Analysis of Arch Dams Using Coupled Trial Load

and Block Element Methods

Sheng-hong Chen1; Ming-yi Xu2; Isam Shahrour3; and Peter Egger4

Abstract: This paper presents the use of the trial load method and the block element method with elastoviscoplastic discontinuities for
analysis of arch dams. The arch dam is considered as an arch-cantilever system and the foundation as a block element system. With the
displacement compatibility condition at the contact surface of the dam and the foundation 共including abutment兲, the governing equations
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

of the arch dam and foundation are established. These methods are used for the analysis of the double curvature arch dam with complex
geology conditions of the Xiaowan Hydroelectric Project in China. The deformation and stress states in both the dam body and the
foundation are determined. Furthermore, the stability safety factors of the foundation and the abutment are calculated at the same time,
which allows for an optimal design of the arch dam considering the strength, the deformation and the stability of the dam and foundation.
DOI: 10.1061/共ASCE兲1090-0241共2003兲129:11共977兲
CE Database subject headings: Dams, arch; Rock mass; Abutments; Deformation; Viscoplasticity; Analysis; China.

Introduction stability analysis of the abutment gives unsatisfied results, it is

There are two basic tasks in the structure design of an arch dam: necessary to return and to revise the dam’s shape. This procedure
The first one concerns the control of the strength and deformation should be repeated several times to obtain an optimal design
of the dam’s body, while the second deals with the control of the which satisfies both the strength condition of the dam’s body and
deformation and stability of the foundation and abutment. the stability condition of the abutment. This procedure is time
The conventional analysis of an arch dam uses the trial load costly and dependent strongly upon the experience.
method proposed by the United States Department of the Interior The problem of the deformation and stability of foundation
Bureau of Reclamation in 1930’s 共Copen et al. 1977兲. This and abutment is of major interest in the arch dam design, but is
method has been improved at various aspects such as the solution very complex due to the presence of many discontinuities. The
techniques 共Suresh and Natarajan 1981; Lin and Yang 1987; Zhu conventional analysis uses the limit equilibrium method because
et al. 1987兲. It considers the arch dam as a system composed of of its simplicity and the high experience obtained by engineers
arches and cantilevers; loads are cocarried by both the arches and from practice. It has, however, certain limitations: The deforma-
cantilevers. Stresses in the arches and cantilevers are calculated tion of the abutment slope can not be solved, and the stability
separately according to the load distribution. Theoretically, this safety factor can be overestimated when the sliding face is com-
method gives rather exact stresses in the arch dams, but it as- posed of more than one discontinuity 共Londe 1965; Guzina and
sumes that the foundation is homogeneous, half-infinite, and elas- Tucovic 1969; Copen et al. 1977; Chan and Einstein 1981; Chen
tic. Based on this assumption the relationship between the defor- 1993a,b兲. The finite element method 共FEM兲 and some relatively
mation of the foundation and the thrust from the dam can be new developed discontinuous models such as the well known
deduced, then the arch dam can be treated as a system of arches digital elevation model 共Cundall 1971; Cundall 1988; Hart et al.
and cantilevers supported on an elastic foundation. This hypoth- 1988兲 and DDA 共Shi 1992兲 allow us to solve the stability and
esis is nonrespected in the arch dams with complex geology con- deformation problems of the arch dam foundation and abutment
ditions. Furthermore, since the stability of the dam abutment is reasonably. The writers of this paper proposed an elastoviscoplas-
very important and depends strongly on the dam’s thrust, the op- tic block element method 共Chen 1993a,b, 1994; Chen et al. 1994;
timal design of the dam with the trial load method without con- Wang and Chen 1998; Xu et al. 2000; Chen et al. 2000; Wang
sideration of the stability of the abutment is not complete. If the and Chen 2001兲 which can deal easily with the abutment slopes
containing many discontinuities of various scales and orienta-
1
Wuhan Univ., Wuhan Hubei 430072, Peoples Republic of China. tions. It offers some facilities such as: a powerful preprocessing
2
Wuhan Univ., Wuhan Hubei 430072, Peoples Republic of China. part of the complicated abutment slope with irregular ground sur-
3 face; a comprehensive mechanical calculation part simulating
Lille Univ. of Science and Technology, Cite Scientifique, 59656
Villeneuve D’Ascq, France. E-mail: isam.shahrour@eudil.fr various loading and construction processes 共seepage influence,
4
Swiss Federal Institute of Technology in Lausanne, Ch-1015, excavation, reinforcement and refilling step by step, etc.兲; an in-
Lausanne, Switzerland. telligent searching part of the potential failure cases formed by
Note. Discussion open until April 1, 2004. Separate discussions must the block combinations; and the corresponding stability safety
be submitted for individual papers. To extend the closing date by one
calibration by both the safety factor method and reliability
month, a written request must be filed with the ASCE Managing Editor.
The manuscript for this paper was submitted for review and possible
method. However, this method needs improvements for more rea-
publication on July 24, 2001; approved on October 2, 2002. This paper is sonable stability analysis of arch dams. The main deficiency of
part of the Journal of Geotechnical and Geoenvironmental Engineer- this method is that it considers the thrust loads coming from the
ing, Vol. 129, No. 11, November 1, 2003. ©ASCE, ISSN 1090-0241/ dam as constant calculated by the trial load method. Actually,
2003/11-977–986/$18.00. following the building up of the deformations of the dam and

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003 / 977

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

 shell theory. Each node has six displacements 关 x ⬘ y ⬘ z ⬘ ␪ ⬘x ␪ ⬘y ␪ z⬘ 兴 T
in the global coordinate system, therefore the total degrees of an
arch-cantilever element are equal to 24 共4 nodes⫻6兲. For each
node there are six forces 关 Q ⬘ N Q M T 0 兴 T in the global coordi-
nate system 共the lateral torsion is assumed to be zero according to
the shell theory兲. For each representative element a local coordi-
nate system is used in the formulation of the method. In this
system, the arch direction is expressed as v a , and the cantilever
direction is expressed as v b , the normal direction of the middle
face of the element is defined as v r ⫽ v a ⫻ v b . In this local coor-
dinate system, every node has three parallel displacements and
two rotate displacements 共according to shell theory, normal rota-
tion is not considered兲. The total local degrees are equal to 20.
For arch or cantilever elements differential slices should be
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

carved along the arch direction or cantilever direction 关shadow

shown in the Fig. 1共a兲兴. Then the equivalent arch and cantilever
Fig. 1. Arch-cantilever element: 共a兲 representative element 共arch or are integrated from the slices.
cantilever element兲; 共b兲 six displacements of node of representative Denoting the nodal displacement increments as 兵 ⌬␦ 其 ac , the
element global displacement increments along the arch and cantilever di-
rections are
兵 ⌬␦ 其 a ⫽ 关 H 兴 a 兵 ⌬␦ 其 ac (1)
abutment, the thrust loads change. Furthermore, since this method
treats the abutment separately from the dam, a real optimal design 兵 ⌬␦ 其 c ⫽ 关 H 兴 c 兵 ⌬␦ 其 ac (2)
is difficult to reach as has been indicated above.
in which 关 H 兴 a and 关 H 兴 c ⫽interpolation functions along the arch
A coupled analysis method is proposed in this paper for the
and cantilever directions. The above displacements can be trans-
analysis of the arch dam and its abutment. The foundation and
formed into the local coordinate system as
abutment are considered as block elements, while arches and can-
tilevers elements are used for the dam. A typical engineering ex- 兵 ⌬␦ 其 a⬘ ⫽ 关 T 兴 a 兵 ⌬␦ 其 a ⫽ 关 T 兴 a 关 H 兴 a 兵 ⌬␦ 其 ac (3)
ample shows the robustness of the proposed method.
兵 ⌬␦ 其 c⬘ ⫽ 关 T 兴 c 兵 ⌬␦ 其 c ⫽ 关 T 兴 c 关 H 兴 c 兵 ⌬␦ 其 ac (4)
where 关 T 兴 a and 关 T 兴 c ⫽transformation matrices.
Formulation of Trial Load Method
At the arch and cantilever directions, the displacement and
load increments are associated by the stiffness matrices as fol-
The modern version of the trial load method can be illustrated as
lows:
follows 共Copen et al. 1977; Suresh and Natarajan 1981; Lin and
Yang 1987; Zhu et al. 1987兲: First, the arch dam is divided into an 兵 ⌬ f 其 ⬘a ⫽ 关 K 兴 ⬘a 兵 ⌬␦ 其 ⬘a (5)
orthogonal arch-cantilever element system along the arch and
cantilever direction; loads are cocarried by both the arches and 兵 ⌬ f 其 ⬘c ⫽ 关 K 兴 ⬘c 兵 ⌬␦ 其 ⬘c (6)
cantilevers according to their stiffnesses; the displacement com-
patibility condition for the nodes of the arches and cantilevers 兵 ⌬ f 其 a⬘ and 兵 ⌬ f 其 ⬘c are then transformed into the global coordinate
leads to a set of equilibrium equations by which the node dis- system as
placements can be determined; next, the displacements are used 兵 ⌬ f 其 a ⫽ 关 T 兴 Ta 兵 ⌬ f 其 ⬘a (7)
to calculate the internal forces according to the planar arch and
cantilever theory; then stresses are calculated and checked by the 兵 ⌬ f 其 c ⫽ 关 T 兴 Tc 兵 ⌬ f 其 ⬘c (8)
permissible tensile strength and compression strength; finally, the
thrust forces at the abutment are calculated from the internal The total load increment is simply the summation of the arch
forces of the arches and cantilevers elements in contact with the and cantilever part according to the basic principle of the trial
abutment and foundation, the thrust forces are then used in the load method:
analysis of the abutment stability. If the strength conditions of the 兵 ⌬ f 其 ac ⫽ 兵 ⌬ f 其 a ⫹ 兵 ⌬ f 其 c ⫽ 关 K 兴 ac 兵 ⌬␦ 其 ac (9)
dam’s body or the stability condition of the abutment are not
satisfied, the shape of the dam should be modified, and the above In which
analysis is repeated. 关 K 兴 ⫽ 关 T 兴 Ta 关 K 兴 ⬘a 关 T 兴 a ⫹ 关 T 兴 Tc 关 K 兴 ⬘c 关 T 兴 c (10)
At the beginning of 1930s, the solution of a large freedom
equation presented a hard task, so the load distribution between Eq. 共9兲 is the equilibrium equation of one arch-cantilever element.
the arches and cantilevers had to be ‘‘guessed’’ first, then the For the whole dam, the similar equation relating the displacement
displacements of the arches and cantilevers were calculated sepa- and load increments at all nodes of the arch-cantilever system can
rately. The displacement compatibility condition was used to be obtained by assemblage around each node, the process is the
check the validity of the guessed load distribution. Usually, sev- same as in the finite element method. For simplicity, Eq. 共9兲 is
eral iterations were necessary to get the displacement compatible used to express the equilibrium equation of the arch-cantilever
load distribution. Therefore, researchers and engineers called this system. When the load increments 共water pressure, temperature,
method ‘‘the trial load method.’’ etc.兲 acting on the dam are transformed into nodal forces, the
Fig. 1 shows an arch-cantilever element; parameters of this displacement increments are determined by Eq. 共9兲. Stresses are
element are endued to the four nodes of its middle face as in the then calculated using the elastic beam and arch theories.

978 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Coordinate systems of rock Block b1

Formulation of Block Element Method

The conventional block element method considers the rock blocks

delimited by the discontinuities as rigid body, while discontinui-
ties are governed by elastoplastic or elastoviscoplastic behavior. Fig. 3. Equilibrium of representative block
The governing equation is formulated considering the force and
moment equilibrium condition of blocks, the deformation com-
patibility condition and the elastoviscoplastic constitutive relation in which if the b1-th block is under the jth plane, the function
at discontinuities 共Chen et al. 1994兲. J( j)⫽⫺1, otherwise J( j)⫽1; the 6⫻3 matrix 关 P 兴 j ⫽function of
There are two kinds of coordinate systems in the deduction of the strike and dip angle of the jth plane, the global coordinate of
the governing equations for the deformation and stability analysis the centroid of the block b1 兵 X 其 b1 ⫽ 关 X b1 Y b1 Z b1 兴 T , and the glo-
共Fig. 2兲. Let the x-axis of the global coordinate system point to bal coordinate of the discontinuity plane j 兵 X 其 j ⫽ 关 X j Y j Z j 兴 T on
the North and keep horizontal; the y-axis point to the West and which the integration calculation is processed 共Chen et al. 1994兲.
keep horizontal, too; the z-axis be upright. For each discontinuity
plane, a local coordinate system is needed for convenience. The Deformation Compatibility Equation
origin of the local coordinate system of the jth plane is at the
plane’s centroid. The z j -axis is perpendicular to the plane, the Under the action of the external load increments at the centroids,
direction of the x j -axis and y j -axis coincide with the strike and blocks will move and rotate inducing deformation at discontinui-
dip directions respectively. Besides, the x j -axis must keep hori- ties. Fig. 4 shows the deformation increments 兵 ⌬␦ 其 j on the jth
zontal 共Fig. 2兲. In the following formulation, the b1-th block is discontinuity plane caused by the displacement increments of the
chosen as representative block, the boundary plane of the block is adjacent b1-th and b2-th blocks.
marked with j, the block which contacts the b1th block through
the jth plane is marked as the b2th block.
On the jth plane the stress and deformation increments are
兵 ⌬␴ 其 j ⫽ 关 ⌬␶ zx j ⌬␶ zy j ⌬␴ z j 兴 T and 兵 ⌬␦ 其 j ⫽ 关 ⌬␦ zx j ⌬␦ zy j ⌬␦ z j 兴 T , re-
spectively. External load increments acting at the centroid
of the b1-th block are expressed in the global coordinate
system as six force and moment components 兵 ⌬F 其 b1
⫽ 关 ⌬F Xb1 ⌬F Y b1 ⌬F Zb1 ⌬M Xb1 ⌬M Y b1 ⌬M Zb1 兴 T . The displace-
ment increments of the b1-th block are also expressed in
the global coordinate system as six parallel and rotation
components at the centroid of the block 兵 ⌬U 其 b1
⫽ 关 ⌬U Xb1 ⌬U Y b1 ⌬U Zb1 ⌬W Xb1 ⌬W Y b1 ⌬W Zb1 兴 T .

Force and Moment Equilibrium Equation

The b1-th block is a representative block element, at its centroid
the external load increment should be equalized by the stress
increments acting on the discontinuities around the block 共Fig. 3兲.
Summing stress increments on each boundary plane of the
b1-th block and combining them to the centroid, the equilibrium
equation for the b1-th block is

兵 ⌬F 其 b1 ⫺ 兺j J 共 j 兲 冕 冕 关 P 兴 j 兵 ⌬␴ 其 j dx j dy j ⫽0 b1⫽1,...B Fig. 4. Deformation of discontinuity j caused by adjacent Blocks b1

and b2
(11)

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003 / 979

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

 By kinematics considerations the relationship between the de- the displacements of both the arch-cantilever and block elements
formation of the j-th plane and the displacements of the b1-th and at this interface. The nodal displacement increment of the arch-
b2-th blocks can be expressed as cantilever element in contact with the block bi is given by
兵 ⌬␦ 其 j ⫽ 兵 ⌬␦ 其 b1
j ⫹ 兵 ⌬␦ 其 j
b2
兵 ⌬␦ 其 ac ⫽ 关 T 兴 兵 ⌬U 其 bi (16)

⫽J 共 j 兲关 L 兴 j 共关 M 兴 b1 兵 ⌬U 其 b1 ⫺ 关 M 兴 b2 兵 ⌬U 其 b2 兲 (12) where the displacement increment 兵 ⌬U 其 bi which comprises three

parallel movements and three rotation movements at the centroid
where the coordinate transformation matrix 关 L 兴 j ⫽function of the
of the bi-th block that contacts the arch and cantilever system of
strike and dip angle of the j-th plane; the 3⫻6 matrix
the dam are solved by the Eq. 共15兲; the transformation matrix
关 M 兴 b1 ⫽function of the strike and dip angle of the j-th plane; the
关T兴⫽function of the centroid coordinate of the block and the co-
global coordinate of the centroid of the block b1 兵 X 其 b1
ordinate of the node of the arch-cantilever element which contacts
⫽ 关 X b1 Y b1 Z b1 兴 T ; and the local coordinate of the discontinuity
the block, if a node does not contact the bi-th block element, then
plane j 兵 x 其 j ⫽ 关 x j y j z j 兴 T on which the deformation is concerned
the transformation matrix is a unit matrix. Similarly, the load
共Chen et al. 1994兲.
increments of the four nodes 兵 ⌬ f 其 ac of the arch-cantilever ele-
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

ment which contact the bi-th block element can be transformed

Constitutive Equation into the centroid of the block bi as
According to the general elastoviscoplastic theory 共Owen and 兵 ⌬F 其 bi ⫽ 关 T 兴 T 兵 ⌬ f 其 ac (17)
Hinton 1980兲 with an explicitគscheme, the deformation and stress
increments at any point of the j-th discontinuity plane are linked Insertion of Eqs. 共16兲 and 共18兲 into Eq. 共9兲 yields the following
by the following expression: relation the relation:
兵 ⌬␴ 其 j ⫽ 关 D 兴 j 共 兵 ⌬␦ 其 j ⫺ 兵 ⌬␦ v p 其 j 兲 (13) 兵 ⌬F 其 bi ⫽ 关 K 兴 cp 兵 ⌬U 其 bi (18)
where the elastic matrix is expressed by the normal and tangential in which the coupled matrix 关 K 兴 cp is
stiffness, the viscoplastic deformation increment is given by the
nonassociated flow rule 共Chen et al. 1994兲. 关 K 兴 cp ⫽ 关 T 兴 T 关 K 兴 ac 关 T 兴 (19)
Now the freedoms of the arch-cantilever nodes which contact the
Equilibrium Equation of Whole Block System block element is replaced by the block’s freedoms, in the assem-
blage of the equilibrium equation of the whole arch dam-
Substituting Eqs. 共13兲 and 共12兲 into Eq. 共11兲, the equilibrium foundation system, Eqs. 共9兲 and 共15兲 will be put together and
equation yields rearranged according to the freedoms of the blocks and the free-
doms of the arch-cantilevers, in this process the special attention
关 K 兴 b1b1 兵 ⌬U 其 b1 ⫹ 兺
b2
关 K 兴 b1b2 兵 ⌬U 其 b2 should be paid to eliminate the freedoms of the nodes of the
arch-cantilever elements in contact with the block element using
⫽ 兵 ⌬F 其 b1 ⫹ 兵 ⌬F v p 其 b1 共 b1⫽1,2,....m 兲 (14) the Eqs. 共16兲–共19兲. The final form of the governing equilibrium
equation of the dam and foundation system is
where b2 runs over all the neighbor blocks of Block b1 in cor-
responding to the discontinuity plane j. 关 K 兴 兵 ⌬U 其 ⫽ 兵 ⌬F 其 ⫹ 兵 ⌬F v p 其 (20)
For other blocks, there exist similar equations as above. As-
sembling all of such equations in the block system, the set of where 关K兴, 兵 ⌬U 其 , 兵 ⌬F 其 , 兵 ⌬F v p 其 ⫽respectively, the stiffness ma-
equilibrium equations for the block system is obtained as 共Chen trix, the displacement increment, the load increment, and the
et al. 1994兲 equivalent viscoplastic load increment assembled from Eqs. 共9兲
and 共15兲 taking into account Eqs. 共16兲–共19兲 for the contact nodes.
关 K 兴 b 兵 ⌬U 其 b ⫽ 兵 ⌬F 其 b ⫹ 兵 ⌬F v p 其 b (15)
in which 兵 ⌬U 其 b ⫽ 关 ⌬U X1 ¯⌬W Z1 ¯⌬U Zm ¯⌬W Zm 兴 and T

兵 ⌬F 其 b ⫽ 关 ⌬F X1 ¯⌬M Z1 ¯⌬F Xm ¯⌬M Zm 兴 T ⫽displacement and Application to Xiaowan Double Curvature Arch Dam
force incremental vectors of the block system 共the subscription b Project
means the block system兲; 关 K 兴 b and 兵 ⌬F v p 其 b ⫽stiffness matrix of
the block system and the equivalent force vector of the viscoplas-
Description of Project
tic deformation increment, respectively. At each time step, first
Eq. 共15兲 is solved; then the displacement increments of the block The 292.5 m high Xiaowan double-curvature arch dam is located
system are used to calculate the deformation and stress incre- on the Lanchang River, Yunnan Province in China. A detailed
ments with Eqs. 共12兲 and 共13兲; next, the yield condition at each study allowed the determination of the geology and the topogra-
point on the discontinuities isគchecked; if yield occurs, the visco- phy of the valley. It showed that the rock mass includes three joint
plastic deformation and corresponding load increments are calcu- sets, four alteration zones caused by the heat of the magma during
lated. the geological movement; 22 faults; and ten loosened fissure
bands caused by the erosion of the corrading river. Fig. 5 shows
the dam’s foundation and the major faults concerned. The faults
Formulation of Coupling Analysis Method denoted by the capital letter F are larger scale 共e.g., F 7 , F 11)
meanwhile the small letter f are, used for the small scale faults
Coupling between the arch-cantilever elements of the dam and the 共e.g., f 32 , f 34). The natural underground water table is about
block element of the foundation and abutment is considered 40– 60 m below the ground surface.
through the contact condition at the dam-foundation interface. The elevation of the dam top is 1,245 m and the normal stor-
The interface is assumed to be rigid, which implies the equality of age water level is 1,240 m. Fig. 6 shows the plan of the dam body.

980 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Geology plan of Xiaowan project

Discrete of Dam and Foundation System Loads

A powerful preprocessing procedure has been developed for the Calculation is carried out for the load combination: self-weight,
discrete of complex rock foundation and abutment slopes with water pressure of the normal storage water level 共ⵜ 1,240 m兲,
irregular ground surface 共Wang and Chen 1998兲. This procedure water pressure of the downstream tail water level 共ⵜ 1,004 m兲, silt
is based on the directed body concept proposed by Ikegawa and pressure of the inactive storage level 共ⵜ 1,097 m兲, temperature
Hudson 共1992兲. It deals with both convex and concave blocks in drop, and seepage.
the same process. The foundation of the dam is discretized into It is conventional to calculate the self-weight, the water pres-
2,231 block elements, while the dam is discretizd into 88 arch- sure, the silt pressure, and the temperature drop in the dam. The
cantilever 共9 arch layers and 17 cantilever columns兲. See Fig. 7. seepage flow has a great influence on the stability and deforma-

Constitutive Parameters
The mechanical parameters of discontinuities are assigned on the
basis of a comprehensive study including: geological investiga-
tions; laboratory and field tests; and engineering analogy. The key
parameters are the hydraulic conductivity, the normal stiffness,
the shear stiffness, the cohesion, and the friction angle.
Elastic behavior is assumed for the concrete of the dam with a
Young’s modulus E⫽20.78 G and a Poison’s ratio ␯⫽0.18.

Fig. 7. Axonometric perspective of dam and foundation discrete

Fig. 6. Plan of Xiaowan arch dam system

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003 / 981

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Displacement at elevation of 1,090 m

tion of the foundation and abutment slopes. A method for the m. Other cross sections show similar displacement patterns. Nu-
discontinuity network seepage analysis has been proposed which merical results indicate that the maximum displacements of the
deals with the unconfined seepage problem and takes into account abutment slope occur at the elevation of 1,090–1,010 m. At the
the seepage control such as the drainage and grout curtains 共Chen left abutment slope the maximum displacement is about 0.027 m,
et al. 2000兲. The seepage problem of the Xiaowan project has while at the right abutment slope the maximum displacement is
been analyzed comprehensively by this method and the result is about 0.020 m. The maximum downstream displacement of the
used directly in this paper 共Chen et al. 2000兲. Since the hydraulic arch dam appears on the crest of the crown cantilever, it is equal
conductivity of discontinuities is assumed constant in this study, to 0.154 m. The displacement at the left bank abutment slope is
the seepage problem is solved independently from deformation. greater than that at the right bank, the reason is that there are two
deeper gullies at the upstream and downstream of the left abut-
ment, the presence of the gullies makes the rock mass to be
Discussion of Main Results
moved more easily. It is, therefore, suggested that more attention
In the following discussion the standard convention used for the should be paid to the left bank in the design of the foundation and
terms of left and right abutments 共or banks兲 refers to the sides of abutment treatment.
the dam looking downstream.
Stresses
Displacement Figs. 9 and 10 show the principal stresses of the downstream and
Fig. 8 shows the displacement and movement of the abutment upstream surfaces. The minimum and maximum stresses of the
slope induced by the impounding at the cross section levels of dam are equal to ⫺16.07 and 3.84 MPa, respectively, they occur
1,090 m, when the reservoir is at the normal storage level 1,240 at the base of the crown cantilever downstream and upstream. The

Fig. 9. Principal stresses of downstream surface Fig. 10. Principal stresses of upstream surface

982 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

 ment. This method is convenient to engineering practice because
of its simplicity and standardization. It takes into consideration
the dam-foundation interaction.
The application of this method to the arch dam of Xiaowan
Hydroelectric Project showed its robustness. Nevertheless, further
development of this method is necessary, in particular for the
consideration of the coupling of seepage/deformation in disconti-
nuities 共Tsang and Witherspoon 1981, Raven and Gale 1985兲, the
rock block deformation 共Chen 1994; Wang and Chen 2001兲 and
rock reinforcement 共Chen 1993a,b兲.

Acknowledgments
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

The work is supported by the Natural Science Foundation of

China 共NSFC-59979021兲. The thanks of the writers go also to the
University of Science and Technologies of Lille in France for
providing a visiting professor position for the first writer from
June to July of 2001 for this research.

Fig. 11. Potential failure Case 4 (K⫽2.1) expressed in three

dimension Appendix. Parameters of the Discontinuities Used
in the Calculation „see Tables 1–16…

Table 1. Geological Feature of Joints

maximum tensile stress is a bit larger than the permissible value,
it means that cracking could occur at the dam’s heel. Further Dip direction Dip angle Spacing
comprehensive analysis should be made by the other available Set Bank 共degrees兲 共degrees兲 共m兲
methods concerning the dam heel cracking problem including First: nearly transverse Left 11 77 0.5
FEM and the physical model test, if necessary, the dam’s shape the river bed Right 12 79 0.5
should be optimized. Second: nearly along Left 267 41 3.0
the river bed,
Safety Factor Against Sliding Right 91 39 3.0
low dip angle
A potential failure case is defined as a potential failure part of the Third: nearly along Left 90 90 0.5
slope formed by one block or by the combination of several the river bed,
blocks whose safety factor is comparatively lower than that given high dip angle Right 275 89 0.5
by the design code. If the slope contains a lot of blocks, theoreti-
cally there will be a lot of possible failure cases, the calibration of
Table 2. Geological Features of Faults
the safeties of all cases will cause huge computing time. To mini-
mize the calculation time caused by the huge number of combi- Faults Dip direction 共degrees兲 Dip angle 共degrees兲 Thickness 共m兲
nation of the blocks in the slope, an intelligent search methodol- F7 357 85 3.00
ogy called ‘‘intelligent search method of the potential failure F 19 12 77 0.50
cases’’ 共Xu et al. 2000兲 has been proposed. The basic idea of the F 11 ⫺10–30 83–90 0.30
method is that only the possible block combinations are to be F5 9 80 0.46
considered. Seventeen main potential failure cases are found F 10 5 85 0.11
where the safety factors are low. All of these potential failure F 20 280 85 0.20
cases are not situated directly under the dam, this means that the
F 27 10 75 0.15
global stability of the dam’s abutment is satisfied, but local failure
f 12 8 83 0.32
could occur in the vicinity of the abutment indicated by potential
f 14 10 80 0.30
failure cases. Analysis shows that the left abutment slope includes
f 11 10 80 0.45
more potential failure cases 共12 cases兲 than at the right abutment
f 21 8 70 0.10
slope 共5 cases兲, because of the existence of deeper gullies at the
f 33 270 90 0.17
left abutment slope, which makes it weaker than the right abut-
f 29 90 87 0.20
ment slope. Fig. 11 shows the location and size of the potential
failure Case 4 which has a stability safety factor K⫽2.1. f 19 200 75 0.06
f 10 10 85 0.20
f 15 95 90 0.15
f 36 245 80 0.10
Conclusions
f 17 5 90 0.15
f 30 300 90 0.11
The coupled method is formulated combining the trial load
f 34 275 88 0.50
method and the block element method with elastoviscoplastic dis-
f 32 65 85 0.20
continuities. It provides a possible way to analyze the interaction
f9 10 82 0.16
of the arch dam structure and its foundation as well as the abut-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003 / 983

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

 Table 3. Geological Features of Alteration Bands Table 8. Hydraulic Conductivity of Loosen Fissure Bands 共m m/s兲
Alteration Dip direction Dip angle Thickness Rock classification
Loosen
bands 共degrees兲 共degrees兲 共m兲
fissure band IV III I⫹II
E1 276 87 0.25–7.0
LD1–LD5 0.15 0.015 0.0008
E5 265–280 90 3.50–20.0 RD1–RD5 0.15 0.015 0.0008
E4 265 87 0.30–13.5
E8 300 90 2.0–12.0

Table 4. Geological Feature of Loosen Fissure Bands Table 9. Stiffness of Joints 共Mpa/m兲

Loosen fissure Dip direction Dip angle Thickness Second set: Third set:
band 共degrees兲 共degrees兲 共m兲 First set: nearly along nearly along
nearly transverse the river bed, the river bed,
LD1 – LD5 267 35 0.05 the river bed low dip angle high dip angle
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

RD1 – RD5 91 35 0.05 Left Right Left Right Left Right

Stiffness Joint bank bank bank bank bank bank
Table 5. Hydraulic Conductivity of Joints 共m m/s兲 Normal IV 92 92 1,677 1,630 169 169
First set: nearly Second set: nearly Third set: nearly Kn III 127 126 2,259 2,196 228 228
transverse the along the river along the river bed, I⫹II 130 130 2,310 2,245 233 233
river bed bed, low dip angle high dip angle Shear IV 34 34 620 603 63 63
Rock Left Right Left Right Left Right Ks III 47 47 836 812 84 84
classification bank bank bank bank bank bank I⫹II 48 48 854 830 86 86
IV 0.19 0.19 7.5e-3 7.8e-3 2.5 2.5
III 3.8e-3 3.8e-3 1.5e-4 1.6e-4 0.05 0.05
I⫹II 7.6e-5 7.6e-5 1.5e-6 1.6e-6 4.0e-4 4.0e-4
Table 10. Stiffness of Faults 共MPa/m兲

Table 6. Hydraulic Conductivity of Faults 共m m/s兲 Normal stiffness Kn Shear stiffness Ks

Rock classification Fault IV III I⫹II IV III I⫹II

Fault IV III I⫹II F7 30 38 152 11 14 56

F19 613 766 1,380 226 283 511
F7 300.0 30.0 3.0
F11 477 596 2,150 177 221 794
F19 2.8 0.28 0.028
F5 275 344 1,030 101 127 381
F11 2.85 0.285 0.0285
F10 613 766 1,380 226 283 511
F5 13.3 1.33 0.133
F20 444 555 1,110 166 208 409
F10 0.616 0.062 0.0062
F27 444 555 1,110 166 208 409
F20 0.2 0.02 0.002
f 12 1,674 2,090 4,760 618 772 1,760
F27 0.15 0.015 0.0015
f 14 2,224 2,780 6,350 824 1,030 2,350
f 12 0.179 0.018 0.0018
f 11 1,488 1,860 4,230 552 690 1,570
f 14 0.168 0.017 0.0017
f 21 2,224 2,780 6,350 824 1,030 2,350
f 11 0.252 0.025 0.0025
f 33 2,224 2,780 6,350 824 1,030 2,350
f 21 0.056 0.0056 0.0006
f 29 2,224 2,780 6,350 824 1,030 2,350
f 33 0.391 0.039 0.0039
f 19 2,224 2,780 6,350 824 1,030 2,350
f 29 0.46 0.046 0.0046
f 10 2,224 2,780 6,350 824 1,030 2,350
f 19 0.034 0.0034 0.0003
f 15 1,912 2,390 5,440 704 880 2,010
f 10 0.112 0.011 0.0011
f 36 1,912 2,390 5,440 704 880 2,010
f 15 0.345 0.035 0.0035
f 17 1,912 2,390 5,440 704 880 2,010
f 36 0.23 0.023 0.0023
f 30 5,008 6,260 16,570 1,552 1,940 6,130
f 17 0.084 0.0084 0.0008
f 34 1,456 1,820 2,240 288 360 830
f 30 0.253 0.025 0.0025
f 32 1,912 2,390 5,440 704 880 2,010
f 34 1.15 0.115 0.0115
f9 4,048 5,060 11,550 1,496 1,870 4,270
f 32 0.46 0.046 0.0046
f9 0.089 0.009 0.0009

Table 7. Hydraulic Conductivity of Alteration Bands 共m m/s兲 Table 11. Stiffness of Alteration Bands 共MPa/m兲
Rock classification Normal stiffness Kn Shear stiffness Ks
Alteration Alteration
band IV III I⫹II band IV III I⫹II IV III I⫹II
E1 20.2 2.07 0.202 E1 5,000 8,000 10,000 1,850 2,960 3,700
E5 67.2 6.72 0.672 E5 5,000 8,000 10,000 1,850 2,960 3,700
E4 39.2 3.92 0.392 E4 5,000 8,000 10,000 1,850 2,960 3,700
E8 700 70.0 7.0 E8 30 40 200 10 15 74

984 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

 Table 12. Stiffness of Loosen Fissure Bands 共Mpa/m兲 Table 16. Strength Parameters of Loosen Fissure Bands
Normal stiffness Kn Shear stiffness Ks Friction angle ␾ 共degrees兲 Cohesion c 共MPa兲
Loosen Loosen
fissure band IV III I⫹II IV III I⫹II fissure band IV III I⫹II IV III I⫹II
LD1–LD5 10,515 11,330 11,585 3,890 4,190 4,285 LD1–LD5 26 35 50 0.121 0.228 1.284
RD1–RD5 10,515 11,330 11,585 3,890 4,190 4,285 RD1–RD5 26 35 50 0.121 0.228 1.284

References
Table 13. Strength Parameters of Joints
Friction angle ␾ 共degrees兲 Cohesion c 共MPa兲 Chan, H. C., and Einstein, H. H. 共1981兲. ‘‘Approach to complete limit
equilibrium analysis for rock wedges the method of ‘artificial sup-
Set IV III I⫹II IV III I⫹II
ports’.’’ Rock Mech., 14共2兲, 59– 86.
First set: nearly 22 38 51 0.094 0.278 1.52 Chen, S. H. 共1993a兲. ‘‘Numerical analysis and model test of rock wedge
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

transverse the river in slope.’’ Proc., Int. Symp. on Assessment and Prevention of Failure
bed Phenomena in Rock Engineering, Istanbul, Turkey, A. A. Balkema,
Second set: nearly 22 38 51 0.094 0.278 1.52 Rotterdam, The Netherlands, 425– 429.
Chen, S. H. 共1993b兲. ‘‘Analysis of reinforced rock foundation using
along the river
elastic-viscoplastic block theory.’’ Proc., 1993 ISRM Int. Symp.-
bed, low dip angle
EUROCK 93, Lisbon, Portugal, A. A. Balkema, Rotterdam, The Neth-
Third set: nearly 22 38 51 0.094 0.278 1.52 erlands, 45–51
along the river Chen, S. H. 共1994兲. ‘‘A new development in elastic-viscoplastic block
bed, high dip angle theory of rock masses.’’ Computer methods and advances in geome-
chanics, A. A. Balkema, W. Va., 2235–2240.
Chen, S. H., Shen, B. K., and Huang, M. H. 共1994兲. ‘‘Stochastic elastic-
viscoplastic analysis for discontinuous rock masses.’’ Int. J. Numer.
Methods Eng., 37共14兲, 2429–2444.
Table 14. Strength Parameters of Faults Chen, S. H., Wang, W. M., She, C. X., and Xu, M. Y. 共2000兲. ‘‘Uncon-
Friction angle ␾ 共degrees兲 Cohesion c 共MPa兲 fined seepage analysis of discontinuous rock slope.’’ J. Hydrodynam.,
12共3兲, 75– 86.
Fault IV III I⫹II IV III I⫹II
Copen, M. D., Lindholm, E. A., and Tarbox, G. S. 共1977兲. ‘‘Design of
F7 11 14 16 0.02 0.025 0.03 concrete dams.’’ Handbook of dam engineering, Alfred R Golze, ed.,
F19 17 22 24 0.03 0.04 0.045 Van Nostrand Reinhold, New York.
F11 19 22 24 0.035 0.04 0.045 Cundall, P. A. 共1971兲. ‘‘A computer model for simulating progressive,
large-scale movement in blocky rock system.’’ Rock fracture, Proc.,
F5 17 19 22 0.03 0.035 0.04
Symp. Int. Society Rock Mechanics, 129–136.
F10 17 22 24 0.03 0.04 0.045 Cundall, P. A. 共1988兲. ‘‘Formulation of three-dimensional distinct element
F20 17 22 24 0.03 0.04 0.045 model. I: A scheme to detect and represent contacts in system com-
F27 17 22 24 0.03 0.04 0.045 posed of many polyhedral blocks.’’ Int. J. Rock Mech. Min. Sci. Geo-
f 12 17 19 22 0.03 0.035 0.04 mech. Abstr., 25共3兲, 107–116.
f 14 17 19 22 0.03 0.035 0.04 Guzina, B., and Tucovic, I. 共1969兲. ‘‘Determining the maximum three
f 11 19 22 24 0.035 0.04 0.045 dimensional stability of a rock wedge.’’ Water Power, 21共10兲, 381–
f 21 19 22 24 0.035 0.04 0.045 385.
Hart, R., Cundall, P. A., and Lemos, J. 共1988兲. ‘‘Formulation of three-
f 33 19 22 24 0.035 0.04 0.045
dimensional distinct element model. II: Mechanical calculations for
f 29 19 22 24 0.035 0.04 0.045 motion and interaction of a system composed of many polyhedral
f 19 19 22 24 0.035 0.04 0.045 blocks.’’ Int. J. Rock Mech. Min. Sci. Geomech. Abstr., 25共3兲, 117–
f 10 19 22 24 0.035 0.04 0.045 125.
f 15 19 22 24 0.035 0.04 0.045 Ikegawa, Y., and Hudson, J. A. 共1992兲. ‘‘A novel automatic identification
F36 19 22 24 0.035 0.04 0.045 system for three-dimensional multi-block system.’’ Eng. Comput.,
f 17 19 22 24 0.035 0.04 0.045 9共2兲, 169–179.
f 30 19 22 24 0.035 0.04 0.045 Lin, S. Z., and Yang, Z. H. 共1987兲. ‘‘Load distribution-stiffness method
for stress analysis of arch dams.’’ J. Hydraul. Eng., 共1兲, 17–25.
f 34 19 22 24 0.035 0.04 0.045
Londe, P. 共1965兲. ‘‘Une method d’analyse o’trois dimensions de la stabi-
f 32 19 22 24 0.035 0.04 0.045 lite d’une rive rocheuse.’’ Ann. Ponts Chaussees, 共1兲, 37– 60.
f9 19 22 24 0.035 0.04 0.045 Owen, D. R. J., and Hinton, E. 共1980兲. Finite elements in plasticity:
Theory and practice, Pineridge Press Ltd., Swansea, U.K.
Raven, K. G., and Gale, J. E. 共1985兲. ‘‘Water flow in a natural rock
fracture as a function of stress and sample size.’’ Int. J. Rock Mech.
Table 15. Strength Parameters of Alteration Bands Min. Sci. Geomech. Abstr., 22共4兲, 251–261.
Friction angle ␾ 共degrees兲 Cohesion c 共MPa兲 Shi, G. H. 共1992兲. ‘‘Discontinuous deformation analysis: a new numerical
Alteration model for the statics and dynamics of deformable block structures.’’
band IV III I⫹II IV III I⫹II Eng. Comput., 9共2兲, 157–168.
E1 39 42 45 0.35 0.45 0.5 Suresh, P., and Natarajan, R. 共1981兲. ‘‘Analysis of laminated composite
E5 39 42 45 0.35 0.45 0.5 shell structures by finite element method.’’ Comput. Struct., 14共3– 4兲,
225–230.
E4 39 42 45 0.35 0.45 0.5
Tsang, Y. W., and Witherspoon, P. A. 共1981兲. ‘‘Hydromechanical behav-
E8 17 22 24 0.03 0.04 0.045
iour of a deformable rock fracture subject to normal stress.’’ J. Geo-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003 / 985

J. Geotech. Geoenviron. Eng. 2003.129:977-986.

 phys. Res. B, 86共B10兲, 9287–9298. puter Methods and Advances in Geomechanics, Arizona.
Wang, W. M., and Chen, S. H. 共1998兲. ‘‘Automatic identification method Xu, M. Y., Wang, W. M., and Chen, S. H. 共2000兲. ‘‘Research on the
for three-dimensional rock block systems.’’ J. Wuhan University of dangerous sliding-block combination of rock slopes.’’ Rock Soil
Hydraulic and Electric Engineering, 31共5兲, 51–55. Mech., 21共2兲, 148 –151.
Wang, W. M., and Chen, S. H. 共2001兲. ‘‘A new implementation of elastic- Zhu, B. F., Rao, B., and Jia, J. S. 共1987兲. ‘‘Stress analysis of noncircular
viscoplastic block theory for rock masses.’’ 10th Int. Conf. on Com- arch dam with variable thickness.’’ J. Hydraul. Eng., 共11兲, 17–25.
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

986 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2003

J. Geotech. Geoenviron. Eng. 2003.129:977-986.