8] Saffarini HS. Ground motion characteristics of the November 1995 Aqaba
assessment of an old masonry tower. Eng Struct 2010;32:1466–78.
earthquake. Eng Struct 2000;22:343–51.
[42] D’Ambrisi A, Mariani V, Mezzi M. Seismic assessment of a historical masonry
[9] Steinbrugge KV, Karl V. Steinbrugge collection. The earthquake engineering
tower with nonlinear static and dynamic analyses tuned on ambient vibration
online archive. UC, Berkeley: NISEE, PEER; 1960.
tests. Eng Struct 2012;36:210–9.
[10] Steinbrugge KV. Earthquake damage and structural performance in the United
[43] CSI. PERFORM-3D V. 4. Nonlinear analysis and performance assessment for 3D
States. In: Wiegel RL, editor. Earthquake engineering. Englewood Cliffs,
structures. Berkeley, California: Computers and Structures; 2006.
NJ: Prentice-Hall; 1970.
[44] Park S, Mosalam KM. Parameters for shear strength prediction of exterior
[11] Steinbrugge KV, Moran DF. An engineering study of the southern California
beam–column joints without transverse reinforcement. Eng Struct 2012;36:
earthquake of July 21, 1952 and its aftershocks. Bull Seismol Soc Am
198–209.
1954;44:201–462.
M. Masoudi et al. / Engineering Structures 39 (2012) 199–209 209

[45] Sharma A, Eligehausen R, Reddy GR. A new model to simulate joint shear [49] ATC-3-06. Tentative provisions for the development of seismic regulations for
behavior of poorly detailed beam–column connections in RC structures under buildings. USA: Applied Technology Council; 1978.
seismic loads, Part I: exterior joints. Eng Struct 2011;33:1034–51. [50] Soong TT, Dargush GF. Tuned liquid dampers. Passive energy dissipation
[46] Ghobarah A, Biddah A. Dynamic analysis of reinforced concrete frames systems in structural engineering. Chichester, New York: John Wiley & Sons;
including joint shear deformation. Eng Struct 1999;21:971–87. 1997.
[47] BSSC. FEMA 450, NEHRP recommended provisions for seismic regulations for [51] BSSC. FEMA P-750, NEHRP recommended provisions for seismic regulations
new buildings and other structures, 2003 ed. Washington, DC: Federal for new buildings and other structures, 2009 ed. Washington, DC: Federal
Emergency Management Agency; 2004. Emergency Management Agency; 2009.
[48] Wilson JL. Earthquake response of tall reinforced concrete chimneys. Eng
Struct 2003;25:11–24.

Das könnte Ihnen auch gefallen