Engineering Failure Analysis 85 (2018) 62–76

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Failure modes and effect analysis of concrete gravity dams

T
subjected to underwater contact explosion considering the
hydrostatic pressure
⁎
Qi Lia, Gaohui Wanga, , Wenbo Lua, Xinqiang Niua,b, Ming Chena, Peng Yana
a
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan 430072, China
b
Changjiang Institute of Survey, Planning, Design and Research, Changjiang Water Resources Commission, Wuhan 430010, China

AR TI CLE I NF O AB S T R A CT

Keywords: To better understand damage characteristics of concrete gravity dams under contact explosion is
Concrete gravity dam a critical issue to evaluate the protective performance of dams. In this paper, a fully coupled
Underwater contact explosion Lagrangian-Eulerian numerical approach, incorporating the detonation process of underwater
Failure mode contact explosion, is performed to predict the damage propagation of a typical concrete gravity
Hydrostatic pressure
dam. In order to verify the validity of the coupled algorithm, damage profiles of a normal
Detonation products
strength reinforced concrete (RC) slab subjected to contact explosion are predicted and compared
with the published experimental results. The hydrostatic pressure of the reservoir is modeled by
using the specific internal energy method. The interaction between detonation products and dam-
foundation-reservoir systems is also considered. The detonation products development processes,
shock wave propagation, and failure modes of the dam subjected to underwater contact explosion
with and without considering the hydrostatic pressure are compared. In order to analyze the
underwater contact explosion effects on failure modes and dynamic responses of the dam, three
positions of the detonation point, i.e., upper blast point, middle blast point, and lower blast point,
are considered in this study. The results show that the initial hydrostatic pressure has a sig-
nificant influence on failure characteristics of the dam subjected to underwater contact explosion.
Underwater contact explosion detonated in the lower zone will cause more serious damage to the
dam heel and threaten the overall stability of the dam. Hence, more attention should be paid to
the deep water contact explosion.

1. Introduction

In recent decades, with the rising of terrorism threats, increasingly more attention is drawn to structural damage under blast loads
[1–6]. During the service life of some crucial infrastructures such as government buildings, subway stations, warships, bridges, and
high dams, the accidental or intentional explosion is a threat to these structures with relatively low probability but disastrous
consequences. High dams are usually designed for power, irrigation, flood control, and drinking water. Due to their significant
political and economic benefits to society, high dams are possible targets for terrorist explosion attacks. In addition, with the rapid
development of precision-guided weapons, dams are more vulnerable to bombing attacks. Blast loads will cause significant damage to
high dams especially when subjected to underwater contact explosion. Therefore, it is very important to understand the damage
mechanism and failure modes of dams subjected to underwater contact explosion, which is of great importance in the blast resistant

⁎
Corresponding author.
E-mail address: wanggaohui@whu.edu.cn (G. Wang).

https://doi.org/10.1016/j.engfailanal.2017.12.008
Received 20 July 2017; Received in revised form 25 September 2017; Accepted 6 December 2017
Available online 07 December 2017
1350-6307/ © 2017 Elsevier Ltd. All rights reserved.
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

design of dams.
Compared to non-contact explosions, contact explosions have higher pressures, shorter load durations, and more temperature
effects [7]. Many researchers have carried out some field tests and numerical simulations about a wide variety of structures under
contact explosion, such as vessels [8], concrete vehicle barriers [7], shell structures [9], concrete slabs [10,11], and beams [12].
However, few researches have focused their attention on the dynamic responses and damage characteristics of concrete gravity dams
subjected to underwater contact explosions. This is likely because that the physical processes during an underwater contact explosion
and the subsequent response of structures are extremely complex, which involve lots of complex issues such as detonation, high-
pressure gas expansion, shock wave propagation, shock wave-structure interaction, and structural response. Hence, studies on dams
subjected to underwater contact explosion are of great significance.
Due to the complicated physical phenomena involved in dam structures subjected to underwater explosion, it is very difficult to
describe this problem by the analytical method. The field tests of concrete dams to explosive loading through full-scale models are
often beyond affordability. However, with the development of the computer technology and numerical techniques, numerical si-
mulation begins to attract scholars' attention and these numerical simulation results are proven to be credible [13,14]. Major de-
velopments in understanding the structural responses and failure modes of concrete dams subjected to blast loads have been made in
recent years. Yu [15] employed the ALE algorithm to simulate the dynamic responses of the concrete dam subjected to underwater
explosion and describe the damage evolution processes of the dam. Wang et al. [16,17] used a fully coupled numerical simulation
approach to predict the damage condition of the typical concrete gravity dam subjected to underwater explosion. They also compared
the damage characteristics of concrete gravity dams subjected to underwater explosion and air blast. Linsbauer [18] studied the
dynamic responses, stability and failure mechanism of concrete gravity dams (with the initial cracks at the upstream surface) when
the charge was detonated at the bottom of the reservoir. Zhang et al. [19] discussed the influence of dam height, standoff distance and
water level on the blast resistant performance of concrete gravity dams. It should be noted that most of the aforementioned studies
are focused on the responses of dams subjected to non-contact underwater explosion. However, relatively less attention has been paid
to the nonlinear dynamic responses of the dam structures subjected to underwater contact explosion. Although the hydrostatic
pressure is usually neglected for its trivial effect on the shock wave pressure [20], this hydrostatic loading will affect the development
of detonation products, especially for contact explosion in deep water. Hence, it is of interest to understand the behavior of concrete
gravity dams subjected to underwater contact explosion with considering the hydrostatic pressure.
The objective of this study is to investigate the damage characteristics of concrete gravity dams subjected to underwater contact
explosion considering the hydrostatic pressure. To achieve this, a fully coupled Lagrangian-Eulerian method is presented. The
computed distribution of damage profiles of a normal strength RC slab subjected to contact explosion is compared with the published
experimental results to verify the validity of the coupled algorithm. The hydrostatic pressure of the reservoir water is considered by
using the specific internal energy method. The influence of the hydrostatic pressure on the detonation products development pro-
cesses, shock wave propagation, and failure modes of the dam subjected to underwater contact explosion is discussed. Three positions
of the detonation point, i.e., upper blast point, middle blast point, and lower blast point, are considered to investigate the underwater
contact explosion effects on failure modes and dynamic responses of the dam.

2. Material models

2.1. Nonlinear dynamic damage constitutive model for concrete material

The dynamic behavior of the concrete material under blast loads is a complex nonlinear and rate-dependent process. In this study,
the Riedel-Hiermaier-Thoma (RHT) [21,22] constitutive model is used in the process of concrete material modeling. In the model, the

Fig. 1. Three strength surfaces of RHT constitutive model.

63
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

variation law of initial yield strength, failure strength and residual strength of concrete is described by the introduction of three
ultimate surfaces, i.e., elastic limit surface, failure surface, and residual failure surface, as shown in Fig. 1.

2.1.1. Failure surface

The failure surface, Yfail∗, is defined as a function of the normalized pressure P, the lode angle θ and strain rate ε ̇
Y ∗fail = σeq
∗ ∗
(P , θ, ε )̇ = YTXC (P ) R3 (θ) FRATE (ε )̇ (1)
∗ ∗
where σeq is the normalized equivalent stress; YTXC is the compression meridian intensity; R3(θ) is introduced to describe the
reduced strength on shear and tensile meridian; FRATE (ε )̇ represents the dynamic increase factor (DIF) as a function of the strain rate.
∗ ∗
YTXC (P ) = A [P ∗ − Pspall FRATE (ε )̇ ]N (2)
where A is the failure surface constant; N is the failure surface index; P∗ is the pressure normalized by fc, where fc is the material
uniaxial compressive strength; Pspall∗ is the normalized spall strength.
(ε /̇ ε ̇ )α,for P > f /3, ε ̇ = 30 × 10−6 s−1
FRATE (ε )̇ = { (ε /̇ ε0̇ )δ,for P < fc /3, ε 0̇ = 3 × 10−6 s−1
0 c 0 (3)
where α is the compressive strain rate factor, δ is the tensile strain rate factor.
1
2(1 − Q22) cos θ + (2Q2 − 1)[4(1 − Q22)cos2 θ + 5Q22 − 4Q2 ] 2
R3 (θ) =
4(1 − Q22)cos2 θ + (1 − 2Q2 )2 (4)
where Q2 is used to describe the tensile and shear meridian dependence.

2.1.2. Elastic limit surface

The elastic limit surface can be described as follows:
∗ ∗
Y ela = Y fail FCAP (P ) (5)
where FCAP(P) is a function that limits the elastic deviatoric stresses under hydrostatic compression.

2.1.3. Residual failure surface

In order to describe the strength of the completely crushed material, a residual failure surface is defined as:
∗
Yres = B × (P ∗) M (6)
where B is the residual failure surface constant, M is the residual failure surface exponent.

2.1.4. Damage evolution

In the RHT constitutive model, the damage D is the accumulation of plastic strain, it can be expressed as:

D= ∑ (ΔεPL/εpfailure) (7)
∗ ∗ D2
where Δ εPL is the equivalent plastic strain increment; εpfailure
= D1(P − Pspall ) ≥ εf , D1, and D2 are damage constants; εf is the min min

minimum failure strain.

The material parameters [23,24] adopted in the present work are based on the typical date for concrete material, as shown in
Table 1.

Table 1
Parameters used in the RHT model for concrete.

Equation of state P alpha Strength RHT concrete

Reference density 2.750e3 (kg/m3) Shear modulus 1.670e10 (Pa)
Porous density 2.314e3 (kg/m3) Compressive strength (fc) 4.000e7 (Pa)
Porous sound speed 2.920e3 (m/s) Tensile strength (ft/fc) 0.100
Bulk modulus A1 3.527e10 (Pa) Shear strength (fs/fc) 0.180
Parameter A2 3.958e10 (Pa) Intact failure surface constant A 1.600
Parameter A3 9.040e9 (Pa) Intact failure surface exponent N 0.610
Parameter B0 1.220 Tens./comp. meridian ratio (Q) 0.6805
Parameter B1 1.220 Brittle to ductile transitio 0.0105
Parameter T1 3.527e10 (Pa) G (elastic)/(elastic-plastic) 2.000
Parameter T2 0 (Pa) Elastic strength /ft 0.700
Reference temperature 300.000 (K) Elastic strength /fc 0.530
Specific heat 654.000 (J/kgK) Fractured strength constant B 0.700
Failure RHT concrete Fractured strength exponent M 0.800
Damage constant D1 0.015 Compressive strain-rate exponent α 0.032
Damage constant D2 1.000 Tensile strain-rate exponent δ 0.036
Minimum strain to failure 8e-4 Maximum fracture strength ratio 1.000e20
Residual shear modulus fraction 0.130 Erosion strain 2.000

64
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a) Dimension and reinforcement of the slab (b) Steel support

Fig. 2. Dimension and support of the specimen test [11].

2.2. Material model for rock mass, explosive, water and air

The Johnson-Holmquist (JH-2) constitutive model, developed by Johnson and Holmquist [25] is used for the modeling of rock
material. This model was originally formulated for the description of the brittle response of ceramics [26,27]. Parameters used in the
JH2 model for rock can be found in [16].
The high-energy explosive material is typically modeled using the standard Jones, Wilkins, and Lee (JWL) equation of state (EOS).
The values of the constants for the explosive are available in Autodyn [24].
Water medium is described by the polynomial EOS, which has different forms in different compression states. The ideal gas EOS is
used for the air. In the simulation, the standard constants of explosive, water and air from the Autodyn material library [24] are used.

3. Validation test

3.1. Verification of the fully coupled contact explosion model

To verify the validity of the coupled algorithm, a contact explosion test of the normal strength RC slab conducted by Li et al. [11]
is simulated. In the test, the concrete slab is reinforced by 9Φ12 mm longitudinal and 11Φ8@200 mm stirrup rebars (Fig. 2(a)). The
weight of the TNT is 1 kg. The dimensions of the slab are: 2000 mm long, 800 mm wide, and 120 mm thick. In order to stabilize the
testing system, the slab was firstly placed on the steel rig using a crane, and then both ends of the slab were bolt fixed with the angle
steel cleats (Fig. 2(b)). More detail information about the test can be found in the literature [11]. A three-dimensional finite element
numerical model, including explosive, air, and RC slab is established to simulate the experiment as shown in Fig. 3. In this numerical
simulation, a fully coupled Lagrangian-Eulerian method is adopted to consider the incorporation of essential processes. The air and
explosive are modeled by the Euler grid, the RC slab is modeled by the Lagrange grid, and the rebar is modeled by beam elements.
The element sizes of the concrete, TNT, air, and steel are 10 mm, 5 mm, 20 mm, and 10 mm, respectively. The whole model consists
of 358,000 elements. For the entire computational domain, a transmission boundary is applied at the truncating boundaries of the air.
The RC slab is considered as fixed in vertical and longitudinal directions.
Comparisons between the damage profiles of numerical results and experimental results are presented in Fig. 4 and Fig. 5. It

Fig. 3. Finite element numerical model.

65
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a) Numerical result (b) Experiment result [11]

Fig. 4. Top surface of the normal strength RC slab under contact explosion.

(a) Numerical result (b) Experiment result [11]

Fig. 5. Bottom surface of the normal strength RC slab under contact explosion.

should be noted that the elements with the damage values between 0.75 and 1.0 are not displayed. The purpose of such processing is
to better show the damage status of the RC slab. This is because that when the damage values of the damage zone are between 0.75
and 1.0, the concrete material has been severely damaged, and lots of macroscopic cracks will appear in the damaged region of the RC
slab [28]. From the numerical and experimental results, it can be easily observed that the concrete crater and spall damage appear on
the top and bottom surfaces of the slab respectively. The diameters of the concrete crater from the simulation and test results are
50 cm and 46 cm, respectively. The spall diameters from the simulation and test results are 75 cm and 82 cm, respectively. The
comparison of the damage characteristics between the numerical simulation and experiment indicates that the numerical result is
highly consistent with the test, which validates the accuracy and reliability of the coupled algorithm. It should be noted that the
presented test is conducted in the air and a better validation may be pursued if the experimental data for the concrete structure
subjected to underwater contact explosion become available.

3.2. Verification of the calculation method of the hydrostatic pressure

To make the consideration of the hydrostatic pressure more convenient and avoid a large amount of calculation in the static
equilibrium process, a method through applying the specific internal energy of water is adopted here.
The water medium can be modeled by the polynomial equation of state, which has different forms in different compression states.
For μ > 0 (compression):

P = A1 μ + A2 μ2 + A3 μ3 + (B0 + B1 μ) ρe (8)

For μ < 0 (tension):

P = T1 μ + T2 μ2 + B0 ρe (9)

66
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

Fig. 6. Schematic diagram of the hydrostatic pressure test.

For μ = 0 (neither compression nor tension):

P = B0 ρe (10)

where μ is the compression ratio, ρ is the initial density; A1, A2, A3, T1, T2 is material parameters; B0 = B1 = 0.28, e is the specific
internal energy.
The hydrostatic pressure at a certain point in the water can also be described as:

P = Pa + ρgh (11)

where h is the depth of a certain point in the water; g is the gravitational acceleration, g = 9.8 m/s2; Pa is the atmospheric pressure,
Pa = 1.013 × 105 Pa.
From Eqs. (10) and (11), the specific internal energy of water can be written as follows:

e = (ρgh + P0)/ B0 ρ (12)

In order to verify the validity of this calculation method, a fully coupled model is established as shown in Fig. 6. A detailed
description of the model is given in Section 4.1. The hydrostatic pressure gradient is taken into account through inputting the
corresponding specific internal energy of the reservoir at the interval of 10 m. Four target points are arranged along the depth of
water. Fig. 7(a) gives the pressure histories from the numerical simulation. The initial hydrostatic pressure of these points can easily
be obtained through the history curves. Fig. 7(b) gives the comparison of numerical and theoretical values of the hydrostatic
pressures at four target points. It can be found that the numerical results of the hydrostatic pressure show a good consistency with the
theoretical results. Hence, the validity of the calculation method is verified and the hydrostatic pressure can be applied through
inputting the corresponding specific internal energy of the water.

(a)Pressure histories and the hydrostatic pressures from numerical simulation (b) Comparison of the hydrostatic pressures

Fig. 7. Comparison of numerical and theoretical values of the hydrostatic pressures at the target points.

67
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a) Schematic diagram of the calculation domain (b) Fully coupled numerical model
Fig. 8. Geometry configuration and fully coupled numerical model of concrete gravity dam-reservoir-foundation systems subjected to underwater contact explosion.

4. Failure characteristics of dams subjected to underwater contact explosion

4.1. Influence of the hydrostatic pressure

4.1.1. Model setup

A typical non-overflow monolith with 120 m high, 15 m wide, and 115 m deep reservoir is selected for characterizing the dynamic
responses and failure modes of concrete gravity dams subjected to underwater contact explosion. In order to investigate the effect of
the hydrostatic pressure on failure modes of concrete gravity dams subjected to underwater contact explosion, a fully coupled contact
explosion model with and without considering the hydrostatic pressure is established. Considering the symmetries of the geometrical
model, a half of the dam, rock mass, water, air, and explosive are modeled to save computation time. The calculation domain of the
non-overflow monolith is shown in Fig. 8(a). A spherical TNT charge of 326 kg (corresponding to the TNT equivalent of GBU-28
guided bomb) is selected for this calculation, representing a major incident. The detonation depth is set as 62.5 m (detonation in the
middle zone). Fig. 8(b) shows the fully coupled numerical model of the dam to underwater contact explosion. It should be noted that
the finite element model is the same for the cases with and without the hydrostatic pressure except for the specific internal energy of
the water. In the central part of the charge, element size is 100 mm [29] for the high explosive and water, and the mesh size increases
gradually away from the charge center. The element size is 100 mm in the middle zone of the dam, and the mesh size increases
gradually toward the dam bottom and dam crest. The whole model consists of 2,051,000 elements. The hydrostatic pressure gradient
is taken into account through inputting the corresponding specific internal energy of water at the interval of 10 m.
In the numerical simulation, a fully coupled Lagrangian-Eulerian method is employed to consider the incorporation of essential
processes. The explosive charge, air, and reservoir water are modeled by the Euler grid. The dam and its foundation are modeled by
the Lagrange grid. The fluid-solid coupling algorithm is adopted to model the dynamic interaction between the reservoir water and
the dam-foundation system. For the entire computational domain, the transmission boundary is applied at the truncating boundaries
to allow for free passage of shock/stress waves. The symmetric boundaries are applied at the symmetry plane, and the bottom of the
bedrock is considered as fixed in all directions.
In order to investigate the influence of the hydrostatic pressure on the dam subjected to underwater contact explosion, the
developing processes of detonation products, shock wave propagation in the water and dam, damage propagation processes, and
dynamic responses of the dam are compared. It should be noted that underwater contact explosion will damage the dam structure by
the shock wave, detonation products, and bubble pulsation. However, this study only considers the effects of the shock wave and
detonation products on the dam and the total calculation time is selected as 60 ms. The reason for this consideration contains three
aspects. On the one hand, the peak value of the first bubble pulsation pressure (only 10%–20% that of the shock wave) is usually
much lower than shock wave peak pressure [30]. On the other hand, the occurrence of the first bubble pulsation is much later than
that of the shock wave [31], and the calculation time will be greatly increased and cause computer memory overflows once the
bubble pulsation is considered [16]. Moreover, the damaged concrete resulting from the action of the shock wave and detonation
products has formed a buffer zone as shown in Fig. 9, and the bubble pulsation propagates via the reservoir water in the opposite
direction of the dam. Therefore, the influence of possible bubble pulsation on the overall response and damage characteristics of the
dam would be small.

4.1.2. Detonation products development and shock wave propagation

To analyze the influence of the hydrostatic pressure on the developing processes of detonation products and shock wave pro-
pagation in the water and dam, six target points 1–6 are arranged as shown in Fig. 10(a). Simultaneously, the explosion in the
reservoir without the dam structure is also considered to be a contrast case. Three target points 1–3 are arranged as shown in
Fig. 10(b). It should be noted that the mesh size, charge weight, and hydrostatic pressure in Fig. 10(b) are the same with Fig. 8.

68
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

Fig. 9. Schematic diagram of bubble pulsation propagation.

(a) Underwater contact explosion (with the dam structure) (b) Underwater explosion in free filed (without the dam structure)
Fig. 10. Explosion in the reservoir with and without the dam structure.

Fig. 11 gives the pressure and overpressure histories of targets 1–3 in the reservoir with the dam structure. It can be seen from
Fig. 11 that the peak pressures of target points in the reservoir with considering the hydrostatic pressure are about 0.6 MPa higher
than that without considering the hydrostatic pressure due to the influence of ambient pressure. The jump of the shock wave pressure
occurs at the target point 1 after 30 ms. This is because that the detonation products expand to this target point. It can also be
observed from the overpressure history in Fig. 11 that the hydrostatic pressure nearly has no influence on the peak overpressures.
However, the attenuation of the peak overpressure appears to be more rapid for the case with considering the hydrostatic pressure,
which is consistent with the experimental results in the literature [20].
Fig. 12 gives the comparisons of shock wave pressure history curves at the selected targets in the reservoir with and without dam
structure when considering the hydrostatic pressure. It can be easily found that the peak pressures of target points in the reservoir
with dam structure are all higher than that without dam structure. This is because of the superposition of the incident wave and
reflected wave due to the existence of the dam structure. The jump of shock wave pressure can also be observed for the case without

(a)Target 1 (b) Target 2 (c) Target 3

Fig. 11. Pressure and overpressure histories at selected targets in the reservoir with dam structure.

69
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a)Target 1 (b) Target 2 (c) Target 3

Fig. 12. Comparisons of shock wave pressure histories in the reservoir with and without the dam structure when considering the hydrostatic pressure.

the dam structure at target point 1.

Fig. 13 shows the radius time histories of detonation products with and without considering the dam structure and the hydrostatic
pressure. It can be found that the dam structure boundary and the hydrostatic pressure have a certain influence on the development
of the detonation products including the radius and expansion speed. In the first 30 ms, the effect is not very significant. However, the
influence will be strengthened after 30 ms. Due to the restriction of ambient pressure, the radius and expansion speed of the deto-
nation products with considering the hydrostatic pressure are smaller than these without considering the hydrostatic pressure. It can
also be observed that the detonation products from the underwater contact explosion with and without considering the hydrostatic
pressure expand to the target point 1 at 32 ms and 33 ms, respectively. However, when the charge is detonated in the free filed with
and without considering the hydrostatic pressure, the times for the detonation products expanding to the target point 1 are 39 ms and
42 ms, respectively. This explains the reasons for the occurrence of jumps in Fig. 11(a) and Fig. 12(a).
Fig. 14 compares the developing processes of the detonation products from underwater contact explosion with and without
considering the hydrostatic pressure. When the explosive is detonated, the detonation products act on the dam directly (Fig. 14(a)).
The pressure of the detonation products is nearly the same for the two cases at t = 2 ms. At t = 10 ms (Fig. 14(b)), the maximum
pressures of the detonation products with and without considering the hydrostatic pressure are 2.735e6 Pa and 2.651e6 Pa, re-
spectively. With the development of the detonation products, the pressure of the detonation products with considering the hydro-
static pressure is still larger than that without considering the hydrostatic pressure (as shown in Fig. 14(c)–(e)). It is worth noting that
the pressure of the detonation products is very small at t = 60 ms. Hence, the interaction between detonation products and the dam
can be neglected after 60 ms.
Fig. 15 gives the pressure time histories of the dam at target points 4–6 with and without considering the hydrostatic pressure. It is
clear that the peak pressures of the dam at the target points with considering the hydrostatic pressure are higher than that without
considering the hydrostatic pressure. This is because that when the hydrostatic pressure is considered, the contact site will be
subjected to an additional stress at the initial stage, and the dynamic interaction between detonation products and the dam will be
more intense.

4.1.3. Structural effects

Fig. 16 shows the damage propagation processes of the dam subjected to underwater contact explosion when detonated in the
middle zone with and without considering the hydrostatic pressure. The contour value ranging from 0 to 1 indicates the damage
degree of the concrete element from intact material to completely damaged material. In order to make the damage distribution more
obvious, the contour value is truncated from 0.1–1. It is clear that the hydrostatic pressure has a significant influence on the damage
propagation of the dam subjected to underwater contact explosion. When considering the hydrostatic pressure, the maximum length
of the radial cracks in the vicinity of the blast point is about 31.5 m. However, the value for the case without considering the

(a) Underwater contact explosion (with the dam structure) (b) Underwater explosion in free field (without the dam structure)

Fig. 13. Radius time histories of detonation products from underwater explosion with and without the dam structure.

70
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

Fig. 14. Developing processes of the detonation products from underwater contact explosion with and without considering the hydrostatic pressure.

(a) Target 4 (b) Target 5 (c) Target 6

Fig. 15. Comparisons of pressure history curves of the dam at selected targets with and without considering the hydrostatic pressure.

Fig. 16. Damage propagation processes of the dam subjected to underwater contact explosion with and without considering the hydrostatic pressure (detonation in the
middle zone).

71
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a) Arrangement of the target points P 1 ~ P 3 (b) Target point P1

(c) Target point P 2 (d) Target point P 3

Fig. 17. Horizontal velocity time histories of the dam subjected to underwater contact explosion with and without considering the hydrostatic pressure.

hydrostatic pressure is about 21.6 m. This is because the dynamic interaction between the detonation products and dam is
strengthened when the hydrostatic pressure is considered. In addition, the dam heel is in a state of tension at the beginning because of
the initial hydrostatic pressure, and the damage on the dam heel is very serious (length of the crack is about 38.8 m). However, there
is no damage on the dam heel when the hydrostatic pressure is not considered.
When the explosive is detonated in the middle zone, the underwater explosion instantly converts the explosive into a hot gas with
a high pressure of approximately 5000 MPa and a high temperature of the order of 3000 °C [32]. Because the instantaneous shock
wave pressure is far more than the dynamic compressive strength of the concrete material, the concrete in the vicinity of the explosive
is crushed instantly. The crushed damage develops continually due to the extrusion effect of the detonation products. Eventually, the
compression crater forms as shown in Fig. 16(a). When the shock wave propagates from the reservoir water to the dam body, the
shock wave will attenuate to a compressive stress wave. This compressive stress wave can cause radial expansion and tangential
tensile damage to the dam, and two obvious cracks appear along the slantingly upward and downward direction at about 45 degrees
as shown in Fig. 16(b). Under the combined action of stress wave and detonation products, radial cracks spread gradually. At
t = 20 ms (Fig. 16(c)), the tensile damage occurs at the dam heel because of the overall response of the dam when considering the
hydrostatic pressure. This damage will expand toward the downstream continuously. However, no damage occurs at the dam heel
when ignoring the hydrostatic pressure. This indicates that the dam heel is more vulnerable to underwater contact explosion when
considering the hydrostatic pressure. When the compressive stress wave propagates to the downstream surface, a strong tensile stress
wave is formed since the wave impedance of the air medium is far less than that of the concrete medium, which causes the tensile
damage to the downstream surface of the dam (Fig. 16 (d–e)).
In order to observe the dynamic responses of the dam, three target points P1–P3 are arranged in the numerical model as shown in
Fig. 17(a). Fig. 17(b)–(d) gives the horizontal velocity time histories of the three target points with the positive velocity toward the
downstream direction. It should be noted that the velocity response is not balanced. This is because the total calculation time is set as
60 ms to reduce the computational cost. However, the subsequent velocity response is not very important. It can be seen from Fig. 17
that the hydrostatic pressure has a significant influence on the velocity response of the middle (target point 2) and lower (target point
3) zone of the dam. The velocity responses of the dam crest (target point 1) with and without considering the hydrostatic pressure are
very similar to each other. The maximum velocities of the target point 2 for the two cases with and without considering the hy-
drostatic pressure are 2.01 m/s and 1.55 m/s, respectively. The velocity response of the target points for the case with considering the
hydrostatic pressure (peak velocity of 0.74 m/s) is significantly different from the case without considering the hydrostatic pressure
(peak velocity of 0.14 m/s).
To summarize, the hydrostatic pressure can affect the development of the detonation products thus causing smaller radius and
higher gas pressure. Accordingly, the dynamic interaction between the detonation products and the dam is also strengthened and
deeper radial cracks appear in the vicinity of the blast position. In addition, the dam heel is in a state of tension at the beginning

72
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a) The schematic diagram (b) t=20ms (c) t=40ms (d) t=60ms
Fig. 18. Damage propagation processes of the dam subjected to underwater contact explosion when detonated in the upper zone.

because of the initial hydrostatic pressure, and the damage on the dam heel is more serious. Therefore, the hydrostatic pressure
should be considered in the study of concrete gravity dams subjected to underwater contact explosion.

4.2. Influence of the blast position

In order to explore the influence of the blast position on the damage characteristics of the dam subjected to underwater contact
explosion, three blast positions (i.e., detonation in the upper, middle, and lower zone) are considered. It should be noted that the
dynamic response of the dam for the detonation in the middle zone has be discussed in Section 4.1. Other two numerical models
(detonation in the upper and lower zone) are established in the same way as described in Section 4.1 except for the blast position. In
the central part of the charge, the element size is 100 mm for the high explosive and water, and the mesh size increases gradually
away from the charge center. The same boundary conditions are applied at the truncating boundaries.

4.2.1. Detonation in the upper zone

Fig. 18 shows the damage propagation processes of the dam subjected to contact explosion when detonated in the upper zone. The
detonation depth is selected as 10 m as shown in Fig. 18(a). It is clear that severe damage occurs in the dam head due to the combined
action of stress waves, detonation products, and rarefaction waves (Fig. 18(b)). In addition, tensile damage occurs at the dam heel
because of the overall response of the dam. This damage develops toward the downstream, and the final crack length is about 20 m.
The damage of the upstream surface of the dam continues to extend to the lower zone of the dam (Fig. 18(c–d)).
Fig. 19 shows the horizontal velocity and acceleration time histories of the three target points. Immediately after the detonation,
the target point P1, which is closest to the blast point, responds rapidly under the action of the detonation products and shock wave
(Fig. 19(a)). Because of the overall response of the dam, the horizontal velocity of the target point 1 increases continuously and
reaches the peak velocity of about 1.4 m/s at t = 60 ms. As for target points P2 and P3, their horizontal velocities are mainly affected
by the overall response of the dam. For the horizontal acceleration as shown in Fig. 19(b), the maximum value appears at the target
point P1, which is up to 1680 m/s2. The maximum acceleration values for the target points 2 and 3 are relatively small.

4.2.2. Detonation in the lower zone

Fig. 20 shows the damage propagation processes of the dam subjected to contact explosion when detonated in the lower zone. The
detonation depth is selected as 115 m as shown in Fig. 20(a). As shown, the damage propagation process and failure modes are
obtained. It can be found that the detonation in the lower zone will cause more serious damage to the lower part of the dam,
especially for the dam heel. A crack initially progresses a long way from the upstream side to the downstream side at the dam heel.
This is because of the existence of the stress concentration and the combined action of the blast shock wave and detonation products.
The final crack length at the dam heel is > 50 m. In addition, the slight damage occurs at the upstream surface of the lower zone of

(a) Horizontal velocity time histories (b) Horizontal acceleration time histories
Fig. 19. Horizontal velocity and acceleration time histories at selected targets when detonated in the upper zone.

73
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a) The schematic diagram (b) t=20ms (c) t=40ms (d) t=60ms
Fig. 20. Damage propagation processes of the dam subjected to underwater contact explosion when detonated in the lower zone.

the dam because of the punching action of shock wave.

Fig. 21 gives the horizontal velocity and acceleration time histories of the three target points when detonated in the lower zone. It
is obvious that the maximum dynamic response occurs at the target point 3. The maximum horizontal velocity and acceleration of the
target point 3 are 1.25 m/s 1250 m/s2, respectively. The horizontal velocity and acceleration responses of the target points 1 and 2
are relatively small. This is because that the dynamic responses of the dam are highly localized in the contact explosion zone. The
global flexural behavior of the dam is expected to be small.

4.2.3. Comparative analysis of cases with different blast positions

In order to investigate the influence of the blast position on failure modes and dynamic responses of the concrete gravity dam
subjected to underwater contact explosion, a comparative study is conducted in this section. Fig. 22 gives the damage characteristics
of the dam subjected to underwater contact explosion with different blast positions. It should be noted that the contour value is
truncated from 0–0.75. The purpose of such processing is to better show the damage status of the dam. It is clear that underwater
contact explosion will cause severe damage to the contact region of the dam, and result in the blasting crater on the blast position. It
can also be found that the blast position has a significant influence on the damage characteristics of the concrete gravity dam
subjected to underwater contact explosion. When the explosive is detonated in the upper zone of the dam as shown in Fig. 22(a),
severe damage occurs at the dam head. However, the risk of the dam break is relatively small. For the detonation in the middle zone,
severe damage occurs at the contact blast zone and dam heel (Fig. 22(b)). However, the underwater contact explosion detonated in
the lower zone will cause extremely serious damage to the dam heel, and threaten the overall stability of the dam (Fig. 22(c)). Hence,
more attention should be paid to the deep water contact explosion.
Fig. 23 gives the velocity time response curves of the selected target points from different blast positions, and Fig. 24 shows the
corresponding acceleration time curves. It is clear that the blast position has a significant influence on the nonlinear dynamic
responses of the dam. The maximum velocity and acceleration of target points occur at the corresponding contact region. When the
explosive is detonated in the upper zone, the horizontal velocity of the target point P1 rises slowly after a sharp rise due to the small
horizontal constraint and large overall response of the dam. However, for target point P2 (detonation in the middle zone) and target
point P3 (detonation in the lower zone), their velocities decrease rapidly after a sharp rise due to the great lateral stiffness in the
middle and lower part of the dam. In addition, obvious negative velocities appear in the dam crest target point P1 (detonation in the
middle zone) and the dam bottom target point P3 (detonation in the upper zone) because of the action of the inertia.

5. Conclusions

The objective of this study is to investigate the failure modes and effect analysis of concrete gravity dams subjected to underwater
contact explosion considering the hydrostatic pressure. A fully coupled Lagrangian-Eulerian numerical approach, incorporating the

(a) Horizontal velocity time histories (b) Horizontal acceleration time histories
Fig. 21. Horizontal velocity and acceleration time histories at selected targets when detonated in the lower zone.

74
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

(a) Detonation in the upper zone (b) Detonation in the middle zone (c) Detonation in the lower zone
Fig. 22. Damage characteristics of the dam subjected to underwater contact explosion with different blast positions.

(a) Target point P 1 (b) Target point P 2 (c) Target point P 3

Fig. 23. Comparison of horizontal velocity time histories of the selected target points from different blast positions.

(a) Target point P1 (b) Target point P2 (c) Target point P3

Fig. 24. Comparison of horizontal acceleration time histories of the selected target points from different blast positions.

detonation process of underwater contact explosion, is performed. Damage profiles of a normal strength RC slab subjected to contact
explosion are predicted and compared with the published experimental results to verify the feasibility of the numerical simulation
and calculation method. The influence of the hydrostatic pressure on the dam subjected to underwater contact explosion is discussed.
Failure modes and dynamic responses of the dam subjected to underwater contact explosions with different blast positions are
compared. The main results and findings are summarized and discussed below.

(1) The coupled Lagrangian-Eulerian method can predict effectively the damage propagation process and failure modes of concrete
gravity dams subjected to underwater contact explosion. The specific internal energy method can effectively simulate the initial
hydrostatic pressure of underwater explosion.
(2) The hydrostatic pressure can limit the development of the detonation products, which will result in a higher pressure for the
detonation products and strengthen the dynamic interaction between the detonation products and dam. The radius and expansion
speed of the detonation products with considering the hydrostatic pressure are smaller than these without considering the hy-
drostatic pressure.
(3) Underwater contact explosion will cause severe damage to the contact region of the dam, and result in the blasting crater on the
blast position. The hydrostatic pressure has a significant influence on the damage propagation process of the dam subjected to
underwater contact explosion. The dam heel is more vulnerable to underwater contact explosion when considering the

75
Q. Li et al. Engineering Failure Analysis 85 (2018) 62–76

hydrostatic pressure.
(4) The blast position has a significant influence on the damage characteristics of the concrete gravity dam subjected to underwater
contact explosion. Although the dam head is basically destroyed when the underwater contact explosion is detonated in the upper
zone, the risk of dam break is relatively small. However, the underwater contact explosion detonated in the lower zone will cause
extremely serious damage to the dam heel, and threaten the overall stability of the dam. Hence, more attention should be paid to
the deep water contact explosion.

Acknowledgement

The authors gratefully appreciate the supports from the National Natural Science Foundation of China (No. 51509189), the
National Key Research Project (No. 2016YFC0402008) and the Natural Science Foundation of Hubei Province of China
(2017CFB666).

References

[1] P. Turgut, M.A. Gurel, R.K. Pekgokgoz, LPG explosion damage of a reinforced concrete building: a case study in Sanliurfa, Turkey, Eng. Fail. Anal. 32 (32) (2013)
220–235.
[2] Z. Zong, Y. Zhao, H. Li, A numerical study of whole ship structural damage resulting from close-in underwater explosion shock, Mar. Struct. 31 (2) (2013) 24–43.
[3] M. Campidelli, M.J. Tait, W.W. El-Dakhakhni, W. Mekky, Numerical strategies for damage assessment of reinforced concrete block walls subjected to blast risk,
Eng. Struct. 127 (2016) 559–582.
[4] S. Yao, D. Zhang, X. Chen, F. Lu, W. Wang, Experimental and numerical study on the dynamic response of RC slabs under blast loading, Eng. Fail. Anal. 66 (2016)
120–129.
[5] S.K. Hashemi, M.A. Bradford, H.R. Valipour, Dynamic response and performance of cable-stayed bridges under blast load: effects of pylon geometry, Eng. Struct.
137 (2017) 50–66.
[6] H. Arora, P.D. Linz, J.P. Dear, Damage and deformation in composite sandwich panels exposed to multiple and single explosive blasts, Int. J. Impact Eng. 104
(2017) 95–106.
[7] A.M. Coughlin, E.S. Musselman, A.J. Schokker, D.G. Linzell, Behavior of portable fiber reinforced concrete vehicle barriers subject to blasts from contact charges,
Int. J. Impact Eng. 37 (5) (2010) 521–529.
[8] F.R. Ming, A.M. Zhang, Y.Z. Xue, S.P. Wang, Damage characteristics of ship structures subjected to shockwaves of underwater contact explosions, Ocean Eng. 117
(2016) 359–382.
[9] Z. Jin, C. Yin, Y. Chen, H. Hua, Graded effects of metallic foam cores for spherical sandwich shells subjected to close-in underwater explosion, Int. J. Impact Eng.
94 (2016) 23–35.
[10] J. Li, C. Wu, H. Hao, Investigation of ultra-high performance concrete slab and normal strength concrete slab under contact explosion, Eng. Struct. 102 (2015)
395–408.
[11] J. Li, C. Wu, H. Hao, Z. Wang, Y. Su, Experimental investigation of ultra-high performance concrete slabs under contact explosions, Int. J. Impact Eng. 93 (2016)
62–75.
[12] D. Zhang, S. Yao, F. Lu, X. Chen, G. Lin, W. Wang, Y. Lin, Experimental study on scaling of RC beams under close-in blast loading, Eng. Fail. Anal. 33 (5) (2013)
497–504.
[13] X.Q. Zhou, V.A. Kuznetsov, H. Hao, J. Waschl, Numerical prediction of concrete slab response to blast loading, Int. J. Impact Eng. 35 (10) (2008) 1186–1200.
[14] Werner Riedel, CMKT, Engineering and numerical tools for explosion protection of reinforced concrete, Int. J. Protective Struct. 1 (1) (2009) 85–101.
[15] T. Yu, Dynamical Response Simulation of Concrete Dam Subjected to Under Water Contact Explosion Load, 2009 World Congress on Computer Science and
Information Engineering, 2009, pp. 767–774.
[16] G. Wang, S. Zhang, Damage prediction of concrete gravity dams subjected to underwater explosion shock loading, Eng. Fail. Anal. 39 (4) (2014) 72–91.
[17] G. Wang, S. Zhang, Y. Kong, H. Li, Comparative study of the dynamic response of concrete gravity dams subjected to underwater and air explosions, J. Perform.
Constr. Facil. 29 (4) (2014) 04014092.
[18] H. Linsbauer, Hazard potential of zones of weakness in gravity dams under impact loading conditions, Frontiers Archit. Civil Eng. China 5 (1) (2011) 90–97.
[19] S. Zhang, G. Wang, C. Wang, B. Pang, C. Du, Numerical simulation of failure modes of concrete gravity dams subjected to underwater explosion, Eng. Fail. Anal.
36 (1) (2014) 49–64.
[20] B.W. Vanzant, R.C. Dehart, Effect of hydrostatic pressure on shock waves from underwater explosions, J. Appl. Phys. 36 (10) (1965) 3116–3117.
[21] W. Riedel, K. Thoma, S. Hiermaier, E. Schmolinske, Penetration of reinforced concrete by BETA-B-500 numerical analysis using a new macroscopic concrete
model for hydrocodes, Proceedings of the 9th International Symposium on the Effects of Munitions with Structures, Berlin-Strausberg, Germany, 1999, pp.
315–322.
[22] W. Riedel, Beton unter dynamischen Lasten: Meso-und makromechanische Modelle und ihre Parameter, Institut Kurzzeitdynamik, Ernst-MachInstitut, der
Bundeswehr Munchen, Freiburg, 2000, p. 210.
[23] Z. Tu, Y. Lu, Evaluation of typical concrete material models used in hydrocodes for high dynamic response simulations, Int. J. Impact Eng. 36 (1) (2009)
132–146.
[24] ANSYS, AUTODYN User Manual Version 13, (2010).
[25] G.R. Johnson, T.J. Holmquist, Response of boron carbide subjected to large strains, high strain rates, and high pressures, J. Appl. Phys. 85 (12) (1999)
8060–8073.
[26] G.R. Johnson, T.J. Holmquist, S.R. Beissel, Response of aluminum nitride (including a phase change) to large strains, high strain rates, and high pressures, J.
Appl. Phys. 94 (3) (2003) 1639–1646.
[27] T.J. Holmquist, G.R. Johnson, Characterization and evaluation of silicon carbide for high-velocity impact, J. Appl. Phys. 97 (9) (2005) 5858–5866.
[28] Y.S. Chung, C. Meyer, M. Shinozuka, Modeling of concrete damage, ACI Struct. J. 86 (3) (1989) 259–271.
[29] G. Wang, Y. Wang, W. Lu, W. Zhou, M. Chen, P. Yan, On the determination of the mesh size for numerical simulations of shock wave propagation in near field
underwater explosion, Appl. Ocean Res. 59 (2016) 1–9.
[30] R. Cole, Underwater Explosions, Dover Publications Inc., New York, 1948.
[31] A.M. Zhang, X.L. Yao, X.B. Yu, The dynamics of three-dimensional underwater explosion bubble, J. Sound Vib. 311 (3–5) (2008) 1196–1212.
[32] G.I. Taylor, The distortion under pressure of a diaphragm which is clamped along its edge and stressed beyond the elastic limit, Underwater Explo. Res. 3 (1950)
107–121.

76