You are on page 1of 12

Computers and Geotechnics 126 (2020) 103737

Contents lists available at ScienceDirect

Computers and Geotechnics


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

Research Paper

Particle size effects on small-scale avalanches and a μ(I) rheology-based T


simulation
Jianbo Feia, Yuxin Jieb, Xiaohui Sunc, , Xi Chend

a
Underground Polis Academy, College of Civil and Transportation Engineering, Shenzhen University, Shenzhen 518060, China
b
State Key Laboratory of Hydroscience and Engineering, Tsinghua University, Beijing 100084, China
c
Underground Polis Academy, College of Civil and Transportation Engineering, Shenzhen University, Shenzhen 518060, China
d
School of Civil Engineering, Beijing Jiaotong University, Beijing 100044, China

ARTICLE INFO ABSTRACT

Keywords: In this paper, we introduce the μ(I) rheology into the traditional depth-averaged continuum framework and
Avalanche include the effect of grain size when describing the normal stresses of a flowing avalanche. We also develop a set
Laboratory experiment of experimental installations for granular flow and use the installations to simulate laboratory avalanches: sand
Dynamic model initially accumulates in various geometries, is then released and slides on inclined planes at different angles of
μ(I) rheology
inclination, and is finally deposited on a horizontal plane. The experimental materials are sand with different
Grain size
particle sizes. With the combination of the experimental and simulation results, the effects of particle size on the
deposit configuration, runout distance, and deposit depth are revealed. The effect on the runout distance de-
creases with the angle of inclination, while the effect on the deposit depth is not obvious. Other factors such as
the initial accumulation geometry and frictional parameters of the μ(I) rheology are carefully considered and
studied.

1. Introduction mass obeys Mohr-Coulomb yield criterion. This idea was further ex-
tended and developed by subsequent studies to form S-H theory (or the
Avalanches are life-threatening natural hazards that can cause many S-H model) (Hutter et al., 1993; Iverson et al., 1997; Gray et al., 1999;
casualties. They often take the form of landslides, debris flows, and mud Iverson and Denlinger, 2001; Pudasaini and Hutter, 2007). Hungr
flows in mountainous regions. The sudden catastrophic collapse of (1995) and Mcdougall and Hungr (2005) modified the earth pressure
waste dumps is a special variety of avalanche, for example, the flow coefficient in S-H theory using a strain-related expression based on
slide at Shenzhen landfill that happened in 2015 (Zhan et al., 2018; Zhu conventional loading and unloading experiments with soil; their studies
et al., 2018). Developing effective and reliable approaches to predict then evolved to form the DAN model. To solve the developed theories
the potential areas endangered by such types of mass movement is for avalanches, several numerical simulation methods have been
necessary. The influence area prediction of avalanches can be modeled adopted, e.g., finite difference method (Hutter et al., 1993; Gray et al.,
using either a discrete (Claessens et al., 2005; Poisel et al., 2008; Wu 1999; Pudasaini and Hutter, 2007), finite element method (Cascini
et al., 2013) or a continuum approach. The continuum approach treats et al., 2014), the mesh-free SPH (smoothed particle hydrodynamics)
the flowing particles as an entity and adopts differential equations in- method (Pastor et al., 2009; Cuomo et al., 2014; Cuomo et al., 2019)
tegrated with various constitutive relationships to describe the flow and SPH-FDM method (Cascini et al., 2016).
process. Dam break and flood routing models (hydrodynamic methods) Apart from the development of theoretical avalanche modeling,
were built to model landslides with high water content or debris flows different scales of experiments have been conducted with varying
with the adaptation of shallow water-type equations and modified basal measurement techniques to validate the theory and demonstrate the
friction laws (Lang and Brown, 1980; Takahashi, 1991; Shieh et al., basic features of granular movements. Gubler (1987), Norem et al.
1996), which were initially developed as continuum methods. Savage (1989), and Issler (2003) used field observations of natural avalanches
and Hutter (1989, 1991) introduced the concept of an earth pressure to investigate the basic patterns and internal mechanisms of natural
coefficient to describe the relative magnitude of the vertical and lateral avalanche events. Field rockfall experiments on an artificial slope were
stresses in a dry cohesionless granular flow by assuming the flowing conducted by Okura et al. (2000) to analyze runout distances and


Corresponding author.
E-mail addresses: feijianbo@szu.edu.cn (J. Fei), jieyx@mail.tsinghua.edu.cn (Y. Jie), sunxiaohui@szu.edu.cn (X. Sun), chenxi@bjtu.edu.cn (X. Chen).

https://doi.org/10.1016/j.compgeo.2020.103737
Received 16 March 2020; Received in revised form 30 June 2020; Accepted 1 July 2020
Available online 23 July 2020
0266-352X/ © 2020 Elsevier Ltd. All rights reserved.
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

individual movements of rockfall block, in which granite slabs were assess hazardous ranges in mountainous areas. In this study, we used
surfaced and overlaid with cubiformed granite blocks. However, field the depth-averaged continuum approach to simulate the granular flow
observations to verify the theoretical models can be challenging. The process and describe the deposit configuration by introducing μ(I)
flow terrain, initial conditions, and material properties of field ob- rheology into the governing equation. Then we established an experi-
servations are complex and hard to control, which motivated us to use mental installation for an avalanche with side confinement and in-
laboratory experiments in this paper. vestigated the effects of the particle size, angle of inclination, and initial
Laboratory experimental approaches are more convenient and ef- geometry of the granular pile on the deposit’s ultimate configuration
fective to study the individual flow behavior of granular flow. The most and runout distance. A combined analysis of the experimental and si-
commonly used laboratorial granular flow experiments can be classified mulation results revealed the avalanche’s flowing and deposit features.
into two categories in terms of the installations. The first category in-
cludes experiments in which the granules flow on a three-dimensional 2. μ(I)-rheology-based depth-averaged continuum model
surface without sidewise confinement with experimental installations
that primarily consist of an inclined plane (Greve et al., 1994; Koch 2.1. μ(I) rheology
et al., 1994), a parabolic surface (Gray et al., 1999; Wieland et al.,
1999), a horizontal runout plane, and a curved transition region. Gray As avalanches are mostly composed of granular materials, recent
and Hutter (1998), Pudasaini et al. (2008), and Iverson et al. (2004) research progress into the mechanism of flowing granular media
modified the aforementioned experimental method in which the talweg highlighted the development of a more physically reasonable dynamic
followed the direction of the sliding surface’s steepest decent, and the avalanche model. In this section, we will first introduce μ(I) rheology
bed profile exhibited both curvature and twist. The other category and related studies. Pouliquen (1999) presented a scaling property of
defines experiments in which granules move within a narrow straight or granular flows down rough inclined planes and proposed an empirical
curved chute situated in a vertical plane: chutes with horizontal beds description of the dynamic friction coefficient of a moving mass on a
were studied by Huber (1980); chutes composed of an inclined plane, a plane. Jop et al. (2005) studied steady uniform flows when granules
horizontal deposit plane, and a curved transition region were assessed were released from a hopper on top of a static pile in a channel and
by Hutter et al. (1988); and chutes with curved beds were examined by developed a theoretical model based on a simple local constitutive law
Hutter and Koch (1991). for granular flow configuration. Endorsed by the numerical simulation
Avalanches’ various properties have been studied in the laboratory results of 2D plane shear (da Cruz et al., 2005), rotating drums (Renouf
for decades. Savage and Hutter (1989), Hogg (2007), and Balmforth and Alart, 2005), 3D numerical simulation and experimental mea-
and Kerswell (2005) conducted experiments on a box-shaped granular surements of inclined plane granular flows (Pouliquen, 1999), and 3D
material behind a lock gate and studied its yielding and moving be- laboratory data obtained from an annular shear cell experiment (Savage
havior on chutes. Manzella and Labiouse (2008) produced a small-scale and Sayed, 1984), the frictional coefficient μ(I) of dense granular flows
physical model for a laboratory simulation of a granular avalanche and can be modeled using the phenomenological expressions (Jop et al.,
studied the effects of the initial volume, fall height, and material 2006):
properties on the runout distance, width, and morphology of the ulti- µ 2 µs
mate deposit. Davies and Mcsaveney (1999) conducted laboratory ex- µ (I ) = µs +
I0 / I + 1 (1a)
periments on granular avalanching with dry sand and gravel, in which a
consistent pattern of runout distance varying with the fall height, fall | |d
I=
slope, and volume of material. Pouliquen (1999), Pouliquen and P/ (1b)
s
Forterre (2002), and Jop et al. (2005) experimentally and theoretically
explored the frictional behavior of a uniform granular flow on an in- where µs and µ 2 are the upper and lower limit values of the frictional
clined plane. coefficient, I0 is an experimental constant, d is the grain diameter, s is
Apart from the Mohr-Coulomb criterion in the S-H model, different the grain density (or solid density), P is the confining pressure, and is
constitutive assumptions were recently made in the development of the strain rate. If the granular flow is incompressible and assuming an
depth-averaged continuum approaches. Faug et al. (2015) assumed that alignment between the stress tensor and strain rate tensor (alignment of
the velocity profile of a thin granular flow on a smooth base could be tensors implies co-axiality) (Silbert et al., 2001; Depken et al., 2007;
approximated by the slip velocity for the basal layer plus a Bagnold Barker et al., 2017; Rauter et al., 2020), the full internal stress tensor of
profile for the upper sliding layer. Gray and Edwards (2014), Edwards a flowing granular mass is expressed as (Jop, 2015):
and Gray (2015), Baker et al. (2016a,b); Viroulet et al. (2018), and µ (I ) P
Rocha et al. (2019) introduced μ(I)-rheology to capture complicated ij = P ij + ij
| | (2)
granular flow phenomena such as roll waves, segregation-induced fin-
gering, and levee formation. Using μ(I) rheology or μ(h,Fr) rheology- where ij = ui / x j + uj / x i is the strain rate tensor and the second
based methods, the particle size effect was considered in several prior invariant of ij is | | = (0.5 ij ij )0.5 .
studies. Pouliquen and Forterre (2002) developed a friction law for
inclined granular flow (μ(h,Fr)-rheology), in which the μ1, μ2, and L 2.2. Plug-like velocity profile assumption
parameters were grain size dependent. Baker et al. (2016b) considered
the effect of particle size on segregation-induced finger formation. Since the earliest laboratory studies on the relationship between
Viroulet et al. (2018) studied the effect of particle size on roll wave stresses and shear rates with annular shear cells (Bagnold, 1954;
instability and found that the frictional properties highly depended on Bagnold, 1962), different types of experiments have been conducted,
the grain size. The effect of the particle size on μ(I)-rheology-based for example, in reference (da Cruz et al., 2005; Campbell, 1990). The
methods is mainly embodied in two frictional parameters, μs and μ2, dependence of the stress on the shear rate was negligible if annular
which are obtained by calibrating inclined plane tests. Despite ex- shear cell tests were conducted at a constant normal stress (Hanes and
tensive research, experiment-theory coupled study on the mechanical Inman, 1985; Savage, 1979; Savage and Mckeown, 1983). The in-
properties (including the direct particle size effects) and flowing fea- dependence of the shear effect motivated the plug-like assumption of
tures of avalanches (from start-up and runout to deposit) is still ne- free surface granular flow modeling (Pudasaini and Hutter, 2007), so
cessary. An accurate and effective depth-averaged continuum approach theories such as the S-H and DAN models (Savage and Hutter, 1989;
is significant because it can help elucidate the nature of avalanche Savage and Hutter, 1991; Hungr, 1995; Mcdougall and Hungr, 2005)
movement, predict the deposition zones of natural avalanches, and were developed based on this assumption. The two types of models

2
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

u
h pxx h µ 2u h u h h x h
dz = gz h ( + ) gz h µ
0 x x | | x2 2 x x x | | x

(4a)
v
h pyy h µ 2v h v h h y h
dz = gz h ( + ) gz h µ
0 y y | | y2 2 y y y | | y

(4b)
pxx and pyy are negatively related to σxx and σyy, respectively. Thus,
we obtain the mass and the momentum conservation equations by in-
troducing the μ(I) rheology (Eqs. (1) and (2)) and integrated terms (Eq.
(4)) into the mathematical governing framework for a shallow granular
flow proposed by Savage and Hutter (1989), Savage and Hutter (1991),
Gray et al. (1999), and Wieland et al. (1999) as
h (hu) (hv )
Fig. 1. A granular avalanche on the move with plug flow and basal shear flow + + =0
t x y (5a)
regimes.
u
(hu) (hu2 ) (huv ) h h
deem that the profile of the flow velocity along an inclined plane is + + = gx h bx gz h + µ (I ) x gz h
t x y x | | x
mainly composed of two factors: an extremely thin shearing layer in
(5b)
which the velocity increases rapidly from zero at the bottom to a certain
value at the top (shear flow) and a relatively thicker sliding layer in v
(hv ) (hv 2) (huv ) h y h
which the velocity is almost identical along the depth (plug flow), as + + = gy h by gz h + µ (I ) gz h
shown in Fig. 1.
t y x y | | y (5c)
This treatment of dense granular flows moving passively above a where gx , gy , and gz are the gravitational acceleration in the x, y, and z
thin fluidized basal layer has been validated in laboratory tests on directions, respectively. bx and by denote the basal friction in the x and
avalanches, large scale natural avalanches and debris flows, and dry y directions. As previously explained, the shear effect is included in a
snow (Gray et al., 1999; Ancey and Meunier, 2003; Keller et al., 1998; thin basal layer in which the velocity increases rapidly from zero at the
Pudasaini et al., 2005). Dent et al. (1998) conducted in situ avalanche bottom to a certain value at the top. The in-plane shears above the
experiments in Montana and drew the velocity profile of an avalanche fluidized basal layer are neglected in our model, and the second in-
along the depth based on the scene photographs. They found that the variant is entirely dominated by the velocity difference between ad-
shearing effect in the thin shearing layer was strong while that in the jacent column-shaped elements,
sliding layer was negligibly small. Laboratory experiments (Hutter
2 2
et al., 1832) confirmed the validity of including all of the shear effect 1 2 2] 1 u v
| |= [( xx ) +( yy ) = +
inside the basal shearing layer when the angle of basal friction ranged 2 2 x y (6)
from 30° to 50° and the basal terrain was relatively smooth. Considering
the reported findings and smoothness of the experimental channel, the Eq. (6) indicates the effect of velocity differences on the internal
present paper assumes a plug-like profile during model deviation, stresses in the model proposed in this paper. The velocity difference
which differs from the classical Bagnold-type shear profile on rough determines whether the avalanche is in the compression or decom-
beds with no slip at the base. pression states, which consistently corresponds to the passive or active
earth pressure coefficient in the traditional S-H model.
To relate the physical characteristics of the velocity profile to the
2.3. Model development mathematical expression, if we further consider the integration process
of the square of the velocity in the x direction u on the left side of Eq.
To develop a depth-averaged continuum model considering a (5b)–(5c),
‘‘shallow’’ granular flow with a scale of lateral spread that is much
1 h
larger than that of the flow depth, we analyze a cubic element in the u2dz = u¯ 2
(7)
h 0
flow with a flow depth of h with average flow velocity components u
and v in the perpendicular directions. By neglecting the internal shear where α is a constant arising during integration, which implies the
stresses above the basal friction layer with the assumption of a plug deviation of the velocity from a uniform profile. α = 1.2 corresponds to
flow velocity (consistent with S-H theory), the granular normal stress (x a parabolic velocity profile with zero basal velocity, which indicates no
and y directions) on the granular flow is expressed as slide at the base while shear takes effect above the basal layer. α ≈ 1
corresponds to a uniform velocity profile, which indicates fully sliding
pxx = P + pxx pyy = P + pyy , (3a) at the base with no in-plane shear above the basal layer. 1 < α < 1.2
corresponds to a power law distribution of the velocity. Detailed in-
µP µP vestigations illustrate that α = 1 does not introduce large errors, and
pxx = pyy = yy,
| | xx
| | (3b) the deviation from α = 1 only affects the motion of an avalanche in
exceptional conditions (Hutter et al., 1994).
The normal stresses between adjacent elements described in this
paper reveal a positive increment when the mass is in the compression 2.4. Discussion
state and a negative increment in the decompression state. This extends
the scope of applications of the μ(I) rheology to the pure compression or The μ(I) rheology was already introduced by Forterre (2006), Gray
decompression states. The integration of the transverse normal stresses and Edwards (2014), Edwards and Gray (2015), Baker et al. (2016a,b),
(pxx and pyy) along the flow depth is obtained by neglecting the second- Viroulet et al. (2018), and Rocha et al. (2019) to capture some complex
order terms: granular phenomena such as roll waves, erosion–deposition waves,

3
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

segregation-induced fingering, and levee formation. Their models and 3.1. Experimental materials and mechanical properties
our proposed model differ regarding the expression of the μ(I) term in
the governing equations, since distinctive hypotheses of the flow ve- The sand is filtered and classified into three groups according to the
locity profile cause separate concerns: these references assume a Bag- particle sizes: coarse (grain size of approximately 10 mm), medium
nold velocity profile, while our proposed model adopts a plug flow (grain size of approximately 2 mm), and fine (grain size of approxi-
profile; their models contain both the velocity difference between ad- mately 0.25 mm). The particle grading curve and mechanical para-
jacent layers (which induces shear stresses) and thickness difference meters (particle size, unit weight, and internal friction angle) of the
between “adjacent columns” (which affects normal stresses), while our three groups of sand are shown in Fig. 3a and Table 1, respectively.
model neglects the velocity difference. In general, the μ(I) rheology in Photographs of the three groups of sand taken at the laboratory are
our proposed model plays a part in describing the normal stresses be- shown in Fig. 3b–d.
tween adjacent elements, while that in these references considers the
effects of normal stresses and shear stresses. 3.2. Experimental procedure
In most granular flow models, the fluidized basal layer is usually
treated as a Coulomb-type boundary condition for simplification, that The experiments held sand in a plate at the top of a chute, and the
is, the basal friction is the product of the basal friction coefficient and sand was released to simulate an avalanche. Overall, 12 independent
normal stress. Similar concepts are also adopted in debris flow mod- trials were conducted with different granular sizes, initial accumulation
eling, for example, in references (Iverson et al., 1997; Iverson and geometries, and inclination angles. After the sand came to rest, the
Denlinger, 2001; Iverson et al., 2004; Pitman et al., 2003; Pitman et al., runout distance, maximum deposit depth, and deposit configuration
2003). In addition, the initial intention of the model proposed in this were measured: the longitudinal dimension of the experimental flume is
paper captures the internal constitutive properties of the upper sliding calibrated by 5 cm and the flexible rule with an accuracy of 1 mm is
block (rather than the basal friction layer) by introducing the μ(I) used for more precise measurements. Vernier calipers with a mea-
rheology. We adopt the simplest Coulomb-type basal friction law with a surement precision of 0.2 mm are used to measure the deposit depth.
constant basal friction angle and neglect the effect of the velocity on The origin of the x coordinate along the longitudinal direction is fixed
basal friction in the simulation. Therefore, the focus of this manuscript at 200 mm downstream from the end of the inclined plane (see Point O
is more obvious, and the adoption of the μ(I) rheology is more readily in Fig. 2). We designed a series of experimental methods with fixed sand
revealed. But this also causes limitations that will be discussed in the masses weighing 7 kg, as shown in Table 2. One set of trials (No. 1) was
conclusion. conducted with various initial accumulation geometries (cuboid and
In the traditional hydrodynamic models, S-H theory and DAN prism, as shown in Fig. 4) but share the same granular property. For the
models are unable to directly depict the influence of the grain diameter other three sets of trials (Nos. 2–4), the initial sand accumulation shape
on granule movement because the parameters in these models do not is similar to a cuboid (Geometry A in Fig. 4), but the granular size and
include particle size. On the contrary, our proposed model directly channel inclination vary during each trial.
embodies the grain diameter parameter. Of note, considering the
u v
granular flow is steady and uniform, we obtain x = y = 0 and the last 4. Comparison between experimental and simulation results
two terms on the right side of Eq. (5b)–(5c) are set to zero in the si-
mulation. In this scenario, the model reduces to a traditional hydraulic 4.1. Numerical simulation
model in which the earth pressure coefficient equals the uniformity.
We use the total variation diminishing (TVD)-MacCormack method
(Liang et al., 2006; Liang et al., 2007) to solve the dynamic equations
3. Experimental design and procedure (Eq. (5a)–(5c)), and develop a fast and precise FORTRAN-based pro-
gram. In the program, we introduce the “wetting/drying algorithm” by
The reference surface of our experimental installation is composed Liang et al. (2006), Liang et al. (2007) that was initially developed to
of two parts (see Fig. 2). The first part is an inclined plane with an treat unreasonably large velocities on steep terrain when solving
inclination that is adjusted by a stand bar, and the second is a hor- shallow water equations. After numerical discretization, the grid points
izontal runout zone. The reference surface is composed of transparent where the flow depths are shallower than a prescribed value are treated
plexiglass, and the side walls are comprised of a transparent HDPE with as without grains with velocities that are set to zero and depths that are
a low friction coefficient (μ < 0.1) so the role of sidewalls in granular assigned unified insignificant values. We then use the developed pro-
surface flows is decreased. gram to compute the experimental trials and compare the simulation

Fig. 2. Schematic of the laboratory installation.

4
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

Fig. 3. Particle size distribution curve (a) and photographs (b–d) for the experimental sands.

Table 1 our experimental equipment, we use the typical values of frictional


Measured properties of the three kinds of experimental sands. coefficients from Pouliquen and Forterre (2002), Jop et al. (2006),
Particle size/(mm) Unit weight/(N·m−3) Angle of internal friction /(°)
Forterre and Pouliquen (2003), and Forterre and Pouliquen (2008);
I0 = 0.3, μs = 0.38, and μ2 = 0.65 in the calculation.
Coarse 10 12.61 32.62
Medium 2 13.54 32.35
Fine 0.25 12.85 30.25
4.2. Effects of particle size

To clarify the effects of the particle size on the configuration and


Table 2 runout of the final deposit, we simulate experiments Nos. 2–4. Overall,
Schemes of independent trials. 9 independent trials were conducted with the initial geometry and
volume of the experimental material fixed, but the grain diameter
NO. of the test set Particle size/(mm) Inclination angle/(°) Initial geometry
varies from 0.25 mm, 2 mm, to 10 mm with different angles of in-
1 2 30 A clination (30°, 40°, and 50°).
2 30 B To analyze the deposit’s dimensionless geometry, the experimental
2 30 C runout distance Rd ==0.25mm
50°
and corresponding height Hd==0.25mm
50°
of the
2 10 30 A
deposit with a grain size d = 0.25 mm and angle of inclination = 50°
2 30 A
0.25 30 A are set as the scales. Thus, the x axis denotes a dimensionless runout
3 10 40 A distance R/ Rd ==0.25mm
50°
and the y axis denotes a dimensionless height
2 40 A H / Hd==0.25mm
50°
, as demonstrated in Fig. 5a–i. Top view photographs of the
0.25 40 A deposition configuration taken during the experiment are shown in
4 10 50 A
2 50 A
Fig. 5j. We provide a non-dimensional deposit configuration in Fig. 5
0.25 50 A because by comparing the x and y axes values with uniformities, we can
elucidate whether the deposit’s runout and depth are greater or less
than those with grain size d = 0.25 mm and angle of inclination
results with the experimental data obtained from the laboratory. For the = 50° . Therefore, the effects of the grain size and angle of inclination
basal friction law, we use the classical Coulomb friction theory with a are more readily revealed.
constant frictional coefficient, in which the basal friction angles for As shown in Fig. 5, the spread of the avalanche with a large grain
sand with different grain sizes were measured and found to be ap- size is greater than that with a small grain size in both the experiment
proximately the same value = 27°. Because the practical parameter and simulation. For an avalanche sliding down the same angle of in-
calibration process requires considerable effort, and the accuracy of the clination, the configuration of its deposit turns plumper and runout
calibrated parameters is hard to control considering the limitations of becomes longer as the particle size increase. This result is consistent

5
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

Fig. 4. Different initial geometries of sand masses before release.

with our analysis: as the avalanche’s dispersion and configuration geometry’s effects is that the geometry affects the barycenter’s position,
transformation are influenced by the terrain gradient term and the term thus influencing the initial potential of the mass. An avalanche with
that contains the friction coefficient μ(I), by analyzing Eq. (1a)–(1b), we greater initial potential flows farther.
know that the inertial number I and friction coefficient μ(I) increase as
the grain diameter increases, so the dispersion effect is more obvious
when the grain diameter increases. In general, by combining the si- 5. Numerical simulation analysis
mulation and experimental results, we can conclude that the grain size
has obvious effect on the deposit’s configuration. 5.1. Tracked avalanche depth and flow velocity data

We tracked the simulated avalanche’s depth and flow velocity in the


4.3. Angle of inclination’s role in grain size effects
x direction at Point B (shown in Fig. 2) for avalanches flowing on planes
with different angles of inclination, as demonstrated in Fig. 8a. The
This section investigates the angle of inclination’s effect on the ab-
figure shows that the flow depths at Point B with different angles of
solute value of the runout distance and deposit depth (their corre-
inclination increase to maximum values within t ≈ 0.5 s, then gradually
sponding correlations are shown in Fig. 6) to further explore the angle
decrease to zero at t ≈ 1.7 s. Comparatively, the simulated avalanche at
of inclination’s role in the grain size effects. As expected, both the si-
Point B is short in height with a large angle of inclination throughout
mulation and experimental results indicate a positive correlation be-
the flow process. Fig. 8b indicates that the flow velocities at Point B
tween the runout distance and angle of inclination and a negative
with different angles of inclination reach high values within t ≈ 0.2 s,
correlation between the deposit depth and angle of inclination. Fig. 6
then remain relatively constant until most grains pass through. The
also demonstrates that, as the angle of inclination increases, the added
avalanche’s termination time is earlier and the magnitude of the
values of the runout distance decrease, a1 > b1 > c1; a2 > b2 > c2,
tracked velocity at Point B increases as the angle of inclination in-
as shown in Fig. 6a. This indicates that the angle of inclination plays a
creases.
significant role in the grain size effects on the runout distance: the ef-
fects decrease the angle of inclination. Of note, the angle of inclination’s
role in the grain size effect in terms of the deposit depth is not obvious,
5.2. Sensitivity analysis of the frictional parameters in the μ(I) rheology
as shown in Fig. 6b.
To clarify how the frictional parameters μs and μ2 in the proposed
4.4. Effects of initial geometry model affect the deposit’s configuration, numerical tests are conducted
that adapt different parameters. Considering the insufficient published
To clarify the effects of the sand pile’s initial geometry on the experimental data on the two frictional parameters, we use 3 sets of
configuration and runout of the final deposit, we use the first set of values in the numerical simulation: one set uses the empirical values in
experiments (No. 1 in Table 2) to simulate a deposit with different in- Refs. (Pouliquen and Forterre, 2002; Jop et al., 2006; Forterre and
itial geometries (geometries A, B, and C in Fig. 4) using our developed Pouliquen, 2008) (the same as in Section 4), µs = 0.38, µ 2 = 0.65; the
program. In our simulation, the particle size is fixed according to the other two sets are µs = 0.55, µ 2 = 0.80 and µs = 0.25, µs = 0.45, re-
test results (d = 0.25 mm). Fig. 7a-c show a comparison of the simu- spectively. Then we compare the computed side views of the ultimate
lated and experimental side views of the ultimate deposit in trials with deposit configuration using different frictional parameters (see Fig. 9a-
different initial geometries. The avalanche depth at Point A (shown in c) and track the evolution of the avalanche depth at Point B during the
Fig. 2) is tracked during the simulated flow process, as illustrated in flow process (see Fig. 9d). Fig. 9a-c shows that the frictional coefficients
Fig. 7d. have a significant impact on the deposit configuration and avalanche
The figure demonstrates that the pile’s initial geometry affects the depth. The configuration of the computational deposit becomes taller
deposit’s configuration: the simulated deposit with an initial geometry and thinner and the avalanche depth at Point B increases as the fric-
of A is the tallest and thinnest while that of C is the shortest and tional parameters decrease. Fig. 9d shows that at the same angle of
plumpest among the three simulation results. The experimental data is inclination, the avalanche at Point B is deeper with small frictional
consistent with the simulation results although the effect of the initial parameters throughout the flow process.
geometry is less pronounced. A possible explanation for the initial

6
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

Fig. 5. Side view comparison of dimensionless deposition configuration with different grain diameters slide from different angle of inclination [(a–i); the deposition
in laboratory experiments is represented by measured dots and the computed deposition is indicated in the contoured graphs], and top view of the deposition
configuration observed in the experiment (j).

7
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

Fig. 6. Experimental and simulation relations between angle of inclination and runout distance (a)/deposit depth (b) with different particle sizes (the simulation
results are represented by dotted dash line, the experimental results are represented by solid line).

5.3. Comparison with the traditional S-H theory the initial geometry of A and grain diameter d = 0.25 mm are fixed,
and the Coulomb friction law and basal frictional parameter are iden-
To elucidate the difference between our proposed model and the tical in the two models. In the S-H model, we calculate the earth
traditional S-H theory, we simulate an avalanche’s flow process using pressure coefficients by introducing the angle of internal friction
the two models and compare the simulation results. In the simulation, = 30.25° and the angle of basal friction δ = 27° into K xact/ pas =

Fig. 7. Side view comparison of ultimate deposition configuration with different initial accumulation geometries [(a–c); the deposit in laboratory experiments is
represented by measured dots and the computed deposition is indicated by the contoured graphs], and the tracked avalanche depth at Point A during the simulated
flow process (d).

8
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

Fig. 8. Tracked data of the simulated flow depth and flow velocity in x- direction at Point B with different inclination angle (bule:θ = 50°; red:θ = 40°;
green:θ = 30°).

Fig. 9. Side views of the calculated configuration of deposit with different frictional parameters (a–c), and tracked avalanche depth at Point B during the flow process
with different angle of inclination and frictional parameters (d).

9
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

Fig. 10. Side views of the numerical calculation results from the proposed model and the S–H model at different time points (t = 0.5 s, t = 1.5 s, t = 3.0 s); the
triangles represent the experimental data of the deposit configuration.

2sec2 (1 (1 cos2 /cos2 )1/2) 1 (x direction) and K yact / pas = particle size increased. The effect of the particle size on the runout
1
{Kx + 1 [(Kx 1) 2
+ 4tan2 (y direction). The internal angle of
]1/2 } decreased with the angle of inclination, while the angle of inclination’s
2
friction is identified as the angle of repose in the experiments, and 5 role in the grain size effect in terms of the deposit depth was not ob-
repeated trials of the same type of sand were conducted to obtain the vious. The initial geometries (cuboid and prisms) also influenced the
mean value of φ = 30.25°. Based on Jop et al. (2006), Pouliquen and deposit configuration: the cuboid pile had the shortest and plumpest
Forterre (2002), and Forterre and Pouliquen (2003, 2008), we learn configuration.
that for most natural granular materials, the frictional coefficient in the Data on the avalanche depth and flow velocity at the inclined plane
μ(I) rheology is in the range 0.3 < µ < 0.7 , and we also know that were tracked during the simulated flow process for analysis. Sensitivity
1.414
u
/| | 1.414 , so the interrelations between the earth analysis of the frictional coefficients of the μ(I) rheology demonstrated
pressure
x
coefficient in the two models are 1<1 that they had a significant impact on the deposit configuration and
u u
µ · x /| | < 1.99 < K xpas = 2.614 when x < 0 and 0 < 1 µ· ux /| | < K x act = runout distance. By comparing the simulation results using our pro-
posed model and the S-H theory, we found that the spread of the ava-
0.946 < 1 when x > 0 . This implies that the spread of the avalanche in
u
lanche in the S-H model was larger than that in the proposed model in a
the S-H model is larger than in the proposed model, which is consistent specific avalanche case. Although laboratory avalanches were ade-
with the results revealed in Fig. 10. By comparing the simulation results
quately simulated using the proposed model, the following limitations
from the proposed model and S-H model to the experimental data, we should not be overlooked:
find that the deposit’s front configuration is simulated better using the
(1) In contrast with rough sliding bed experiments by Baker et al.
S-H model, the height of the deposit is captured better using our pro- (2016a,b) and Rocha et al. (2019), the base of the experimental chute in
posed model, and the shape of the back of the deposited material is
this study was relatively smooth. Thus, we used simplified assumptions
modelled equally well by both models. (plug-like velocity and lithostatic pressure) during the model develop-
ment. These simplified treatments were effective for the reproduction of
6. Conclusion the laboratory results reported in this paper, but a more sophisticated
model that considers a more detailed velocity profile is essential.
To model an avalanche’s flow and deposition process, we in- (2) The inclusion of second-order terms (representing the shear ef-
troduced the μ(I) rheology into the depth-averaged continuum gov- fect in the upper sliding mass) in the momentum equation was proved
erning equations, simplified the model by assuming a plug flow velocity (Gray and Edwards, 2014; Edwards and Gray, 2015; Baker et al., 2016;
along the depth, and neglected the second-order (viscous) terms during Baker et al., 2016; Viroulet et al., 2018; Rocha et al., 2019) to be crucial
the derivation of the equations. A laboratory installation was also es- for regularizing models that can properly simulate the morphological
tablished that was composed of a confined horizontal runout zone and a properties of an avalanche and capture segregation-induced fingering
confined inclined plane with an adjustable angle of inclination. Piles of (Baker et al., 2016) and levee formation (Rocha et al., 2019). But this
accumulated sand with different particle sizes were released from the was not the main focus of the model proposed in this paper. The pro-
top of inclined planes at different angles of inclination to simulate posed model intended to describe the particle size effect on the spread
avalanches. Both the simulation and experimental results revealed that of the final deposition revealed in our laboratory experiments. There-
the grain diameter affected the ultimate deposit’s configuration: the fore, although the second-order viscous terms were important, they
configuration of deposit turned plumper and runout was longer as the

10
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

were neglected considering that the final runout and deposition con- Process Landf. 30, 461–477.
figuration were the main objectives of the present study. Cuomo, S., Pastor, M., Cascini, L., Castorino, G.C., 2014. Interplay of rheology and en-
trainment in debris avalanches: a numerical study. Can. Geotech. J. 51 (11),
(3) Baker et al. (2016) generalized the depth integration procedure 1318–1330.
to develop a two-dimensional frame invariant approach to retain in- Cuomo, S., Moretti, S., Aversa, S., 2019. Effects of artificial barriers on the propagation of
plane shear stresses. But the proposed model and S-H model are not debris avalanches. Landslides 16, 1077–1087. https://doi.org/10.1007/s10346-019-
01155-1.
frame invariant, which limits the modeling of granular flows over da Cruz, F., Emam, S., Prochnow, M., Roux, J.N., Chevoir, F., 2005. Rheophysics of dense
general terrain. granular materials: discrete simulation of plane shear flows. Phys. Rev. E 72 (2),
(4) The initial intention of our proposed model was to capture the 254–271.
Davies, T.R., Mcsaveney, M.J., 1999. Runout of dry granular avalanches. Can. Geotech. J.
constitutive properties of dense granular flow moving passively above a 36, 313–320.
thin fluidized basal layer by introducing the μ(I) rheology, so only a Dent, J.D., Burrell, K.J., Schmidt, D.S., Louge, M.Y., Adams, E.E., Jazbutis, T.G., 1998.
simple Coulomb-type basal friction law was employed in the model as a Density, velocity and friction measurements in a dry-snow avalanche. Ann. Glaciol.
26, 247–252.
boundary condition, in which the friction coefficient was assumed to be
Depken, M., Lechman, J.B., van Hecke, M., Saarloos, W.V., Grest, G.S., 2007. Stresses in
constant. Simulation of sophisticated cases requires a more precise smooth flows of dense granular media. Europhys. Lett. 78, 417–429.
basal friction law that can reveal its dependence on the flow velocity, Edwards, A.N., Gray, J.M.N.T., 2015. Erosion–deposition waves in shallow granular free-
avalanche depth, and material properties. surface flows. J. Fluid Mech. 762, 33.
Faug, T., Childs, P., Wyburn, E., Einav, I., 2015. Standing jumps in shallow granular flows
(5) It was assumed that the volume remained unchanged during the down smooth inclines. Phys. Fluids 27, 073304.
flow process in our model as we neglected bed erosion and other im- Forterre, Y.L., 2006. Kapiza waves as a test for three-dimensional granular flow rheology.
ported geomaterials along the flow path. In simulations of real landslide J. Fluid Mech. 563, 123–132.
Forterre, Y., Pouliquen, O., 2003. Long-surface-wave instability in dense granular flows.
cases, a more sophisticated model should be developed to include the J. Fluid Mech. 486, 21–50.
effects of bed erosion, such as the model developed by Braun et al. Forterre, Y., Pouliquen, O., 2008. Flows of dense granular media. Annu. Rev. Fluid Mech.
(2018). 40, 1–24.
Gray, J.M.N.T., Edwards, A.N., 2014. A depth-averaged μ(I)-rheology for shallow gran-
ular free-surface flows. J. Fluid Mech. 755, 503–534.
CRediT authorship contribution statement Gray, J.M.N.T., Hutter, K., 1998. Physik granularer. Lawinen Physikalische Blatter 54,
37–43.
Gray, J.M.N.T., Wieland, M., Hutter, K., 1999. Gravity-driven free surface flow of gran-
Jianbo Fei: Conceptualization, Software, Writing - original draft, ular avalanches over complex basal topography. Proc, R Soc, London, Ser, A 455,
Methodology. Yuxin Jie: Supervision, Funding acquisition. Xiaohui 1841–1874.
Sun: Writing - review & editing. Xi Chen: Supervision, Project ad- Greve, R., Koch, T., Hutter, K., 1994. Unconfined flow of granular avalanches along a
partly curved surface I: Theory. Proc. R. Soc. London, Ser. A 445, 399–413.
ministration.
Gubler, H., 1987. Measurements and modelling of snow avalanche speeds. Iahs Publ. 162,
405–420.
Declaration of Competing Interest Hanes, D.M., Inman, D.L., 1985. Observations of rapidly flowing granular-fluid materials.
J. Fluid Mech. 150, 357–380.
Hogg, A.J., 2007. Two-dimensional granular slumps down slopes. Phys. Fluids 19 (9),
The authors declare that they have no known competing financial 093301.
interests or personal relationships that could have appeared to influ- Huber, A., 1980. Schwallwellen in Seen aus Folge von Felsstürzen, PhD thesis,
Eidgenossische Technische Hochschule, ETH, Zürich, https://doi.org/10.3929/ethz-
ence the work reported in this paper. a-000207623.
Hungr, O., 1995. A model for the runout analysis of rapid flow slides, debris flows, and
Acknowledgments avalanches. Can. Geotech. J. 32 (4), 610–623.
Hutter, K., Koch, T., 1991. Motion of a granular avalanche in an exponentially curved
chute: experiments and theoretical predictions. Philos. T. R. Soc. A 334 (1633),
This study was funded by the National Natural Science Foundation 93–138.
of China (NSFC) (Grant No. 41790434, 51938008) and the Key Hutter, K., Wang, Y., Pudasaini, S.P., 1832. The Savage-Hutter avalanche model: how far
can it be pushed? Philos. T. R. Soc. A 2005 (363), 1507–1528.
Research and Development Program of China Railway (Grant No. Hutter, K., Plüss, C., Maeno, N., 1988. Some implications deduced from laboratory ex-
K2019G033). periments on granular avalanches. Hydrol. Glaziol. der ETH 323–344.
Hutter, K., Siegel, M., Savage, S.B., Nohguchi, Y., 1993. Two-dimensional spreading of a
granular avalanche down an inclined plane part I Theory. Acta Mech, 100 (1–2),
References
37–68.
Hutter, K., Svendsen, B., Rickenmann, D., 1994. Debris flow modeling: a review.
Ancey, C., Meunier, M., 2003. Estimating bulk rheological properties of flowing snow Continuum. Mech. Therm. 8, 1–35.
avalanches from field data. J. Geophys. Res. 109, F01004. https://doi.org/10.1029/ Issler, D., 2003. Experimental Information on the Dynamics of Dry-snow Avalanches,
2003JF000036. Dynamic Response of Granular and Porous Materials Under Large and Catastrophic
Bagnold, R.A., 1954. Experiments on gravity-free dispersion of large solid spheres in a Deformations. Springer, Berlin, Heidelberg.
Newtonian fluid under shear. Proc. R. Soc. A-Math Phys. 225, 49–63. Iverson, R.M., Denlinger, R.P., 2001. Flow of variably fluidized granular masses across
Bagnold, R.A., 1962. Auto-suspension of transported sediment; turbidity currents. Proc. three-dimensional terrain: 1 Coulomb mixture theory. J, Geophys, Res,-Sol, Ea 106
R. Soc. A-Math Phys. 265, 315–319. (B1), 537–552.
Baker, J., Barker, T., Gray, J.M.N.T., 2016a. A two-dimensional depth-averaged μ(I)- Iverson, R.M., Reid, M.E., LaHusen, R.G., 1997. Debris-flow mobilization from landslides.
rheology for dense granular avalanches. J. Fluid Mech. 787, 367–395. Annu, Rev, Earth Pl, Sc, 25 (1), 85–138.
Baker, J.L., Johnson, C.G., Gray, J.M.N.T., 2016b. Segregation-induced finger formation Iverson, R.M., Matthew, L., Denlinger, R.P., 2004. Granular avalanches across irregular
in granular free-surface flows. J. Fluid Mech. 809, 168–212. three-dimensional terrain: 2 Experimental tests. J. Geophys. Res.-Earth 109 (F1)
Balmforth, N.J., Kerswell, R.R., 2005. Granular collapse in two dimensions. J. Fluid Mech. (F01015), 1–16.
538, 399–428. Jop, P., 2015. Rheological properties of dense granular flows. CR Phys. 16 (1), 62–72.
Barker, T., Schaeffer, D.G., Shearer, M., Gray, J.M.N.T., 2017. Well-posed continuum Jop, P., Forterre, Y., Pouliquen, O., 2005. Crucial role of sidewalls in granular surface
equations for granular flow with compressibility and μ(I)-rheology. Proc. R. Soc. A flows: consequences for the rheology. J. Fluid Mech. 541, 167–192.
473 (2201), 20160846. https://doi.org/10.1098/rspa.2016.0846. Jop, P., Forterre, Y., Pouliquen, O., 2006. A constitutive law for dense granular flows.
Braun, A., Cuomo, S., Petrosino, S., Wang, X., Zhang, L., 2018. Numerical SPH analysis of Nature 441 (7094), 727–730.
debris flow run-out and related river damming scenarios for a local case study in SW Keller, S., Ito, Y., Nishimura, K., 1998. Measurements of the velocity distribution in ping-
China. Landslides 15, 535–550. https://doi.org/10.1007/s10346-017-0885-9. pong-ball avalanches. Ann. Glaciol. 26, 259–264.
Campbell, C.S., 1990. Rapid granular flows. Ann. Rev. Fluid Mech. 22, 57–92. Koch, T., Greve, R., Hutter, K., 1994. Unconfined flow of granular avalanches along a
Cascini, L., Sorbino, G., Cuomo, S., Ferlisi, S., 2014. Seasonal effects of rainfall on the partly curved surface Part II: Experiments and numerical computations. Proc. R Soc.
shallow pyroclastic deposits of the Campania region (southern Italy). Landslides 11, London, A 445, 415–435.
779–792. https://doi.org/10.1007/s10346-013-0395-3. Lang, T.E., Brown, R.L., 1980. Snow avalanche impact on structures. J. Glaciol. 25 (93),
Cascini, L., Cuomo, S., Pastor, M., Rendina, I., 2016. SPH-FDM propagation and pore 445–455.
water pressure modelling for debris flows in flume tests. Eng. Geol Liang, D., Falconer, R.A., Lin, B., 2006. Comparison between TVD-MacCormack and ADI-
S0013795216302472. type solvers of the shallow water equations. Adv. Water Resour. 29 (12), 1833–1845.
Claessens, L., Heuvelink, G.B.M., Schoorl, J.M., Veldkamp, A., 2005. DEM resolution ef- Liang, D., Lin, B., Falconer, R.A., 2007. Simulation of rapidly varying flow using an ef-
fects on shallow landslide hazard and soil redistribution modelling. Earth Surf. ficient TVD–MacCormack scheme. Int. J. Numer. Meth. Fl. 53 (5), 811–826.

11
J. Fei, et al. Computers and Geotechnics 126 (2020) 103737

Manzella, I., Labiouse, V., 2008. Qualitative analysis of rock avalanches propagation by 2019–2041.
means of physical modelling of non-constrained gravel flows. Rock Mech. Rock Eng. Rocha, F., Johnson, C.G., Gray, J.M.N.T., 2019. Self-channelisation and levee formation
41 (1), 133–151. in monodisperse granular flows. J. Fluid Mech. 876, 591–641.
Mcdougall, S., Hungr, O., 2005. Dynamic modelling of entrainment in rapid landslides. Savage, S.B., 1979. Gravity flow of cohesionless granular materials in chutes and chan-
Can. Geotech. J. 42 (5), 1437–1448. nels. J. Fluid Mech. 92, 53–96.
Norem, H., Irgens, F., Schieldrop, B., 1989. Simulation of snow-avalanche flow in run-out Savage, S.B., Hutter, K., 1989. The motion of a finite mass of granular material down a
zones. Ann. Glaciol. 13, 218–225. rough incline. J, Fluid Mech, 199, 177–215.
Okura, Y., Kitahara, H., Sammori, T., Kawanami, A., 2000. The effects of rockfall volume Savage, S.B., Hutter, K., 1991. The dynamics of avalanches of granular materials from
on runout distance. Eng. Geol. 58 (2), 109–124. initiation to run out. Acta Mech, 86 (1–4), 201–223.
Pastor, M., Haddad, B., Sorbino, G., Cuomo, S., Drempetic, V., 2009. A depth-integrated, Savage, S.B., Mckeown, S., 1983. Shear stresses developed during rapid shear of con-
coupled SPH model for flow-like landslides and related phenomena. Int. J. Numer. centrated suspensions of large spherical particles between concentric cylinders. J.
Anal. Meth. Geomech. 33, 143–172. https://doi.org/10.1002/nag.705. Fluid Mech. 127, 453–472.
Pitman, E.B., Nichita, C.C., Patra, A.K., Bauer, A.C., Bursik, M., Weber, A., 2003. A nu- Savage, S.B., Sayed, M., 1984. Stresses developed by dry cohesionless granular materials
merical study of granular flows on erodible surface. Discrete Contin. Dyn. Syst. Ser. B. sheared in an annular shear cell. J. Fluid Mech. 142, 391–430.
3, 589–599. Shieh, C.L., Jan, C.D., Tsai, Y.F., 1996. A numerical simulation of debris flow and its
Pitman, E.B., Nichita, C.C., Patra, A., Bauer, A., Sheridan, M., Bursik, M., 2003. application. Nat, Hazards 13 (1), 39–54.
Computing granular avalanches and landslides. Phys. Fluids 15, 3638. Silbert, L.E., Ertas, D., Grest, G.S., Halsey, D., Levine, S.J., 2001. Plimpton granular flow
Poisel, R., Preh, A., Hungr, O., 2008. Run out of landslides–continuum mechanics versus down an inclined plane: Bagnold scaling and rheology. Phys. Rev. E Stat. Nonlin. Soft
discontinuum mechanics models. Geomech. Tunn. 1 (5), 358–366. Matter Phys. 64, 051302.
Pouliquen, O., 1999. Scaling laws in granular flows down rough inclined planes. Phys. Takahashi, T., Debris flow, IAHR Monograph Series A.A. Balkema, Rotterdam, the
Fluids 11 (3), 542. Netherlands, 1991.
Pouliquen, O., Forterre, Y., 2002. Friction law for dense granular flows: application to the Viroulet, S., Baker, J.L., Rocha, F.M., Johnson, C.G., Kokelaar, B.P., Gay, J.M.N.T., 2018.
motion of a mass down a rough inclined plane. J. Fluid Mech. 453, 133–151. The kinematics of bidisperse granular roll waves. J. Fluid Mech. 848, 836–875.
Pudasaini, S.P., Hutter, K., 2007. Avalanche Dynamics. Springer, Berlin, Heidelberg. Wieland, M., Gray, J.M.N.T., Hutter, K., 1999. Channelized free-surface flow of cohe-
Pudasaini, S.P., Hsiau, S.S., Wang, Y., Hutter, K., 2005. Velocity measurements in dry sionless granular avalanches in a chute with shallow lateral curvature. J. Fluid Mech.
granular avalanches using particle image velocimetry technique and comparison with 392, 73–100.
theoretical predictions. Phys. Fluids 17, 093301. Wu, J.H., Chen, J.H., Lu, C.W., 2013. Investigation of the Hsien-du-Shan rock avalanche
Pudasaini, S.P., Eckart, W., Hutter, K., 2008. Gravity-driven rapid shear flows of dry caused by typhoon Morakot in 2009 at Kaohsiung county. Taiwan Int. J. Rock Mech.
granular masses in helically curved and twisted channels. Math. Models Methods Min. 60, 148–159.
Appl. Sci. 13 (07), 1019–1052. Zhan, L.T., Zhang, Z., Chen, Y.M., Chen, R., Zhang, S., Liu, J., Li, A.G., 2018. The 2015
Rauter, M., Barker, T., Fellin, W., 2020. Granular viscosity from plastic yield surfaces: The Shenzhen catastrophic landslide in a construction waste dump: reconstitution of
role of the deformation type in granular flows. Comput. Geotech. 122, 103492. dump structure and failure mechanisms via geotechnical investigations. Eng. Geol.
https://doi.org/10.1016/j.compgeo.2020.103492. 238, 15–26.
Renouf, M., Alart, P., 2005. Conjugate gradient type algorithms for frictional multi-con- Zhu, C., Huang, Y., Zhan, L.T., 2018. SPH-based simulation of flow process of a landslide
tact problems: applications to granular materials. Comput. Method Appl. M 194 (18), at Hongao landfill in China. Nat. Hazards 93, 1113–1126.

12

You might also like