27

Chinese Journal of Rock Mechanics and Engineering

2008 8

Vol.27 No.8
Aug.2008

INFLUENCES OF HYDRAULIC
UPLIFT PRESSURES ON STABILITY OF
GRAVITY DAM
UTILI Stefano1 2YIN Zhenyu 2JIANG Mingjing3

(1. Department of Civil EngineeringPolitecnico di MilanoMilan 20133Italy

2. Department of Civil EngineeringUniversity of StrathclydeGlasgow G4ONGUnited Kingdom
3. Department of Geotechnical EngineeringTongji UniversityShanghai 200092China)
AbstractA study of the influences of the hydraulic uplift pressures underneath the base of a typical concrete gravity dam
on its stability is presented. The dam is located at Cumbidanovu(SardegnaItaly). The foundation of the dam is made of
heavily fractured rock. Firstlyanalytical calculations about the equilibrium of the dam as a free body have been employed
to evaluate the maximum hydraulic pressure before collapsing and to assess the impact of an effective drainage system on
the stability of the dam in a simple way. Secondlynumerical analyses by the distinct element method(DEM) using the
code UDEC have been carried out to evaluate the hydraulic flow taking place within the fractured rock foundationthe
uplift pressure distribution generated by the calculated flowand its influence on the stability of the dam. For design
purposesit emerges that availability of reliable data on the hydraulic permeability of rock foundations and a
computationally advanced distinct element modeling might lead to the acceptance of loads significantly higher than the
more conservative estimations obtained from equilibrium analyses.
Key wordshydraulic engineeringdistinct element method(DEM)hydraulic upliftdam stabilitygravity dam
CLC numberTV 642

Document codeA

Article ID10006915(2008)08155415

UTILI Stefano1 2 2 3

(1.

201332.

3.

G4ONG

200092)

Cumbidanovu

Received date20080222Revised date20080520

Foundation itemNational Science Foundation Project(50679057)Pujiang Talents ProjectShanghaiChina(06PJ14088)
Corresponding authorUTILI Stefano(1976)malePh. D.graduated from Politecnico di Milano in 2000. He is now at Strathclyde Universityand his main
research interests are mainly covered in DEMslope stabilityand limit analysis. E-mailmjjiang@excite.com

 27 8

UTILI Stefano
et al. Influences of Hydraulic Uplift Pressures on Stability of Gravity Dam

1555

INTRODUCTION

stability since it may give rise to a significant vertical

force making the dam overturning about its toe and

The gravity dam investigated is located at

decreasing its frictional resistance against sliding[3 4].

The following two methods will be used to evaluate

CumbidanovuSardegnaItalyand it rests on a

the influence of the uplift thrust on dam stability

heavily fractured rock foundation. The data related to

the foundation have been taken from the results

equilibrium analysis of the dam as a rigid free

body(single wedge analysis) and the distinct element

where a detailed collection of the geological and

method(DEM). The former has been used to achieve

geotechnical

a first rough estimation of the maximum reservoir

level and some insight on the influence of the

[1]

data

for

the

site

is

available.

Concerning with the dam body instead(see Fig.1)

the geometry has been taken from the proposal A2 of
the 1999 ICOLD(International Committee of Large
Dams) held at Denver [2]USAwhich is therefore
slightly different from the real one. This assumption
however has no influence on the final results
obtained since the global size of the dam is the same
and the uplift pressure distribution underneath the
dam base is not affected by the shape of the dam
body. The objective of the ICOLD proposal was to
benchmark numerical and analytical methods for the
evaluation of the maximum sustainable reservoir
level before dam collapsing and the evaluation of the
uplift pressure distributions acting along the dam
base.

effectiveness of drainage system on stability of

the dam. The usefulness of this method for

stability calculations[4 5] for gravity dams is well

known. From a theoretical viewpointan alternative
formulation by the upper bound limit analysis
method could be adopted. According to this method
instead of the cardinal equations of statics the
balance between external work and dissipated energy
is imposed[6]but in this paperthe single wedge
mechanism formulation has been preferred for sake
of simplicity. The DEM has been used to perform a
more refined analysis calculating the uplift thrust
acting along the base of the damas a result of the
actual hydraulic flow taking place into the fractured
rock foundation for both cases of effective and
ineffective drainage system. The use of the DEM for
dam stability analyses[7] is not new but its
application to the study of seepage within the rock
foundation as a coupled hydro-mechanical problem
is quite recent in the literature since seepage in the
foundation is often calculated by finite element
analyses[8] which can not directly model the
hydraulic flows within the joints in the foundation.
For gravity damsthe most common drainage
system is made by an inspection gallery running
longitudinally above the dam base and located as
close as technologically achievable to the foundation
and several drainage holes drilled from the gallery
floor into the rock foundation(for details of the

drainage system see USACE regulations[9 10]).

Fig.1

Dam geometry

2
In the literatureit is known that hydraulic uplift
thrust due to water percolating underneath the dam
base is an important factor to be assessed for dam

EQUILIBRIUM ANALYSES OF
DAM
In this sectionSome calculations based on the

 1556

2008

performed to determine the maximum sustainable

But the assessment of the effectiveness of the drains

after dam construction and reservoir impoundment is

reservoir level. According to the method adoptedthe

a difficult task. Thereforeit is common practice to

dam assumed to be rigid and static equilibrium has

make assumptions on their effectiveness taking

been imposed taking into account all the forces

values recommended by regulations issued by

acting on itthe pull of gravity upon the dam mass

designated engineering national bodies although

the lateral thrust due to the water impounded in the

rules are notin generalstrictly prescriptivein

reservoir and the pore water pressure distributed

Europe and the US at least.

equilibrium of the dam as a free rigid body will be

along all or a portion of the boundaries of the dam body.

According to the related Italian and USACE

Regulations[11

12]

The foundation has been considered impermeable

whereas the joint between dam and foundation has

the uplift pressure along the dam base may be assumed

been assumed fully permeable. To work out the

to vary linearly from the value of hydrostatic pressure

following calculationsthe upper right triangular part

generated by the reservoir underneath the dam heel

of the dam crest(see Fig.2) has been neglected since

which is w in our caseto the hydrostatic pressure

it represents only 0.52% of the total area given by

underneath the dam toe due to the water level

in case of absent drainage system

where B and H are the width of dam base and dam

downstream of the dam that is zero in our case. If a

part of the dam baseadjacent to the upstream

height respectively.

face(marked as a in Fig.2)is not compressed against

Atot = BH / 2

(1)

the foundation because of a high overturning

moment acting on the damthe uplift pressure p
under the non-compressed part must be assumed
equal to the hydrostatic pressure corresponding to
the reservoir water level(see Fig.2(a))
p = w

(2)

where and w are the reservoir water level and the

water unit weight respectively.
In case of a properly working drainage system
the distribution recommended by the regulations is
(a) Absent or ineffective drainage system

bilinearconstant under the non-compressed part of

the dam a and linearly decreasing in the
compressed part. The uplift pressure variation along
the compressed part of the dam is characterized by a
steeper gradient upstream of the drains line and a
flatter one downstream of it(see Fig.2(b)). This type
of uplift distribution is also assumed in related
results[13]. Underneath the drains linethe regulations
of many countries recommend to assume a value of
pressure ranging from 0.25 to 0.6 times the
difference between the pressure underneath the

(b) Effective drainage system

Fig.2

Uplift water pressure distribution under the dam

upstream facep = win our caseand the pressure

underneath the toep = 0 in the case. According to
the Italian Regulations[11]this value must be 0.35

Drainage is the single most effective mean of

times the aforementioned difference. Up taking this

significantly reducing the amount of uplift thrust.

recommendation in our calculations the uplift

 27 8

UTILI Stefano
et al. Influences of Hydraulic Uplift Pressures on Stability of Gravity Dam

1557

pressure acting under the drains line results to be p =

0.35w. If the drains are not suitable to guarantee
such a pressure reductionstability analyses must be
performed assuming no drainage system. When the
non-compressed zone extends downstream of the
drain line(ad)thusthe recommended pressure
distribution becomes constant along the noncompressed zone and linear in the compressed one
hence falling again into the case shown in Fig.2(a).
In the following sectionthe stability of the dam
will be examined with regard to two possible
collapse mechanismsoverturning about the toe and
sliding along the dam-foundation interface joint.
Punching of the dam into the foundation or the
development of a failure surface throughout the
foundation has been disregarded since the rock
constituting the foundation is highly resistant to
crushing and shear failure. The two mechanisms
overturning and slidingtherefore will be considered
separately.
2.1 Overturning mechanism
Overturning of the dam about its toe occurs
when the overturning moment due to the lateral
thrust of the water impounded in the reservoir and to
the uplift forces acting on its base overcomes the
resistance of the stabilizing moment due to its
dead-weight. Considering first a low reservoir level
the dam rests on the foundation lying with its entire
base in contact with it. If the reservoir level is then
continuously increasedthe dam will be progressively
uplifted with an increasingly smaller part of its base
in contact with the foundation until the contact reduces
to one pointits toeand overturning effectively occurs.
In this casethe vertical net reaction force exchanged
between dam and foundation N passes through
the only point of contact O(see Fig.3). In this case
the uplift pressure along the dam base isaccording
to the regulationsuniform and equal to w since a =
B and therefore ad. In order to determine the
reservoir level responsible of overturningoverturning
it is enough to impose the equilibrium of moments
around the point O
H
H
B 2
+ S2
+ U WB = 0
(3)
3
2
2 3
where S1 and S 2 are the triangular and rectangular
S1

Fig.3

Forces acting on the dam at incipient overturning

(sliding is not considered)

components of the lateral thrust respectivelyU is the

uplift thrustand W is the dam dead-weight.
After calculation of the forces and some passages
(see Appendix)the reservoir level at incipient overturning is achieved

overturning

c B 2
2 H + 1

w H
=
= 80.21 m
2
B

3 + 1
H

(4)

where c is concrete unit weight c = 24 kN/m3.

Considering now a reservoir level overturning
but high enough to uplift a part of the dam base from
the foundation(a0)the position of N is unknown
and therefore the problem becomes statically
indeterminate. To solve the problemit is necessary
to introduce a hypothesis on the distribution of the
effective normal stresses along the compressed part
of the dam against the foundation. Firstif the dam is
considered as a rigid bodyand the reaction offered
by the foundation by means of Winklers elastic
springs with constant stiffnessthe distribution of the
normal stresses is triangular(see Fig.4(a)). But in
realitythe more the dam is subjected to a higher
overturning moment under an increasing reservoir
levelthe more its toe will be compressed against the
foundationand therefore the zone of the dam near
the toe will be subjected to high compressive stresses

 1558

(a) Triangular

Fig.4

2008

(b) Real

(c) Uniform

Stress distributions along the compressed zone of the

dam base with x/3cx/2

leading to plastic stress redistribution. In factit is

(a) Absent drainage system

well known that concrete undergoing high compressive

stresses exhibits a stress-strain relationship progressively
drifting away from linearity until crushing occurs.
This implies that along the dam zone in compression

x = B a the normal stresses will vary nonlinearly

as shown in the distribution in Fig.4(b). To find out

Triangular stress distribution with

effective drainage
Uniform stress distribution with
effective drainage

the real stress distribution is a complex problem that

would require nonlinear 3D finite element analyses.
To bracket the real stress distributiontwo distributions

(b) Effective drainage system

were assumeda triangular and a uniform one. In the

former case N is applied at x/3 from the dam toe

Fig.5 Reservoir level vs. non-compressed zone length

whereas in the latter one at x/2 with the true line of

action of N lying somewhere in between x/3 and

obtained in case of triangular stress distribution are

x/2.

distributionsince the destabilizing moment due to

N is smaller because of a shorter lever arm(x/3

To determine the extension of the uplifted part

of the dam baseagiven a known reservoir level
it is necessary to impose the equilibrium of moments
for the forces acting on the dam assuming point O
as the pole

overturning

= M stabilising

(5)

In Appendixthe analytical calculations are

higher than those obtained in case of uniform

instead of x/2).
Looking now at Fig.5(a)the gap between the
two curves reduces with increasing auntil it becomes
nil at a = B = 60 mwhen overturning occurs. Since
the dam is in contact with the foundation only in one
pointits toeand the location of N becomes

shown. Depending on the two distributions adopted

independent of the assumed stress distribution.

two different polynomial expressions were found

In case of effective drainage systemthe analytical

expressions = (a) obtained for ad give rise to

triang = ( a)

(6)

higher values than the case of absent drainssince

Since the real normal stress distribution has

effective drainage allows a significant reduction of

uplift pressures making the dam much more stable

uniform = 1 ( a)

been bracketed by the two distributions assumed

also the real relationship real = (a) will be bracketed
by the two functions in Eq.(6). In Fig.5the real
function will lie in the region of space enclosed by
the two linesthe gray onecorresponding to triang
and the black one corresponding to uniform. In Fig.5(a) it
is shown the case of ineffective or absent drainage
system whereas in Fig.5(b) the case of effective
drainage. In the graphsthe reservoir levels triang

(compare Fig.5(a) with Fig.5(b) for a d ) . But

when the non-compressed zone goes beyond the

drains linethe drains must be considered ineffective
according to the regulations[11

12]

and therefore uplift

water pressures suddenly increase to the value they

assume in case of absent drains(see Fig.6). This is
due to the fact that when the dam is uplifted beyond
the drains line an effective water interception by the
drainage system is no longer possible. In Fig.5(b)

 27 8

Region 1

UTILI Stefano
et al. Influences of Hydraulic Uplift Pressures on Stability of Gravity Dam

1559

Region 2

Fig.6 Uplift pore pressure distributions in case of a1d

(region 1) and a2d(regions 1 + 2)

the dashed curves for ad are the same curves as

the solid ones plotted in Fig.5(a) for the case of absent
drainage system. They are dashed to point out that
they are only theoretical ones. In factconsidering
the dam subjected to an increasing reservoir levelas
soon as the uplifted zone goes beyond the drains
line a sudden increase of a occurs at constant
reservoir level. In case of uniform distributionthe
reservoir level may continue to be increased even
after the uplifted zone has gone beyond the drains
line until the dam overturns around its toe(a = B)
whereas in case of triangular distribution the
reservoir level cannot be further increased since the
reservoir level reached at a = d = 10 m is larger than
overturning. Hencein case of uniform distributionthe
presence of drains increases the load bearing capacity
of the dam only for values of a17 mwhereas in
case of triangular distributionfor any value of a.
2.2 Sliding mechanism
Resistance against horizontal sliding is given by
a frictional component proportional to the net vertical
force exchanged between dam and foundation N
and a cohesive one. This force opposes the lateral
thrust due to the water impounded in the
reservoir(see Fig.7). If Tres is the resisting forceit is
given by
Tres = N tan + cx

calculation for the sliding mechanism

From Eq.(8)two polynomial functions = (a)

are obtained one in case of absent drainage
system(see Fig.8(a)) and the other in case of
effective drains(see Fig.8(b)). In both cases the
reservoir level decreases with increasing a. This is
because the higher the value of a the less the
cohesive part of the resisting forceand therefore the
lower the value of .

(a) Ineffective drains

(7)

where and c are the friction angle and the

cohesion along the dam base with N calculated by
imposing the vertical equilibrium of forces.
The relationship between the reservoir level
and the extension of the non-compressed zone ais
found imposing the horizontal equilibrium(details of
the calculations are given in Appendix)
Tres = Tact

Fig.7 Forces acting on the dam taken into account in the

(8)

(b) Effective drainage system

Fig.8 Reservoir level vs. uplifted zone in case of ineffective

drains and effective drainage system

 1560

2008

2.3 Determination of the dam bearing capacity

this condition does not lead to collapse since the

To determine the bearing capacity of the dam

reservoir level can still be increased prior to the onset

the two investigated mechanisms must be analyzed

of sliding. In realitythe normal stress distribution

together. In Fig.8a graph is shown in terms of

reservoir level versus length of the uplifted zone

will be neither linear nor uniform but something in

along the dam basea. The curves related to the

analyses performed if the attainment of the condition

overturning mechanism are given by solid lines

whereas the curves related to sliding are given by

a = d leads to the collapse of the dam or there is still

cross lines. The real collapse load is given by the

system is presentedhigher reservoir loads can be

between so it is not clear from the equilibrium

a margin of safety. Howeverif an effective drainage

minimum collapsing load determined for each

mechanism taken separately. Thereforeconsidering

sustained by the dam in so far as the drainage system

remains efficient i.e. the uplifted zone remains

the dam under an initially low and then increasing

upstream of the drains linead.

reservoir levelthe more critical failure mechanism

is overturning since it is associated with the lower

curve. This means that the dam will be gradually

uplifted because of increasingly higher overturning
moments(see curves with pointed arrows in the

DISTINCT ELEMENT ANALYSES

The specifications of the A2 proposal of the

ICOLD Benchmark 99 do not provide data related to

graphs). The dam upliftin turnwill progressively

the dam foundation. These have been taken from the

decrease the resistance against sliding along the dam

foundation interface(see the descending curves in

foundation of Cumbidanovus dam[1]. Insteadthe

geometry of the dam and its mechanical properties

Fig.8 related to the sliding mechanism). When the

have been assumed according to the Benchmark

curve related to the overturning mechanism and the

curve related to sliding meet in point A for the case

specifications that are consistent with the equilibrium

analyses. These are given in Table 1.

of triangular distribution and point B for the case of

Concerning the jointsa Mohr-Coulomb shear-

uniform distributionthe collapse of the dam occurs.

In correspondence of those pointsit happens that

strength relation has been adopted. The dam-foundation

Tres becomes smaller than Tact and the dam starts

joint is characterized by a softening behavior as

sliding downstream.
The collapse load calculated by the performed

prescribed by the ICOLD specifications(see Fig.9). The

equilibrium varies within a precise rangefrom 76.5

mechanical parametersalso reported in Table 1are

given byc = 0.7 MPa = 30 = 10(dilation

to 78 m in case of absent/ineffective drainage system

(line AB in Fig.8(a))from 76.5 to 81.8 m in case of

angle) f t = 0(tensile resistance) cres = 0(residual

= 30(residual friction angle). The joints
cohesion) res

effective drainage system(line AB in Fig.8(b)).

In case of linear stress distributionas soon as

in the foundations are characterized by the following

parameters[1]c = 0 MPa = 30 = 0 f t = 0

the uplifted zone goes beyond the drains linesliding

MPa. Water viscosity K j has been assumed equal to

occurs whereas in case of uniform stress distribution

K j = 300 Pas 1.

Table 1
Methods

Summary of the physico-mechanical parameters used in calculation

Dam

Foundation

Dam-foundation joint

c/(kNm 3)

Ec/GPa

rock/(kNm 3)

Erock/GPa

rock

/(
)

c/MPa

/(
)

ft

res/(
)

cres

Equilibrium analysis

24

24.5

0.17

30

0.7

DEM

24

24.5

0.17

26

20

0.3

30

0.7

10

30

Methods

Joints in foundation
Kj/(Pas 1)

ha0 /mm

300

0.5

5104

5102

/(
)

c/MPa

/(
)

ft/MPa

Equilibrium analysis

DEM

30

hares /mm

KN/(kNm 1)

KS/(kNm 1)

 27 8

UTILI Stefano
et al. Influences of Hydraulic Uplift Pressures on Stability of Gravity Dam

1561

In order to perform a meaningful analysis the

choice of the hydraulic conductivities along the
joints is critical. To verify that the chosen values
ha0 = 1 mm and hares = ha0 / 2 are adequate for the

problemthe simpler case of non-fractured foundation

has been analyzed in order to check the agreement
Tension t

Fig.9

with the equilibrium analyses performed in Section 2.

Shear strength of the dam-foundation joint according

[2]

to the Mohr-Coulomb criterion

In Fig.11it is shown the pressure distribution at

incipient collapse with athe length of the uplifted
part of the dam base beyond the drains line. This

tangential mechanical joint stiffnessesK N and K S

distribution is well in agreement with that taken from

the referenced regulations and assumed in the

respectivelyand the joint hydraulic apertures ha0

equilibrium analyses(see Fig.2(a)for the case a

Other parameters such as the normal and

and hares have been selected by the authors.

The regulations taken as a reference for the

d ). This match corroborates the goodness of the

hydraulic apertures adopted in the UDEC analyses.

equilibrium analyses only specify the reduction of

uplift pressure operated by the drains without any
indication in regard to the hydraulic flow along the
drains. Therefore the same hydraulic condition
assumed in Section 2 has been implemented into the
distinct element analyses by assigning a fixed pore
pressure value p = 0.35w at the point of
intersection between the drains line and the
dam-foundation joint.

Fig.11 Pore pressure distribution underneath the dam

In Fig.10it is shown the relationship between

hydraulic aperture and normal stress adopted in the
code UDEC[14

assuming intact foundation(reservoir level

= 80 m)

15]

the maximum hydraulic aperture

along a joint is reached when the compressive stress

goes to zero ha0 in the figurethen the hydraulic
aperture varies linearly to a residual valuehares with
m given by

K N and K S the mechanical stiffnesses of the

jointsrule the amount of mechanical overlapping

allowed between two adjacent blocks but they also
greatly influence the time needed to run a simulation
since the stability of the integration scheme used to

m = 1 / K N

(9)

solve the equations of motion is explicit and the time

step adopted by the code is inversely proportional
to the square root of K. Here there are two
conflicting objectives which can be labeled asfast
computations andrealistically deformable joints.

Tension t

After some preliminary analysesK N = 5104 kN/m

1
and K S =
K N have been proven suitable for the
100
problem.
3.1 Analyses

Fig.10 Hydraulic aperture of jointhavs. stress normal to

jointnwithin the rock foundation

[14]

The effect of the inclination and spacing of the

joints has been studied assuming four different

 1562

geometries. The total of the presented models is

eight for each set of joints both cases of absent and
effective drainage system have been considered(see
Table 2). In the first configurationthe joints form a
+45 angle with the horizonthe spacing between
joints beingin model a6 m for the first set and 12 m
for the second setin model b12 m for the first set
and 6 m for the second set. In model cthe first set of
joint is +15 inclined over the horizon with 6 m
spacing and the second set 69with 12 m spacing
whereas in model dthe first set is +15inclined over
the horizon with 12 m spacing and the second is

69

with 6 m spacing.
Table 2

Summary of the models analyzed

Joint
inclination
Model
Drains
to the
)
horizon/(
a
b
c
d

UDEC
LE analysis
Block
analysis bearing
bearing
size
/(mm) capacity /m capacity /m

a1

+45

No

126

87

76.578.5

a2

45

Yes

126

91

76.581.8

b1

+45

No

612

85

76.578.5

b2

45

Yes

612

87

76.581.8

c1

+15

No

126

77

76.578.5

c2

69

Yes

126

80

76.581.8

d1

+15

No

612

75

76.578.5

d2

69

Yes

612

80

76.581.8

2008

The loads(dam weight and reservoir level) were

applied progressively in successive steps in order to
achieve equilibrium conditions in a quasi-static way
avoiding to introduce any inertial effects. Firstthe
load of the dam was applied to the foundation
increasing the dam dead-weight by stepsand then
the reservoir load was assigned by steps until the
dam collapsed. After each stepa number of cycles
large enough for the system to reach static
equilibrium was run. Some preliminary analyses
were performed to determine the minimum number
of steps for which the influence of the loading
procedure on the final results is negligible.
The reservoir level was increased by 5 m steps
up to the collapse of the dam. Thenthe analysis was
resumed from the last step of loading before collapse
and smaller increments of reservoir level were assigned
to identify with better accuracy the reservoir level
causing collapse. Collapse was detected looking at
the displacement undergone by the dam. After the
total displacement of the crest exceeded 0.5 mthe
analysis was stopped.
In all the simulations runa small numerical
damping was used. This damping is given by decreasing
block accelerations and increasing block decelerations

Each block of rock and the dam have been

assumed linearly elastically deformable. To impose
the elastic laweach rock block in the foundation has
been subdivided into four regular triangles whereas
the dam has been subdivided into non-regular triangular
elements(see gray lines in Fig.12). Thenstresses and
strains were calculated by UDEC in the nodes of
each mesh every time-step.

by a factor proportional to the unbalanced forces

acting on each block. It was used only to shorten the
computational time needed to run a simulation and
therefore it has no particular meaning but its use is
justified as long as it does not affect the global
response of the assembly of blocks to the mechanical
and hydraulic actions imposed. Some preliminary
tests with zero damping were run to verify that the
introduced damping had a negligible effect in terms
of the analyzed global variables.
3.2 Results
For all the joint configurations analyzedthe
maximum hydraulic flow takes place along the
dam-foundation joint and it has been noted that the
greater the distance of a joint from the dam base is
the smaller the flow is. This is due to the fact that
the hydraulic conductivity of the joints depends on
the compressive stresses. For increasing depths

Fig.12

Domain of analysis(model c)black lines indicate

compressive stresses increase and the hydraulic

edges of blocksgray lines indicate the mesh contour

conductivity decreases as a result.

 27 8

UTILI Stefano
et al. Influences of Hydraulic Uplift Pressures on Stability of Gravity Dam

1563

In Fig.13 it is shown the hydraulic flow

monitored along the dam-foundation joint with the

the damwhen a = 0 the distribution is quite in

dam at incipient failureeach point in the graph

In case of absent/ineffective drainagesit is linear

indicates the flow along one hydraulic domain

extending between two consecutive rock joints

(see Fig.15(a))whereas in case of effective drains in

operationthe distribution changesit is still linear

intersecting the dam-foundation joint. It can be noted

upstream of the drains but becomes parabolic-like

that in case of absent drainage systemthe flow is

almost constant. In case of effective drainage system
insteadthe flow along the upstream part of the joint
is more than twice as high as the flow in the previous
casewhereas the flow along the downstream part
of the jointfrom the drains line to the dam toe
remains lower than the previous case with a jump in
correspondence of the drainage point. Moreoverthe

agreement with what prescribed by the regulations.

downstream of them(see Fig.15(b)) slightly different

from what prescribed by the regulations. After the
dam starts upliftinga0the obtained pressure
distributions strongly depend on the geometry of the
rock jointsdrifting substantially from the prescriptions
of the regulations.
Dam base/m

the hydraulic flow in a large area of the foundation.

Depending on the inclination of the rock jointsit
may also happen that the flow along the damfoundation interface goes upstream for a short
length(see Fig.14just before the drains line).

Pore pressure/kPa

presence of the drainage system greatly influences

0
100
200
300
400
500
600
700
800

10

20

30

40

50

60

Model b
Model d
Regulations

h = 85 mm(no drainage system)

h = 87 mm(drainage system)

20

Dam base/m

15
10
5
0

10

20

30

40

50

60

Dam-foundation joint/m

Fig.13

Hydraulic flow along the dam-foundation interface

at incipient collapseblack line in case of effective
drainage system and gray one in case of absent
drainage system

Pore pressure/kPa

Hydraulic flow /(Ls 1)

(a) No drainage
25

0
100
200
300
400
500
600
700
800

10

20

30

40

50

60

Model b
Model d
Regulations
(b) Active drainage

Fig.15 Uplift pressures with the dam base all in contact

against the foundationa = 0in case of models
b and d with reservoir level = 70 m

The load bearing capacity of the dam varies a

lotdepending on the geometry of the joints in the
foundation(see Table 2). Making a comparison with
the equilibrium analyses of Section 2it is possible
to say that in case of joints inclined 45on the
horizontal linethe resulting bearing capacity ison
average7 m higher than the equilibrium analysis
Fig.14

Hydraulic flow directions underneath the dam

with drainage system in operation

With regard to the uplift pressure underneath

predictionswhereasin case of joints inclined +15/

69 on the horizontal line(cases c and d) the
bearing capacity is within the range predicted by
equilibrium analyses. The physical explanation is as

 1564

2008

followsin models a and bgiven a known reservoir

level a smaller length a and lower water pressures

Fig.17 provides the horizontal displacements of

the left edge of the dam crest during the reservoir

underneath the dam were generated leading to a

filling until the last step of loading before failure.

larger resistance against sliding. On the contraryin

models c and dgiven the same reservoir level a

The trend shown in Fig.17 is reasonable at the

beginning the crest moves upstreambeing the dam

larger length a and higher uplift pressures were

rotating counterclockwiseand then moves downstream

generated leading to a smaller resistance against

sliding. In factresistance against sliding depends

with the dam rotating clockwise. The horizontal

asymptote in the figure indicates the onset of the

for its cohesive termon the extension of the zone in

sliding of the dam along the foundation.

contact with the foundation B-a and for its

frictional termon the amount of N (see Eq.(7))
that in turn depends on the uplift pressures.
Using UDECit is also possible to investigate
the displacements undergone by the damotherwise
not achievable by the equilibrium analyses performed
in Section 2. During its process of construction and
until the reservoir level becomes half full(about 35 m)
the dam tends to tilt towards the reservoir. Then
further increases of the reservoir level lead to a

Fig.17

Horizontal displacements recorded at the dam crest

vs. reservoir level

change of direction the dam tilting towards the

opposite side(see Fig.16).

CONCLUSIONS
By means of static equilibrium analysesit has

been possible to achieve analytical expressions for

the bearing capacity of a typical Italian gravity dam
resting on a rock foundation. To this endthe uplift
thrust due to the hydraulic pressures underneath the
dam and the beneficial effect of an effective drainage
system were taken into account by assuming uplift
(a) = 35 m

pressure distributions recommended by the Italian

and USACE regulations.
Secondlya more refined analysis was carried
out by the DEM. The hydraulic flow taking place
within the fractured foundation and the influence of
different geometries of rock joints on the uplift thrust
andas a resulton the dam bearing capacity was
investigated. With this numerical methodit was also
possible to monitor the displacements undergone by
the dam.
Both methods are shown to be viable ways of

(b) = 85 m

Fig.16

investigating the influence of the hydraulic uplift

Dam and rock foundation for model b with reservoir

pressures on the stability of dams. Comparing the

level(deformations magnified by 100 times)

obtained resultsequilibrium analyses give rise to

 27 8

UTILI Stefano
et al. Influences of Hydraulic Uplift Pressures on Stability of Gravity Dam

more conservative predictions in terms of dam

[6]

1565

bearing capacity whereas the DEM is a more

[J]. 2006 25(3)533 538.(XU

powerful numerical tool which can be relied on to

Qianjun LI Xu CHEN Zuyu. Three-dimensional stability

accept less conservative predictionssince it can also

analysis of dam foundation of Baise hydro-junction[J]. Chinese

model the hydraulic flow within fractured rock

Journal of Rock Mechanics and Engineering200625(3)533

foundations.

538.(in Chinese))

For design of new dams or evaluation of existing

[7]

onesin light of the analyses performedit emerges

[J]. 200625(6)1 226

the importance of having both reliable data about the

1 232.(ZHANG ChongHOU YanliJIN Fenget al. Analysis of

hydraulic permeability of rock foundations and a

arch dam-abutment stability by 3D deformable distinct elements[J].

computationally advanced modelinge.g. DEMso

Chinese Journal of Rock Mechanics and Engineering200625(6)

that loads significantly higher than the estimations

1 2261 232.(in Chinese))

obtained from equilibrium analyses might be accepted.

[8]

200726( 1)3 0173 024.(XIANG

ACKNOWLEDGEMENTS The authors would like

Yan WU Zhongru. Analysis theory and method of crossfeed

to thank the staff at Politecnico di Milano(Milan

between concrete dam body and foundation[J]. Chinese Journal of

Italy)in particular Professor R. Nova and Professor F.

Rock Mechanics and Engineering200726(Supp.1)3 017

Calvetti for their support. The support of Itasca C.G.

is also gratefully acknowledged.

3 024.(in Chinese))
[9]

1995.
[10]

BONINI M. Studio delle condizioni di flusso idraulico nellammasso

Department of the Army1986.

[11]

1999.
[2]

n.]1982.
[12]

LIU J FENG X T DING X L. Stability assessment of the Three

modelingpart IInumerical modeling[J]. International Journal of

[5]

U.S. Army Corps of Engineers. Gravity dam design engineer

manual 111022200[S]. WashingtonD. C.Department of the

Gorges dam foundation China using physical and numerical

[4]

MINISTERO L P. Norme tecniche per la progettazione e la

costruzione delle dighe di sbarramento(DM No.44)[S]. Roma[s.

ICOLD. A2 benchmark proposal for numerical and analytical

methods for gravity dams[R]. DenverUSA[s. n.]1999.

[3]

U.S. Army Corps of Engineers. Seepage analysis and control for

damsengineer manual 111021901[S]. Washington D. C.

roccioso di fondazione di una diga a gravita mediante il metodo

degli elementi distinti[Ph. D. Thesis][D]. Politecnico Torino[s. n.]

U. S. Army Corps of Engineers. Grouting technologyengineer

manual 111023506[S]. WashingtonDepartment of the Army

References()
[1]

. [J].

Army1995.
[13]

Rock Mechanics and Mining Sciences200340(5)633652.

[J]. 2007 26(8)

YU XZHOU Y FPENG S Z. Stability analyses of dam abutments

1 594 1 602.(CHANG Xiaolin JIANG Chunyan ZHOU Wei

by 3D elastoplastic finite-element methoda case study of Houhe

et al. Equal safety factor method and its reliability analysis for rock

gravity-arch dam in China[J]. International Journal of Rock

foundation of dam[J]. Chinese Journal of Rock Mechanics and

Mechanics and Mining Sciences200542(3)415430.

Engineering200726(8)1 5941 602.(in Chinese))

[14]

Itasca Consulting Group Inc.. Universal distinct element codeuser

[J]. 200527(6)657661.(HOU

manual(Version 3.0)[R]. Denver Itasca Consulting Group Inc.

YanliZHANG ChongZHANG Chuhanet al. Simulation of

1999.

upward-sliding failure of interface in arch dams by deformable

[15]

CUNDALL P A. UDECa generalized distinct element program

distinct elements[J]. Chinese Journal of Geotechnical Engineering

of modeling jointed rock[R]. [S. l.]European Research Office

200527(6)657661.(in Chinese))

US Army1980.

 1566

2008

APPENDIX
A.1 Overturning mechanism
In the followingit is shown the detail of the
derivation of Eq.(4) from Eq.(3). The forces in Eq.(3)
are given by

H2
S1 = w

S 2 = w ( H ) H
(A1)

U = w B

BH

W =c

2
Thereforesubstituting them in Eq.(3)it can be
obtained
H3
H2
B2
H
w
+ w ( H )
+ w
= c B2
(A2)
6
2
2
3
Thenit can be written
B2 H 2
H wH3
= c B 2
+
+
w

2
3
3 (A3)
2

3 w ( B 2 + H 2 ) = 2 c B 2 H + 2 w H 3

(a) ad

And finally
B 2
2 H c + 1
w H

=
2
B

3 + 1
H

(A4)

Hereafterit is shown the derivation of the analytical

functions i = i (a) giving rise to the graphs plotted
in Fig.5. Lets start with the case of effective drainage
system and ad as shown in Fig.A1(a). In this case
Hthe forces acting on the dam are

Imposing the equilibrium of moments about point

O(see Fig.A1(a))leads to

f = 0.35

U 1 = w a

1
U 2 = w (1 f ) ( d a )
2

U 3 = w f ( d a )

1
U 4 = w f ( B d )

BH

W =c
2

S1 = w

(b) ad

Fig.A1 Forces acting on the dam overturning mechanism

S1

(A5)

a
2

+ U 1 B + U 2 (d a ) + B d +
3
2
3

2
1

U 3 (d a ) + B d + U 4 ( B d ) +
3
2

2
(A6)
N' x W B = 0
3
Imposing the vertical equilibrium N can be
expressed by
N = W U1 U 2 U 3 U 4

(A7)

Let us consider first the case of uniform stress

 27 8

distribution x =

UTILI Stefano
et al. Influences of Hydraulic Uplift Pressures on Stability of Gravity Dam

Ba
. Substituting N and x into
2

Eq.(A6)
S1

a
d

+ U1 B + U 2 a + B +
2
3
3

1567

The function f (a) = 0 may be rewritten as a

third degree polynomial function
A0 3 + A1 2 + A2 + A3 = 0
(A12)
with the coefficients Ai depending on the parameter a
A0 = 1.67

A1 = 0

A2 = 0.541 7a 2 + 164.2a + 2 033

A3 = 28 800a 576 000

2
B a
2

a d
U 3 + B + U 4 B d +W
3
2 2
3

2 2
B a
B a
B a
U1 U 2 U 3
2 2
2 2
2 2
2
B a
U4 W B = 0
3
2 2

(A8)

Then

(A13)

A third degree function always presents three

solutionsbut in this case only one is real. Sousing
the available well known algebraic formulaeafter

a B a
B a

2 d
S1 +U1 B + +U 2 a + B + +
3
2 2 2
2 2

3 3

some more calculations the sought expression

= (a) with B = 2 168.6 10 3 a 43 373 10 3 and

B a
B a
2
2
a d
U3 + B + + U4 B d + +
2 2
3
2 2
3
2 2

C = 2.713 9a 2 + 822.44a + 10 186 can be obtained

B a 2
W B = 0
2 2 3

(A9)

+U1

B
a d B
+U 2 + +
2
6 3 2

case of ad the pore pressure distribution changes

radically(see Fig.A1(b)). The forces involved in the
equilibrium of moments are the same as before except
(A10)

for the pore water pressure. The forces due to the

water pressure are(see Fig.5(b))

Now all the forces can be replaced by their

analytical expressions obtaining

U 1 = w a

1
U 2 = w (B a )
2

B
1
+ w a + w (1 f ) (d a )
6
2
2

cB

as illustrated aboveanother third degree polynomial

function is achieved which again has only one real root

1
a B 2
( B d ) + d +
2
2 6 3

H a B
= 0
2 2 2

(A15)

Thenfollowing the same calculation procedure

d B
a d B
+ + w f ( d a) + +
2 2
6 3 3

w f

B + B 2 + 4C 3
2

black solid curve in the right side(ad) of Fig.5(b). In

a B 2
d B
U3 + + U4 + d +
2 6 3
2 2
a B
W = 0
2 6

(A14)
The function achieved in Eq.(A14) is plotted as a

And then
S1

1 B + B 2 + 4C 3
C
5
2

then applying again the algebraic formulae with B =

2 168.6 10 3 a 43 373 10 3 and C = 2.713 9a 2 +

(A11)

Substituting the known parameters with their

numerical valuesan expression of the type f(a)=0
can be obtained.
This expression Eq.(A11) can be solved either in

822.44a + 10 186

12
6

B + B 2 + 4C 3
C
2

2
3
B + B + 4C
2

(A16)
The function = (a) achieved in Eq.(A16) is

a or . In the following calculations has been

considered as a variable and a as a parameter so that a

plotted as a black partly solid and partly dashed curve

function of the type = (a) is obtained consistently

drainage systemthe forces acting on the dam are

with Fig.5where is plotted against a.

in the left side(ad) of Fig.5(b). In case of absent

those shown in Fig.A1(b). Imposing the equilibrium of

 1568

2008

moments leads to Eq.(A16) whichin this caseis valid

w a and U 2 = w ( B a) / 2 . Therefore

for any value of a. The graphs obtained are visible in

N = W U 1 U 2 = c BH / 2

Fig.5(a).
In case of triangular stress distributionthe same
calculations are performed. The only difference is the
Ba
value of x which becomes x =
. In case of da
3
it happens that H and therefore the lateral thrust

w a w ( B a) / 2

It is now possible to get the two forces Tact and

Tres

Tres

w H / 2 and S2 = ( H ) w H . The obtained functions

2

have been plotted as gray curves in Fig.5(a) and

Fig.5(b).
Let start with the case of effective drains. Two
different expressions hold true depending on the
extension of awhich can be larger or smaller than d.
For the sake of spacehere only one case will be
shownadbeing this the case for which sliding
effectively takes place(see Fig.8). The derivation
shown hereafter is also valid for the case of absent/
ineffective drainages since the uplift forces acting on
the dam are the same(see Fig.2(a)).
Considering the case of H the horizontal
forces due to the lateral thrust caused by the reservoir
are S1 = w H 2 / 2 and S 2 = ( H ) w H . Defining
the horizontal force causing sliding Tact and the one
opposing sliding Tres it can be written
Tact = S1 + S 2
Tres

= N tan + c ( B a)

(A17)

N is achieved by the vertical equilibrium as

shown for the overturning mechanism. The forces due

to the water pressureas shown in Fig.7are U 1 =

BH
= tan c
tan w a
2

( B a)
tan w
+ c ( B a )

Tact = w H w

is no longer expressed by one force but two S 1 =

A.2 Sliding mechanism

(A18)

H2
2

(A19)

Imposing the horizontal equilibrium Tact = Tres

leads to

w H w
tan w

H2
B

= tan w a tan w +
2
2

a
BH
+ c ( B a)
+ tan c
2
2

(A20)

Rearranging with the objective of getting a

function of the type = (a)

w H + tan w a + tan w

B
a
tan w =
2
2

H2
BH
+ tan
+ cB ca
2
2
and then

(A21)

2ca + 2cB + w H 2 + tan c BH

=
tan w a + 2 w H + tan w B
1 400 a + 214 470
5.77a + 1 946

(A22)

In case of Hthe same calculations hold true

with the only difference due to Tact that becomesTact =
2
w
.
2