Mechanics Research Communications 73 (2016) 25–30

Contents lists available at ScienceDirect

Mechanics Research Communications

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

A three-dimensional yield-criterion-based flow model for avalanches

Jian-bo Fei a , Tie-lin Chen b , Yu-xin Jie a,∗ , Bing-yin Zhang a , Xu-dong Fu a
a
State Key Laboratory of Hydroscience and Engineering, Tsinghua University, Beijing 100084, China
b
Tunnel and Underground Engineering Research Center of Ministry of Education, Beijing Jiaotong University, Beijing 100044, China

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, a flow model for avalanches based on the three-dimensional yield criterion is presented in
Received 10 December 2014 an attempt to allow the relaxation of the assumption of lateral confinement pressure that is adopted in the
Received in revised form traditional three-dimensional Savage–Hutter model (S–H model). One of the advantages of this model is
14 December 2015
that a simplified constitutive relationship for granular flow, which could reveal the internal mechanism
Accepted 3 February 2016
of avalanches, is adopted. Additionally, another advantage is that the strength parameters used in the
Available online 11 February 2016
proposed model are readily available for natural materials. The flow properties of avalanches are influ-
enced by the generalized friction coefficient, which is a parameter that can be assessed by introducing the
Keywords:
Avalanche
three-dimensional yield criterion. By comparing the results obtained by numerical simulations using the
Granular flow model proposed in this paper and laboratory experiments, a reasonably good agreement can be reached
Yield criterion with regard to the prediction of the moving process of avalanches.
Generalized friction coefficient © 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Avalanches, landslides, debris flows, and mudflows are danger-
ous natural hazards that always happen in mountainous regions,
which could cause large numbers of casualties. Therefore, it is
vital to develop reliable methods to simulate such types of mass
movement. The flow process of avalanches can be modeled by
either a discrete or a continuum approach. For a discrete approach,
such as the distinct element method (DEM), the global proper-
ties of the granular material are studied based on an analysis
of individual particles that obey the basic laws of motion [1–4].
Another method of modeling is to treat the flowing particles as an
entity and adopt differential equations with integrated constitu-
tive relations to describe the flow process. The governing equations
for most methods in this category are derived by integrating the
Navier–Stokes equations from the free surface to the base, and
they can be distinguished based on the basal friction terms and
the adopted constitutive relations.
Various dynamic models have been developed since the 1980s.
In the early years, dam break and flood routing models with modi-
fied basal friction terms were used to model landslides and debris
flows, e.g., Lang and Brown [5], Jeyapalan et al. [6], Takahashi [7,8],
Voight and Sousa [9], and Shieh et al. [10]. These hydrodynamic
methods adopted shallow water-type equations to simulate the
∗ Corresponding author. Tel.: +86 10 62785593; fax: +86 10 62785593.
E-mail address: jieyx@tsinghua.edu.cn (Y.-x. Jie).
debris flow process, i.e., the internal stress was assumed to satisfy
the hydrostatic state. However, the basal friction terms differed in
each of these models and most of them were empirically derived.
Generally, there are two types of mass flow in nature: quasi-
static mass flow and dynamic mass flow. This classification is
based on the ratio between the inertial grain collision stress and
quasi-static Coulomb friction stress. Iverson et al. [11] introduced
the Savage number to indicate the relative magnitude between
the inertial grain collision stress and quasi-static Coulomb friction
stress in granular flow. In the theory, the critical threshold value for
whether a mass flow is quasi-static or dynamic mass is 0.1. It is indi-
cated that most natural mass flows (e.g., avalanches, landslides, and
debris flows) are quasi-static flows in which the Coulomb friction
is dominant, i.e., the flows can be treated as frictional geotechnical
materials. This frictional property has led to the wide adoption of
the yield criterion to describe the state of stress in a moving mass. It
has been proposed by several researchers such as Savage and Hut-
ter [12], Iverson et al. [11], and Frenette et al. [13] that granular
material begins to “flow” when its yielding condition is reached.
It is also presumed in Bingham fluid theory that the mass will not
flow until the stress reaches a certain value, i.e., the yield stress.
Savage and Hutter [12] assumed that the internal stresses in a
moving mass obey the Mohr–Coulomb yield criterion in the direc-
tion of the dominating motion. Furthermore, the basal friction force
was assumed to take the form of Coulomb friction, i.e., the shear
stress was equal to the product of the normal stress and the fric-
tion coefficient. The derivations in this model were the active and
passive earth pressure coefficients, which were used to describe

http://dx.doi.org/10.1016/j.mechrescom.2016.02.004
0093-6413/© 2016 Elsevier Ltd. All rights reserved.
26 J.-b. Fei et al. / Mechanics Research Communications 73 (2016) 25–30

the ratio of the horizontal to vertical stress. Hungr [14] presented

different terms for the basal flow resistance force for different
material rheologies (i.e., plastic flow, friction flow, Newtonian lam-
inar flow, Bingham flow, Coulomb viscous flow, and Voellmy fluid),
and for internal stress, the earth pressure coefficient was assigned
based on the tangential strain prevailing at each mass block. In
their model, Chen and Lee [15] used the earth pressure coefficient
derived by Savage and Hutter [12] to describe the internal stress
and introduced the principle of effective stress when evaluating
the basal friction force.
To simulate avalanches flowing over three-dimensional terrain,
Hutter et al. [16] extended the Savage–Hutter (S–H) theory by
assuming that the stresses in the downslope direction dominate.
Gray et al. [17] made a further improvement to the S–H theory
for modeling avalanches flowing over irregular three-dimensional
terrain. Other modifications of the S–H theory include two mod-
els proposed by Refs. [18,19]: one that relaxes the restrictions of
small slope variations, and the other that takes erosion processes
into account. Additionally, Fernández-Nieto et al. [20] proposed a
new S–H type theory to simulate submarine avalanches and the
associated generated tsunamis.
Although granular flow experiments show that the S–H the-
ory depicts the flow of granular material well, some disadvantages
remain. It is assumed in the S–H theory that the lateral confinement
pressure is equivalent to a principal stress. However, this assump-
tion is valid only under conditions when the “downhill” velocity
and its variation are much bigger than the lateral velocity and its
variation at each specific point in the flowing mass. Therefore, the
S–H theory might have difficulty in theoretical verification when
modeling avalanches flowing over complicated topographies. The
S–H theory indicates a simple constitutive law such that the lateral
stress depends only on the normal stress and the earth pressure
coefficient. However, granular experiments [21] have revealed that
the velocity and strain rate are factors that affect the internal stress
of granular flow.
Jop et al. [22] proposed a constitutive relationship (i.e., (I)-
rheology) for granular material movement under conditions of
shear. In their theory, internal stresses are closely connected with
strain rates, and this proposed constitutive relationship has been
used to interpret experimental results of dry granular flows [21].
However, if we apply this theory to the modeling of natural mass
flows (i.e., landslides and debris flow), parameters such as grain size
and the upper/lower bound friction angles are difficult to determine
because the moving mass is a mixture of sediments with divergent
grain sizes, clay, and water. Moreover, the method proposed by Jop
et al. [22] is difficult to apply, in which a large number of experi-
ments have to be conducted for the calibration of the parameters
in the constitutive law. In this paper, we find the solution for this
problem by importing the three-dimensional yield criterion into
the simplified constitutive relationship. In this way, the physical
parameters that are difficult to determine could be substituted by
strength parameters, which are fundamental benchmarks used to
evaluate the mechanical properties of soil. Furthermore, the use of
the three-dimensional yield criterion might contribute to the relax-
ation of the assumption of the lateral confinement pressure, which
would mean that the modified model could be applied to simulate
avalanches flowing over complex topographies.
Fig. 1. Diagram of the basic element taken from the moving mass for analysis.
the x-coordinate and the horizontal plane is  and that the angle
between the y-coordinate and the horizontal plane is , we can
obtain the angle between the inclined plane and horizontal plane

2
from ˇ = arc cos( 1 − (sin )2 − (sin ) ). If the elevation of the
free surface and the bottom of the moving mass are z = b(x, y, t)
and z = f(x, y, t), respectively, then the depth of the mass flow is
h(x, y, t) = f(x, y, t) − b(x, y, t).
2.1. Simplified constitutive relations
If the lateral scale of the avalanche is much bigger than the
flow depth, and the lateral velocity is much bigger than the ver-
tical velocity, then the avalanche can be treated as “shallow” flow.
Let us consider a cubic element of the entire flowing mass, clinging
to the inclined plane, as the basic element for analysis (Fig. 1). As
the gradient of vertical velocity is small in shallow granular flow,
the shear stresses on the x–z plane and y–z plane can be neglected.
The remaining stresses are shown in Fig. 2, where  xz and  yz are
shear stresses generated by basal friction and pxx , pyy and pzz are
normal stresses in three directions.
Jop et al. [22] proposed a phenomenological frictional viscoplas-
tic constitutive law, which is usually referred to as (I)-rheology.
Based on dimensional analysis, (I)-rheology is proposed by ana-
lyzing large amount of experimental and numerical data sets. It
reveals a common framework describing a wide range of behav-
ior of granular materials [23]. The rheology is different from the
description developed in other viscoplastic materials in two main
aspects: (1) the effective viscosity depends on the pressure, not just
on the shear rate; (2) the flow threshold is not a material constant
but also depends on the pressure [24]. And it is indicated that the
internal stresses are closely related to the isotropic pressure, strain
rate, and second invariant of the strain rate [22].
We introduce the generalized friction coefficient  into the con-
stitutive relation to describe the property of the moving mass;
the definition of the generalized friction coefficient will be dis-
cussed in the following. While for the basal friction, it is assumed
to be Coulomb friction and bx and by denote the basal frictional

2. Governing equations

To describe a moving mass flowing on an inclined plane, a three-

dimensional Cartesian coordinate system is established as follows:
the z-coordinate follows the normal direction of the plane, and
the x-coordinate and y-coordinate are two orthogonal axes par-
allel to the inclined plane (Fig. 1). Given that the angle between Fig. 2. Diagram of stress analysis for the basic element.
 J.-b. Fei et al. / Mechanics Research Communications 73 (2016) 25–30 27

  
coefficients in the x- and y-directions, respectively. When the ele- ∂ 1 2 (∂u/∂x)
ment moves on a plane surface whose curvature is indicated as x g cos  dxdy h 1−
∂x 2 ||
˙
and y in both directions, taken from Gray et al. [17], Nx = x u2 /g and  
Ny = y v2 /g are dimensionless enlarged parts of the basal pressure, ∂h (∂u/∂x) ∂h
i.e., the “centrifugal” contributions due to the downslope curvature ≈ gh cos  dxdy − (4)
∂x ||
˙ ∂x
of the bed, where g is gravitational acceleration, u and v represent
the components of velocities along x- and y- axes, respectively. The inclusion of second-order terms is crucial in simulating
Thus, the stresses can be expressed in the following form: some complex granular problems such as the roll waves and the
erosion-deposition waves (e.g., Refs. [25,26]). However, different
pzz = P
from those detailed morphological study for granular motion, here
P we particularly focus on the range of the movement and runout dis-
pxx = P −  · ˙ xx
||
˙ tance in the analysis of avalanche. Since the second-order terms in
P the momentum equation are not the controlling factors in describ-
pyy = P −  · ˙ yy (1)
||
˙ ing the transverse spread of the avalanche, they are neglected in
 the expression of lateral differential pressure in this paper.
2
xz = ( 1 − (sin )2 − (sin ) + Nx ) · bx Therefore, given the Coulomb-type basal friction term and the
 gravity component term, the continuity equation and the momen-
2
yz = ( 1 − (sin )2 − (sin ) + Ny ) · by
tum equation can be finally written as:

 P is the normal pressure, which can be expressed as P =

where ∂h ∂(hu) ∂(hv)
+ + =0 (5a)
2
gh 1 − (sin )2 − (sin ) and ˙ xx = ∂u/∂x and ˙ yy = ∂v/∂y are ∂t ∂x ∂y
strain rates in two directions.
 The parameter ||
˙ is defined as ||
˙ =
 2 2 √
2 2
(˙ xx ) + (˙ yy ) = (∂u/∂x) + (∂v/∂y) , which is 2 magnitude ∂(hu) ∂(hu2 ) ∂(huv)
+ +
of the second invariant of the strain rate tensor. Substituting these ∂t ∂x ∂y
expressions into Eq. (1), we obtain:  
x u2
 = gh sin  − sgn(u) · bx gh
2
1 − (sin )2 − (sin ) +
pzz = gh 1 − (sin )2 − (sin )
2 g

     
 ∂u 2 ∂h ∂u/∂x ∂h
pxx = gh 1 − (sin )2 − (sin )
2
1− · − gh 1 − (sin )2 − (sin ) · −· · (5b)
||
˙ ∂x (2) ∂x ||
˙ ∂x
  
2  ∂v ∂(hv) ∂(hv2 ) ∂(huv)
pyy = gh 1 − (sin )2 − (sin ) 1− ·
||
˙ ∂y + +
∂t ∂y ∂x
 
2 2 y v2
2.2. Integrated equations = gh sin  − sgn(v) · by gh 1 − (sin ) − (sin ) +
g
By assuming that the moving mass is incompressible, we can   
2 2 ∂h ∂v/∂y ∂h
introduce the Navier–Stokes equation [12]. Since the component of − gh 1 − (sin ) − (sin ) · −· · (5c)
∂y ||
˙ ∂y
velocity along z-coordinate is relatively small compared with the
other two components, it can be neglected. In addition, the small
gradient of vertical velocity assumption indicates that  xz = 0. Thus
3. The generalized friction coefficient
by expanding the Navier–Stokes equations to the three coordinates,
we can get simplified equations as
The above mentioned viscoplastic constitutive law for granular
∂u ∂v flow by Jop et al. [22] is based on shearing experiments with dry
+ =0 granular material of uniform particle size [21]. However, because
∂x ∂y
  natural avalanches, landsides, and debris flows are mixtures of sed-
∂u ∂u ∂u 1 ∂pxx ∂pxz iments with divergent grain sizes, clay, and water, the formula that
+ u+ v = g sin  − + (3)
∂t ∂x ∂y ∂x ∂z includes a parameter for unique grain size might be ineffective.
  In this paper, the unknown generalized friction coefficient  in
∂v ∂v ∂v 1 ∂pyy ∂pyz
+ u+ v = g sin  − + Eqs. (5b) and (5c) can reveal the mechanism of the flowing mate-
∂t ∂x ∂y ∂y ∂z rial, and we introduce yield criteria (e.g., the von Mises criterion
and the Drucker–Prager criterion) to obtain our expression for the
In addition, by assuming that the velocity components in the x-
generalized friction coefficient, which only involves the strength
and y-directions are identical along the depth of the moving mass,
and frictional parameters that are usually available for natural
the velocities can be integrated from the free surface to the base,
materials.
and the magnitude of the shape-factor is assumed to be 1. By analyz-
The von Mises criterion is a strain energy theory pertaining
ing the mass inflow and outflow fluxes of the basic element within
to classical theorem of strength in which the octahedral stress is
an avalanche, we can get the conventional depth-averaged mass
evaluated to estimate whether a material might yield or not. This
conservation equation, that is, ∂h/∂t + ∂(hu)/∂x + ∂(hv)/∂y = 0
criterion can be expressed by three principal stresses as shown in
(whose derivation process was introduced by Ref. [12]).
the following.
As to the depth-averaged momentum equation along the x-
axis, the normal stress applied on the cubic basic element in the 2
J2 = km (6)
x-direction is pxx = gh cos [1 −  · (∂u/∂x)/||].
˙ Then by inte-
grating the normal stress along the flow depth and neglecting the then, modifying Eq. (6) by introducing the second invariant of devi-
second-order terms, we obtain the lateral differential pressure atoric stress J2 , we can acquire
28 J.-b. Fei et al. / Mechanics Research Communications 73 (2016) 25–30

2 · ||
2km ˙ 2
= 2 + ˙ 2 )
− 2(2bx + 2by )
P 2 (˙ xx yy
In which, km is the parameter of material strength.
The Drucker–Prager criterion could reveal the effect of the mean
principal stress on the shear strength and it is widely applied in
numerical calculations for geotechnical problems. The expression
for the Drucker–Prager criterion can be written as (for most gran-
ular material, km = 0)

J2 − ˛m I1 − km = 0
Here, I1 is the first stress invariant. ˛m and km are material
constants that are determined from experiments. As the stress ten-
sor in the mass flow can be divided into a spherical tensor and a
deviator tensor. By substituting the first invariant of principal stress
and the second invariant of deviator stress into Eq. (8), it evolves
into
 
1 2˙ xx ˙ yy 6˛2m (˙ xx + ˙ yy )
2 ˛2m − + +·
2 ||˙ 2 ||
˙ diminishing term, is added to eliminate oscillations in high-
+ 9˛2m − 2bx − 2by = 0
If the first invariant of principal stress is approximately equal
to 3P, we can obtain
 a simpler form of generalized friction coef-
ficient, i.e.,  = 2(9˛2m − bx 2 − by 2 ). This brief expression is
convenient for numerical calculation.
When the Drucker–Prager criterion is adopted, the parameters
(˛m and km ) can be evaluated by fitting the stress space of the
Drucker–Prager criterion to that of the Mohr–Coulomb criterion.
Cohesion (c) and internal friction angle (ϕ) are two basic physi-
cal parameters in the Mohr–Coulomb criterion that describe the
strength of soil whose magnitude can be obtained by conventional
triaxial tests.
4. Basal friction and numerical schemes
4.1. Basal friction
In the Coulomb friction theory, it is assumed that the basal fric-
tion is controlled by the normal stress and the frictional coefficient.
The basal frictional coefficient is usually defined as a constant, e.g.,
Morgenstern [27], Körner [28], and McLellan and Kaiser [29]. How-
ever, the results of experiments on granular material indicate that
flow depth, flow velocity, and the properties of the moving mate-
rial also influence the frictional force. A comprehensive expression
for the basal friction coefficient has been proposed by Jop et al. [30]
based on the experimental results of Pouliquen and Forterre [31].
Supposing that the granular material flows at speed u with depth
h, the formula can be written as
start − stop
bx/by = stop +  (10)
(ˇh gh/uL0 ) + 1
where ˇ and L0 are constants. For granular material flowing on an
inclined plane, two critical angles of inclination are defined:  start
and  stop . When the angle of inclination increases from zero to
 start , the granular material begins to flow. At this point, to immo-
bilize the flowing mass, the angle of inclination has to be reduced
to the second threshold,  stop . Then by relating these two types
of experimental inclination angles with the friction coefficient,
an empirical friction law can be developed with two calibration
parameters, i.e., start and stop (for detailed calibration process,
see Refs. [30,31,32]).
Besides the above mentioned granular basal friction theory,
there are several other types of basal friction laws that can be
adopted to avalanche modeling [14]. In fact, we can also use a sim-
(7) ple basal friction theory (e.g., S–H theory), or the friction law for
plastic flows in the simulation. For example, in Section 5.2, we used
the following friction law for plastic flows which is controlled by
the undrained shear strength, i.e.,
T = fA (11)
in which T is the basal flow resistance force, f is the undrained shear
strength, A is the base area [14].
(8) 4.2. Numerical schemes
The total variation diminishing (TVD)–MacCormack scheme
is adopted in this paper to solve the dynamic Eqs. (5a)–(5c).
This scheme was initially proposed to solve shallow water equa-
tions (SWEs) with second-order accuracy in both time and
space. In the TVD–MacCormack scheme, the MacCormack explicit
predictor–corrector scheme is the fundamental scheme and the
TVD step, which includes a five-point symmetric total variation
gradient regions. During the computation process, we used the
(9)
non-conservative form of Eq. (5) to enhance efficiency and an adap-
tive time step to reduce the computational cost further [33,34].
5. Numerical simulation examples
5.1. Granular material flowing on three-dimensional topography
Experiments with granular material flowing on an inclined
plane without side confinements have been conducted previously,
such as the tests in Hutter [35], Lang [36], Gray et al. [17], and
Wieland et al. [37]. These experiments without constraint could
reveal the flow properties of three-dimensional mass flows better;
thus, we chose the experiments performed with quartz in Wieland
et al. [37] to test the validity of our model. In the experiments, the
reference surface was composed of three parts. The first part was
an inclined plane whose angle of inclination was 40◦ , the second
was a horizontal run-out zone, and the third part was a cylindrical
transition zone that connected the other two. Other specific defini-
tions of the topography can be found in Wieland et al. [37]. Initially,
sand with a hemi-spherical initial free surface was held at the top
of the chute. When the tests commenced, the sand was released
and a sequence of images taken.
In our simulation, the constitutive properties of the flowing
mass are described by adopting the Drucker–Prager criterion. The
Drucker–Prager criterion could show the effect of the mean princi-
pal stress on the shear strength and its shape in the principal stress
space is a smooth conical surface. Thus, the Drucker–Prager crite-
rion was employed to deduce the expression for the generalized
friction coefficient in our simulation. As the experimental mate-
rial was cohesionless with an internal angle of 40◦ , the strength
parameter ˛m for the Drucker–Prager criterion could be evaluated
by fitting the stress space of Drucker–Prager criterion to that of the
Mohr–Coulomb criterion. By adopting the exterior angle circum-
scribed fitting pattern in the three-dimensional stress state, the
strength parameter was determined as
2 sin ϕ
˛m = √ = 0.315 (12)
3(3 − sin ϕ)
For the basal friction, when a constant bed friction angle was
adopted, it was found that the tail of the computed flow using
the S–H model moved too slowly in comparison with the experi-
ments [37], as shown in Fig. 3. To align the experiments and theory,
Wieland et al. [37] introduced an empirical bed friction angle that
changed with the relative position in the flow. In our simulation,
 J.-b. Fei et al. / Mechanics Research Communications 73 (2016) 25–30 29

Fig. 4. The experimental installation (unit: cm).

Fig. 3. Comparison of numerical simulations and laboratory experiments (the posi-

tion of the mass flow’s edge in laboratory experiments is represented by a thick solid
line and the computed position of the mass flow’s edge is indicated by a dotted line)
(unit: cm).

Eq. (10) was employed to calculate the basal frictional coeffi-

cient. Since the basal friction parameters are not measured in the
experiment, thus we approximately take the typical friction param-
eters for glass beads in our simulation by referring to Pouliquen
and Forterre [31], i.e., L0 /d = 1.65, ˇ = 0.136,  start = 32.76◦ , and
 stop = 20.9◦ .
The results of numerical simulations using the model proposed
in this paper are shown in Fig. 3. Furthermore, for comparative anal-
ysis, Fig. 3 also illustrates the computation results obtained using
the S–H model and the results of laboratory experiments. By com-
paring the computation results using the modified model and the
experimental results, we found that the mass flow was simulated
well before 1.00 s, but thereafter the simulated flow appeared to
move more slowly than the experimental one. In the experiment,
the mass flow came to rest after 1.79 s, but the simulated flow con-
tinued to move until 1.85 s. Compared with using the S–H model, it
can be seen from the graphs that the flow simulated by the modified
model could reproduce the experimental results better.

5.2. Deposition zones prediction

It is significant to predict the deposition zones of avalanches

because it can help us to assess the hazardous range and risk level
in mountainous areas. Liu et al. [38] conducted laboratory experi-
ments to study the relationships between initial flow velocity and
the ultimate deposition in avalanches. The experimental installa- Fig. 5. The shapes of the deposition range with different inflow velocities (the
tion is shown in Fig. 4. In the experiments, 9 independent trials positions of the deposition edges in the laboratory experiments are represented
by dotted line, and the contoured graphs are the computed deposition edge)
were carried out with the volume of experimental material fixed at
(unit: cm).
11,000 cm3 , but the inflow velocity (as the initial condition) varies
from 1.00 m/s to 2.25 m/s. The unit weight of the soil is 1.817 g/cm3 .
In this example, we adopt Von Mises criterion to get the general- inflow velocities, and the simulation results fit much better when
ized friction coefficient. As to the basal friction, we used Eq. (11) to the inflow velocity increases.
make the simulation.
 Then the equation for the generalized
 friction
√
coefficient is  = 2 /P 2 ) − (2f 2 /P 2 ) = ( 2/P) ·
(2km 2 − f 2 . The
km 6. Conclusions
Von Mise strength parameter and the frictional parameter are set
as km = 12.0 kPa and f = 6.5 kPa, respectively. For numerical calcu- In this paper, to relax the assumption of lateral confinement
lation, the boundary condition was defined at the supply gateway pressure in the three-dimensional S–H model, a flow model based
with different inflow velocities, i.e., v0 = 1.00, 2.00, and 2.25 m/s, on the three-dimensional yield criterion was proposed for the sim-
respectively. ulation of avalanches. One of the main advantages of this model
The experimental and computational results of the ranges of is that it can reveal the internal mechanisms of avalanches more
the depositions with diverse inflow velocities are shown in Fig. 5. reasonably by adopting a simplified constitutive relationship for
In general, the modified model could well predict the runout dis- granular flow. Another advantage is that the strength parameters
tances and deposition zones. But it is obvious that the model fails used in the proposed model can be acquired easily for natural mate-
to accurately show the transverse spreading of avalanches at low rials. The generalized friction coefficient defined in this paper is
30 J.-b. Fei et al. / Mechanics Research Communications 73 (2016) 25–30
one of the most important factors in describing the flow prop-
erties of avalanches. The expression for the generalized friction
coefficient was defined according to the von Mises criterion and
the Drucker–Prager criterion. Although the values of the physical
parameters in the granular basal friction law are taken from another
experimental result, numerical simulation results indicate that the
model can well capture the flow process of the avalanche and their
deposition zone.
It should be noted that, apart from the two criteria used in
this paper, the application of the Mohr–Coulomb criterion and
Matsuoka–Nakai criterion [39] to the dynamic equations was also
attempted in our simulation. In the evaluation of the generalized
friction coefficient during the numerical simulations, the problem
became one of solving a quadratic equation with one unknown
for the Mohr–Coulomb criterion, and solving a cubic equation
with one unknown that included a third-order term of velocity
for the Matsuoka–Nakai criterion. Numerical tests for the first
example using these two criteria showed that numerical oscil-
lations occurred during the simulation, and that it was difficult
to make the iteration converge. Therefore, superior numerical
schemes are required to solve the dynamic equations with the inte-
grated Mohr–Coulomb and Matsuoka–Nakai criteria and further
work should be done on this topic in the future.
Additionally, since the avalanche dynamic model proposed in
this paper focuses on describing the flow process and the deposi-
tion zone of the avalanche, the second-order terms are neglected
in the momentum equation. Though a reasonably good agreement
between simple numerical simulation and realistic observation
can be achieved, more sophisticated governing equations should
be considered when dealing with other complicated granular
problems.
Acknowledgements
The support of the National Basic Research Program of China
(973 Program 2013CB036402), Natural Science Foundation of China
(51279085), and the Project of Zhongtiaoshan tunnel of Mengxi
railway is gratefully acknowledged.
References
[1] R. Bharadwaj, C. Wassgren, R. Zenit, The unsteady drag force on a cylinder
immersed in a dilute granular flow, Phys. Fluids 18 (2006) 043301.
[2] L. Claessens, G.B.M. Heuvelink, J.M. Schoorl, A. Veldkamp, DEM resolution
effects on shallow landslide hazard and soil redistribution modelling, Earth
Surf. Process. Landf. 30 (2005) 461–477.
[3] D.M. Hanes, O.R. Walton, Simulations and physical measurements of glass
spheres flowing down a bumpy incline, Powder Technol. 109 (2000)
133–144.
[4] H. Teufelsbauer, Y. Wang, M.C. Chiou, W. Wu, Flow–obstacle interaction in
rapid granular avalanches: DEM simulation and comparison with experiment,
Granul. Matter 11 (2009) 209–220.
[5] T.E. Lang, R.L. Brown, Snow avalanche impact on structures, J. Glaciol. 25 (1980)
445–455.
[6] J.K. Jeyapalan, J.M. Duncan, H.B. Seed, Analyses of flow failures of mine tailings
dams, J. Geotech. Eng. ASCE 109 (2) (1983) 150–171.
[7] T. Takahashi, Debris Flow, Balkema, Rotterdam, 1991.
[8] J.S. O’Brien, P.Y. Julien, W.T. Fullerton, Two-dimensional water flood and mud-
flow simulation, J. Hydraul. Eng. 119 (1993) 244–261.
[9] B. Voight, J. Sousa, Lessons from Ontake-san: a comparative analysis of debris
avalanche dynamics, Eng. Geol. 38 (1994) 261–297.
[10] C.L. Shieh, C.D. Jan, Y.F. Tsai, A numerical simulation of debris flow and its
application, Nat. Hazards 13 (1996) 39–54.
[11] R.M. Iverson, M.E. Reid, R.G. LaHusen, Debris-flow mobilization from landslides,
Annu. Rev. Earth Planet. Sci. 25 (1997) 85–138.
[12] S.B. Savage, K. Hutter, The motion of a finite mass of granular material down a
rough incline, J. Fluid Mech. 199 (1989) 177–215.
[13] R. Frenette, T. Zimmermann, D. Eyheramendy, Unified modeling of fluid or
granular flows on dam-break case, J. Hydraul. Eng. 128 (2002) 299–305.
[14] O. Hungr, A model for the runout analysis of rapid flow slides, debris flows, and
avalanches, Can. Geotech. J. 32 (1995) 610–623.
[15] H. Chen, C.F. Lee, Numerical simulation of debris flows, Can. Geotech. J. 37
(2000) 146–160.
[16] K. Hutter, M. Siegel, S.B. Savage, Y. Nohguchi, Two-dimensional spreading of
a granular avalanche down an inclined plane, part I. Theory, Acta Mech. 100
(1993) 37–68.
[17] J. Gray, M. Wieland, K. Hutter, Gravity-driven free surface flow of granular
avalanches over complex basal topography, Proc. R. Soc. Lond. 455 (1999)
1841–1874.
[18] F. Bouchut, A. Mangeney-Castelnau, B. Perthame, J. Vilotte, A new model of
Saint Venant and Savage–Hutter type for gravity driven shallow water flows,
C. R. Math. 336 (2003) 531–536.
[19] F. Bouchut, E.D. Fernández-Nieto, A. Mangeney, P.-Y. Lagrée, On new erosion
models of Savage–Hutter type for avalanches, Acta Mech. 199 (2008) 181–208.
[20] E.D. Fernández-Nieto, F. Bouchut, D. Bresch, M.J. Castro Díazd, A. Mangeney,
A new Savage–Hutter type model for submarine avalanches and generated
tsunami, J. Comput. Phys. 227 (2008) 7720–7754.
[21] F. da Cruz, S. Emam, M. Prochnow, J.N. Roux, F. Chevoir, Rheophysics of dense
granular materials: discrete simulation of plane shear flows, Phys. Rev. E 72
(2005) 021309.
[22] P. Jop, Y. Forterre, O. Pouliquen, A constitutive law for dense granular flows,
Nature 441 (2006) 727–730.
[23] P.Y. Lagrée, L. Staron, S. Popinet, The granular column collapse as a continuum:
validity of a two-dimensional Navier–Stokes model with a (I)-rheology, J.
Fluid Mech. 686 (2011) 378–408.
[24] B. Andreotti, Y. Forterre, O. Pouliquen, Granular Media: Between Fluid and Solid,
Cambridge University Press, 2013.
[25] J. Gray, A. Edwards, A depth-averaged (I)-rheology for shallow granular free-
surface flows, J. Fluid Mech. 755 (2014) 503–534.
[26] A. Edwards, J. Gray, Erosion–deposition waves in shallow granular free-surface
flows, J. Fluid Mech. 762 (2015) 35–67.
[27] N.R. Morgenstern, Submarine Slumping and the Initiation of Turbidity Currents,
Marine Geotechnique, Urbana University of Illinois Press, Urbana, 1967, pp.
189–220.
[28] H.J. Körner, Reichweite und Geschwindigkeit von Bergstürzen und
Fließschneelawinen, Rock Mech. 8 (1976) 225–256.
[29] P.J. McLellan, P.K. Kaiser, Application of a two-parameter model to rock
avalanches in the Mackenzie Mountains, in: Proceedings of the 4th Interna-
tional Symposium on Landslides, Toronto, Ontario, 1984, pp. 135–140.
[30] P. Jop, Y. Forterre, O. Pouliquen, Crucial role of sidewalls in granular surface
flows: consequences for the rheology, J. Fluid Mech. 541 (2005) 167–192.
[31] O. Pouliquen, Y. Forterre, Friction law for dense granular flows: application to
the motion of a mass down a rough inclined plane, J. Fluid Mech. 453 (2002)
133–151.
[32] Y. Forterre, O. Pouliquen, Long-surface-wave instability in dense granular flows,
J. Fluid Mech. 486 (2003) 21–50.
[33] D. Liang, R.A. Falconer, B. Lin, Comparison between TVD–MacCormack and ADI-
type solvers of the shallow water equations, Adv. Water Resour. 29 (2006)
1833–1845.
[34] D. Liang, B. Lin, R.A. Falconer, Simulation of rapidly varying flow using an
efficient TVD–MacCormack scheme, Int. J. Num. Methods Fluids 53 (2007)
811–826.
[35] K. Hutter, Two-and Three Dimensional Evolution of Granular Avalanche Flow –
Theory and Experiments Revisited. Aeolian Grain Transport 1, Springer, Vienna,
1991, pp. 167–181.
[36] R.M. Lang, An Experimental and Analytical Study on Gravity-driven Free Surface
Flows of Cohesionless Granular Media (Ph.D. thesis), Department of Mechanics,
Technische Hochschule Darmstadt, Germany, 1992.
[37] M. Wieland, J. Gray, K. Hutter, Channelized free-surface flow of cohesionless
granular avalanches in a chute with shallow lateral curvature, J. Fluid Mech.
392 (1999) 73–100.
[38] J.F. Liu, Q. Guo, Y. You, Experimental research on velocity and deposition mode
of debris flow, Res. Soil Water Conserv. 13 (2006) 120–121 (in Chinese).
[39] H. Matsuoka, On the significance of the spatial mobilized plane, Soils Found. 16
(1976) 91–100.