EARTHQUAKE ENGINEERlNG AND STRUCTURAL DYNAMICS, VOL.

13, 13-31 (1985)

EFFECTS OF RESERVOIR BOTTOM ABSORPTION AND

DAM-WATER-FOUNDATION ROCK INTERACTION ON
FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE
GRAVITY DAMS

GREGORY FENVES* AND ANIL K . CHOPRAT

Department of Civil Engineering, University of California, Berkeley, California, U.S.A.

SUMMARY
The linear response of an idealized concrete gravity dam monolith to harmonic horizontal or vertical ground motion is
presented for a range of the important system parameters that characterize the properties of the dam, foundation rock,
impounded water and reservoir bottom materials. Based on these frequency response functions, the effects of alluvium
and sediments at the reservoir bottom on the response of the dam, including its interaction with the impounded water
and foundation rock, are investigated. It is shown that the partial absorption of hydrodynamic pressure waves by the
reservoir bottom materials has an important effect on the dynamic response of concrete gravity dams.

INTRODUCTION
Previous investigations have shown that the response of concrete gravity dams to earthquake ground motion
is affected by: interaction between the dam and impounded water,', compressibility of the impounded
water,' and interaction between the dam and flexible foundation rocka3Considering only the fundamental
vibration mode, it was recently demonstrated that the sediments invariably present at the bottom of actual
reservoirs can have a significant effect on the earthquake response of dams4 The alluvium and sediments at
the reservoir bottom were approximately modelled by a reservoir bottom that partially absorbs incident
hydrodynamic pressure waves. The absorptive reservoir bottom was shown to provide an important energy
radiation mechanism through refraction of pressure waves into the foundation medium. An analytical
procedure, based on the substructure method, was developed and implemented in a computer program5 to
evaluate the earthquake response of concrete gravity dams, considering all the significant natural vibration
modes, and including the effects of reservoir bottom absorption and dam-water-foundation rock interaction.
Presented in this paper is the dynamic response of an idealized concrete gravity dam monolith obtained
from this analytical procedure5 for a wide range of the important system parameters that characterize the
properties of the dam, foundation rock, impounded water and reservoir bottom materials. The response of
the idealized dam to harmonic horizontal or vertical ground motion is presented in the form of frequency
response functions. Based on these response results, the effects of reservoir bottom absorption on the
response of the dam, including its interaction with the impounded water and foundation rock, are
investigated.

SYSTEM, CASES ANALYSED AND RESPONSE QUANTITIES

Dam-water-foundation rock system
The idealized monolith considered as representative of concrete gravity dams has a triangular cross-
section with a vertical upstream face and a downstream face slope of 0.8 to 1. The dam is assumed to be
* Currently Assistant Professor of Civil Engineering, University of Texas, Austin, Texas.
t Professor of Civil Engineering.
oO98-8847/85/0l0013-19$01.90 Received 19 December 1983
@ 1985 by John Wiley & Sons, Ltd. Revised 25 M a y 1984
14 G . FENVES AND A. K. CHOPRA

homogeneous and isotropic with linear elastic properties for the mass concrete: Poissons ratio = 02, unit
weight = 155 lb/ft, and the Youngs modulus E , is varied over a practical range, E , = 2,4 and 5 million psi.
Energy dissipation in the dam is represented by constant hysteretic damping with a hysteretic damping factor
of qs = 010. This value corresponds to a viscous damping ratio of 0.05 in all natural vibration modes of the
dam on rigid foundation rock with an empty reservoir.
The finite element idealization of the dam monolith, shown in Figure 1, consists of twenty quadrilateral
elements and twenty-six nodal points. This idealization has forty-two degrees of freedom if the foundation
rock is assumed to be rigid, and fifty-two degrees of freedom if foundation rock flexibility is considered.

NODAL POINT NO.

ELEMENT NO.

Figure 1. Finite element idealization of the dam monolith

The dam monolith is supported on the surface of foundation rock idealized by a homogeneous, isotropic,
viscoelastic half-plane. The material properties of the foundation rock are: Poissons ratio = 1/3; unit
weight = 165 lb/ft3; and the Youngs modulus E, is varied to cover a wide range of foundation materials. In
particular, EJE, ratios of 00 (rigid foundation rock), 2, 1 and 1/4 are considered in the analyses presented
here. Energy dissipation in the flexible foundation rock is represented by constant hysteretic damping with a
hysteretic damping factor of qf = 0.10.
The impounded water in the reservoir is idealized by a fluid domain of constant depth H extending to
infinity in the upstream direction. The unit weight of water is 62.4 lb/ft3, and the velocity of pressure waves in
water is C = 4,72Oft/s. Two values for the depth of the water are considered: an empty reservoir (H/& = 0)
and a fuli reservoir ( H / H , = l), where H,is the height of the upstream face of the dam. The natural vibration
frequencies CL$ of the impounded water aid in interpreting the response of dams to harmonic ground motion.
For a rigid reservoir bottom, they are given by w: = (2n - 1) nC/2H, n = 1,2, ....4
The dam and foundation rock are assumed to be in a state of generalized plane stress. This assumption,
though not strictly appropriate for the foundation rock, is dictated by the expected behaviour of the non-
keyed joints between the dam monoliths.

Absorptive reservoir bottom

The bottom of a reservoir upstream of a dam may consist of highly variable layers of exposed bedrock,
alluvium, silt and other sediments. The effects of these reservoir bottom materials are approximately
modelled by a reservoir bottom that partially absorbs incident hydrodynamic pressure waves. Reservoir
bottom absorption is characterized by the wave reflection coefficient u, defined as the ratio of the amplitude
of the reflected hydrodynamic pressure wave to the amplitude of a vertically propagating pressure wave
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 15

incident on the reservoir bottom, and is given by the following expression:6-

1 -qc
a=-
1+qc
where q = p/p, C,, C is the velocity of pressure waves in water, p is the density of water, C , = ,/(E,/p,), E , is
the Youngs modulus and p , is the density of the reservoir bottom materials. The wave reflection coefficient CI
may range within the limiting values of 1 and - 1. For rigid reservoir bottom materials, C , = co and q = 0,
resulting in CI = 1. For very soft reservoir bottom materials, C , approaches zero and q = co, resulting in
a = - 1. It is believed that a values from 1 to 0 would cover the wide range of materials encountered at the
bottom of actual reservoirs.
The principal objective of this study is to investigate effects of reservoir bottom absorption on the
earthquake response of dams. For this reason, the wave reflection coefficient a is varied over a wide range. Y . 7

independently of the foundation rock properties. The values considered are: a = 1.0 (rigid reservoir bottom),
0.75, 0.50 and 0.

Cases analysed
The responses of the several dam-water-foundation rock systems shown in Table I are presented. These
systems are defined by the chosen values for the important system parameters, E,, E,/E,, H / H , and a. The
responses of the various systems and their interpretation are organized to elucidate the effects of reservoir
bottom absorption on dam-water-foundation rock interaction, and ultimately on dam response.

Table I. Cases of idealized dam-water-foundation rock system analysed

Dam Foundation rock Impounded water Reservoir bottom

Es
Case (million psi) Condition EJE, Condition HIH, Condition U

any* rigid none 0 - -

1 00
2 4 rigid 00 full 1 rigid 1.o
3 4 rigid a2 full 1 absorptive 0.75
4 4 rigid co full 1 absorptive 0.5
5 4 rigid 00 full 1 absorptive 0
6 5 rigid a2 full 1 rigid 1.o
7 5 rigid co full 1 absorptive 0.5
8 2 rigid a2 full 1 rigid 1.o
9 2 rigid 00 full 1 absorptive 0.5
10 any* rigid co full, incompressible 1 rigid any?
any* flexible 1 none 0 - -
11
12 4 flexible 1 full 1 rigid 1.o
13 4 flexible 1 full 1 absorptive 0.75
14 4 flexible 1 full 1 absorptive 0.5
15 4 flexible 1 full 1 absorptive 0
16 4 flexible 2 full 1 rigid 1.o
17 4 flexible 2 full 1 absorptive 0.5
18 4 flexible 114 full 1 rigid 1.o
19 4 flexible 114 full 1 absorptive 0.5

* Response results for these cases, when presented in normalized form, are valid for all E,.
t Response results for the case neglecting water compressibility are independent of 8.

Response quantities
The complex-valued frequency response functions presented here are dimensionless response factors that
represent the acceleration at the dam crest due to unit harmonic free-field ground acceleration. Additionally,
when dam-foundation rock interaction is included, frequency response functions for the horizontal
acceleration at the centre of the base of the dam and vertical acceleration at the upstream edge of the base are
16 G . FENVES AND A. K. CHOPRA

presented. The frequency response functions are for acceleration relative to the free-field ground motion; they
are nor direct measures of deformation.
These functions, describing the response to harmonic horizontal or vertical ground motion. were
computed using the analytical procedure described in Reference 5 with the excitation frequency w varied over
a relevant range of interest. In computing the response, five generalized co-ordinates were included in the
analytical procedure if the foundation rock was assumed to be rigid, and ten generalized co-ordinates were
included if dam-foundation rock interaction arising from foundation rock flexibility was considered, With
these generalized co-ordinates, the resulting frequency response functions essentially converged for excitation
frequencies up to approximately five times the fundamental natural frequency w 1 of the dam or1 rigid
foundation rock with an empty reservoir.
For each case in Table I the absolute value (or modulus) of the complex-valued frequency response
function for acceleration is plotted against the normalized excitation frequency parameter w / w l . When
presented in this form, the response results apply to dams of any height with the idealized triangular cross-
section and chosen values for Poisson's ratio, E,, E,/E,, H/H,and a. Furthermore, these response results are
independent of E , and a if compressibility of the impounded water is neglected, or the reservoir is empty.'

DAM-WATER INTERACTION EFFECTS

EfSeects ojreservoir bottom ubsorption
The effects of interaction between the dam and the impounded water, including reservoir bottom
absorption, on the dam response to horizontal or vertical ground motion are shown in Figure 2, where the
results from the analyses of Cases 1 to 5 (Table I) are plotted. The response of the dam with an empty
reservoir (Case 1) is characteristic of a multi-degree-of-freedom system with frequency-independent mass,
stiffness and damping properties. The response of the dam with a full reservoir (Cases 2 to 5), however, is
affected by frequency-dependent hydrodynamic terms in the equations of motion for the dam. The
hydrodynamic terms can be interpreted as modifying the properties of the dam by introducing an added force
(different for horizontal and vertical ground motion), an added mass, and an added damping, all of which
depend on the wave reflection coefficient a s 5
The response of the dam with a full reservoir and rigid reservoir bottom (Case 2), described in detaiI in an
earlier work,3 is summarized here for convenient reference. The response is especially complicated for
excitation frequencies near o1 and w:, the fundamental natural vibration frequencies of the dam and
impounded water, respectively, where a double resonant peak is observed. The fundamental resonant
frequency i& of the dam is reduced below both w1 and w; by the added mass of the water. Dam-water
interaction leads to a larger fundamental resonant peak over a narrower frequency bandwidth because of the
added force, and reduction in the effective damping ratio indirectly due to the added mass, and because of
interaction between the two uncoupled frequencies w1 and w ; . The response of the dam-water system has
unusual characteristics when pressure waves are completely reflected at the reservoir bottom (c1 = l), because
at excitation frequencies equal to w;, the natural vibration frequencies of the impounded water, the added
force and added mass are infinite. The dam response to harmonic vertical ground motion at these excitation
frequencies is unbounded because the added force for vertical ground motion approaches infinity faster than
the added mass as w approaches w;. In contrast, the response to horizontal ground motion is bounded
because the added force for horizontal ground motion and the added mass approach infinity at the same rate
as w approaches w i . These limiting values appear as local dips or increases in the response function for
horizontal ground motion. Dam-water interaction somewhat decreases the frequencies of the higher
resonant peaks, but not as much as the frequency of the fundamental resonant peak. The amplitudes of the
higher resonant peaks of the dam with a full reservoir are much smaller than those for the dam with an empty
reservoir because, for excitation frequencies greater than wi, the energy radiation due to propagation of'
hydrodynamic pressure waves in the upstream direction results in large added damping.
The effects of reservoir bottom absorption on the response of the dam with full reservoir (Cases 3 to 5) are
shown in Figure 2. With increasing wave absorption at the reservoir bottom, i.e. decreasing wave reflection
coefficient a, the fundamental resonant peak due to horizontal ground motion decreases in amplitude while
 FREQUENCY RESPONSE FUNCTlONS FOR CONCRETE GRAVITY DAMS 17

30
H O R I Z O N T A L GROUND MOTION

CURVE (I WATER
I 100
2 0.75 FULL
3 050
4 0
5 - NONE

H/H, = I I I I I I

w;/ w , w;/wl w;/wl w:/wl

Figure 2. Hydrodynamic effects in response of dams to harmonic ground motion. Results presented for full reservoir with varying values
of the wave reflection coefficient a (Cases 2, 3, 4 and 5 of Table I), and for no water (Case 1)

the second, smaller peak increases, resulting in a single fundamental resonant peak at an intermediate
frequency value. Reservoir bottom absorption reduces the added mass and added forces for both ground
motion components to finite values when the excitation frequency equals o;. Consequently, at this excitation
frequency the unbounded response to vertical ground motion is eliminated, along with the dip in the
response function for horizontal ground motion. In general, reservoir bottom absorption smoothes the
frequency response function near the fundamental resonant peak, reduces its amplitude and widens the
frequency bandwidth. These effects of reservoir bottom absorption on the response for excitation frequencies
up to and slightly beyond the fundamental resonant frequency, observed from the computed response
including the contributions of several (five) vibration modes of the dam, are consistent with the conclusions
reached in Reference 4 from the response considering only the fundamental vibration mode of the dam.
Because the contributions of several vibration modes are included in the response results of Figure 2, the
effects of reservoir bottom absorption at higher excitation frequencies can now be discussed. Reservoir
bottom absorption has little effect on the response to horizontal or vertical ground motion for excitation
frequencies greater than w;+xcept locally near a:.This is because the energy radiation due to refraction of
hydrodynamic pressure waves into the absorptive reservoir bottom is small, compared to the energy
radiation at excitation frequencies greater than w; due to pressure waves propagating in the upstream
direction. For excitation frequencies equal to of reservoir bottom absorption eliminates the unbounded
18 G . FENVES AND A. K. CHOPRA

response values due to vertical ground motion and the dips in the response function for horizontal ground
motion.
Thus, reservoir bottom absorption primarily affects the dam response for excitation frequencies less than
01, where material damping in the dam concrete is normally the only energy dissipation mechanism present
if the foundation rock is assumed to be rigid. At higher frequencies the upstream radiation of energy
dominates the energy radiation into the absorptive reservoir bottom, essentially eliminating its effect.

Influence of Youngs modulus E ,

The frequency response function for the dam, when presented in dimensionless form, is independent of the
Youngs modulus E , for the dam concrete if there is no impounded water or its compressibility is neglected.
The frequency response functions for Cases 2,4,6, 7, 8 and 9 due to horizontal and vertical ground motion,
presented in Figures 3 and 4, respectively, demonstrate that the E , value affects the response when water
compressibility is included. Most affected is the fundamental resonant frequency Gr and the response in the
neighbourhood of this frequency, and the response of the dam with rigid reservoir bottom due to vertical
ground motion at excitation frequencies close to wi-because of the response singularities discussed earlier.
As E , increases, the normalized fundamental resonant frequency &/u,of the dam decreases due to dam-
water interaction. This decrease in resonant frequency also depends on reservoir bottom absorption, being
less pronounced for a wave absorptive reservoir bottom than for a rigid reservoir bottom [compare Figure
3(a) to (b), and Figure 4(a) to (b)].

30
I 112
RIGID RESERVOIR BOTTOM, a = 1.0

ES
+
v)
CURVE iMlLLlON PSI 1 WATER

w
4
$ 20
COMPRESSIBLE

I ANY INCOMPRESSIBLE
a
n
5
z
0 10
5
IY
2
W
J
W
V
0
Q O I I I I I
J
2 30
z ABSORPTIVE RESERVOIR BOTTOM. a =0.5
0
4
N :1

[I: i4 !I
0
I ii

o i
0
I
I 2
I I
3
I
4
I
5 6
w/w,
Figure 3. Influence of Youngs modulus E , of dam concrete on response of dams with full reservoir to harmonic horizontal ground
motion. Results presented for rigid reservoir bottom (Cases 2,6 and 8 of Table I), absorptive reservoir bottom (Cases 4, 7 and 9) and
incompressible water (Case 10)
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 19

30

I-
v)
W
Lz
v 20
I
U
n
G
g
Z
I0

Lz
W
-1
W
V
W
a 0
-1
2 30
z ABSORPTIVE RESERVOIR BOTTOM, a = 0.5
0
EJ E S
Lz CURVE I M I L L I O N PSI 1 WATER
0
I I 5
b 20 2
3
4
2
COMPRESSIBLE

W
3 4 ANY INCOMPRESSIBLE
-1
3 4
W
I-
3 10
-1
0
v)
m
a

0
0 I 2 3 4 5
W/W,
Figure 4.Influence of Youngs modulus E , of dam concrete on response of dams with full reservoir to harmonic vertical ground motion.
Results presented for rigid reservoir bottom (Cases 2, 6 and 8 of Table I), absorptive reservoir bottom (Cases 4, 7 and 9) and
incompressible water (Case 10)

Reservoir bottom absorption affects the amplitude and frequency bandwidth of the fundamental resonant
peak in an especially significant way. The response to horizontal ground motion shows that increasing E ,
causes larger resonant response over a narrower bandwidth for a rigid reservoir bottom [Figure 3(a)], but
causes smaller resonant response over a wider bandwidth for a wave absorptive reservoir bottom [Figure
3(b)]. This opposite response behaviour results from the manner in which the effective damping at resonance
is influenced by reservoir bottom a b ~ o r p t i o n .If~ the reservoir bottom is rigid, dam-water interaction
provides no added damping at the fundamental resonant frequency, because pressure waves cannot refract
through the rigid bottom, nor do they propagate in the upstream direction at this frequency because it is less
than w;. Explained in Reference 4 is the observation that, as the frequency ratio w ; / w l decreases (due to
increasing E,, for example), the frequency-dependent added mass evaluated at the fundamental resonant
frequency increases, as does the added force. Because of the assumed constant hysteretic damping in the dam
concrete, the added mass does not affect the overall damping at the fundamental resonant frequency, but the
increased added force leads to larger resonant response. If reservoir bottom absorption is included, however,
there is added damping at the fundamental resonant frequency due to refraction of pressure waves through
the absorptive reservoir bottom and propagation of pressure waves upstream. Also explained in Reference 4
is the observation that, as the ratio w ; / w , decreases, more natural vibration modes of the impounded water
contribute to radiation of energy upstream at the fundamental resonant frequency, increasing the added
damping while the magnitude of the added force does not increase very much; and the resonant response is
reduced because of the increase in the effective damping ratio.
20 G . FENVES AND A. K. CHOPRA

Figure 4 indicates different trends for the response of the dam to vertical ground motion. If the reservoir
bottom is rigid, increasing E , causes a dramatic increase in the resonant response [Figure 4(a)] because the
resonant frequency of the dam-water system approaches co: where, as previously mentioned, the response is
unbounded. If reservoir bottom absorption is included, however, the amplitude of the resonant response is
affected little by the value of E , [Figure 4(b)], in the range of E , values considered. This is because as E ,
increases, the decrease in resonant response amplitude due to the increasing added damping, discussed earlier
for horizontal ground motion, is offset by the increase in response amplitude due to the increasing added
force for vertical ground motion.
Comparison of Figures 3(a) and 4(a) indicates that, if the reservoir bottom is rigid, E , has little effect on the
resonant response to horizontal ground motion, but has substantial influence on the response to vertical
ground motion. As displayed by Figures 3(b) and 4(b), the comparison is reversed for a wave absorptive
reservoir bottom, i.e. E , has considerable effect on the resonant response to horizontal ground motion, but
little influence on the response to vertical ground motion. As seen above, the value of E, can significantly
affect the fundamental resonant frequency and the response function for excitation frequencies near it. For
larger excitation frequencies, however, the frequency response functions for both components of ground
motion are less affected by the E , value, irrespective of whether the reservoir bottom is rigid or absorptive.

Effects qj' wuter compressibility

To understand how compressibility of the impounded water affects the dam with a full reservoir, the
response obtained by neglecting water compressibility (Case 10) is also plotted in Figures 3 and 4. If water
compressibility is neglected dam-water interaction results in frequency-independent, real-valued added force
and added mass; there is no added damping, With decreasing E,, the effects of water compressibility on the
fundamental resonant response become smaller. Although this trend is straightforward for horizontal ground
motion (Figure 3), it is more complicated for the response to vertical ground motion (Figure 4) because of the
unbounded response peaks that exist at excitation frequencies equal to 0;if the reservoir bottom is rigid.
Contrary to an earlier recommendation based on experiments conducted in Japan,8 in the range of E,
values encountered in concrete gravity dams, the effects of wave absorption at the reservoir bottom on dam
response are not properly represented by an analysis that neglects water compressibility [Figures 3(b) and
4(b)]. Although such an analysis provides a good approximation to the fundamental resonant frequency Gr,
the fundamental resonant response to horizontal ground motion is overestimated because incompressible
water does not allow for radiation of energy upstream or through the reservoir bottom; and the amplitudes of
the higher resonant peaks are overestimated by even a greater margin. Therefore, water compressibility
should be considered in the earthquake analysis of concrete gravity dams.

Comparison of responses to horizontal and vertical ground motion components

Comparing the response of the dam with an empty reservoir to horizontal and vertical ground motion
(Figure 2) it is apparent-consistent with common view-that the fundamental resonant response to vertical
ground motion is relatively small. The presence of impounded water, however, leads to considerably different
response depending upon the component of ground motion. If the reservoir bottom is rigid, dam--water
interaction increases the fundamental response of the dam to both ground motion components. If water
compressibility is included, the increase in response to vertical ground motion is greater because the added
hydrodynamic force is large compared to the relatively small effective earthquake force associated with the
mass of the dame4 Reservoir bottom absorption somewhat reduces the ratio of fundamental resonant
response to vertical ground motion compared to the resonant response to horizontal ground m ~ t i o n In .~
general though, the fundamental resonant response to vertical ground motion is more important if
hydrodynamic effects are included with either a rigid or an absorptive reservoir bottom.
In contrast to the preceding observations regarding the fundamental resonant response, the resonant
response to vertical ground motion corresponding to a natural vibration mode with dominantly vertical
displacements is large compared to the resonant response to horizontal ground motion when the reservoir is
empty (Figure 2). Consequently, the response of the dam with full reservoir to vertical ground motion may be
significant at some higher resonant frequencies, whether the reservoir bottom is rigid or absorptive. The
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 21

hydrodynamic effects of a full reservoir reduce the higher resonant peaks due to horizontal and vertical
motion by roughly the same percentage.

DAM-WATER AND DAM-FOUNDATION ROCK INTERACTION EFFECTS

Effects of dam-foundation rock interaction
To introduce the effects of reservoir bottom absorption on the response of dams including interaction
between the dam, impounded water and foundation rock, results are presented in Figures 5 and 6 for four
systems: dam on rigid foundation rock with no water (Case 1); dam on flexible foundation rock with no water
(Case 11); dam on rigid foundation rock with full reservoir (Cases 2 and 4); and dam on flexible foundation
rock with full reservoir (Cases 12 and 14). As noted earlier,3 interaction between the dam and flexible
foundation rock affects the response of the dam in a simpler manner than does dam-water interaction
(compare curve 2 to 3). This is because the impedances of the half-plane idealization for the foundation rock
region are slowly-varying, smooth functions of excitation frequency without resonant frequencies, whereas
the added hydrodynamic force, mass and damping are frequency-dependent functions with peaks at mi,
Dam-foundation rock interaction reduces the fundamental resonant frequency Gfof the dam, and reduces

30
1 3
RIGID RESERVOIR BOTTOM, cI = 1.0

L,
I 1 I CURVE
I
WATER
NONE
FOUNDATION
ROCK
RIGID
W 2 NONE FLEXIBLE
LL
0 20 3 FULL RIGID
4 FULL FLEXIELE
2
a
0
t-
Q
2
0
l-
10
a
[L
W
-J
W
0
C)
p o
J

ABSORPTIVE RESERVOIR BOTTOM, a ~ 0 . 5

LL
P
:: I ,

t; 20
I
8I ,,
I ,

I :I::
W 0 ,

3
-I
2 -

9
W
I-
3 10
0
Q
m
a.

I I I 1
0
0 I 2 3 4 5
w/wl
H/H,=I L I I I I
w;/w, w; / w , w; /wl w; /wl
Figure 5. Response of dams to harmonic horizontal ground motion for four conditions: dam on rigid foundation rock with no water
(Case 1 of Table I); dam on flexible foundation rock with no water (Case 11); dam on rigid foundation rock with full reservoir (Cases 2
and 4); and dam on flexible foundation rock with full reservoir (Cases 12 and 14)
22 G. FENVES AND A. K. CHOPRA

30

z
g
U
10

W
J
W
u
0
4
-I
0
2
Z
30
ABSORPTIVE RESERVOIR BOTTOM, a = 0.5

FOUNDATION
CURVE WATER ROCK
I NONE RIGID
2 NONE FLEXIBLE
3 FULL RIG10
4 FULL FLEXIBLE

"
0 I 2 3 4 5 6
w/w,
H/HS=I I I I I I
w;/w, w; /w, w; /w, w: / w ,
Figure 6. Response of dams to harmonic vertical ground motion for four conditions: dam on rigid foundation rock with no water (Case
1 of Table I); dam on flexible foundation rock with no water (Case 11); dam on rigid foundation rock with full reservoir (Cases 2 and 4);
and dam on flexible foundation rock with full reservoir (Cases 12 and 14)

the amplitude of the fundamental resonant peak and increases the bandwidth at resonance beca.use of
radiation and material damping in the foundation rock region. Similarly, dam-foundation rock interaction
reduces the higher resonant frequencies, although to a lesser degree than the fundamental resonant
frequency, and substantially reduces the amplitude of the higher resonant peaks. As shown earlier,3 for
decreasing values of E,/E,, which for fixed E , means decreasing foundation rock modulus, each resonant
frequency of the dam decreases; the resonant amplitude at each of these frequencies decreases and the
bandwidth at resonance increases, implying an increase in the apparent damping of the structure. Dam-
foundation rock interaction, as also shown earlier,3 affects the response of the dam to horizontal and vertical
ground motions in a similar manner.

Hydrodynamic and reservoir bottom absorption eflects

The effects of dam-water interaction and of dam-foundation rock interaction on the dynamic response of
the dam can be observed from the remaining response functions presented in Figures 5 and 6. The effects of
dam-water interaction on the dam response to either ground motion component are qualitatively similar for
rigid and flexible foundation rock, whether the reservoir bottom is rigid [Figures 5(a) and 6(a)] or absorptive
[Figures 5(b) and 6(b)]. Dam-water interaction leads to almost the same percentage reduction in the
fundamental resonant frequency irrespective of the foundation rock condition. This observation leads to the
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 23

following equation:
bl
- --
--
b, bf
w1 0 1 a1

where o1is the fundamental resonant frequency of the dam alone (without water, supported on rigid
foundation rock), which reduces to br due to dam-water interaction, to b, due to dam-foundation rock
interaction, and to bl due to both types of interaction simultaneously.
The trends in the amplitude of the fundamental resonant peak depend on the contribution of damping
from dam-foundation rock interaction, dam-water interaction and reservoir bottom absorption, and on the
added hydrodynamic forces for horizontal and vertical ground motion. The effects of the various damping
contributions in the response of the dam to horizontal ground motion can be identified from Figure 5. If the
reservoir bottom is rigid [Figure 5(a)], dam-water interaction increases the added force at the fundamental
resonant frequency, resulting in increased fundamental resonant response whether the foundation rock is
rigid or flexible. This increased added force has less influence on the dam response if the foundation rock is
flexible than if the foundation rock is rigid (compare the change from curve 2 to 4 with the change from curve
1 to 3) because of the large radiation damping permitted by foundation rock flexibility. If the effects of
reservoir bottom absorption are included [Figure 5(b)], the opposite trends occur: dam-water interaction
increases the effective damping at the fundamental resonant frequency, resulting in reduced resonant
response whether the foundation rock is rigid or flexible. This increase in effective damping has less influence
on the dam response if the foundation rock is flexible than if it is rigid (compare the change from curve 2 to 4
with the change from curve 1 to 3) also because of the large radiation damping permitted by foundation rock
flexibility.
The amplitude of the fundamental resonant peak due to vertical ground motion (Figure 6) is mainly
affected by the added force, and less by the previously discussed trends in added damping. If the reservoir
bottom is rigid [Figure 6(a)], dam-water interaction dramatically increases the added force at the
fundamental resonant frequency, leading to increased resonant response and a double resonant peak, of
which one peak is unbounded, whether the foundation rock is flexible or rigid. If the reservoir bottom is
absorptive [Figure 6(b)], dam-water interaction introduces added damping, but the fundamental resonant
response still increases because of the large added force for vertical ground motion, whether the foundation
rock is flexible or rigid. The increase in resonant response is less pronounced for flexible foundation rock than
for rigid foundation rock (compare the change from curve 2 to 4 with the change from curve 1 to 3).
The conclusions deduced in the previous section concerning the response to vertical ground motion from
response results for dams supported on rigid foundation rock are confirmed by Figures 5 and 6 in which the
effects of dam-foundation rock interaction are included. The significance of the response of the dam to
vertical ground motion, relative to the response to horizontal ground motion, increases because of
hydrodynamic effects irrespective of whether the foundation rock is rigid or flexible. However, as noted
before, reservoir bottom absorption reduces the relative significance of the response to vertical ground
motion.
The effects of reservoir bottom absorption on the response of the dam, supported on flexible foundation
rock, due to horizontal and vertical ground motion are shown in Figures 7 and 8, respectively. The response
of systems with moduli ratio E J E , = 1 is presented for four values of the wave reflection coefficient: a = 1.0
(rigid reservoir bottom), 0.75,05and 0 (Cases 12, 13, 14 and 15). Reservoir bottom absorption mainly affects
the fundamental resonant peak due to horizontal ground motion (Figure 7), reducing its amplitude as a
decreases with little change in the resonant irequency; and it essentially has no effect on the response at higher
excitation frequencies, an observation noted earlier from the results for rigid foundation rock (Figure 2). The
most pronounced effects of reservoir bottom absorption on the response to vertical ground motion (Figure 8)
are at excitation frequencies near 05 where the unbounded response peaks reduce to bounded values that
decrease with a. The effects are relatively small at excitation frequencies not near mi.
Reservoir bottom absorption has less effect on the response of the dam supported on flexible foundation
rock (Figures 7 and 8) compared to the response of the dam on rigid foundation rock (Figure 2). This is
explained by consideration of the damping due to dam-foundation rock interaction and reservoir bottom
24 G. FENVES AND A. K. CHOPRA

DAM CREST

I
30
HORIZONTAL ACCLN VERTICAL L CCLN

CURVE
__.- a
I 100 30
20 2 075 20 -
3 050
4 0

v)
z
0
;10
a
Lz lot
w I .I
_1
w
V
I
2 0 0
I 2 3 4 5 0 I 2 3 4 5
LL
0
w
3 DAM B A S E
2 2
> H O R I Z O N T A L A C C L N AT CENTER V E R T I C A L A C C L N AT U/S EDGE
W
I-
3
_1
0
(0
m
a
I I

C
I 2 3 4 5 6
w/w, w/w,
I I I I 1 I I I 1 I
w;/wl w; /WI w;/w, w:/wl w:/w, w;/w, w; /w, w:/w,
Figure 7. Influence of wave reflection coefficient a on response of dams on flexible foundation rock with full reservoir to hdrmonic
horizontal ground motion (Cases 12, 13, 14 and 15 of Table I)

absorption. As mentioned in the previous section, where the foundation rock was assumed rigid, the primary
effect of reservoir bottom absorption on dam-water interaction is to increase the effective damping at the
fundamental resonant frequency, where normally no damping exists other than the material damping, in the
dam concrete. If foundation rock flexibility is included, however, substantial damping exists at all excitation
frequencies due to radiation of energy in the foundation rock region, so the additional damping that arises
from reservoir bottom absorption is not as effective in further reducing the response.
The frequency response functions for the horizontal and vertical motion at the base of the dam, relative to
the free-field ground motion, permitted by foundation rock flexibility are presented in Figures 7 and 8.
Reservoir bottom absorption reduces the amplitude of the first resonant peak of motion at the dam base,
with negligible effect at higher excitation frequencies4xcept to eliminate the unbounded response to
vertical ground motion at excitation frequencies w = a$,.

Injuenre of moduti ratio E , : E ,

To understand how reservoir bottom absorption effects are influenced by dam-foundation rock
interaction, the response of the dam with full reservoir to horizontal and vertical ground motion is presented
in Figures 9 and 10, respectively, for E J E , = co,2, 1 and 1/4. Response results are included for dam on rigid
foundation rock with rigid reservoir bottom (Case 2); dam on rigid foundation rock with absorptive reservoir
bottom ( a = 0.5, Case 4); dam on flexible foundation rock with rigid reservoir bottom (Cases 12, 16 and 18);
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 25

DAM CREST
30
HORIZONTAL ACCLN VERTICAL ACCLN
t t
m m
I I
~

100 k k
20

10
I I
W
-I
W .J-
V
0 0
- 0 -
a 0 I 2 3 4 5 6 0 ' I 2 3 4 5 6
LL
0
w DAM BASE
3 ~~ ~ ~

J W
a m HORIZONTAL ACCLN ATCENTER t V E R T I C A L ACCLN AT U/S EDGE
>
w
k
3
-I
I m
t
0
v)
m
a

I 2 3 4 5 0 I 2 3 4 5 6
w/wl w/w,
L I I I I [ I I 1 I

w;/w, w; /w, w; /wl w:/wl w ;/wl w;/w, w ; /w, w:/w,

Figure 8. Influence of wave reflection coefficient ct on response of dams on flexible foundation rock with full reservoir to harmonic
vertical ground motion (Cases 12, 13, 14 and 15 of Table I)

and dam on flexible foundation rock with absorptive reservoir bottom (c1 = 0.5, Cases 14, 17 and 19). The
response functions for the first two systems are repeated in parts (a),(b) and (c) of Figures 9 and 10; they show
the effects of reservoir bottom absorption on the response of the dam with full reservoir on rigid foundation
rock, which were discussed in the previous section (see Figure 2). The remaining curves in Figure 9 for
response to horizontal ground motion show that as the moduli ratio Ef/E, decreases, which for fixed E ,
means increasingly flexible foundation rock, reservoir bottom absorption has less effect on the response of the
dam, especially at higher excitation frequencies. The effects of reservoir bottom absorption are hardly
noticeable at excitation frequencies greater than the fundamental resonant frequency, even if the foundation
rock is rather stiff with Ef/E, = 2 [Figure 9(a)]. Reservoir bottom absorption effects become negligible, even
at the fundamental resonant frequency, if the foundation rock is very flexible, e.g. Ef/E, = 1/4 [Figure 9(c)].
Similarly, as E,/E, decreases, reservoir bottom absorption has less effect on the response to vertical ground
motion, except at excitation frequencies near the natural vibration frequencies w ; of the impounded water
(Figure 10). The reduced importance of reservoir bottom absorption as E,/E, decreases was previously
explained by consideration of the damping contributions from dam-foundation rock interaction, dam-water
interaction and reservoir bottom absorption. In particular, the effects of reservoir bottom absorption are
most significant if the foundation rock is rigid because, except for material damping in the dam, there is no
other damping mechanism at the fundamental resonant frequency of the dam-water system. As the
foundation rock becomes more flexible, more energy radiates through the foundation rock region because of
26 G. F E N W S A N D A. K. CHOPRA

30
FOUNDATION
I -
CURVE ROCK a
I RlGlO 10
2 RIGID 05
__ --
3 FLEXIBLE I 0
4 FLEXIBLE 0 5
20

10
I-
v)
W
[L
V

LL
0
W
3 4 - Cf/Cs' I

sw
1 0

30
I I I I

I-
3
-I
0
v)
m
U
20

10

0
O/WI

W;/W, w; /w, w; / w , wyw,

Figure 9. Effects of reservoir bottom absorption on response of dams with full reservoir to harmonic horizontal ground motion for
various values of the moduli ratio E J E ,
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 27

30
FOUNDATION
CURVE ROCK a
I RIG10 10
2 RIGID 0 5
3 FLEXIBLE I 0
4 FLEXIBLE 0 5
20

10
t-
cn
W
[L
0
I
a
n o
5 30 al
z t
0 I.:
t-
a
n
W

g 20
0
a

t
0 I
Y 10
n
0
I
LL
0
W

'
3
1 0
30
W
t-
3
-J
0
cn
m
a
20
7
10

Figure 10. Effects of reservoir bottom absorption on response of dams with full reservoir to harmonic vertical ground motion for various
values of the moduli ratio E J E ,
28 G . FENVES AND A. K. CHOPRA

dam-foundation rock interaction, so that the additional damping due to reservoir bottom absorption is not
as effective in further reducing the response.
The curves in Figures 9 and 10 are replotted in Figures 11 and 12 to show further the influence of the
moduli ratio E,/E, on the response of the dam. As E J E , decreases, which for a fixed E, means an increasingly
flexible foundation rock, the fundamental resonant frequency decreases, the dam response at this frequency
decreases and the frequency bandwidth at resonance increases. These trends are the same for a rigid reservoir
bottom [Figures 1l(a) and 12(a)] and an absorptive reservoir bottom [Figures 1l(b) and 12(b)]. It is apparent
from Figures 1 1 and 12 that the effects of decreasing moduli ratio E J E , on the fundamental resonant
response of the dam are qualitatively similar, whether the reservoir bottom is rigid or absorptive; but
quantitatively, the relative decrease in amplitude of the fundamental resonant peak depends on the wave
reflection coefficient a, being less pronounced for an absorptive reservoir bottom.

30
RIGID RESERVOIR BOTTOM, a= 1.0
'I
CURVE EtlE,

~m
* 20 2
3
2
1
4 1/4
U
n

2
z
30
0 ABSORPTIVE RESERVOIR BOTTOM, a = 0.5
N

0 I 2 3 4 5 6
w/w,

W;/W, w; /wl w; /wl w; / w l

Figure 11. Influence of moduli ratio E,/E, on response of dams with full reservoir to harmonic horizontal ground motion

As the moduli ratio EJEs decreases, dam-foundation rock interaction introduces increased radiation
damping at the higher resonant peaks, in addition to the damping from hydrodynamic effects, thus reducing
the amplitude of the higher resonant peaks. For relatively flexible foundation rock, e.g. E,/E, = 1/4, the
higher resonant peaks are almost completely suppressed.
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 29

ABSORPTIVE RESERVOIR BOTTOM, a = 0.5

"
0 I 2 3 4 5 6
w/w,
L I I I I
w;/w, w;/w, w: / w , (0: /w I
Figure 12. Influence of moduli ratio E,/E, on response of dams with full reservoir to harmonic vertical ground motion

CONCLUSIONS
Based on the frequency response functions for the horizontal crest acceleration of concrete gravity dams,
presented for a wide range of the basic system parameters, the effects of reservoir bottom materials on dam
response were investigated. It has been shown that the partial absorption of hydrodynamic pressure waves
due to alluvium and sediments at the reservoir bottom may have an important effect on the dynamic
response of dams. Specifically, this investigation leads to the following conclusions:
1. Reservoir bottom absorption primarily affects the dam response for excitation frequencies less than w;,
where material damping in the dam concrete is the only damping mechanism present if the foundation
rock is assumed to be rigid. At higher excitation frequencies the radiation of energy through upstream
propagation of hydrodynamic pressure waves dominates the energy radiation into the absorptive,
reservoir bottom, essentially eliminating its effect. However, because of reservoir bottom absorption, the
response to vertical ground motion at excitation frequencies equal to 0:is bounded.
2. Dam-water interaction with an absorptive reservoir bottom (a < 1) and rigid foundation rock reduces
the fundamental resonant frequency of the dam to a value below wl, but not as much as for a rigid
reservoir bottom (a = I), where the fundamental resonant frequency is reduced to a value below
both w1 and w;. Reservoir bottom absorption smooths the frequency response function near the
fundamental resonant peak, reduces its amplitude and widens its frequency bandwidth.
30 G. FENVES AND A. K. CHOPRA

3. The effects of dam-water interaction on the dam response are qualitatively similar for rigid and flexible
foundation rock, whether the reservoir bottom is rigid or absorptive. The fundamental resonant
frequency decreases because of dam-water interaction and dam-foundation rock interaction with
the two reductions being cumulative if both effects are considered simultaneously. In particular,
dam-water interaction leads to almost the same percentage reduction whether the foundation rock
is rigid or flexible.
4. Reservoir bottom absorption has less effect on the response of a dam supported on flexible foundation
rock than on the response of a dam supported on rigid foundation rock. The effects of reservoir bottom
absorption are most important for E,: E, = 00 (rigid foundation rock); they decrease as the moduli ratio
E , :E , decreases, becoming negligible if the foundation rock is very flexible, e.g. E, : E, = 1 :4.
5. The significance of the dam response to vertical ground motion relative to the response to horizontal
ground motion increases because of hydrodynamic effects, irrespective of whether the foundation rock
is rigid or flexible. However, reservoir bottom absorption reduces the relative significance of the
response to vertical ground motion.
6. Neglecting the compressibility of the impounded water does not properly represent the effects of
reservoir bottom absorption in the response of the dam to horizontal ground motion, although it may
provide a good approximation to the fundamental resonant frequency. Therefore, the compressibility of
the impounded water should be considered in the earthquake analysis of concrete gravity dams.
The results presented herein, and the conclusions from previous have demonstrated that the
response of concrete gravity dams to earthquake ground motion is affected by: interaction between the dam
and impounded water, compressibility of the impounded water, interaction between the dam and flexible
foundation rock, and the materials at the reservoir bottom. All these effects can be efficiently included in
practical analyses of dams utilizing the analytical procedure and computer program reported earlier.5

ACKNOWLEDGEMENTS
This research investigation was supported by Grant CEE-8120308 from the National Science Foundation
for which the authors are grateful.

APPENDIX

velocity of pressure waves in water

= J(EI/PI)
Youngs modulus of the foundation rock
Youngs modulus of the reservoir bottom materials
Youngs modulus of the dam concrete
depth of impounded water
height of the upstream face of the dam
= p / p , Cr, admittance or damping coefficient of the reservoir bottom materials
+
= (1 - qC)/(1 qC), wave reflection coefficient for vertical pressure waves incident on the reservoir
bottom
constant hysteretic damping factor for the foundation rock
constant hysteretic damping factor for the dam concrete
density of water
density of the foundation rock
density of the reservoir bottom materials
harmonic excitation frequency
fundamental resonant frequency of the dam including effects of dam-water interaction and
dam-foundation rock interaction
fundamental natural vibration frequency of the dam on rigid foundation rock without water
 FREQUENCY RESPONSE FUNCTIONS FOR CONCRETE GRAVITY DAMS 31

Gf fundamental resonant frequency of the dam without water including dam-foundation rock
interaction effects
w; = (2n - 1) 71C/2H, the nth natural vibration frequency of the impounded water with rigid reservoir
bottom
Gr fundamental resonant frequency of the dam supported on rigid foundation rock including
dam-water interaction effects

REFERENCES
1. A. K. Chopra, Earthquake behavior of reservoir-dam systems, J . eng. mech. dio. ASCE 94, 1475-1500 (1968).
2. P. Chakrabarti and A. K. Chopra, Hydrodynamic effects in earthquake response of gravity dams, J. struct. d i n A S C E 100, 1211-
1224 (1974).
3. A. K. Chopra and S. Gupta, Hydrodynamic and foundation interaction effects in frequency response functions for concrete gravity
dams, Earthquake eng. struct. dyn. 10, 89-106 (1982).
4. G. Fenves and A. K. Chopra, Effects of reservoir bottom absorption on earthquake response of concrete gravity dams, Earthquake
eng. struct. dyn. 11, 809-829 (1983).
5. G. Fenves and A. K. Chopra, Earthquake analysis of concrete gravity dams including reservoir bottom absorption and dam-water-
foundation rock interaction, Earthquake eng. struct. dyn. 12, 663-680 (1984).
6. E. Rosenblueth, Presion hidrodinamica en presas debida a a1 aceleracion vertical con refraccion en el fondo, 2nd Congreso Nacional
ingenierib sismica, Veracruz, Mexico (1968).
7. J. F. Hall and A. K. Chopra, Two-dimensional analysis of concrete gravity and embankment dams including hydrodynamic effects,
Earthquake eng. struct. dyn. 10, 305-332 (1982).
8. T. Hatano, An examination of the resonance of hydrodynamic pressure during earthquakes due to elasticity of water, Technical
Report C-65001, Central Research Institute, Electric Power Industry, Tokyo, Japan (1965).