Professional Documents
Culture Documents
www.elsevier.com/locate/engstruct
Abstract
The seismic behavior of concrete gravity dams with cracks that penetrate the monoliths has rarely been studied. The treatment of dynamic
contact conditions at the cracks is the major difficulty. Not only all the sliding, rocking and drifting modes of motion have to be properly handled,
but also specific attention should be focused on impact when the crack is closing. This study employs the FEM and adopts the IDCE (incremental
displacement constraint equations) model to deal with all the modes of motion along the cracks. For impact, equivalent damping is introduced
based on the concept of coefficient of restitution that is used in collision between point masses. After the IDCE damping model is verified in
dynamic contact situations for rigid and flexible bodies, a typical concrete gravity dam with full reservoir water and three cases of crack is
investigated under El Centro 1940 NS earthquake excitation. Calculations reveal very interesting phenomena, such as occurrence of rocking and
jumping, coupling between the direction of residual sliding and the direction of peak rocking, and the large damping effect of multi-cracks on
peak residual sliding, rocking and jumping.
c 2006 Elsevier Ltd. All rights reserved.
Keywords: Seismic behavior; Concrete gravity dams; FEM; Cracks; Impact damping
1. Introduction
Due to previous earthquakes or for other reasons, concrete
gravity dams normally have cracks in practical service. Apart
from those cracks with limited depth that exist on both
the upstream and downstream faces and which will possibly
develop under static or dynamic conditions, some cracks may
have developed so much that they almost penetrate, or have
already penetrated, through the monoliths. In all these cases,
the concrete gravity dams will behave non-linearly under
earthquake conditions.
Non-linear response of cracked concrete gravity dams has
been of great interest in engineering. Many studies [15]
focused on the propagation of cracks in the dam, which is
accompanied by opening and closing of the cracks. However,
in the case of penetrated cracks the non-linearity in seismic
response becomes more complex, since the dam is no longer
a structure but a system of blocks separated by one or more
Corresponding author. Tel.: +1 514 848 2424x7809; fax: +1 514 848 7965.
337
338
C=
ci .
(4)
as [17]
s
C = 2 ln r
M1 M2 K
(M1 + M2 )[ 2 + (ln r )2 ]
(5)
kn,i
C.
K
(6)
339
Fig. 4. Energies in square steel block: (a) r = 0.3; (b) r = 0.5; and (c) r = 1.0.
according to Eqs. (5) and (6), each of the dampers will have a
damping coefficient given by
If bodies that come into contact are flexible, the energy of the
system always consists of potential, strain energy and kinetic
energy. Regarding the kinetic energy, part of it will transform
to strain energy during impact, which is stored in the contact
elements and in the flexible bodies. Since the equivalent damper
will only dissipate part of the kinetic energy associated with the
IDCE damping contact elements, it is reasonable to expect that
less energy will be dissipated than in the rigid case.
To examine the effect of the IDCE damping contact model
in the flexible case, the rectangular concrete block in Fig. 5 is
studied. Similar to the steel block, it is also placed 0.2 m above
a frictionless rigid surface and then dropped without initial
velocity. The concrete block is discretized with a 10 40 mesh.
Each of the normal contact stiffnesses of the IDCE dampers
takes the value of the vertical stiffness at the corresponding
node, which is 1.5 1010 N/m at the corners and 3.0
1010 N/m at the interior nodes. Taking M1 = 20,000 kg for the
concrete block and M2 = for the rigid surface, respectively,
the equivalent damping coefficients are
c= p
7746
(/ ln r )2 + 1
kN s/m
(at corners)
(8)
kN s/m
(9)
and
c= p
386
(/ ln r )2 + 1
kN s/m.
(7)
c= p
15,492
(/ ln r )2 + 1
340
Fig. 8. Rocking of rectangular steel block with initial angle of 0.1 rad.
Fig. 7. Energies in concrete block: (a) r = 0.0; (b) r = 0.5; and (c) r = 1.0.
341
The flexible block keeps jumping after the first impact, which
is indicated by straight lines in the Fig. 10. This suggests that it
is more possible for a flexible block than a rigid block to enter
the free flight mode of motion when impact occurs.
4. Seismic behavior of typical cracked concrete gravity dam
To reveal the seismic behavior of cracked concrete gravity
dams, especially with cracks at both the base and a height, a
typical dam 60 m high with full reservoir water is studied for
three cases of crack, i.e., (A) at the base; (B) at 48 m height,
where the cross section of the dam changes abruptly; and (C) at
both the base and the height of 48 m. A 1.0 m thickness of
the dam is considered. The FEM mesh for case A is shown
in Fig. 11, in which the IDCE dampers along the crack are
also depicted. A relatively high coefficient of friction of 1.0
is assumed for all the cases to consider the effective interlock
in the cracks. The El Centro 1940 NS earthquake record is
applied with scaled peak accelerations Pa . Westergaard virtual
mass [18] is employed to include the dynamic effect of the
water. For comparison, both the purely elastic impact (r = 1)
and partially plastic impact (r = 0.5) are calculated. The time
step for all the cases is 1 105 s.
342
Table 1
Sliding and rocking of the typical dam cracked at the base
Pa (g)
0.5
0.6
0.7
0.8
0.9
1.0
r = 1.0
Residual
sliding (mm)
r = 0.5
Residual
sliding (mm)
Peak upstream
rocking (103 rad)
83.9
114.2
224.3
449.8
489.5
634.1
1.49
2.55
2.70
4.28
4.82
5.75
0.58
1.08
1.49
1.88
1.87
1.69
108.1
66.9
135.0
278.0
351.0
248.2
2.66
3.56
4.63
4.94
7.52
8.11
1.45
1.82
2.33
2.52
3.51
3.25
Table 2
Sliding and rocking of the typical dam cracked at the height
Pa (g)
0.5
0.6
0.7
0.8
0.9
1.0
r = 1.0
Residual
sliding (mm)
r = 0.5
Residual
sliding (mm)
Peak upstream
rocking (103 rad)
220.2
545.1
308.0
339.4
1809.6
574.5
8.49
37.45
17.62
5.12
50.69
14.149
3.83
2.17
13.85
114.89
33.51
135.05
136.7
204.0
787.6
155.6
740.9
1670.1
7.86
9.29
7.93
16.57
42.93
22.01
4.03
13.96
24.42
39.52
48.02
41.95
Fig. 12. Drifting of the typical dam cracked at the base induced by scaled El
Centro 1940 NS with r = 0.5.
343
0.5
0.6
0.7
0.8
0.9
1.0
r = 1.0
Residual
sliding (mm)
r = 0.5
Residual
sliding (mm)
Peak upstream
rocking (103 rad)
387.1
689.3
130.4
369.3
1165.4
1867.9
11.86
18.76
33.44
27.88
60.66
37.76
18.72
7.91
62.60
20.10
18.93
28.98
204.2
293.8
398.1
851.5
645.3
47.0
8.50
16.89
19.49
33.73
38.83
34.01
25.74
24.88
14.01
20.49
61.39
68.11
Fig. 14. Comparison of jumping height between two single crack cases with
r = 0.5.
the crack at the base, is observed again in this case for the crack
at the height. For all cases listed in Table 2, no matter in which
direction the peak rocking is larger, the residual sliding will be
in the opposite direction.
4.4. Cracked at both the base and the height
Fig. 13. Configuration of the typical dam cracked at the height under El Centro
1940 NS with Pa = 1.0g.
344
Table 4
Sliding and rocking of the lower portion of the typical dam for multi-crack case
Pa (g)
0.5
0.6
0.7
0.8
0.9
1.0
r = 1.0
Residual
sliding (mm)
r = 0.5
Residual
sliding (mm)
Peak upstream
rocking (103 rad)
8.8
3.0
4.5
33.6
31.89
51.24
1.62
1.58
2.03
1.96
2.97
5.13
0.79
1.12
1.29
1.44
1.59
2.03
1.2
0.8
3.6
27.0
20.4
57.4
0.87
1.65
2.20
2.21
5.15
4.56
0.64
1.08
1.25
1.54
2.70
2.01
Fig. 15. Comparison of residual base sliding between single crack case at the
base and multi-crack case with r = 0.5.
Fig. 16. Comparison of base jumping between single crack case at the base and
multi-crack case with r = 0.5.
jumping of the dam above the crack are observed for most of
the studied cases. The residual sliding at the base is generally in
the downstream direction, whereas it could be in the upstream
direction for a crack at the height. The amounts of sliding,
rocking and jumping are non-linear with respect to the peak
acceleration of the earthquake.
The direction of peak rocking is coupled with the direction
of the residual sliding in single crack cases as well as for
base sliding in the multi-crack case. In all these cases, the
downstream residual sliding is associated with larger peak
rocking in the upstream direction, and vice versa. Therefore
rocking motion is of importance for overall sliding when a high
coefficient of friction is adopted, even if the angle may be small,
as seen in many of the studied cases.
Jumping could happen even with low peak acceleration as of
0.5g. For the same Pa , the jumping height at the upper crack is
much higher than at the base crack, being possibly more than
200 mm for elastic impact or 100 mm for r = 0.5. Because
of jumping, the cracked dam may experience drifting in the
upstream or downstream direction, depending on the details of
the earthquake. This enhances the non-linearity of the seismic
response, especially for strong earthquakes.
In the multi-crack case, both the residual sliding and peak
rocking of the dam at the base are much reduced compared
with the case of single crack at the base, owing to the large
damping effect of the upper crack. Regarding the upper crack,
the residual sliding and peak rocking show high non-linearity
and randomness in both the amount and direction. However,
they have the same order of magnitude as in the case of a single
crack at the height.
5. Conclusions
Equivalent damping is introduced through the coefficient
of restitution to consider the plasticity during impact in the
seismic analysis of concrete gravity dams with penetrated
cracks. The verifications show that the impact damping can
properly dissipate energy for rigid and flexible bodies in rightimpact cases as well as under rocking conditions.
The seismic analysis of the typical cracked concrete gravity
dam indicates that the impact damping is important in the
estimation of the seismic response. With impact damping
equivalent to coefficient of restitution of 0.5, the amount of
345