You are on page 1of 12

Engineering Structures 207 (2020) 110287

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Study on the nonlinear deformation and failure mechanism of a high arch T


dam and foundation based on geomechanical model test

Zhuofu Tao, Yaoru Liu , Qiang Yang, Shouguang Wang
State Key Laboratory of Hydroscience and Hydraulic Engineering, Tsinghua University, Beijing 100084, China

A R T I C LE I N FO A B S T R A C T

Keywords: Mengdigou double-curvature arch dam, located in the southwest of China, is a 200-meter-high dam with
Geomechanical model test complex geological conditions. Failure analysis of the arch dam coupled with the foundation is important for
Arch dam evaluation of the stability. In this regard, a geomechanical model test was carried out. The data pertaining to
Failure mechanism displacements, stresses, and acoustic emissions, and the entire process of the failure occurring during over-
Nonlinear deformation
loading were obtained. During the overloading process, crack initiates in the upstream dam heel. Then, a crack
appears downstream at the dam toe as the load increases. After the damage of the dam toe, cracks occur in the
faults around the abutment. Meanwhile, cracks develop rapidly in the arch dam. Eventually, the dam-foundation
system loses its ultimate bearing capacity and fails. The results of the model test were compared with those of
numerical calculation. Then, the nonlinear deformation and failure mechanism was analyzed. Finally, the safety
factor of cracking of the arch dam, the safety factor of the initial nonlinear deformation, and the safety factor of
the ultimate bearing capacity were obtained, which reveal that the design of the arch dam is reasonable.

1. Introduction method [10–12], discontinuous deformation analysis method [13], and


numerical manifold method [14] are widely used in the analysis of arch
An arch dam is a multiple statically indeterminate structure with its dams. Through these numerical methods, the arch dam and foundation
own capability of stress adjustment. During the loading process, the can be analyzed as a whole, and local cracking can also be carefully
failure of arch dams is gradual. After local cracking, an arch dam still studied. In this context, geomechanical model test can simulate the
retains a certain bearing capacity until ultimate failure occurs [1,2]. entire failure process of arch dams during overloading, and the local
The geological conditions of arch dams are complex and the loads huge. cracking, ultimate failure, and failure mode can be investigated. The
Therefore, it is necessary to carry out a detailed stability analysis before geomechanical model test displays unique advantages in the stability
the construction of an arch dam.Table 1. analysis of arch dams.
Mengdigou double-curvature arch dam is located on the Yalong In the 1960 s, Fumagalli in Italy first proposed the geomechanical
River in the southwest of China. The dam crest elevation is EL 2259 m model test method [15] and applied it to the stability analysis of arch
and the bottom elevation EL 2059 m, with the dam height being 200 m. dams [16,17]. After that, a large number of geomechanical model tests
The installed capacity of this hydropower station is 2.4 × 106 kW and were carried out in different countries [18–21], and the technology
the average annual power generation is 9.29 × 109 kWh. The geological associated with the model test also developed. Owing to the structural
conditions of the foundation are complex. There are several faults, and loading characteristics and gradual failure of high arch dams, it is
joints, and alteration zones in the foundation of the arch dam that may difficult to evaluate the overall stability of arch dams based on a single
affect the overall stability of the dam to some extent. index. Based on the results of a series of model tests on the arch dams in
Traditional analysis methods of arch dams, such as trial load China, Zhou et al. [22] proposed three safety factors to evaluate the
method [3] and rigid body limit equilibrium method [4], are based on overall stability of arch dams: the safety factor of cracking K1, the safety
linear elasticity calculations and do not fully consider the interaction factor of initial nonlinear deformation K2, and the safety factor of the
between an arch dam and its foundation. With the development of ultimate bearing capacity K3.
numerical methods, finite element method [5], boundary finite element Although we can obtain a comprehensive understanding of the
method [6,7], extended finite element method [8,9], discrete element failure modes of high arch dams, the mechanism of nonlinear


Corresponding author.
E-mail address: liuyaoru@tsinghua.edu.cn (Y. Liu).

https://doi.org/10.1016/j.engstruct.2020.110287
Received 26 June 2019; Received in revised form 23 December 2019; Accepted 22 January 2020
0141-0296/ © 2020 Elsevier Ltd. All rights reserved.
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Table 1 deformation and failure of high arch dams is still unclear. Additionally,
Similarity coefficients of geomechanical model test. during the overloading process, the arch dam and foundation interact
Similarity coefficient Value Similarity coefficient Value with each other. Therefore, it is necessary to analyze the failure me-
chanism associated with the coupling between an arch dam and its
Cγ 1.0 CE 200 foundation. In this research, geomechanical model test of an arch dam
CL 200 Cμ 1.0 was carried out. The deformation of the arch dam and foundation, the
Cε 1.0 Cf 1.0 stress distribution in the dam, and the failure mode were obtained.
Cσ 200 Cτ 200
Then, through the evolution law of displacements and cracks, the me-
Cc 200 Cδ 200
chanism of nonlinear deformation and failure of high arch dams was
analyzed in detail.

2. Similarity theory and model test technique

2.1. Principles of similarity

In order to obtain a response similar to that of the prototype, the


physical model must satisfy some basic conditions such as similar
geometric sizes, similar constitutive relations, similar mass or gravity
relations, and similar initial and boundary conditions. The similarity
coefficient between the prototype and the model is defined as follows
[23]:
ip
Ci =
im (1)

where ip represents the physical quantity of the prototype and im


represents that of the model. In this equation, i can represent different
physical quantities, including geometric size (L), displacement (δ ), co-
hesion (c), shear strength (τ), stress (σ), friction coefficient (f), strain (ε),
Poisson's ratio (μ), unit weight (γ), and Young's modulus (E).
Fig. 1. Similarity of stress–strain process between model and prototype. Based on principles of similitude[16], for linearly-elastic static
models, the following conditions must be satisfied.
According to the equilibrium equation, it can be deduced that

=1
Cγ CL (2)

According to the constitutive conditions, it can be deduced that


Cε CE
Cμ = 1, =1
Cσ (3)

According to the geometric equation, it can be deduced that


Cε CL
=1
Cδ (4)

The geomechanical model test belongs to the failure test. After the
linear elastic stage, the stress–strain relationship between the prototype
and the model should remain similar, as shown in the Fig. 1. In addition
to the above conditions, the following conditions shall also be met in
the test.
Cε = 1 (5)

CE
=1
Cσ (6)

CL
=1
Cδ (7)

According to the Mohr Coulomb strength theory, the strength


parameters of the model and the prototype should also be similar.
Cσ = Cτ = Cc (8)

Cf = 1 (9)
Fig. 2. Model of Mengdigou arch dam: (a) upstream; (b) downstream and
According to the above conditions, it can be deduced:
slopes.
CE = Cγ ·CL (10)

Cμ = Cε = Cf = Cγ (11)

2
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Table 2
Mechanical parameters of concrete and rock masses.
Rock type Deformation modulus E0 (MPa) Friction coefficient f Cohesion c (MPa) Poisson’s ratio μ Uniaxial compressive strength R (MPa)

3
Prototype × 10 Model Prototype Model Prototype Model Prototype Model Prototype

Concrete 24 0.12 1.6 1.6 2.5 0.0125 0.27 0.27 35–40


II 20–25 0.10–0.125 1.2–1.4 1.2–1.4 1.7–2.3 0.0085–0.0115 0.22–0.24 0.22–0.24 70–100
III III1 10–13 0.05–0.065 1.0–1.2 1.0–1.2 1.0–1.5 0.005–0.0075 0.25–0.29 0.25–0.29 50–100
III2 4.0–6.0 0.02–0.03 0.80–1.0 0.80–1.0 0.70–1.0 0.0035–0.005 0.30–0.33 0.30–0.33 40–70
IV 1.5–3.0 0.0075–0.015 0.60–0.80 0.60–0.80 0.30–0.60 0.0015–0.003 0.33–0.35 0.33–0.35 20–60
V 0.25–0.55 0.00125–0.00275 0.40–0.55 0.40–0.55 0.05–0.25 0.00025–0.00125 > 0.35 > 0.35 < 20

Note: The classification of rock types is based on the classification standard of engineering rock mass in China, GB50218-94 [34].

Fig. 3. Layout of the rock types and faults.

Table 3
Physical properties and mechanical parameters of faults.
Location Geological feature Thickness E0 (MPa) f c (MPa)

Prototype (m) Model (cm) Prototype ×103 Model Prototype Model Prototype Model

Left bank f1 0.15 0.075 2.5 12.5 0.40 0.40 0.06 0.0003
f2 0.20 0.10 2.5 12.5 0.40 0.40 0.06 0.0003
f3 0.225 0.1125 2.5 12.5 0.40 0.40 0.06 0.0003
f6 0.20 0.10 2.5 12.5 0.40 0.40 0.06 0.0003
dominant fissures ① – – – – 0.60 0.60 0.10 0.0005
dominant fissures ④ – – – – 0.70 0.70 0.15 0.00075
HL02 0.02 0.01 1.5 7.5 0.50 0.50 0.05 0.00025
Right bank f4 1.5 0.75 2.5 12.5 0.225 0.225 0.006 0.00003
f5 1.25 0.625 2.5 12.5 0.325 0.325 0.03 0.00015
AZR02 2.5 1.25 1.5 7.5 0.475 0.475 0.15 0.00075
AZR03 0.75 0.375 0.25 1.25 0.475 0.475 0.15 0.00075
AZR04 0.80 0.40 0.25 1.25 0.325 0.325 0.03 0.00015
AZR05 1.0 0.50 0.25 1.25 0.475 0.475 0.15 0.00075
AZR06 0.45 0.225 0.25 1.25 0.475 0.475 0.15 0.00075
AZR07 1.0 0.50 0.25 1.25 0.325 0.325 0.03 0.00015
HL01 0.02 0.01 1.5 7.5 0.50 0.50 0.05 0.00025

Cσ = Cτ = Cc = Cδ = CE (12) Thousands of small blocks were used to construct the model layer-by-
layer. In different zones of the model, the shape and size of the small
where CE , Cγ , CL , Cμ , Cε , Cf , Cσ , Cτ , Cc , andCδ are the similarity
blocks differ in order to simulate the geological condition. These small
coefficients of Young's modulus, unit weight, geometric size, Poisson's
blocks were combined by a tailor-made glue that can simulate the
ratio, strain, friction coefficient, stress, shear strength, cohesion, and
connectivity and surface friction of rock mass.
displacement, respectively. Generally, Cγ is considered as 1.0, while CL
With regard to the model material, a great amount of research has
is determined by the practical experimental conditions. In this test,
been carried out [24–26]. According to the principles of similarity, the
CL = 200 . The other similarity coefficients can be easily obtained.
material used in the model needs to meet the requirements of high
density, low strength, and low modulus [24]. In this test, a kind of
2.2. Masonry method and similarity materials mixed powder, composed of barite powder, bentonite, and glue, was
used. The mixture was processed into a small block of a certain shape
Small block masonry technique [23] was adopted in this test.

3
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Fig. 4. Schematic diagram of the upstream water load.

Fig. 5. Arrangement of the loading equipment.

with the help of a special machine, and then, the dam and its founda-
tion were constructed block-by-block. Actually, Young's moduli of the
dam and the foundation are different, while Young's modulus of the
rock mass in different regions of the foundation also varies. Therefore,
groups of small blocks with different densities were fabricated and
uniaxial compression tests were carried out to measure Young's moduli
of the small blocks at different densities, and then, the corresponding
densities for the dam and the rock mass in different regions were se-
lected.

2.3. Method of evaluation of the overall stability

In this test, an overloading process was employed. The higher water


density method was adopted, which involved increasing the water
Fig. 6. Layout of strain gauges and displacement transducers: (a) upstream
density while keeping the water head constant. The model test involves
surface of the dam; (b) downstream surface of the dam; (c) foundation.
a multi-step incremental loading process, and the overloading factor K
can be defined as
displacement curve of the crown cantilever. When a variation appears
P in the displacement curve of the crown cantilever, it indicates that the
K=
P0 (13) structure has entered the stage of nonlinear deformation. The value of
where P is the current load imposed on the dam and P0 is the load K3 can be obtained directly during the experiment. When the load
under normal water. Under normal conditions, the overloading factor cannot be increased, it shows that the structure has reached the ulti-
K = 1.0 , which implies that the current load equals the design load. mate bearing capacity.
Based on the results of several model tests on arch dams, Zhou et al.
[22] proposed three safety factors to evaluate the overall stability of 3. Design of geomechanical model test
arch dams, which are correlated with the overload factor K. With the
help of these three safety factors, the entire failure process of an arch 3.1. Simulation of the dam and foundation
dam during overloading can be described. They are K1, the safety factor
of cracking; K2, the safety factor of initial nonlinear deformation; and This test was carried out in the hydraulic structural laboratory of
K3, the safety factor of the ultimate bearing capacity. Generally, K1 is Tsinghua University. In order to fully simulate the influence of the
determined from the strain and acoustic emission data obtained from geological conditions and reduce the boundary effect, the model con-
around the dam heel. When a variation appears on the strain curve or sidered large-scale simulation. The simulation range was 100 m up-
the first effective signal due to acoustic emission is detected, it indicates stream, 400 m downstream, 100 m above the dam crest, 100 m below
that cracking has occurred near the dam heel. At this time, the over- the riverbed, and 400 m on both sides of the dam. The model is shown
loading factor is K1. The value of K2 can be determined from the in Fig. 2.

4
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Table 4
Layout of internal displacement transducers.
Number Elevation Monitoring area

No.1 2057 m upstream dam heel


No.2 2059 m upstream of fault f3 on the left bank
No.3 2059 m alteration zone AZR02 on the right bank
No.4 2090 m upstream of dominant fissures ④ on the left bank
No.5 2090 m intersection of dominant fissures ④ and fault f3
No.6 2090 m downstream of alteration zone AZR03 on the right bank
No.7 2090 m intersection of alteration zones AZR03 and AZR04
No.8 2120 m intersection of dominant fissures ④ and fault f2
No.9 2120 m downstream of dominant fissures ④ on the left bank
No.10 2120 m upstream of alteration zone AZR04 on the right bank
No.11 2150 m intersection of dominant fissures ④ and fault f2
No.12 2150 m intersection of alteration zones AZR05 and AZR07
No.13 2150 m intersection of fault f1 and fault f6
No.14 2180 m fault f2 near the dam toe on the left bank
No.15 2180 m fault f4 on the right bank
No.16 2180 m intersection of alteration zone AZR07 and fault f4
No.17 2210 m fault f1 on the left bank
No.18 2240 m upstream of fault f6 near the dam toe
No.19 2240 m fault f1 near the dam toe on the left bank

Table 5
Displacements of downstream dam under normal condition obtained by the
model test (mm).
Elevation Right arch Right arch Crown Left arch Left arch
abutment cantilever abutment

2259 m X: −3.6 X: −5.3 X: −0.5 X: 4.4 X: 4.0


Y: 4.3 Y: 22.5 Y: 60.1 Y: 32.9 Y: 5.5
2210 m X: −2.9 X: −4.3 X: 2.2
Y: 18.0 Y: 50.1 Y: 30.7
2150 m X: −5.4 X: −3.0 X: 3.1
Y: 9.2 Y: 38.4 Y: 8.5
2120 m X: −5.4 X: 4.8
Y: 7.0 Y: 13.0
2090 m X: −6.4 X: −2.5 X: unavailable
Y: 4.9 Y: 21.2 Y: 7.8
2059 m X: 2.2
Y: 2.7

Fig. 7. Relationship curves between the displacements (mm) of crown canti-


The properties of rock mass in different regions are different. The lever and the overloading factor K: (a) displacements along the river; (b) dis-
mechanical properties of the dam and the rock mass are listed in placements along the dam axis.
Table 2. The rock mass in the foundation was divided into five cate-
gories for simulation and the classification of rock types was based on
to form an approximate triangularly distributed load. Meanwhile, jacks
the classification standard of engineering rock mass in China, GB50218-
are connected to the oil pumps, and the pressure is controlled by ac-
94. The test simulated the actual geology condition as much as possible
curate pressure gauges. Cushion blocks and shims are arranged to avoid
in order to comprehensively reveal the interaction between the arch
inhomogeneous loading. The force and the corresponding internal
dam and the foundation.
pressure of each group were calculated based on the equivalences of the
There are faults f1, f2 , f3 , f4 , f5 , f6 ; structural planes HL01, HL02;
magnitude and moment of the hydrostatic pressure. The model test
dominant fissures ①, dominant fissures ④; and alteration zones AZR02,
involves a multi-step incremental loading process in which the loading
AZR03, AZR04, AZR05, AZR06, AZR07 in the foundation of Mengdigou arch
increment is 50% of the design load per step.
dam, and some of them are shown in Fig. 3. These structures directly
affect the safety of the dam, and were all simulated in the test. Their
mechanical properties are presented in Table 3. Reinforcement mea- 3.3. Sensors and data acquisition
sures for the foundation were also taken into consideration in the test.
These measures mainly included concrete replacement of faults f4 and In order to comprehensively monitor the stress and deformation of
alteration zones AZR03, AZR04 and AZR07. When constructing the model, the dam and foundation, strain gauges, resistive displacement trans-
according to the design scheme, the blocks with the corresponding ducers, acoustic emission sensors and cameras were arranged. A total of
characteristic parameters were used in the replacement area. 247 strain gauges were pasted on the upstream and downstream sur-
faces of the dam to monitor the strain during loading, and then, the
3.2. Loading mode stresses can be obtained throughout based on generalized Hooke's Law.
A total of 116 external displacement transducers were fixed at 54
In this test, only the upstream hydraulic pressure was simulated and measuring points on the downstream surface and foundation to monitor
overloading method was adopted, which was achieved by increasing the absolute displacements of the arch dam and slope. The arrangement
the specific weight of water. As shown in Fig. 4 and Fig. 5, thirty-six of the strain gauges and displacement transducers is shown in Fig. 6.
tailor-made jacks are used for loading, which are divided into six layers The thin black lines represent the strain gauges, whereas the solid black

5
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Fig. 8. Finite element model of Mengdigou arch dam.

Fig. 10. Displacement curves of the internal displacement transducers in fault


f1.

Fig. 9. Distribution of dam displacements (mm) obtained by numerical calcu-


lation: (a) displacements along the river; (b) displacements along the dam axis.

Table 6
Displacements of downstream dam under normal condition obtained by nu-
merical calculation (mm).
Elevation Right arch abutment Crown cantilever Left arch abutment

2259 m X: −3.3 X: −0.7 X: 2.4


Y: 4.3 Y: 51.8 Y: 3.7
2210 m X: −0.8
Y: 56.1
2150 m X: −4.0 X: −0.7 X: 3.0
Y: 9.6 Y: 46.2 Y: 10.3
2120 m X: −4.0 X: 2.4
Y: 9.7 Y: 9.8
2090 m X: −3.0 X: −0.2 X: 1.7
Y: 8.1 Y: 20.1 Y: 7.9
2059 m X: −0.2 Fig. 11. Displacement curves of the internal displacement transducers in fault
Y: 5.5 f2 .

squares represent the external displacement transducers. In addition, to shear deformation of the fault fissures in the abutment rock mass. The
analyze the influence of the geological structures on the overall stabi- layout of the internal displacement transducers is detailed in Table 4.
lity, 19 internal displacement transducers were set up to monitor the Meanwhile, 15 acoustic emission sensors were arranged on the dam and
foundation to monitor the internal damage during loading with the help

6
Z. Tao, et al. Engineering Structures 207 (2020) 110287

right bank and concrete replacement is carried out. However, it is much


more complicated in the foundation of the left bank, where several
faults such as f1, f2 , f3 , and f6 are developed. Perhaps, this is the reason
for the larger deformation observed on the left side of the dam.
Fig. 7 shows the relationship curves between the overloading factor
K and the displacements along the river and dam axis of the measuring
points on the downstream crown cantilever. With the increase in K, the
displacements increase gradually. Before K reaches 4.5, the displace-
ments vary linearly with the increase in K. After that, the displacements
increase faster and nonlinear deformation appears.

4.1.2. Comparisons with the numerical results


For Mengdigou arch dam, a three-dimensional nonlinear finite ele-
ment analysis was also carried out. MSC Patran was utilized for mod-
eling. Eight-node hexahedron elements with a small number of six-node
pentahedron elements were used in the model. As with the model test,
the complex geological conditions of the arch dam were fully simulated
through numerical calculation. The finite element model is shown in
Fig. 12. Displacement curve of the internal displacement transducer in fault f3 . Fig. 8.
TFINE [35] program developed by the Department of Hydraulic
Engineering at Tsinghua University was used for the numerical calcu-
lation. The characteristics of high arch dams and foundations can easily
lead to abnormal grids and highly uneven material zoning, which ex-
hibit a considerably adverse impact on the convergence. The TFINE
program adopts a robust iterative algorithm and integral strategy based
on Drucker-Prager criterion, which improves the stability of numerical
calculation. The ideal elastoplastic model was adopted as the con-
stitutive model and the Drucker-Prager criterion was used as the yield
condition, as shown in Eq. (6).

f = αI1 + J2 − k ≤ 0 (14)

I1 = σ1 + σ2 + σ3 (15)

1
J2 = [(σ1 − σ2 )2 + σ2 − σ32 + (σ3 − σ1 )2]
6 (16)

where σ1, σ2 and σ3 are the principal stresses.α and k are obtained by
fitting the Mohr-Coulomb criterion.
The displacements of the downstream dam under normal condition
(K = 1.0) obtained by numerical calculation are shown in Fig. 9 and
Fig. 13. Displacement curve of the internal displacement transducer in fault f4 .
Table 6 [27]. In Fig. 9, we can see that the distribution of the dam
displacements is reasonable. Comparing the data in Table 5 and
of acoustic signals. Eleven cameras were placed around the arch dam to Table 6, the values of the displacements along the river are close, except
monitor the entire cracking process in real time. for those near the dam crest. In the model test, the displacement along
UCAM-70A, manufactured by Kyowa Electronic Instruments, and the river is the largest at the top of the crown cantilever (60.1 mm),
PXIe-1075, manufactured by National Instruments, were employed for whereas the displacement along the river is the largest at EL 2210 m
data acquisition; they can feedback the strain and displacement data in (56.1 mm) in the numerical calculation. This may be attributed to the
real time. loading mode of the model test. As for the displacements along the dam
axis, the value obtained from the model test is slightly larger than that
4. Analysis of the nonlinear deformation of the arch dam and obtained from the numerical calculation. Generally, the results of the
foundation model test and numerical calculation are in good agreement.

4.1. Displacements (referred to the prototype) of the dam 4.2. Deformation (referred to the prototype) of the faults

4.1.1. Results of the model test Fig. 10 shows the monitoring curves obtained from the internal
According to the results of the geomechanical model test, the dis- displacement transducers in fault f1. We can see from the figure that
placements of the downstream dam obtained under normal condition before K reaches 5.5, the increase in the displacements is slow and
(K = 1.0) are shown in Table 5. The X-direction is along the dam axis linear. Subsequently, the increase is nonlinear for the displacement
(positive for deformation to the left bank) and the Y-direction is along curves. It is indicated that cracks occur in fault f1 at EL 2210 m up-
the river (positive for deformation downstream). The displacement stream and EL 2240 m downstream when K reaches 5.5.
along the river is the largest at the top of the crown cantilever Fig. 11 shows the monitoring curves obtained from the internal
(60.1 mm) and the smallest at the bottom of the crown cantilever displacement transducers in fault f2 . When K reaches 6.0, a nonlinear
(2.7 mm). The displacements on the two sides of the dam are asym- increase is observed for the curve of the internal displacement trans-
metric. Except for the left side at EL 2150 m, the displacements on the ducer at EL 2150 m. Furthermore, a variation is observed in the curve of
left side are slightly larger than those on the right side. In terms of the the internal displacement transducer at EL 2180 m when the over-
geological condition, only fault f4 is developed in the foundation of the loading is about 7.0 times. It is illustrated that cracks occur in fault f2 at

7
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Fig. 14. Photographs of the final failure mode of the arch dam: (a) upstream; (b) downstream.

EL 2150 m and EL 2180 m near the downstream dam toe when K 5. Failure mechanism of the dam-foundation
reaches 6.0 and 7.0, respectively.
Fig. 12 shows the monitoring curve obtained from the internal 5.1. Failure evolution during overloading process
displacement transducer in fault f3 . No.2 internal displacement trans-
ducer is in fault f3 near the upstream dam heel at EL 2059 m. Before Fig. 14 is the photographs of the final failure mode of the arch dam
overloading 4.5 times, the displacement increases linearly. Afterward, a obtained by the model test. The schematic diagrams are shown for three
strong nonlinearity appears, which suggests that this displacement different stages, that is, when the upstream dam heel cracked, when the
transducer may be damaged at that time. It is observed that cracks downstream dam toe cracked, and when the dam finally failed
occur in fault f3 at EL 2059 m when K reaches 4.5. (Fig. 15). The solid line represents the crack, whereas the number re-
Fig. 13 shows the monitoring curve obtained from the internal presents the overloading factor K when a crack is initiated.
displacement transducer in fault f4 . When K reaches 5.0, nonlinear in- It is obvious in Fig. 15(a) that a crack first occurs on the left side of
creases are observed in most of the curves. It is demonstrated that the upstream dam heel when K reaches 1.5P0. At this point, no cracks
cracks occur in fault f4 near the downstream dam toe at EL 2180 m are observed in other areas upstream or downstream. As the displace-
when K reaches 5.0. However, as the curve shows, the displacement of ment curve of the crown cantilever in Fig. 7 shows, the increase in the
fault f4 is not large. In the model test, cracks in fault f4 do not extend a displacements is still linear with K for a period after the dam heel cracks
long distance, which also proves that the concrete replacement of fault and the dam is still in the elastic stage. Therefore, the cracking of the
f4 is effective. dam heel is not the cause of the nonlinear deformation.
Overall, the damage of the foundation is not particularly serious in As shown in Fig. 15(b), when K just reaches 4.5P0, apart from the
the model test. Cracks occur mainly on the faults near the dam toe and dam heel, cracks also occur in the area around the dam near the faults
dam heel and no serious sliding failure occurs in the faults. on the upstream surface of the dam. However, as analyzed in section
4.2, the cracking of faults is local and no obvious sliding failure occurs
in the faults. On the downstream surface of the dam, the dam toe is the
first and only area to be damaged. Afterward, with the increase in load,
the cracks on the dam increased significantly both upstream and
downstream. When the arch dam reaches its ultimate bearing capacity,

8
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Fig. 15. Schematic diagrams of the failure evolution of the arch dam: (a) when the overloading factor reaches K1; (b) when the overloading factor reaches K2; (c)
when the overloading factor reaches K3 (left column: upstream, right column: downstream).

Fig. 16. Schematic diagram of the final failure mode of the foundation.

it fails, as shown in Fig. 15(c). It can be inferred from the model test bearing capacity K3 is 10.0 to 11.0.
that the damage of the dam toe is a variation of the overloading process
of the arch dam. After the damage of the downstream dam toe, the dam
5.2. Nonlinear deformation and failure of the arch dam and foundation
and the foundation enter the stage of accelerated failure.
Based on the geomechanical model test, the safety factor of cracking
The arch dam-foundation system by itself displays a certain cap-
of Mengdigou arch dam K1 is 1.5, the safety factor of the initial non-
ability of stress adjustment. During the process of overloading, the
linear deformation K2 is 4.5 to 5.5, and the safety factor of the ultimate
deformations of the arch dam and foundation coordinate with each

9
Z. Tao, et al. Engineering Structures 207 (2020) 110287

5.3. Comparison with analysis based on acoustic emissions

Acoustic emissions are effective in detecting internal cracks that


cannot be observed otherwise. Thus, it is an important supplement to
the above analysis. Some similar conclusions have been drawn from
acoustic emission monitoring data [30].
In this test, 15 acoustic emission sensors were placed on the dam
and slopes, as shown in Fig. 17. These sensors were connected to signal
amplifiers and then connected to the analyzer. The analyzer used in this
test was PCI-2 acoustic emission system produced by Physical Acoustics
Corporation and the model of the sensor was RFAT-100, which is sui-
table for geotechnical materials.
Hits per second is an index that is widely used in parameter analysis
Fig. 17. The acoustic emission sensors placed on the dam. of acoustic emission results; it can be directly recorded by the analyzer
[28–29]. When a sensor receives one signal that exceeds the threshold,
other in order to maintain the structural integrity as far as possible. one hit is recorded. In this test, the threshold was set to 40 dB.
When K reaches 1.5, the stress adjustment capacity of arch dams is Therefore, during the test, the greater the hits per second, the more
not enough to maintain the integrity of all areas of the structure. serious is the cracking.
Affected by faults f3 , which represent the weak area of the foundation, a Fig. 18 shows the chart of hits per second versus the overloading
crack first occurs on the left side of the upstream dam heel near fault f3 . factor K. It is apparent that the hits per second reaches the first peak
Although the dam heel has cracked, in this period, the arch dam and value when K is 4.5 to 5.0 and then maintains a relatively high value all
foundation as a whole are still in the stage of elastic deformation, when the time, while the safety factor of the initial nonlinear deformation is
the arch dam still exhibits a relatively surplus bearing capacity. When K 4.5 to 5.5. The hits per second reaches its maximum when K is around
reaches 4.5, the downstream dam toe is destroyed. The damage of the 11.0, while the safety factor of the ultimate bearing capacity is 10.0 to
dam toe greatly reduces the integrity between the arch dam and the 11.0. The conclusion obtained from acoustic emissions is totally con-
foundation. sistent with the result of nonlinear deformation analysis.
Fig. 16 shows a schematic diagram of the final failure mode of the
foundation. It can be seen that both upstream and downstream cracks 5.4. Failure analysis based on comparisons with similar engineering systems
occur first around the riverbed, that is, in the area where the arch dam
and foundation are coupled. Moreover, the cracks in the foundation are Other geomechanical model tests of arch dams like Baihetan [31],
mainly observed around the abutment or along the faults. What is even Xiluodu [23], and Jinping-I [33] also reveal similar conclusions. Fig. 19
more interesting is that all downstream cracks along the faults originate shows the schematic diagrams of the final failure modes of these arch
near the abutment when K is about 5.0 to 6.0. Note that the down- dams. The safety factors K1 of Baihetan, Xiluodu, and Jinping-I arch
stream dam toe is damaged at an overloading of about 4.5 times and dams are 1.4 to 2.0, 1.8 to 2.0 and 2.5, respectively, and the safety
that the subsequent cracking of faults is only small when nonlinear factors K2 of Baihetan, Xiluodu, and Jinping-I arch dams are 3.0 to 4.0,
deformation has occurred in the arch dam. 4.5 and 4.0 to 5.0, respectively. It can be seen that all the cracks are
During the process of overloading, the arch dam and the foundation initiated in the upstream dam heel when the overloading factor K is
deform together and interact with each other. The weak area in the relatively low and that the dam toe cracks first in all the downstream
foundation affects the cracking of the arch dam. Subsequently, the areas. In addition, after the cracking of the downstream dam toe, the
cracking of the arch dam extends to the foundation, which reduces the number of cracks in the arch dam increases sharply. These observations
integrity of the foundation, and the damage of the foundation further are in good agreement with the conclusion of Mengdigou model test.
impacts the arch dam. In practice, for high arch dams, the value of the safety factor of the
initial nonlinear deformation, K2, can be efficiently improved by setting
anchorage, concrete replacement, and other reinforcement measures
around the downstream dam toe. For example, according to the model

Fig. 18. Chart of hits per second versus the overloading factor K.

10
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Fig. 19. Schematic diagrams of the final failure modes of arch dams: (a) Baihetan; (b) Xiluodu; (c) Jinping-I (left column: upstream, right column: downstream).
[23,31,33]

test of Dagangshan arch dam, the value of the safety factor K2 is im- fails. The safety factor of cracking of arch dam K1 is 1.5, the safety
proved to 5.5 after reinforcement of the downstream dam toe, whereas factor of the initial nonlinear deformation K2 is 4.5 to 5.5 and the safety
the value is 4.5 without the reinforcement [32]. Additionally, in the factor of the ultimate bearing capacity K3 is 10.0 to 11.0. In general, the
design specifications for arch dams, the stress safety factor of concrete design of Mengdigou arch dam is safe and reasonable.
is set to 4.0, which is close to the value of the safety factor K2 obtained Based on nonlinear deformation analysis, faults f1, f2 , f3 , and f4 crack
by the model test. when the overloading is 6.0, 7.0, 4.5, and 5.0 times, respectively. The
Based on the above analysis, it can be deduced that the damage of cracks observed in the foundation are mainly around the abutment or
the dam toe is the key cause of the overall nonlinear deformation of along the faults. During the overloading process, the arch dam and
arch dams and that the failure mode corresponding to the safety factor foundation deform together and interact with each other. Most of the
of the initial nonlinear deformation K2 is only the damage of the dam cracks initiate from the coupled area of the arch dam and foundation.
toe. The conclusion drawn from the acoustic emissions is consistent with
the result of the analysis of the nonlinear deformation. The hits per
6. Conclusion second reaches the first peak value when the overloading is 4.5 times to
5.0 times and reaches its maximum at an overloading of about 11.0
In this study, a geomechanical model test of Mengdigou arch dam times, which are in accordance with the values of the safety factors K2
was conducted. In the test, the arch dam, its foundation, and the re- and K3, respectively.
inforcement measures of faults were simulated. An overloading process Through analysis of the nonlinear deformation and failure me-
was adopted and the deformation, stress, and acoustic emission data chanism, it is indicated that the damage of dam toe is the main cause of
were obtained. Furthermore, the overall stability of the arch dam and the nonlinear deformation and failure of the arch dam, that is, the
the nonlinear deformation and failure mechanism were analyzed. Some safety factor K2 corresponds to the damage of the downstream dam toe.
of the conclusions made are as follows: Similar conclusions were drawn from the model tests conducted on
According to the model test, when the overloading factor K reaches several other arch dams and, in practice, reinforcement of the dam toe
1.5, a crack first occurs on the left side of the upstream dam heel; when has proved to be effective, which therefore requires greater attention
K reaches 4.5, the downstream dam toe is damaged and a nonlinearity during the design of arch dams.
is observed in the displacement curve of the crown cantilever. When K
is 10.0 to 11.0, the dam reaches its ultimate bearing capacity and finally

11
Z. Tao, et al. Engineering Structures 207 (2020) 110287

Acknowledgements [17] Fumagalli E. Geomechanical model of the dam foundation. In: Proceedings of the
international colloquium on physical geomechanical models, Bergamo. 1979. p.
29–30.
This research was supported by the National Key R&D Program of [18] Lemos JV. Modelling of arch dams on jointed rock foundations. Proceedings of
China under Project Nos. 2018YFC0407005, National Basic Research ISRM international symposium—EUROCK, Turin. 1996. p. 519–26.
Program of China with Grant No. 2015CB057904 and the State Key [19] Meguid MA, Saada O, Nunes MA, Mattar J. Physical mod- eling of tunnels in soft
ground: a review. Tunn Undergr Sp Tech 2008;23(2):185–98.
Laboratory of Hydroscience and Hydraulic Engineering under Grant No. [20] Fishman YA. Features of shear failure of brittle materials and concrete structures on
2019-KY-03. rock foundation. Int J Rock Mech Min Sci 2008;45(6):976–92.
[21] Chen Yuan, Zhang Lin, Yang Baoquan, Dong Jianhua, Chen Jianye. Geomechanical
model test on dam stability and application to Jinping High arch dam. Int J Rock
Appendix A. Supplementary data Mech Min Sci 2015;76:1–9.
[22] Zhou WY, Yang RQ, Liu YR. Research on geomechanical model of rupture tests of
Supplementary data to this article can be found online at https:// arch dams for their stability. J Hydro Eng 2005;24(1):53–8. (in Chinese).
[23] Liu YR, Guan FH, Yang Q, et al. Geomechanical model test for stability analysis of
doi.org/10.1016/j.engstruct.2020.110287.
high arch dam based on small blocks masonry technique. Int J Rock Mech Min Sci
2013;61:231–43.
References [24] Silveira Da, et al. High density and low strength material for geomechanical models.
Proceeding of the international col- loquium on geomechanical model, Bergamo.
1979. p. 115–31.
[1] Zingoni A. Shell structures in civil and mechanical engineering. London: Thomas [25] Fei WP, Zhang L, Zhang R. Experimental study on a geo-mechanical model of a high
Telford Publishing; 1997. arch dam. Int J Rock Mech Min Sci 2010;47:299–306.
[2] Fu X, Gu C, Qin D. Deformation Features of a Super-high Arch Dam Structural [26] Li ZK, Liu H, Dai R, Su X. Application of numerical analysis principles and key
System. Optik – Int J Light Electron Optics 2016. technology for high fidelity simulation to 3D physical model tests for underground
[3] Copen MD, Lindholm EA, Tarbox GS. Design of concrete dams. In: AR Golze (ed.) caverns. Tunn Undergr Sp Tech 2005;20:390–9.
Handbook of dam engineering, Van Nostrand Reinhold, New York; 1977. [27] Lin C, Yang Q, et al. Research on numerical simulation of Mengdigou arch dam
[4] Baker R, Garber M. Theoretical analysis of the stability of slopes. Geotech based on nonlinear finite element method. Rock Soil Mech 2016;37(9):2624–30. (in
1978;28(4):395–411. Chinese).
[5] Clough RW. 1960. The finite element method in plane stress analysis, Proc. 2nd [28] Smith A, Dixon N, Fowmes GJ. Early detection of first-time slope failures using
ASCE. Conf. on Electronic Computation, Pittsburg, Pa. acoustic emission measurements: large-scale physical modelling. Geotech
[6] Burczytiski T. Stochastic boundary value problems of elastostatics in terms of the 2017;67(2):138–52.
boundary integral equation method, Papers of Institute Civil Engineering. WrocIaw [29] Dixon N, Smith A, Flint JA, et al. An acoustic emission landslide early warning
Technical University; 1981. p. 61–8. system for communities in low-income and middle-income countries. Landslides
[7] Guminiak M. Free vibrations analysis of thin plates by the boundary element 2018;15:1631–44.
method in non-singular approach. Sci Res Inst Math Comput Sci 2007;6:75–90. [30] Wang SG, Liu YR, Tao ZF, et al. Geomechanical model test for failure and stability
[8] Daux C, Moes N, Dolbow J, et al. Arbitrary branched and intersecting cracks with analysis of high arch dam based on acoustic emission technique. Int J Rock Mech
the extended finite element method. Int J Numer Meth Eng 2000;48:1741–60. Min Sci 2018;112:95–107.
[9] Belytschko T, Moes N, Usui S, et al. Arbitrary discontinuities in finite elements. Int J [31] Song ZH, Liu YR, Yang Q. Experimental and numerical investigation on the stability
Numer Meth Eng 2001;50(4):993–1013. of a high arch dam with typical problems of nonsymmetry: Baihetan Dam, China.
[10] Cundall PA. A computer model for simulating progressive, large-scale movement in Bull Eng Geol Environ 2016;75:1555–70.
blocky rock system. Rock fracture, Proc. of Symp. Int Soc Rock Mech 1971:129–36. [32] Zhang L, Liu YR, Yang Q. Evaluation of Reinforcement and Analysis of Stability of a
[11] Cundall PA. Formulation of three-dimensional distinct element model. I: a scheme High-Arch Dam Based on Geomechanical Model Testing. Rock Mech Rock Eng
to detect and represent contacts in system com- posed of many polyhedral blocks. 2015;48(2):803–18.
Int J Rock Mech Min Sci 1988;25(3):107–16. [33] Tao ZF, Liu YR, Cheng L, et al. Failure and Stability Analysis of Jinping-I Arch Dam
[12] Hart R, Cundall PA, Lemos J. Formulation of three- dimensional distinct element Based on Geomechanical Model Test and Nonlinear Numerical Analysis. Rock Mech
model. II: mechanical calculations for motion and interaction of a system composed Rock Eng 2019;4.
of many polyhedral blocks. Int J Rock Mech Min Sci 1988;25(3):117–25. [34] Ministry of Water Resources of the People’s Republic of China. Standard for en-
[13] Shi GH. Discontinuous deformation analysis: a new numerical model for the statics gineering classification of rock masses (GB50218-94). China Planning Press,
and dynamics of deformable block structures. Eng Comput 1992;9(2):157–68. Beijing; 1995 (in Chinese).
[14] Shi GH. Manifold Method of Material Analysis. Triangle Park NC: Army Research [35] Liu YR, Wu ZS, Yang Q, Leng KD. Dynamic stability evaluation of underground
Office Research; 1992. tunnels based on deformation reinforcement theory. Adv Eng Softw
[15] Fumagalli E. Stability of arch dam rock abutments. Proceedings of the 1st ISRM 2018;124:97–108.
congress, Lisbon. 1966. p. 503–8.
[16] Fumagalli E. Statical and geomechanical models. New York: Springer; 1973.

12

You might also like