You are on page 1of 46

Accepted Manuscript

A general numerical model for surface waves generated by granular material intruding
into a water body

Pengfei Si, Huabin Shi, Xiping Yu

PII: S0378-3839(18)30169-8
DOI: 10.1016/j.coastaleng.2018.09.001
Reference: CENG 3421

To appear in: Coastal Engineering

Received Date: 3 April 2018


Revised Date: 4 September 2018
Accepted Date: 8 September 2018

Please cite this article as: Si, P., Shi, H., Yu, X., A general numerical model for surface waves
generated by granular material intruding into a water body, Coastal Engineering (2018), doi: 10.1016/
j.coastaleng.2018.09.001.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT

A General Numerical Model for Surface Waves Generated by


Granular Material Intruding into a Water Body

Pengfei Si a, Huabin Shi a, Xiping Yu a*

PT
a
State Key Laboratory of Hydroscience and Engineering, Department of Hydraulic

RI
Engineering, Tsinghua University, Beijing, 100084, China.
*

SC
Corresponding Author
Emails: spf14@mails.tsinghua.edu.cn (P Si)
shihuabin@mail.tsinghua.edu.cn (H Shi)
yuxiping@tsinghua.edu.cn (X Yu)
U
AN
M
D
TE
C EP
AC

1
ACCEPTED MANUSCRIPT

1 A General Numerical Model for Surface Waves Generated by


2 Granular Material Intruding into a Water Body
3 Pengfei Si1, Huabin Shi1, Xiping Yu2

4 Abstract

PT
5 Surface water waves generated by large-scale landslides have been a major concern of many

6 geoscientists and coastal engineers because they may result in disastrous consequences. This

RI
7 study presents an advanced two-phase model for dry granular material intruding into an

8 otherwise still water body as well as their resulting waves. The water-air interface both within

SC
9 and outside the granular material is captured by the volume of fluid method. The

10 inter-granular stresses are formulated based on a general collisional-frictional law developed

11

U
for underwater granular flows and a modified k - e model is adopted to describe the
AN
12 turbulence effect of the ambient fluid. Phase interaction is characterized by the drag force

13 caused by the relative motion between the granular particles and the fluid. The effect of the
M

14 ambient fluid on the restitution coefficient of granular particles is also considered. The newly

15 proposed two-phase model is validated following a reasonable agreement of the numerical


D

16 results with measured data from small-scale laboratory tests on surface waves caused by
TE

17 collapse of a subaerial granular column and by intrusion of a landslide into water. Generation

18 and propagation of the waves, as well as the motion and deformation of the granular body, are
EP

19 all adequately represented by the numerical model. A relatively more general applicability of

20 the proposed model to the study of waves generated by granular landslides can thus be
C

21 expected.
AC

22 Keywords: Two-phase model; Underwater granular flow; Water waves; Collisional-frictional

1
State Key Laboratory of Hydroscience and Engineering, Department of Hydraulic
Engineering, Tsinghua University, Beijing, 100084, China.
2
Corresponding Author, State Key Laboratory of Hydroscience and Engineering,
Department of Hydraulic Engineering, Tsinghua University, Beijing, 100084, China; Phone:
+86-10-62776777; Email: yuxiping@tsinghua.edu.cn
1
ACCEPTED MANUSCRIPT

23 stress; Volume of fluid method.

24 1. Introduction
25 Landslide generated waves may cause massive damage to coastal facilities or the dam of a

26 reservoir, and even result in serious fatalities (Miller, 1960). In addition, underwater spreading

PT
27 of the landslide materials may destroy submarine cables and seabed infrastructure (Masson et

28 al., 2006). Since many deformable landslides can be treated as granular flows, establishment

RI
29 of a general model for granular material intruding into an open or enclosed water body and

SC
30 their resulting waves is thus of great practical importance.

31 Recent experimental studies (Fritz et al., 2003; Ataie-Ashtiani and Najafi-Jilani, 2008;

U
32 Mohammed and Fritz, 2012; McFall and Fritz, 2016; McFall et al., 2018) showed that the

33 waves generated by deformable slides involve more complex dynamics as compared to the
AN
34 rigid-slide generated waves. Numerical models for landslide generated waves that take into

35 account the effect of landslide deformation have been extensively developed in recent decades.
M

36 Most of the well established models regard the landslide and the water as two different
D

37 continuums. Compared to discrete element model which treats the landslide as the assembly

38 of a large number of dispersed particles (Zhao et al., 2015; Tan and Chen, 2017), the
TE

39 continuum models have a better computational efficiency and are more applicable to practical
40 problems. By further simplification, various continuum models including depth-averaged
EP

41 models (Heinrich et al., 2001; Kelfoun et al., 2010; Yavari-Ramshe and Ataie-Ashtiani, 2017),
42 non-hydrostatic models (Ma et al., 2015; Si et al., 2018a) and Navier-Stokes equation models
C

43 (Abadie et al., 2010; Basu et al., 2010) have been proposed. The depth-averaged models based
AC

44 on the hydrostatic law neglect the vertical accelerations of the fluid motion and are not valid

45 in the cases where granular flows are rapid (Ma et al., 2015). The non-hydrostatic models

46 correct the pressures at the landslide-water interface and yield better results on various

47 occasions.

48 For a general description of the highly unsteady process of landslide generated waves, a
49 Navier-Stokes equation model is advantageous in principle. An effective Navier-Stokes

2
ACCEPTED MANUSCRIPT

50 equation model for granular slide generated waves, however, must be developed with

51 particular emphases on effective formulation for rheology of granular materials and on

52 accurate description of interactions between granular particles and their ambient fluid. The

53 non-Newtonian continuum rheologies originating from hydromechanics and soil mechanics

54 (Pastor et al., 2004; Ancey and Cochard, 2009; Ionescu et al., 2015; Shi et al., 2016) have

PT
55 been widely used in the last few decades to describe the deformable features of the granular

56 slide. These rheological models, however, cannot cover the dynamic behavior of a granular

RI
57 slide at different stages of its motion. In fact, landslides composed of granular matter may

58 vary from solid-like to fluid-like in properties, depending on the magnitude of the

SC
59 inter-granular stresses and the bulk deformation rate of the granular mass (Jaeger et al., 1996).

60 A granular slide experiences both contact friction and instant collision from initial quasi-static

61
U
state, through rapid motion, to deceleration and final deposition (Iverson and George, 2014).
AN
62 Thus, it is critical to adequately deal with the inter-granular stresses.

63 To develop an advanced model for waves generated by granular slides it is also


M

64 important to accurately describe the interactions between granular particles and their ambient

65 fluid (Pailha et al., 2008; Rondon et al., 2011; Wang et al., 2017; Si et al., 2018b). When the
D

66 granular slide runs into a water body, the ambient water is lifted and pushed away due to
TE

67 displacement of the slide (Fritz et al., 2009; Heller and Hager, 2010), causing a complex fluid

68 flow around the granular body. While at the same time, the ambient water will not only resist
EP

69 but also penetrate into the granular mass (Miller et al., 2017). Strong mixing may also occur

70 near the surface of the granular body. In general, the interaction between granular particles
C

71 and their ambient fluid in a two-phase flow model is often characterized by the interphase
AC

72 forces, including the drag, the inertia force, the lift force, and Basset force. In a dense granular

73 problem, however, only the drag is predominant, especially in the flows with relatively low

74 Reynolds number (Zhao et al., 2015). Effect of fluid turbulence on the particle motion has

75 rarely been considered previously in such a problem. Compared to the inter-granular stresses

76 and the interphase drag force, fluid turbulence has been realized to play a negligible role when

77 the granular concentration is large (Lubin et al., 2006; Lee and Huang, 2018).
78 In this study, an effective two-phase Navier-Stokes equation model is proposed for

3
ACCEPTED MANUSCRIPT

79 description of the water waves generated by granular slides. For simplicity, the granular slides

80 are assumed to be assemblies of monodisperse spherical particles. A recently established

81 collisional-frictional relation is employed to represent the constitutive law of granular

82 materials. The water-air interface both within and outside the granular body is captured by the

83 volume of fluid method. The present two-phase model thus has a fairly good generality in

PT
84 principle and can capture the entire process of landslide generated waves, including the

85 motion and deformation of the landslide, as well as the generation and propagation of the

RI
86 water waves. Validation of the numerical results by laboratory experiments is also performed.

SC
87 2. Mathematical Formulation

U
88 2.1 Basic equations
AN
89 The physical problem of water waves generated by a granular flow is illustrated in Figure 1. If

90 a complete description of the problem is required, we need to deal with three phases, i.e., the
M

91 water, the air, and the granular material. For simplicity, however, a particle-fluid two-phase

92 approach is adopted in this study. That is, the water and the air are treated as a single fluid
D

93 phase with variable properties. Theoretically, such a two-phase approach can represent the
TE

94 original three-phase problem without any loss of accuracy but can be much more efficient.

95 This is essentially because the water and the air are immiscible and the properties of the fluid
EP

96 phase can be readily determined as long as the water-air interface is captured.


97
C
AC

98

99 FIGURE 1. PHYSICAL DESCRIPTION OF WATER WAVES GENERATED BY A GRANULAR FLOW


100

101 In computational fluid dynamics, a water-air interface can be captured through

4
ACCEPTED MANUSCRIPT

102 numerically solving an advection equation in terms of the volume-of-fluid (VOF) function

103 (Hirt and Nichols, 1981). For the problem of our present interest, we introduce the liquid

104 saturation function x , which is defined as the volume fraction of the water to the water-air as

105 a whole. Therefore, the pore volume in the neighborhood of a spatial point is fully occupied

106 by water if x = 1 , or air if x = 0 at this point. If 0 < x < 1 , the point should be located on

PT
107 the air-water interface. From the mass conservation of the water phase, the governing

108 equation for x can be written as

RI
¶ (a f x )
109 + Ñ ×(a f x u f ) = 0 (1)

SC
¶t

110 where, a f is the volume fraction of the fluid phase, u f is the averaged velocity of the

U
111 fluid phase. Note that the fluid in an element must be treated as water-air mixture if the local
AN
112 value of x is a decimal fraction. In such an element, the density and viscosity of the fluid

113 are often assumed to be


M

114 r f = xr w + (1 - x )r a (2)
D

115 n f = xnw + (1 - x )na (3)


TE

116 where r w and r a are the density of the water and the air; n w and na are kinetic
EP

117 viscosity of the water and the air, respectively. Eq. (3) may not be physically correct, but the

118 effect of this assumption is limited to a narrow range in the vicinity of the water surface as
C

119 long as the sharpness of the water-air interface is maintained.


AC

120 The basic equations for the two-phase flow can be derived by spatially averaging the
121 Navier-Stokes equations over the fluid and granular phases, respectively. By assuming that

122 there is no mass exchange between the two phases, the mass and momentum conservation
123 laws then lead to

¶ (a s r s )
124 + Ñ ×(a s r s us ) = 0 (4)
¶t

5
ACCEPTED MANUSCRIPT

¶ (a s r s us )
125 + Ñ ×(a s r s us us ) = - a s Ñ p f - Ñ ps + Ñ ×t s + a s r s g + F (5)
¶t

126 for the granular phase, and

¶ (a f r f )
127 + Ñ ×(a f r f u f ) = 0 (6)
¶t

PT
¶ (a f r f u f )

RI
128 + Ñ ×(a f r f u f u f ) = - a f Ñ p f + Ñ ×t f + a f r f g - F (7)
¶t

SC
129 for the fluid phase, where the subscripts s and f denote values associated with the

130 granular phase and the fluid phase, respectively; a is the volume fraction satisfying

U
131 a s + a f = 1 ; r is the density; u is the velocity; p is the pressure; t is the stress; F
AN
132 is the interphase force; g is the acceleration due to gravity. The interphase force is usually

133 expressed as a function of the relative motion between the two phases. The stress must be
M

134 determined by a constitutive law. Note that the fluid shear stress includes both the viscous

135 effect of the fluid and the turbulence effect, while the granular stress includes the effects of
D

136 particle collision as well as the friction due to particle contact. It is also worthwhile to
TE

137 mention that surface tension, which may play an important role if there are very short waves

138 on the water-air interface so that capillary effects on the problem of interest cannot be
EP

139 neglected, we may add a special term in the momentum equation for the fluid phase to treat it

140 (Liu and Yu, 2016). In the present study, the surface tension is omitted.
C

141 2.2 Constitutive law for granular phase


AC

142 Following previous studies (e.g., Johnson and Jackson, 1987; Si et al., 2018b), both the

143 pressure and the shear stress in the granular phase are assumed to be the direct addition of a

144 collisional part and a frictional part so that

145 t s = t cs + t sf (8)

146 ps = psc + psf (9)

6
ACCEPTED MANUSCRIPT

147 where the superscripts c and f represent the collisional and frictional component of the

148 granular stresses, respectively. A constitutive relation for the collisional shear stress may be

149 expressed as

150 t cs = 2msc Ss (10)

PT
151 where, msc is the granular viscosity, Ss is the tensor of the deviatoric rate of granular strain

152 defined by

RI
1é Tù 1
153 Ss = êÑ us + (Ñ us ) ú- (Ñ ×us )I (11)

SC
2 ëê ûú 3

154 where, the superscript T denotes transpose of tensor. Following the kinetic theory of Lun et

155

U
al. (1984), the collisional pressure psc and the granular shear viscosity msc can be derived as
AN
8 Q
156 psc = a s r s Q (1 + 4ha s R )- r sds ha s 2R (Ñ ×us ) (12)
3 p
M

é æ 8 öæ ö ù
ê 5 p ç ÷ç 8 ÷ 8 2 ú
ç1 + h (3h - 2)a s R ÷
c
157 m = r sds Qê ç1 + ha s R ÷ ÷+ ha s R ú (13)
D

÷
ê96h (2 - h )R çè ÷ç
s
5 øè 5 ø÷ 5 p ú
ë û
TE

158 where, ds is the particle diameter; Q is the granular temperature representing the kinetic

159 energy of the granular phase due to velocity fluctuations caused by particle collisions;
EP

160 R = (2 - a s ) [2(1 - a s )3 ] is the radial distribution function characterizing the spatial


161 correlation among particles; h = (1 + e ) 2 with e being the restitution coefficient of
C

162 particle collisions. To consider the effect of ambient fluid on collisions of granular particles,
AC

163 e takes the following form (Armanini et al., 2009)

164 e = e ¢- 2.85 St - 0.5 (14)

165 where St = r sds Q 0.5 (18r f nm ) is the particle Stokes number with n m being the molecular

166 viscosity of the fluid, e ¢ is the restitution coefficient of dry granular particles, which is often

167 suggested to be 0.9 for glass beads (Lorenz et al., 1997). Eq. (14) has been proven to be

7
ACCEPTED MANUSCRIPT

168 effective for various types of granular materials (Armanini et al., 2009).

169 The governing equation for the granular temperature Q , taking into account the

170 interaction between the granular and fluid phases (Gidaspow, 1994), can be written as

é ù
3 ê¶ (a s r s Q )
171 ê + Ñ ×(a s r s us Q )úú= Ñ ×(k s ÑQ ) + (- psc I + t cs ) : Ñ us - a s r sJ s + G (15)
2 ê ¶t ú
ë û

PT
172 where, the first term on the right side represents the diffusion of the granular temperature with

RI
173 k s being a diffusion coefficient; the second term is the production of the granular

174 temperature; J s represents the energy dissipation due to inelastic particle collisions; and G

SC
175 reflects the production or dissipation due to granular-fluid interaction. Based on the kinetic

176 theory of Lun et al. (1984), the diffusion coefficient k s and the dissipation rate J s are

U
177 expressed as
AN
é æ öæ 12 ö 4 ù
25 p çç1+ 12 ha R ÷
178 k s = r sds Q êê s ÷
÷çç1+ h (4 h - 3 a R ÷
) s ÷÷
÷+ ha s
2 ú
R ú (16)
êë16h (41 - 33h )R çè 5 ÷
øèç 5 ø p úû
M

48 a sR
179 Js = h (1 - h ) Q 3/ 2 (17)
D

p ds
TE

180 According to Gidaspow (1994) and Agrawal (2001), G can be given by


EP

2
81a s r f 2nm 2 u f - us
181 G = - 3K Q + (18)
R ds 3 r s p Q
C

182 where, K is a generalized drag coefficient to account for the interaction between the
AC

183 granular and fluid phases, which is discussed in Section 2.4.

184 The frictional stress in the granular phase originates from contacts between granular

185 particles. For cohesionless granular materials, the frictional stress may be generally expressed

186 as (Schaeffer, 1987; Tardos, 1997)

187 t sf = 2msf Ss (19)

8
ACCEPTED MANUSCRIPT

188 where msf is the viscosity due to inter-particle friction. The formulation of msf in this study

189 is based on a statistical relation between macroscopic stress and the bulk rate of strain

190 proposed by Christoffersen et al. (1981):

æ æcos2 f ö÷ cos2 f ö
f pf çç 2 ç ÷
÷
191 m = s sin f + ç Ñ ×u s ÷÷+ 2Sˆ Ñ ×u s ÷ (20)
s
Sˆ ççèç çè 2Sˆ ÷
ø ÷
÷
ø

PT
192 where, Sˆ = 2Ss : Ss ; f is the internal friction angle of the granular material. If

RI
193 Ñ ×u s = 0 , i.e., the granular material being at the critical state, Eq (20) reduces to the

194 classical Mohr-Coulumb law. The effect of dilatancy or contraction, which has been proven to

SC
195 be important in the initial stage of a granular slide (Pailha et al., 2008), is also represented by

196 this relation. Following previous studies (Johnson and Jackson, 1987; Josserand et al., 2005),

U
197 we adopt an empirical expression for the frictional pressure psf :
AN
ìï g1
ïï (a s - a min )
198
ï c r gd
psf = ïí s s (a g2 (a min
< a s < a max )
(21)
ïï - as )
M

max
ïï
ïî 0 (a s
< a min )
D

199 where a max is the close-packed volume fraction and a min is the loose-packed volume
TE

200 fraction of the granular material; g is the magnitude of the acceleration due to gravity; c ,

201 g1 and g 2 are empirical constants related to the properties of the granular material. For
EP

202 spherical glass beads it is suggested to take c = 4 ´ 10- 4 , g 1 = 2 and g 2 = 5 (Johnson

203 and Jackson, 1987). It is worthwhile to point out that the values of a max and a min vary with
C

204 the shape and size distribution of the granular particles. Also note that the frictional pressure
AC

205 vanishes when the volume fraction a s is less than a min . In other words, the frictional stress

206 occurs only when a s exceeds a min .

207 2.3 Constitutive relation for fluid phase

208 The shear stress in the fluid phase includes a viscous stress and a turbulent stress, and in

209 general, may be expressed as (Rodi, 1993)

210 t f
= 2a f r f (nm + nt )S f (22)

9
ACCEPTED MANUSCRIPT

211 where, nt is the eddy viscosity due to fluid turbulence, S f is the tensor of the deviatoric

212 rate of fluid strain:

1é Tù 1
213 Sf = êÑ u f + (Ñ u f ) ú- (Ñ ×u f )I (23)
2 êë úû 3

PT
214 In most of the gravity-driven dense-granular flow problems, the effect of the fluid

215 turbulence is known to be insignificant (Lubin et al., 2006; Lee and Huang, 2018). However,

RI
216 the turbulence effect may not be negligible if the volume fraction of the granular phase is

217 locally reduced, especially near the free surface of a granular slide (Lee et al., 2017). In the

SC
218 present study, the fluid turbulence is considered for completeness. Following Hsu et al. (2004)

219 and Cheng et al. (2017), a modified k - e model is employed to describe the fluid

U
220 turbulence in the granular-fluid mixture. The eddy viscosity nt is then expressed as a
AN
221 function of the fluid turbulent kinetic energy k f and its dissipation rate ef :

kf 2
M

222 nt = C m (24)
ef
D

223 where C m = 0.09 is a constant. The governing equations for k f and ef are
TE

¶ (a f r f k f ) é n ù
+ Ñ ×(a f r f u f k f ) = t f : Ñ u f + Ñ ×êêa f r f t Ñ k f úú
224 ¶t sk (25)
ëê ûú
- a f r f ef - 2a s K (1 - b )k f
EP

¶ (a f r f ef ) ef é n ù
C

+ Ñ ×(a f r f u f ef ) = C 1e t f : Ñ u f + Ñ ×êêa f r f t Ñ ef úú
¶t kf êë se úû
225 (26)
AC

ef ef
- C 2ea f r f ef - C 3e 2a s K (1 - b )k f
kf kf

226 where C 1e = 1.44 , C 2 e = 1.92 , C 3 e = 1.2 , s k = 1.0 , s e = 1.3 are constants. Except

227 for the last term on the right side of Eqs. (25) and (26), other terms are similar to those in the

228 conventional k - e model for clear water flows. This special term represents dissipation due

229 to phase interaction. The parameter b is introduced to characterize the ability of the fluid

10
ACCEPTED MANUSCRIPT

230 turbulence to carry granular particles (Cheng et al., 2017).

231 2.4 Interphase force

232 The interphase force F in Eqs. (5) and (7) governs the momentum exchange between the

233 fluid phase and the granular phase. Since phase interaction is particularly important in the

PT
234 region where the volume fraction of the granular phase takes relatively large values in the

235 problem of our interest, we consider only drag force acting on the granular particles by the

RI
236 ambient fluid for simplicity. Thus, F can be expressed as

æ n ö÷

SC
237 F = K çççu f - us + t Ñ a s ÷÷ (27)
çè sc ø÷

U
238 where, K is a generalized drag coefficient, s c = 1.0 is the Schmidt number representing
AN
239 the ratio of the eddy viscosity of the fluid phase to the diffusivity of the granular phase. In Eq.

240 (27), the mean drag force due to velocity difference between the fluid and granular phases and

the turbulent dispersion, which results from the correlation of granular volume fraction and
M

241

242 fluid velocity fluctuation, are both included. It is also worthwhile to note that the buoyancy
D

243 has been included into the pressure gradient term of the fluid phase.
TE

244 Taking into account the particle group effect, the generalized drag coefficient K can be

245 determined following Gidaspow (1994):


EP

ìï r f a s u f - us - 1.65
ïï 3
ïï
4
CD
ds
af (a s
£ 0.2)
246 K = í ï (28)
C

ïï 150a s 2 r f nm 1.75r f a s u f - us
ïï
ïï a f 2ds 2
+
a f ds
(a s
> 0.2)
AC

247 where the drag coefficient C D is given by

ìï 24
ïï (
1 + 0.15 Res 0.687 ) (Re < 1000)
248 C D = ïí Res s
(29)
ïï
ïîï
0.44 (Re s
³ 1000)

249 in which, Res = u f - us ds nm is the particle Reynolds number.

11
ACCEPTED MANUSCRIPT

250 2.5 Boundary conditions

251 For a specific solution of the basic equations described above, the boundary conditions need

252 to be specified. As shown in Figure 1, there are basically two types of boundaries in the

253 problems of the granular slide generated waves: the solid bottom boundary and the artificial

254 top boundary. In addition, Eq. (1) should be adequately treated to capture the interface

PT
255 between the air and the water both within and outside the granular body. In particular, a

256 non-slip boundary condition is applied on the bottom boundary for both granular and fluid

RI
257 phases. The top boundary is assumed to be a rigid lid allowing slip of fluid while the fluid

SC
258 pressure is set to be zero. Zero-gradient condition is applied to the granular volume fraction

259 a s , the liquid saturation x and the granular temperature Q at all boundaries, which has

U
260 been proven to be effective for various granular flows (Lee et al., 2015). To specify the

bottom boundary conditions for k f and ef , a wall function is introduced so that the normal
AN
261

262 gradient of k f vanishes while ef = C m3/ 4k f 3/ 2 (k d), in which k » 0.4 is Karman constant

263 and d is the distance from the wall.


M

264 2.6 Numerical methods


D

265 The basic equations to be numerically dealt with in this study are Eqs. (4) – (7). The granular
TE

266 stress appeared in the equations is a combined expression of the kinetic theory of Lun et al.

267 (1984) and the frictional law of Christoffersen et al. (1981). The turbulence stress is
EP

268 determined by a modified k - e model. The interphase force includes only drag force

269 between the particle and the fluid phases as shown by Eq. (27). The granular temperature, the
C

270 turbulence kinetic energy and its dissipation rate, which are key parameters in the constitutive
AC

271 relations, are solved with Eqs. (15), (25) and (26).

272 The numerical model is developed on the open-source platform, OpenFOAM®, which

273 offers a variety of numerical schemes for discretization. The present work utilizes the finite

274 volume method (FVM). Detailed solution procedure is outlined in Figure 2. In particular,

275 initial conditions for the volume fraction of granular phase, liquid saturation, granular and

276 fluid velocities, fluid turbulence and granular temperature must be provided. The volume

277 fraction of the granular phase a s is solved by a transport-based approach with flux

12
ACCEPTED MANUSCRIPT

278 correction (Zalesak, 1979), called MULES (Multidimensional Universal Limiter with Explicit

279 Solution) integrator in OpenFOAM®. The pressure-velocity coupled momentum Eqs. (5) and

280 (7) are then solved based on a combination of PISO and SIMPLE methods, called PIMPLE in

281 OpenFOAM® (Passalacqua and Fox, 2011). The convective terms in both momentum and

282 granular temperature equations are discretized with a limited second order central scheme

PT
283 (Sweby, 1984). The diffusion terms are treated with the second-order central difference

284 scheme. To increase computational efficiency, adaptive time step determined by the

RI
285 Courant-Friedrichs-Lewy (CFL) condition is employed. More details on the numerical

286 schemes may be referred to Passalacqua and Fox (2011).

SC
287

U
AN
M
D
TE
C EP
AC

288

289 FIGURE 2. FLOWCHART OF NUMERICAL PROCEDURE.

290 3. Numerical results

291 3.1 Waves generated by collapse of a granular column

13
ACCEPTED MANUSCRIPT

292 We apply the numerical model developed in this study to the simulation of waves generated

293 by collapse of a granular column, which was experimentally investigated by Robbe-Saule et

294 al. (2017). The physical problem is shown in Figure 3. The experiment was conducted in a

295 rectangular tank of 2 m long, 0.3 m high and 0.15 m wide. Initially, a column of granular

296 particles with height H i = 14.5 cm and width Li = 8 cm , confined by a sliding gate, is

PT
297 placed just above the still water surface in the tank with constant water depth h 0 = 5.5 cm .

298 The granular column is filled with monodisperse dry glass beads with density

RI
299 r s = 2500 kg / m 3 , diameter ds = 5 mm and initial volume fraction a 0 = 0.64 . Other
300 parameters required by the numerical model, which are not measured directly in the

SC
301 experiment, are specified according to the suggested values for glass beads in the literature

302 (Johnson and Jackson, 1987; Josserand et al., 2005). It is then determined that f = 20° ,

303
U
e ¢= 0.9 , a min = 0.58 and a max = 0.68 . Collapse of the granular column occurs
AN
304 immediately after the gate is rapidly lifted. For simplicity, numerical simulation is performed

305 under the two-dimensional condition, since the motion of both the granular mass and the
M

306 generated wave does not vary significantly in the transverse direction due to high aspect ratio

307 of the experimental flume. A computational grid with uniform size of 2.5 mm ´ 2.5 mm
D

308 is used to obtain the numerical results.


TE

309
C EP
AC

310

311 FIGURE 3. SKETCH OF THE EXPERIMENTAL PROBLEM OF ROBBE-SAULE ET AL. (2017).


312

313 Figure 4 provides a comparison of the snapshots of both the granular motion and the

314 water waves obtained from the experimental observation and the numerical simulation at

14
ACCEPTED MANUSCRIPT

315 different instants. Note that the water-air interface in the numerical results is represented by

316 the contour of the liquid saturation with x = 0.5 . It is demonstrated that the observed

317 collapsing process of the granular column, plunging of the granular material into the water

318 and the wave generation are qualitatively well reproduced by the numerical model. An

319 impulse wave occurs when the granular mass impacts the water body. The amplitude of so

PT
320 generated leading wave remains nearly unchanged as the wave propagates away from the

321 impact source. It is also worthwhile to point out that the pore space of the dry granular

RI
322 material cannot be filled with water immediately after the granular mass intrudes into the

323 water and a dry portion of the granular body exists for a short period in the experiment. As

SC
324 time goes on, the ambient water then penetrates into the initially dry granular mass. Most of

325 the previous two-phase models can hardly represent this phenomenon because of their

326
U
limitation in dealing with the free surface within the granular material. The present model,
AN
327 however, depicts this dynamic process much better.
328
M
D
TE
C EP
AC

329
330 FIGURE 4. SNAPSHOTS OF THE EXPERIMENT (LEFT PANEL) OF ROBBE-SAULE ET AL. (2017) AND
331 PRESENT NUMERICAL RESULTS (RIGHT PANEL) AT t = 0.1 s, 0.3 s AND 0.6 s .
332

333 Shown in Figure 5 are comparisons between measured and computed amplitude of the

334 leading wave and run-out distance of the granular mass. Figure 5(a) presents the variation of

15
ACCEPTED MANUSCRIPT

335 the amplitude of the generated leading wave during propagation while Figure 5(b) presents

336 the temporal variation of the run-out distance of the granular mass. It is demonstrated that the

337 computed run-out distance of the granular mass matches the experimental observation very

338 well, but the amplitude of the leading wave is slightly overestimated by the numerical model.

339 The discrepancy in the leading wave amplitude may be caused by the inaccurate simulation of

PT
340 the granular mass spreading in the water. As it can be seen in Figure 4, the simulated granular

341 mass has a relatively larger front, resulting in an excessive elevation of the upper water body

RI
342 and thus a higher wave crest. We also performed a test of grid-independence by comparing the

343 numerical results obtained with a finer grid of 1 mm ´ 1 mm and a coarser grid of

SC
344 5 mm ´ 5 mm . No significant differences, in terms of wave amplitude and run-out distance

345 of the granular mass are observed in Figure 5 when varying the grid size, which guarantees

346 the reliability of the numerical results.


U
AN
347
M
D
TE
EP

348
349 FIGURE 5. TIME-VARIATION OF (A) LEADING WAVE AMPLITUDE A AND (B) RUN-OUT DISTANCE OF THE
C

350 GRANULAR MASS L s . THE LINES REPRESENT NUMERICAL RESULTS WITH DIFFERENT GRID SIZES AND THE
351 DOTS REPRESENT EXPERIMENTAL DATA OF ROBBE-SAULE ET AL. (2017).
AC

352

353 3.2 Waves generated by granular slide over inclined surface

354 The numerical model developed in this study is also applied to the simulation of waves

355 generated by granular slide over an inclined surface, which has been investigated in

356 laboratory by Viroulet et al. (2013, 2014). The experimental setup of Viroulet et al. (2013,

357 2014) is shown in Figure 6. Initially, a wedge-shaped mass of granular material of 14 cm

16
ACCEPTED MANUSCRIPT

358 high and 14 cm wide is placed on the inclined surface with a slope of 45o . The granular

359 material used in the experiment consists of glass beads with r s = 2500 kg / m 3 ,

360 ds = 1.5 mm , a 0 = 0.6 and f = 26o . Other parameters involved in the numerical model

361 are taken to be e ¢= 0.9 , a min = 0.58 and a max = 0.68 following previous studies

362 (Johnson and Jackson, 1987; Josserand et al., 2005). Once the gate is lifted, the granular mass

PT
363 slides downwards and intrudes into the water body with a relatively large depth of

364 h 0 = 15 cm . The motion of the granular slide generated waves is measured with 4 wave

RI
365 gauges, located at x = 0.45 m , 0.75 m , 1.05 m , 1.35 m , respectively. For numerical

366 simplicity, this problem is also treated as a two-dimensional one. The computational domain

SC
367 is discretized into 85918 unstructured triangle elements with size ranging from 2 m m

368 near the bottom of the wave flume and the surface of the slope, to 8 mm far above the

369
U
water surface. Grid independence of the numerical results is also verified similar as in the
AN
370 previous computation.
371
M
D
TE

372

373 FIGURE 6. SKETCH OF THE EXPERIMENTAL PROBLEM OF VIROULET ET AL. (2013).


EP

374

375 Shown in Figure 7 are comparisons between the measured and computed flow velocity
C

376 of the water phase as well as the volume fraction of the granular phase. Numerical results are
AC

377 shown to agree with the experimental observation in general. At the beginning (i.e.,

378 t = 0.23 s ), a well-rounded front with larger thickness is formed after the granular slide

379 intrudes into the water body, as shown in Figure 7(a). The phenomenon can be explained by

380 the drag force acted on the granular particles by the ambient water, which causes an

381 appreciable volume of water to be elevated and the water in front of the sliding mass to be

382 pushed away with a higher velocity. In Figure 7(b), the generated leading wave has

383 propagated away from the impact source, and the wave crest is ahead in front of the granular

17
ACCEPTED MANUSCRIPT

384 slide, which agrees with the experimental observation. It is worthwhile to mention that the

385 front of the wave crest has a larger speed as compared to the rear of the wave crest, resulting

386 in a varying wave profile with an increasing wavelength and a decreasing wave amplitude

387 (McFall and Fritz, 2016). In addition, a vortex is formed over the surface of the granular slide.

388 Figure 7(c) shows that at t = 0.52 s , the granular slide has experienced a large deformation

PT
389 and its front has passed the turning point of the slope. Over the upper part of the granular slide

390 at this stage, a back flow can be observed both in experimental and numerical results.

RI
391 However, the vortex over the granular slide that is clearly observed in the experiment is not

392 simulated by the numerical model. This is probably due to the morphological difference in the

SC
393 front of slide between the computed and measured results, since the front of slide in the

394 experiment is evidently steeper, which leads to a notable separation of the ambient fluid and a

395
U
vortex is thus formed. Fortunately, this vortex does not seem to have a significant effect on
AN
396 the characteristics of the leading wave, which is often the major concern in the study of

397 landslide generated waves.


M

398
D
TE
C EP
AC

18
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M

399
D

400 FIGURE 7. SNAPSHOTS OF THE EXPERIMENT (LEFT PANEL) OF VIROULET ET AL. (2013) AND PRESENT
401 NUMERICAL RESULTS (RIGHT PANEL) AT t = 0.23 s, 0.41 s and 0.52 s.
TE

402

403 A comparison between the computed and the measured wave profiles at four different
EP

404 positions, i.e., at x = 0.45 m , 0.75 m , 1.05 m , 1.35 m , is shown in Figure 8.

405 Considering the complexity of the phenomena of landslide intrusion into water, our simulated
C

406 results appear to be fairly reasonable for the leading wave, although the wave amplitude is
AC

407 slightly overestimated (the relative error is less than 20%). The following waves, however, are

408 not similarly well simulated. In particular, an evident phase lag exists in the numerical results.

409 The phase errors in the following waves may be closely related to the inaccurate numerical

410 simulation of the spreading process of the granular particles as they intrude into the water.
411

19
ACCEPTED MANUSCRIPT

PT
412

RI
U SC
AN
413
414 FIGURE 8. COMPARISONS OF COMPUTED (SOLID LINES) AND MEASURED (DASHED-DOTTED LINES) WAVE
M

415 PROFILES AT (a) x = 0.45 m , (b) x = 0.75 m , (c) x = 1.05 m , (d) x = 1.35 m .
D

416 4. Discussions
TE

417 The constitutive law for description of the granular flow in this study is essentially a

418 modification of the relevant relation for dry granular materials by considering the effect of the
EP

419 fluid phase. The modifications are definitely important but most of them are empirical. It is

420 thus of interest to discuss the performance of the key parameters involved in the
C

421 modifications.
AC

422 4.1 Restitution coefficient

423 The restitution coefficient plays an important role in the kinetic theory for granular flows and

424 is thus a critical parameter in the expression for collisional stress. To take into account the

425 effect of the ambient fluid, the restitution coefficient e is expressed as a function of the

426 Stokes number St in this study. The empirical expression is in agreement with the observed

427 facts in standard experiments for collisions of two particles, i.e., energy loss during the

20
ACCEPTED MANUSCRIPT

428 collision process would increase if the granular particles are immersed in a viscous fluid

429 (Armanini et al., 2008). If the density or the viscosity of the ambient fluid is very small, or

430 St becomes very large, e tends to the value of e ¢, i.e., the restitution coefficient for dry

431 granular particles. To identify the effect of the ambient fluid, we compare the numerical

432 results obtained in the previous section with those obtained by letting e = e ¢.

PT
433 As shown in Figure 9(a) and 9(b) for the experimental case of Robbe-Saule et al. (2017),

434 the increased dynamics of the granular material by neglecting the effect of ambient water on

RI
435 the restitution process leads to an obviously overestimated amplitude of the leading wave and

436 run-out distance of the granular mass. Figures 9(c) and 9(d), however, show that the water

SC
437 wave profiles in the experimental case of Viroulet et al. (2013) is not significantly affected by

438 the effect of the water on the restitution coefficient. This is because the amplitude of the

439
U
impulse wave in the experimental case of Viroulet et al. (2013) is mainly determined by the
AN
440 vertical motion of the granular slide. When the granular slide moves over the inclined surface

441 with a large slope, the collision-induced inter-granular stress plays a much less important role
M

442 in the vertical motion of the granular mass as compared to the gravity. In the experimental

443 case of Robbe-Saule et al. (2017), however, the effect of the collisional stress becomes
D

444 relatively important since the wave is mainly generated by the horizontal movement of
TE

445 granular mass over the bottom.


446
C EP
AC

447

21
ACCEPTED MANUSCRIPT

PT
448
449 FIGURE 9. EFFECTS OF RESTITUTION COEFFICIENT ON (A) WAVE AMPLITUDE AND (B) RUN-OUT DISTANCE

RI
450 OF GRANULAR MASS IN THE EXPERIMENTAL CASE OF ROBBE-SAULE ET AL. (2017) AND WAVE PROFILES
451 OF (C) AT x = 0.45 m AND (D) AT x = 0.75 m IN THE EXPERIMENTAL CASE OF VIROULET ET AL.

SC
452 (2013).
453

U
454 4.2 Internal friction
AN
455 The internal friction angle is a critical parameter in the expression of the frictional stress in

456 the granular phase. It has been known that the internal friction angle of the granular material
M

457 is about 5 degrees smaller if they are immersed in the water (Richefeu et al., 2006). However,

458 it is not very clear that such a decrease of the internal friction angle is due to wetting of the
D

459 particles or due to pore pressure change in response to contraction of the granular skeleton
TE

460 when subjected to shear motion. To avoid a double count of the shear contraction effect, we

461 used the internal friction angle for dry granular materials in this study. Nevertheless, a brief
EP

462 discussion on the sensitivity of the numerical results to variation of the internal friction angle

463 of the granular material is still meaningful.


C

464 Shown in Figure 10(a) and 10(b) are comparisons of the computed amplitude of the
AC

465 leading wave and run-out distance of the granular mass in the experimental case of

466 Robbe-Saule et al. (2017), as the internal friction angle is decreased by 5 degrees for those

467 part of the granular material that entered the water. Figures 10(c) and 10(d) show the wave

468 profiles at x = 0.45 m and x = 0.75 m in the experimental case of Viroulet et al. (2013),

469 obtained in the same way. It is evident that in both cases, the numerical results are not

470 significantly affected by reducing the internal friction of the granular material in water.
471

22
ACCEPTED MANUSCRIPT

PT
472

RI
U SC
AN
473
M

474 FIGURE 10. EFFECTS OF INTERNAL FRICTION ON (A) WAVE AMPLITUDE AND (B) RUN-OUT DISTANCE OF
475 GRANULAR MASS IN THE EXPERIMENTAL CASE OF ROBBE-SAULE ET AL. (2017) AND WAVE PROFILES OF
476 (C) AT x = 0.45 m AND (D) AT x = 0.75 m IN THE EXPERIMENTAL CASE OF VIROULET ET AL.
D

477 (2013).
TE

478

479 4.3 Drag force


EP

480 After the granular slide intrudes into the water, the relative motion between the granular

481 particles and their ambient water causes a drag force, which slows down the motion of the
C

482 granular material and drives the water around the granular mass to move away. To
AC

483 demonstrate the important role of the drag force in generation and propagation of the water

484 waves, as well as the motion and deformation of the granular mass, we compare the numerical

485 results obtained in the previous section with what they would be if the drag force is omitted

486 (i.e., the numerical results at K = 0 ). Figure 11 presents the simulated snapshots of the

487 granular collapse and its resulting wave in the experimental case of Robbe-Saule et al. (2017)

488 with K = 0 . By comparing to Figure 4, it is found that the granular mass collapses with a

489 larger velocity and finally deposits with a longer run-out of the granular mass if the drag force

23
ACCEPTED MANUSCRIPT

490 is omitted. In addition, the ambient water penetrates more easily into the granular material

491 after the granular mass plunges into the water. The granular slide generated impulse wave has

492 a slightly smaller amplitude without phase interaction represented by the drag force.
493

PT
494

RI
SC
495

U
AN
M

496 FIGURE 11. SIMULATED SNAPSHOTS OF THE EXPERIMENT OF ROBBE-SAULE ET AL. (2017) AT
497 t = 0.1 s, 0.3 s AND 0.6 s AS DRAG FORCE IS OMITTED.
D

498
TE

499 The simulated snapshots of the experiment of Viroulet et al. (2013) with K = 0 is

500 shown in Figure 12. By comparing to Figure 7, the granular slide is found to move with a
EP

501 larger velocity. The configuration of the granular mass is also significantly different if the

502 drag force is omitted. Eventually, the thinner granular mass with higher mobility elevates less
C

503 volume of water and results in a smaller amplitude of the leading wave. This is phenomenally

in agreement with the results of previous studies (Miller et al., 2017; Mulligan and Take, 2017;
AC

504

505 Si et al., 2018a) that the velocity and front shape of the granular mass are key factors to

506 determine the amplitude of the generated waves.


507

24
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M

508
D

509 FIGURE 12. SIMULATED SNAPSHOTS OF THE EXPERIMENT OF VIROULET ET AL. (2013) AT
510 t = 0.23 s, 0.41 s and 0.52 s AS DRAG FORCE IS OMITTED.
TE

5. Conclusions
EP

511

512 This paper presents an advanced two-phase Navier-Stokes equation model for study of the
C

513 entire process of the granular slide intruding into an otherwise still water body. The VOF

514 method is employed to track the water-air interface inside and outside the granular material.
AC

515 The granular phase is modeled with a general rheological relation that takes into account both

516 collisional and frictional effects. The fluid phase is treated as a viscous water-air mixture and

517 the turbulence effect is described by a modified k - e model.

518 The proposed two-phase model is applied to the study of waves generated by collapse of

519 a subaerial granular column and of waves generated by intrusion of a granular slide into water.

520 The motion and deformation of the granular slide, as well as the generation and propagation

25
ACCEPTED MANUSCRIPT

521 of the water waves are well captured by our model. Effects of the restitution coefficient,

522 internal friction and interphase drag force on the deformation of the granular slide as well as

523 on the evolution of the generated waves are also investigated. It is found that, if neglecting the

524 effect of ambient fluid on the restitution coefficient, the model may overestimate the

525 dynamics of the granular slide as well as the amplitude of the leading wave. Reducing the

PT
526 internal friction of the immersed granular material does not seem to cause a significant

527 difference of the numerical results. The interphase drag force, however, has an important

RI
528 effect on the motion and deformation of the granular material, as well as on the wave

529 generation and propagation. Neglect of the drag force may result in an overestimated run-out

SC
530 of the granular mass and an underestimated amplitude of the leading wave.

531 It may also be worthwhile to point out that the laboratory data used to validate the

532
U
numerical model in this study were all obtained in small-scale experiments. Although the
AN
533 waves generated are not so short in the problems studied so that capillary effects appear in the

534 measured results, small-scale experimental data are usually considered to be less accurate
M

535 mainly because the relative importance of some undesirable factors in the phenomenon of our

536 primary interest, such as the bottom boundary layer effect on the wave motion, may increase.
D

537 To critically validate the numerical model, comparison of computational results with
TE

538 larger-scale physical model data may still be necessary in future studies.
EP

539 Acknowledgement
540 This research is supported by National Natural Science Foundation of China (NSFC) under
C

541 grant No. 11732008. A special acknowledgement is also given to OpenFOAM® developers
AC

542 for their continuous efforts to manage this open-source platform.

543 References
544 Abadie, S., Morichon, D., Grilli, S., Glockner, S., 2010. Numerical simulation of waves
545 generated by landslides using a multiple-fluid Navier-Stokes model. Coastal Engineering
546 57(9), 779–794.

26
ACCEPTED MANUSCRIPT
547 Agrawal, K., Loezos, P.N., Syamlal, M., Sundaresan, S., 2001. The role of meso-scale
548 structures in rapid gas-solid flows. Journal of Fluid Mechanics 445, 151–185.
549 Ancey, C., Cochard, S., 2009. The dam-break problem for Herschel-Bulkley viscoplastic
550 fluids down steep flumes. Journal of Non-Newtonian Fluid Mechanics 158(1–3), 18–35.
551 Armanini, A., Fraccarollo, L., Gambarotto, S., Larcher, M., 2008. Experimental determination
552 of the restitution coefficient for collisional granular-liquid flows. River Flow, Grafiker,

PT
553 1717–1724.
554

RI
Armanini, A., Larcher, M., Fraccarollo, L., 2009. Intermittency of rheological regimes in
555 uniform liquid-granular flows. Physical Review E 79(5), 051306.

SC
556 Ataie-Ashtiani, B., Najafi-Jilani, A., 2008. Laboratory investigations on impulsive waves
557 caused by underwater landslide. Coastal Engineering 55(12), 989–1004.

U
558 Basu, D., Das, K., Green, S., Janetzke, R., Stamatakos, J., 2010. Numerical simulation of
559
AN
surface waves generated by a subaerial landslide at Lituya Bay Alaska. Journal of
560 Offshore Mechanics and Arctic Engineering 132(4), 041101.
561 Cheng, Z., Hsu, T-J., Calantoni, J., 2017. SedFoam: A multi-dimensional Eulerian two-phase
M

562 model for sediment transport and its application to momentary bed failure. Coastal
563 Engineering 119, 32–50.
D

564 Christoffersen, J., Mehrabadi, M.M., Nematnasser, S., 1981. A micromechanical description
TE

565 of granular material behavior. Journal of Applied Mechanics 48(2), 339–344.


566 Fritz, H.M., Hager, W.H., Minor, H-E., 2003. Landslide generated impulse waves. 1.
EP

567 Instantaneous flow fields. Experiments in Fluids 35, 505–519.


568 Fritz H.M., Mohammed, F., Yoo, J., 2009. Lituya Bay landslide impact generated
C

569 mega-tsunami 50th anniversary. Tsunami Science Four Years after the 2004 Indian
AC

570 Ocean Tsunami, pp 153–175.


571 Gidaspow, D., 1994. Multiphase flow and fluidization: continuum and kinetic theory
572 descriptions. Academic press, San Diego.
573 Heinrich, P., Piatanesi, A., Hebert, H., 2001. Numerical modelling of tsunami generation and
574 propagation from submarine slumps: the 1998 Papua New Guinea event. Geophysical
575 Journal International 145(1), 97–111.
576 Heller, V., Hager, W.H., 2010. Impulse product parameter in landslide generated impulse

27
ACCEPTED MANUSCRIPT
577 waves. Journal of Waterway, Port, Coastal, and Ocean Engineering 136(3), 145–155.
578 Hirt, C.W., Nichols, B.D., 1981. Volume of fluid (VOF) method for the dynamics of free
579 boundaries. Journal of Computational Physics 39(1), 201–225.
580 Hsu, T.J., Jenkins, J.T., Liu, P.L.F., 2004. On two-phase sediment transport: sheet flow of
581 massive particles. Proceedings of the Royal Society A: Mathematical, Physical and
582 Engineering Sciences 460(2048), 2223–2250.

PT
583 Ionescu, I.R., Mangeney, A., Bouchut, F., Roche, O., 2015. Viscoplastic modeling of granular
584

RI
column collapse with pressure-dependent rheology. Journal of Non-Newtonian Fluid
585 Mechanics 219, 1–18.

SC
586 Iverson, R.M., George, D.L., 2014. A depth-averaged debris-flow model that includes the
587 effects of evolving dilatancy. I. Physical basis. Proceedings of the Royal Society A:
588 Mathematical, Physical and Engineering Sciences 470(2170), 20130819.
589
U
AN
Jaeger, H.M., Nagel, S.R. and Behringer, R.P., 1996. Granular solids, liquids, and gases.
590 Reviews of Modern Physics 68(4), 1259–1273.
591 Johnson, P.C., Jackson, R., 1987. Frictional-collisional constitutive relations for granular
M

592 materials, with application to plane shearing. Journal of Fluid Mechanics 176, 67–93.
D

593 Josserand, C., Lagrée, P.Y., Lhuillier, D., 2005. Granular pressure and the thickness of a layer
594 jamming on a rough incline. Europhysics Letters 73(3), 363–369.
TE

595 Kelfoun, K., Giachetti, T., Labazuy, P., 2010. Landslide-generated tsunamis at Réunion
596 Island. Journal of Geophysical Research: Earth Surface 115, F04012.
EP

597 Lee, C.H., Huang, Z., Chiew, Y.M., 2015. A three-dimensional continuum model
598 incorporating static and kinetic effects for granular flows with applications to collapse of
C

599 a two-dimensional granular column. Physics of Fluids 27(11), 113303.


AC

600 Lee, C.H, Huang, Z., 2017. A two-phase flow model for submarine granular flows: with an
601 application to collapse of deeply-submerged granular columns. Advances in Water
602 Resources 115, 286–300.
603 Liu, Y., Yu, X., 2016. A coupled phase-field and volume-of-fluid method for accurate
604 representation of limiting water wave deformation. Journal of Computational Physics
605 321, 459–475.
606 Lorenz, A., Tuozzolo, C., Louge, M.Y., 1997. Measurements of impact properties of small,

28
ACCEPTED MANUSCRIPT
607 nearly spherical particles. Experimental Mechanics 37(3), 292–298.
608 Lun, C.K.K., Savage, S.B., Jeffrey, D.J., Chepurniy, N., 1984. Kinetic theories for granular
609 flow: inelastic particles in Couette flow and slightly inelastic particles in a general
610 flowfield. Journal of Fluid Mechanics 140, 223–256.
611 Ma, G., Kirby, J.T., Hsu, T-J., Shi, F., 2015. A two-layer granular landslide model for
612 tsunami wave generation: theory and computation. Ocean Modelling 93, 40–55.

PT
613 Masson, D.G., Harbitz, C.B., Wynn, R.B., Pedersen, G., Løvholt, F., 2006. Submarine
614

RI
landslides: processes, triggers and hazard prediction. Philosophical Transactions of the
615 Royal Society A: Mathematical, Physical and Engineering Sciences 364(1845), 2009–
616

SC
2039.
617 McFall, B.C., Fritz, H.M., 2016. Physical modelling of tsunamis generated by
618 three-dimensional deformable granular landslides on planar and conical island slopes.
619

U
Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
AN
620 472(2188), 20160052.
621 McFall, B.C., Mohammed, F., Fritz, H.M., Liu, Y., 2018. Laboratory experiments on
M

622 three-dimensional deformable granular landslides on planar and conical slopes.


623 Landslides 15(9), 1713–1730.
D

624 Miller, D.J., 1960. Giant waves in Lituya Bay Alaska. USGS Professional Paper 354-C,
TE

625 Shorter Contributions to General Geology.


626 Miller, G.S., Take, W.A., Mulligan, R.P., McDougall, S., 2017. Tsunamis generated by long
EP

627 and thin granular landslides in a large flume. Journal of Geophysical Research: Oceans
628 122(1), 653–668.
C

629 Mohammed, F., Fritz, H.M., 2012. Physical modeling of tsunamis generated by
630
AC

threedimensional deformable granular landslides. Journal of Geophysical Research:


631 Oceans 117, C11015.
632 Mulligan, R.P., Take, W.A., 2017. On the transfer of momentum from a granular landslide to
633 a water wave. Coastal Engineering 125, 16–22.
634 Pailha, M., Nicolas, M., Pouliquen, O., 2008. Initiation of underwater granular avalanches:
635 influence of the initial volume fraction. Physics of Fluids 20(11), 111701.
636 Passalacqua, A., Fox, R.O., 2011. Implementation of an iterative solution procedure for

29
ACCEPTED MANUSCRIPT
637 multi-fluid gas-particle flow models on unstructured grids. Powder Technology 213(1–3),
638 174–187.
639 Pastor, M., Quecedo, M., González, E., Herreros, M.I., Fernández-Merodo, J.A., Mira, P.,
640 2004. Simple approximation to bottom friction for Bingham fluid depth integrated
641 models. Journal of Hydraulic Engineering 130(2), 149–155.
642 Richefeu, V., El Youssoufi, M.S., Radjai, F., 2006. Shear strength properties of wet granular

PT
643 materials. Physical Review E 73(5), 051304.
644

RI
Rodi, W., 1993. Turbulence models and their application in hydraulics: A state of the ar
645 review. Taylor and Francis, New York.

SC
646 Rondon, L., Pouliquen, O., Aussillous, P., 2011. Granular collapse in a fluid: role of the initial

647 volume fraction. Physics of Fluids 23, 073301.

U
648 Robbe-Saule, M., Morize, C., Bertho, Y., Sauret, A., Gondret, P., 2017. Experimental study of
AN
649 wave generation by a granular collapse. EPJ Web of Conferences, EDP Sciences 140,

650 14007.
M

651 Schaeffer, D.G., 1987. Instability in the evolution equations describing incompressible
652 granular flow. Journal of Differential Equations 66(1), 19–50.
D

653 Shi, C., An, Y., Wu, Q., Liu, Q., Cao, Z., 2016. Numerical simulation of landslide-generated
654
TE

waves using a soil-water coupling smoothed particle hydrodynamics model. Advances in


655 Water Resources 92, 130–141.
656 Si, P., Aaron, J., McDougall, S., Lu, J., Yu, X., Roberts, N., Clague, J., 2018a. A
EP

657 non-hydrostatic model for the numerical study of landslide-generated waves. Landslides
658 15(4), 711–726.
C

659 Si, P., Shi, H., Yu, X., 2018b. Development of a mathematical model for submarine granular
AC

660 flows. Physics of Fluids 30(8), 083302.


661 Sweby, P.K., 1984. High resolution schemes using flux limiters for hyperbolic conservation
662 laws. SIAM Journal on Numerical Analysis 21(5), 995–1011.
663 Tan, H., Chen, S., 2017. A hybrid DEM-SPH model for deformable landslide and its
664 generated surge waves. Advances in Water Resources 108, 256–276.
665 Tardos, G.I., 1997. A fluid mechanistic approach to slow, frictional flow of powders. Powder
666 Technology 92(1), 61–74.

30
ACCEPTED MANUSCRIPT
667 Viroulet, S., Sauret, A., Kimmoun, O., Kharif, C., 2013. Granular collapse into water: toward
668 tsunami landslides. Journal of Visualization 16(3), 189–191.
669 Viroulet, S., Sauret, A., Kimmoun, O., 2014. Tsunami generated by a granular collapse down
670 a rough inclined plane. Europhysics Letters 105(3), 34004.
671 Wang, C., Wang, Y., Peng, C., Meng, X., 2017. Dilatancy and compaction effects on the
672 submerged granular column collapse. Physics of Fluids 29(10), 103307.

PT
673 Yavari-Ramshe, S., Ataie-Ashtiani, B., 2017. A rigorous finite volume model to simulate
674

RI
subaerial and submarine landslide-generated waves. Landslides 14(1), 203–221.
675 Zhao, T., Utili, S., Crosta, G. B., 2016. Rockslide and impulse wave modelling in the Vajont

SC
676 reservoir by DEM-CFD analyses. Rock Mechanics and Rock Engineering 49(6), 2437–
677 2456.

U
678
AN
M
D
TE
C EP
AC

31
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
ACCEPTED MANUSCRIPT

Highlights

○ A general two-phase model is proposed for description of landslide generated


waves.

○ An effective collisional-frictional law is adopted for the granular phase and the
drag force is considered to represent the phase interaction.

PT
○ Numerical results are successfully verified by experimental observations.

RI
U SC
AN
M
D
TE
C EP
AC

You might also like