Documente Academic
Documente Profesional
Documente Cultură
16
Nonlinear Seismic Analyses
of a High Gravity Dam with
and without the Presence
of Reinforcement1
Yuchuan Long *, Chuhan Zhang y, Yanjie Xu y
*
School of Civil Engineering, Chongqing University, Chongqing
y
State Key Laboratory of Hydroscience and Engineering, Tsinghua
University, Beijing
O U T L I N E
1
Engineering Structures 2009; 31: 2486e94. Ó 2009 Elsevier Ltd.
Seismic Safety Evaluation of Concrete Dams Copyright Ó 2013 Tsinghua University Press. Published
http://dx.doi.org/10.1016/B978-0-12-408083-6.00016-7 369 by Elsevier Inc. All rights reserved.
370 16. NONLINEAR SEISMIC ANALYSES OF A HIGH GRAVITY DAM
16.1 INTRODUCTION
To meet the enormous energy demands, several high gravity dams are
being constructed in high seismic risk zones of south-western China, e.g.,
Longtan (192 m), Xiangjiaba (161 m) and Jin’anqiao (160 m). The peak
ground accelerations of these dams calculated from current seismic-
resistant codes are 0.20, 0.222 and 0.399 g, respectively, with the design
criterion of 2% probability of exceedance occurring in 100 years. A strong
earthquake is therefore a great challenge to Chinese high dam construc-
tion, especially after the Wenchuan earthquake that occurred on May 12,
2008 (Richter scale: 8.0).
Although minor damage to high concrete dams in the form of limited
cracking is acceptable, extensive cracking resulting in structural insta-
bility and reservoir loss must be prevented even under extremely strong
earthquakes. Thus, it is necessary to model earthquake-induced cracking
behavior when evaluating the seismic safety of high gravity dams. To
improve the seismic-resistant capacity of concrete dams, Chinese engi-
neers have proposed using reinforcement in potential cracking zones, but
its effectiveness needs further investigation.
So far, two different approaches have been used to analyze the
nonlinear cracking behavior of concrete dams. One approach, called the
discrete crack model, simulates the discrete cracking observed in plain
concrete based on the fracture mechanics concept and uses an automatic
remeshing technique to model the crack propagation [1]. Javanmardi et al.
[2] combined the discrete crack model with a theoretical formulation for
transient water pressure along a tensile seismic concrete crack to study the
seismic stability of concrete gravity dams accounting for dynamic varia-
tion in the uplift force. Based on the concept of a discrete crack, Arabshahi
and Lotfi [3] investigated the earthquake response of gravity dams
including the damefoundation interface nonlinearities using interface
elements which can model the behavior of crack opening/closing and
sliding at the dam base. These two works can conveniently utilize the
discrete crack model without remeshing because the potential cracking
path is definitely at the damefoundation interface. If the cracking path is
unknown, a remeshing technique should be used to model the crack
propagation, and the computational cost will increase enormously.
Therefore, the discrete crack model is inefficient for conducting a seismic
cracking analysis of high gravity dams.
The alternative approach, called the smeared crack model, uses the
blunt crack band theory proposed by Bazant and Oh [4] to simulate the
concrete cracking behavior, and it is easily implemented into standard
finite element procedures. Using this model, Vargas-Loli and Fenves [5]
conducted a nonlinear seismic analysis of a gravity dam including tensile
cracking of the concrete and the interaction between the dam and
compressible water. Their results show that concrete cracking is an
important nonlinear phenomenon and extensive cracking due to strong
earthquakes may affect the stability of the dams. Bhattacharjee and Leger
[6] developed a smeared crack model which considers the strain-rate
sensitivity of concrete fracture parameters, the softening initiation crite-
rion under biaxial loading and the behavior of crack opening/closing
under cyclic loading conditions. They applied the modified model to
investigate the seismic fracture and energy response of gravity dams and
discussed the significance of viscous damping models accounting for the
non-cracking structural energy dissipation mechanisms. Based on the
model proposed by Bhattacharjee and Leger [6], Ghaemian and Ghobarah
[7] investigated the effects of the damereservoir interaction on the
nonlinear seismic response of gravity dams in the time domain. To
conduct a seismic analysis of arch dams, Lotfi and Espandar [8] combined
the discrete crack approach with a non-orthogonal smeared crack method
to model the opening/closing behavior of contraction joints and concrete
cracking in the monolith. Although the smeared crack model has been
widely applied to seismic analysis of concrete dams, it is still difficult to
define rigorous evolution rules under multiaxial cyclic loading conditions
and model concrete cracking with other inelastic phenomena near the
fracture process zone [9].
Some researchers have used continuum damage mechanics to
investigate the seismic cracking behavior of gravity dams. Considering
the demand of thermodynamic consistency, Cervera et al. [10] devel-
oped an isotropic damage model which split the stress tensor into tensile
and compressive components and utilized respective damage evolution
laws about these two stress states to simulate the different damage
behavior of concrete in tension and compression. Moreover, this model
is extended to include rate dependence, which can benefit regularizing
the ill-posed problem of strain-softening materials [11]. Nevertheless, it
is still hard to model the damage behavior of concrete under cyclic
loading conditions because there is no representation of the inelastic
strain. Subsequently, Lee and Fenves [12] proposed a plastic-damage
model for concrete subjected to cyclic loading based on a general
damage concept that combines the degradation of elastic stiffness with
the evolution of inelastic strain and includes the effect of stiffness
recovery when the crack closes. This model has been used to analyze the
seismic damage response of the Koyna dam [13] and it obtained
a cracking pattern for the prototype dam that is comparable with the
experimental results [14e16].
To evaluate the effectiveness of the reinforcement measure, Long
et al. [17] developed a modified embedded-steel model by combining
the approach of stiffening reinforced steel [18] with the zoning method
of lightly reinforced concrete [19]. While this model can simulate the
nonlinear seismic damage response of concrete dams with reinforce-
ment, it ignores the influence of the slip between the reinforced steel
and its surrounding concrete on the cracking response of the dams.
Kwak and Filippou [20,21] developed a reinforced steel model with
bondeslip which is embedded into a concrete element and includes the
bondeslip effect by incorporation of the equivalent steel stiffness. This
model is easily implemented into standard FE procedures and can be
used to investigate the influence of the reinforcement slip on the dam
responses.
This chapter first explains the constitutive models of concrete and
reinforced steel used herein. Then, nonlinear analyses are conducted to
investigate the seismic response of a 160 m high gravity dam. These
analyses account for several nonlinearities, for example tensile cracking,
plastic offset strain in tension, stiffness recovery when the crack closes
and the bondeslip effect. Numerical results are analyzed in great detail to
study the seismic damage mechanics of the gravity dam and evaluate the
effect of the reinforcement.
sðsc Þ ¼ s0 þ ð1 s0 Þrðsc Þ
8
>
> 0 sc ¼ 0
>
>
>
>
>
<P 3
(16.2)
rðsc Þ ¼ hsic i
>
> i¼1
>
> 3
otherwise
>
> P
>
: sc
i
i¼1
where Dt and Dc are the tensile and compressive damage variables; kt and
kc are the ratios of dissipating energy to fracture energy in tension and
compression, respectively; sic , i ¼ 1,2,3 denotes the principal value of the
effective stress; and h,i is the McCauley bracket as hxi ¼ (x þ jxj)/2.
Herein, the crack opening/closing behavior is modeled by multiplying
the parameters, which represents the elastic stiffness recovery during the
elastic unloading process from the tensile state to compression. s0 is
a constant to set the minimum value of s, and r is a weighted parameter
determined by the principal values of the current effective stress vector.
Figure 16.1 shows the uniaxial constitutive relation of concrete under
cyclic loading. s0 ¼ 1 represents the concrete fully recovering the initial
undamaged stiffness while s0 ¼ 0 ignores the stiffness recovery.
The difficulty involved in this model is the lack of a theoretical
definition of the tensile plastic offset for concrete. This offset occurs when
rough cracking surfaces come into contact during unloading and do not
realign because of shear slip along the cracking surfaces. This phenom-
enon is demonstrated by concrete tests [22,23]. Palermo and Vecchio [24]
proposed a plastic offset relation as
p 2
εc ¼ 146ðεm m
c Þ þ 0:0523εc (16.3)
p
εc
where is the tensile plastic offset; and εm
c is
the maximum tensile strain
in the loading history. This proposed relation has been implemented into
the constitutive model of concrete to define the evolution of the elasto-
plastic surface.
Moreover, the degradation variable D is not convenient for evaluating
the damage degree of a concrete structure, and therefore the equivalent
crack opening displacement w defined in Eq. (16.4) is adopted as
a damage index for a plain or reinforced concrete structure in the
following numerical analyses:
w ¼ lc ecr ¼ lc Tn εcr
c
h i (16.4)
L1
εcr
c ¼ εc Ec sc ; Tn ¼ a2x a2y ax ay
where ecr is the cracking strain; lc denotes the crack band width of
a cracked element, which is determined by the element dimension and
crack normal [4]; and ax and ay are directional cosines of the crack normal
corresponding to global axes x and y.
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1r 1 r εs εf
a ¼ ; b ¼
1 r þ nr r εt εf
(16.5)
1 Es
fscr ¼ 1 þ n ft ; n ¼
r Ec
where ss and εs are the axial stress and strain of the reinforced steel; Es and
Ec denote the elastic modulus of the steel and concrete, respectively; r is
the effective ratio of reinforcement to concrete estimated according to the
CEB-FIP model code 1990 [25]; and ft and fy are the concrete tensile
strength and steel yielding strength, respectively. The characteristic
strains εt, εf and εy represent three key states: reaching the tensile strength
of the concrete, complete damage of the concrete and steel yielding at
the cracking section. The solid line shown in Fig. 16.2 is the tensile
stressestrain relation defined by Eq. (16.5).
the rock foundation and dam structure. Herein, the maximum size of the
dam mesh is limited to 2.5 m for modeling the concrete cracking.
Moreover, the shading zone in Fig. 16.4 represents the dam intake whose
cross-stream thickness is 0.4 times as thick as the monolith.
To reduce the opening and extension of concrete cracks, three layers of
rebar are placed near the upstream and downstream faces with a stream
spacing of 300 mm. As shown in Fig. 16.4(c), the major reinforced steel is
36 mm in diameter and parallel to the dam slope with a cross-stream
spacing of 200 mm. Therefore, the ratio of the major reinforcement to
concrete is 0.025% for the monolith. The horizontal reinforcement, used as
a constructional steel bar, is 20 mm in diameter with a vertical spacing of
500 mm. Herein, the horizontal steel is ignored in the simulation because
it is parallel to the cracking surface and has little influence on the seismic
response of the dam. The method of reinforcement embedment is omitted
here because it can be found in the work of Long et al. [17] and Kwak and
Filippou [20,21].
Figure 16.5 shows the constitutive relations of the concrete, reinforced
steel and steeleconcrete interface, each with its own material constants. In
particular, only tensile nonlinearities are modeled for the concrete and
reinforced steel because the linear analysis shows that the compressive
stresses are much less than their compressive strength.
After imposing the two static loads of gravity and hydrostatic pressure,
the design earthquake motions are input at the truncated boundary of the
foundation to conduct seismic analyses. The acceleration components of
the design earthquake and their response spectra are shown in Fig. 16.6.
In addition, the damereservoir interaction is modeled using the added
mass at the upstream face nodal points, and the energy dissipation of the
monolith is considered by the Rayleigh damping method with 5%
damping ratio. The traditional massless foundation approach is used
herein to account for the damefoundation interaction. Although this
model overestimates the dam response because it ignores the energy
dissipation resulting from the radiation damping at the far field, it is still
an important method for the seismic safety evaluation of concrete dams.
From an engineering point of view, the benefits due to radiation damping
are preferable to be used as the safety margin of a concrete dam. The
elastic modulus of the rock is 19.5 GPa in the following analyses.
which indicates that concrete cracking may initiate at the two slope-
change points near the upstream face.
16.4 CONCLUSIONS
Nonlinear seismic analyses of a high gravity dam with and without the
presence of reinforcement are conducted to investigate its dynamic
damage response and evaluate the effectiveness of the strengthening
measure. The analyses account for several nonlinearities such as the
tensile cracking, the plastic offset strain of the cracked concrete in tension,
the stiffness recovery behavior when the crack closes and the bondeslip
effect. Based on the comparison of these results, it is concluded that:
1. The stiffness recovery constant s0 has a significant influence on the
seismic response of the Jin’anqiao dam, e.g., the stream displacement
history, residual deformation and cracking patterns. Residual
displacement will occur in the dam owing to the combined effects of
tensile plastic offset strain and stiffness recovery existing in the
concrete. To accurately simulate the dam response, cyclic tests of the
mass concrete should be conducted to calibrate the stiffness recovery
constant and to obtain the rules of the tensile plastic offset strain.
2. There is little difference in the dam response calculated from the
reinforcement analyses with and without bondeslip. Therefore, the
modified embedded-steel model used herein is applicable to
evaluating the effectiveness of the reinforcement strengthening.
3. Although the current strengthening plan has a slight effect on
decreasing the seismic deformation of the dam, the reinforcement can
reduce the maximum opening and extension of the major crack. What
is more, it is beneficial in maintaining the integrity of the dam and
preventing the upper part from destabilizing.
4. The presence of the secondary cracks resulting from the reinforcement
has some drawbacks from a structural point of view, e.g., uplift pressure
may develop on the cracking surfaces and cause dam instability.
However, this can be solved by impermeable rehabilitation after the
earthquake.
Although the reinforcement is beneficial in improving the seismic-
resistant capacity of the Jin’anqiao dam, the effect of the current strength-
ening plan seems insufficient from an engineering point of view. Therefore,
further studies should be conducted to investigate the effect of the
integrated earthquake-resistant measures by combining the reinforcement
with other techniques, e.g., crack-inducing joints and crest dampers.
Acknowledgments
Financial support for this study was provided by the National Natural Science Foundation of
China (grant nos 90510018 and 90715041) and the National Basic Research Program of China
(973) (grant no. 2007CB714100). This support is gratefully acknowledged. The authors would
also like to thank Drs Jin Feng and Zhou Yuande for their suggestions in the study.
References
[1] Ayari ML, Saouma VE. Fracture mechanics based seismic analysis of concrete gravity
dams using discrete cracks. Eng Fract Mech 1990;35(1e3):587e98.
[2] Javanmardi F, Leger P, Tinawi R. Seismic structural stability of concrete gravity dams
considering transient uplift pressures in cracks. Eng Struct 2005;27(4):616e28.