Professional Documents
Culture Documents
Experimental Study On The Effects of Physical Conditions On The Interaction Between Debris Flow and Baffles
Experimental Study On The Effects of Physical Conditions On The Interaction Between Debris Flow and Baffles
Haiqing Yang (杨海清 ), Md. Emdadul Haque, and Kanglei Song (宋康磊 )
Robust and unstable axisymmetric vortices, including neutral vortices, of a new two-
dimensional vortex family
Physics of Fluids 33, 054103 (2021); https://doi.org/10.1063/5.0048128
Airborne transmission of COVID-19 and mitigation using box fan air cleaners in a poorly
ventilated classroom
Physics of Fluids 33, 057107 (2021); https://doi.org/10.1063/5.0050058
© 2021 Author(s).
Physics of Fluids ARTICLE scitation.org/journal/phf
Haiqing Yang (杨海清),1,2,a) Md. Emdadul Haque,1,2,b) and Kanglei Song (宋康磊)1,2,c)
AFFILIATIONS
1
State Key Laboratory of Coal Mine Disaster Dynamics and Control, School of Civil Engineering, Chongqing University,
Chongqing 400045, China
2
National Breeding Base of Technology and Innovation Platform for Automatic-Monitoring of Geologic Hazards, Chongqing
Engineering Research Center of Automatic Monitoring for Geological Hazards, Chongqing 400042, China
a)
E-mail: yanghaiqing06@163.com and yanghaiqing@cqu.edu.cn
b)
Electronic mail: emdadulhaque@cqu.edu.cn
c)
Author to whom correspondence should be addressed: sklei@cqu.edu.cn
ABSTRACT
The gravitational debris flow, such as the agent forming alluvial cones in the mouths of mountain canyons, could bring about devastating disaster
to downstream structures in mountainous areas. In the present study, a series of model tests were conducted on the sand and the ceramsite to
systematically explore the interaction between debris flow and baffles. During the runout process, the impact force exerted by debris flow was
measured by dynamometers. The runout distance, velocity of the flow, and flow depth were monitored by a video camera and a high-speed cam-
era in a real time. The dynamic interaction under different particle sizes of dry granular materials, slop angles, and baffle configurations was simu-
lated. Experimental results show that the smaller size material is favorable for the frictional energy dissipation during the sliding process, giving
rise to the smaller runout distance. The present findings provide important references for the debris flow control engineering.
Published under license by AIP Publishing. https://doi.org/10.1063/5.0046670
I. INTRODUCTION suppress overflow, while the second staggered row of short, staggered
Large volume landslides frequently carry disastrous high-speed baffles could not affect in reducing the kinetic energy of debris. On the
debris flows, which causes loss of much property and fatality.1–5 basis of experiments and numerical simulations, Chen et al.25,26 con-
Available data have indicated that annual global landslide disaster cluded that the increases in length and width of the energy dissipation
would cause 1000 deaths and $4 million property loss,6,7 almost 70% structure improve the energy dissipation ratio. Furthermore, Wang
of which are attributed to the high-speed and long-distance rock ava- et al.27 studied the velocity reduction effects and energy dissipation
lanches.8 To avoid such hazardous phenomenon, rigid or flexible bar- capacities of deceleration baffles through a series of indoor experi-
riers9 combined with isolated baffles are commonly employed.10 Such ments. In the experiments, the influences of debris-flow density, baffle
debris-resisting structure can promote the energy loss of debris flow shape, and row spacing were systemically investigated. For the com-
and suppress further flow of debris.11–17 bined baffle-avalanche wall system, Bi et al.28 numerically investigated
Baffle configurations have significant effects on the mobility of the effects of the size of the retaining wall and layout of the baffle-
debris flow.18,19 A simple baffle may alter the original flow direction avalanche wall system on energy dissipation capacity. However, the
only in the vicinity of the contact surface, while an array of baffles can above-mentioned investigations mainly focus on the deflecting dams
change the flow pattern such that the gravity-driven flow slows down in the runout zone. Actually, retarding structures in the deposition
as it approaches each baffle and then accelerates toward the next row zone are also widely been used. By comparison, retarding structures
to facilitate energy dissipation.20–24 Therefore, the energy dissipation also have a high capacity in dissipating the kinetic energy of debris
characteristics of baffle with different layouts and shapes have been flows. Therefore, Fei et al.29 explored the effects of baffle pile layouts
widely investigated. Choi et al.20 found that tall baffles are beneficial to on their energy dissipation capacity by experiment and numerical
simulation. Baffle piles were installed on the horizontal runout zone. A. Experimental materials and devices
But the impact force acting on piles and velocity of debris flow were The sand and ceramsite were selected to simulate the debris flow
still unclear. in the experiment. The sand has a bulk unit weight of 16.3 kN=m3
It is widely accepted that debris flows are multi-phase medium with the average particle size of 0.25 mm. The ceramsite has a bulk
with particle–fluid,30–35 which means that the flow mechanisms of the unit weight of 9.28 kN=m3 with the average particle size of approxi-
debris are difficult to predict. Previous research has extensively aimed mate 10 mm. The volume of the material used in each test is 0.002 m3 .
at the pore pressure characteristic36–38 and the frictional character- During the landslide, the bulk density of the sand flow ranges from
istic39–41 of the debris during the sliding process, while the debris–baf- 1300 to 1700 kg=m3 , while the bulk density of the ceramsite flow
fle interaction mechanisms, which are important to evaluate debris varies from 300 to 1000 kg=m3 . According to Iverson,39 the Savage
mobility, have not yet received enough attention. Although the move- number and the Bagnold number of the sand are 0.2 and 400, respec-
ment mechanism of debris is difficult to predict, laboratory experi- tively. The values of the ceramsite are estimated to be 0.03 and 120,
ments can help us to better understand the physical processes. respectively. That is to say, the sand represents the collisional debris
Combined with model tests, researchers have also begun to flow (Sav > 0.1), and frictional stresses from grains are more impor-
explore the effects of a series of factors such as landslide material, the tant (Bag > 200). The ceramsite represents the frictional debris flow
volume of landslide on debris flow. Different dry granular materials (Sav < 0.1), and viscous stresses from grains are more important
releasing to inclined chute were usually applied to simulate the debris (Bag > 200). As shown in Fig. 1(a), the experimental devices mainly
flow.42 For example, Okura et al.43 developed an almost real-size slope consist of an inclined chute, a storage container, a horizontal chute,
model to study the process of landslide fluidization under torrential dynamometers, a video camera, and a high-speed camera. The sizes of
rain conditions; it was found that the three-stage characteristic of land- inclined chute and horizontal runout plate are set as 0.26 0:5 m2 and
slide fluidization occurs simultaneously. Lu et al.44 employed four 0.7 mm 0:9 m, respectively. Dynamometers were employed to mea-
types of dry granular materials with particle size ranging from 2 to sure the impact force of the debris flow acting on the baffles. While the
80 mm and studied the effect of particle size and volume of the slide velocity of the flow and flow depth can be monitored by a video cam-
material on the entrainment process. The experimental results indi- era and a high-speed camera.47 According to Ng et al.,19 dynamic sim-
cated that the energy of debris flow increases with the growth of parti- ilarity is obtained by the Froude number, Fr , which is given by
cle size and volume. Thus, the particle size of the sliding material
should be one of the crucial parameters when analyzing the energy dis- Fr ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; (1)
sipation during sliding process, especially when evaluating the gh cos ðhÞ
debris–baffle impact mechanism. Furthermore, a new model was pro-
where is the frontal velocity in m=s; g is the gravitational accelera-
posed to estimate the overrun area and volume in terms of known
tion in m=s2 ; h represents the flow depth in m, and h stands for the
mobility data by Pudasaini and Miller.8 However, due to the lack of channel inclination. In the experiment, the Froude number is set as 3,
energy dissipation mechanism during the sliding process, the impact which stands for a frontal velocity of 10 m=s and a flow depth of 1 m.
momentum of the landslide could not be accurately calculated.45 In addition, a black line is marked inside the storage container, so
Considering this limitation, a landslide model experiment involving that the same volume of the sliding body was determined in each test.
the impact energy dissipation was constructed by Wu et al.46 The A movable plate was designed at the bottom of the storage container
impact momentum of a landslide was calculated. The results indicate to release the granular material stored at the top of the inclined chute.
that the impact energy increases remarkably with the increase in the Once the movable plate is removed in short time (1–2 s) by hand,
debris particle size. which, to a great extent, avoids the potential influence of the removing
In this study, a series of model tests were conducted on the sand process on the experimental results, then the sliding body runs out of
and the ceramsite to study the interaction between debris flow and baf- the storage container and starts to slide along the inclined chute. Both
fles. The influences of dry granular materials, slop angles and baffle of the inclined chute and the horizontal chute are marked with 0.02
configurations on runout distance, impact force, and debris frontal 0:02 m2 grid cells for the sake of determining the location of the
velocity were investigated. sliding body in real-time during each experiment. We chose cube glass
blocks with the edge length (w) of 0.05 m to simulate the baffles and
II. MODEL TEST SETUP OF GRANULAR–STRUCTURE- used dry sand and ceramsite as an analogue to the sliding material.
INTERACTION A video camera and a high-speed camera were used to capture
Baffles are commonly installed in front of barriers (e.g., Lantau the dynamic process of debris flow. The high-speed camera used in
Island, Hong Kong) or in open stream courses (e.g., Kennedy Town, the experiment is Basler daa2500 14 um made in Germany, which
Hong Kong) according to Choi et al.12 However, the design of baffles can capture 5 106 pixels of images and 14 frames of images per sec-
is typically designed using empirical methods, so that the debris–baffle ond. The two cameras were, respectively, installed above and aside the
interaction mechanism is not well understood. So, investigating on the model, which work synchronously with a timer placed on a flat plate
effect of baffle configuration in terms of different particle sizes, slope to record the sliding process on the inclined chute and the overrun
angles, and installation distances plays a vital role in the field baffle phenomenon on the horizontal chute. According to dynamic photo-
design and arrangement. Therefore, a series of laboratory model tests graphs of the landslide, the velocity of the sliding body can be calcu-
were carried out to study the baffle–debris interaction mechanism. lated by comparing the same point of the sliding body at different
Details on experimental devices and test program are illustrated in times. To measure the impact force, each glass block is connected with
Secs. II A and II B. a dynamometer. The dynamometer used in the experiment is
FIG. 2. Dry granular materials employed in the experiment: (a) sand; (b)
ceramsite.
inclined chute and the front of baffles are set: 0.05, 0.1, and 0.15 m.
Each case of the baffle configuration is schematically delineated in Fig.
1(b). It should be noted that the arrangement of one single baffle can
be used as a control group when comparing the influence of the baffles
beside the middle baffle on the debris flow, although the single baffle
may be rare in the field. Besides, to analyze the effect of slope angle on
the sliding process of debris flow, the inclined angles of slope were set
as 30 ; 45 ; and 45 , respectively.
The experimental results are summarized accordingly in Table I.
We used the “Sand_40 _5 cm” symbols as an example to explain
FIG. 1. Schematic diagram of experimental setup: (a) 3D view with scaled ratio the meaning of these test IDs. “Sand” represents the sand group with
1:100; (b) top view of the array of baffles. only one baffle, “40 ” represents the slope angle is 40 , and “5 cm”
represents the installation distance is 0.05 m (5 cm) from the inclined
chute.
MARK-10-M7, with a range of 0–400 N and a minimum scale of
0.05 N. The accuracy can meet the experimental requirements. III. RESULTS AND DISCUSSIONS
Due to the existence of prepositioned baffles, the mobility of
B. Testing program debris flow will be slowed after contacting the baffles, which simulta-
23 groups of model tests were carried out by releasing dry granu- neously results in different runout distance and velocity changes at dif-
lar materials (dry sand and ceramsite, as shown in Fig. 2) onto the ferent points in the sliding body. For the convenience of presentation
inclined chute. To analyze the influence of baffles on debris mobility, and analysis, we defined front, middle, and tail of the sliding body
three kinds of the baffle configurations were presented: a single baffle with the following specific meaning:48 front refers to the leading edge
(one glass block), one row of baffles (three blocks in one line), and two of the landslide body; middle represents the middle part of landslide
rows of baffles (two glass blocks in front and three glass blocks behind, body that slides out of the storage container at half of the entire sliding
referring to Ng et al.19); also, three initial distances (D) between the distance; tail is the trailing edge of the landslide body.
Test ID Baffle Slope angle (deg) Baffle distance (cm) Material type
A. Runout distance Especially, the runout distance of sand is up to 23 % less than that of
Runout distance is an essential physical parameter for evaluating ceramsite in the front part of the sliding body. Nevertheless, in addi-
the hazardous consequences and risks from landslides and debris tion to the difference in particle size, the unit weight of the ceramsite
avalanches.6–8 This parameter substantially reflects the final form of and the sand differs greatly. To highlight the influence of particle size
energy dissipation of the debris flow after it experiences gravity-driven on runout distance, the ratio of runout distance to the unit weight was
falls, frictional slides on slope surface, interactions with obstacles in calculated, as shown in Fig. 3(b). Generally, the trends of ratios in this
flow path, and dispersive flows in deposit zone.48 During this dynamic figure are similar with that in Fig. 3(a). Besides, it is also observed that
process, runout is also affected by the coefficient of restitution, which the maximum ratio for the ceramsite at middle part is more than that
governs energy dissipated in the flow.22 Since the inclined and hori- for the sand at front part. The maximum ratio for the ceramsite at tail
zontal chutes are both installed with fixed surface roughness, we can part is more than that for the sand at middle part. That is to say, the
mainly focus on other dominant parameters such as particle size of effect of particle size is more influential. This phenomenon is due to
granular material, slope angle, baffle location, and configuration to the fact that the average diameter of the ceramsite (10 mm) is much
explore the mobility of debris flow. In order to study the problem of greater than that of the sand (0.25 mm). Under the same volume, the
the sliding distance after the landslide body interacts with the baffle, specific surface area of the ceramsite is smaller than that of the sand.
the following discussion is presented. In the case of identical slope angles, the kinetic energy loss of the
ceramsite caused by friction is smaller than that of the sand during the
sliding process. In addition, as the particle size of the sand is smaller
1. Size of granular materials
than that of ceramsite, after the sand hits the baffle, the collision fre-
To explore the influence of particle size on runout distance of quency of the sand must be higher than that of the ceramsite, which
sliding body, the ceramsite and the sand were employed in the experi- consumes more kinetic energy. Therefore, the runout distance of
ment. The slope angle is set as 45 . Experimental results were depicted ceramsite is longer. Similar findings are provided by Li et al.49 who
in Fig. 3. As illustrated in Fig. 3(a), the runout distance increases with reported that the fine-grained granular material has a large specific
the increase in sliding time despite different particle sizes (ceramsite surface area and a significant larger frictional energy consumption
for Ceramsite_45 and sand for Sand_45 ), then the distance remains than the coarse-grained granular material, which can result in serious
constant. The maximum runout distance of the ceramsite at different energy dissipation inside the sliding body, inhibit particle collision and
parts is bigger than those of the sand at corresponding parts. leap, and therefore limit the impact of the sliding body on baffles.
FIG. 4. (a) The contour line of the deposit area; (b) actual distribution of the ceramsite’s deposit area; (c) actual distribution of the sand’s deposit area.
3. Distribution of baffles results in earlier energy dissipation, so the sliding distance becomes
decreased. Additionally, the rank order of sliding distance for different
Specific locations and configurations of baffles have pivotal role part of the landslide body is as follows: the front part > the middle
in attenuating the mobility and flow energy of debris avalanches.20 In part > the tail part. The main reason for this phenomenon is that dif-
the experiment, the distribution of baffle involves the installation dis- ferent positions of the baffles have different effects on the frontal speed
tance from the inclined chute (D) and the rows of baffle. In detail, the (Figs. 13 and 14), which results in different sliding distances. When
length of the cubic glass block is w ¼ 0.05 m, and the slope angle is the slope angle is 40 , the baffle is increased from one row [Fig. 13(a):
40 . The installation distances were set as w, 2, and 3w (corresponding Sand_40 _5 cm] to two rows [Fig. 14(a): Sand_40 _2 or 5 cm], the
to Sand_40 _5 cm, Sand_40 _10 cm, and Sand_40 _15 cm in Table I, speed of the frontal part is reduced by 14 %, which implies the signifi-
respectively). Additionally, baffle rows were adopted as no baffle, sin- cant role of baffle arrangement in controlling debris flow. Of course,
gle baffle, one row of baffles (three baffles), and two rows of baffles Fig. 13 also contains very rich information, which will be analyzed in
(five baffles), as shown in Fig. 5. detail in later chapters.
Figure 6 shows the variations of runout distance (L) for different Figure 7 shows the variations of runout distance of the front of
parts of the sand sliding body and installation distance (D). It is clear the flow for different baffle configurations. Obviously, in the case of
that during the sliding process, the similar movement trends of the sand and ceramsite, the variations of sliding distance show the similar
sliding front, middle, and tail parts are found for different installation trend. Sliding distance increases gradually with time at first. After
distances, in general: the runout distance increases with the increase in about 0.75 s, the value reaches a stable state. In addition, except for
the installation distance but not systematically. The closer the installa- some individual cases, the fewer the baffles, the longer the runout dis-
tion distance, the smaller the runout distance along the horizontal tance. Specifically, in the case of the sand, runout distances reduce by
unit. Moreover, the closer the installation distance is, the earlier the 15 %, 8%, and 19% for single, one row, and two rows of baffles, respec-
sliding body collides with the baffle model, which simultaneously tively [Fig. 7(a)]. As for the ceramsite particles [Fig. 7(b)], the runout
distances under single, one row, and two rows of baffles reduce by
10%, 14%, and 27%, respectively.
It is worth noting that lines in Figs. 7(a) and 7(b) are not simply
parallel, but there are crossovers. The reason for this is that the land-
slide body speed varies as time goes on. According to the Choi et al.,12
debris flow goes through the transient state and the quasi-steady state
sequentially in the impact process. Specifically, after the collision, the
flow is highly incoherent, causing the overflow and backwater effects.
As a result, more debris is stagnant upstream of the baffles, and the
kinetic energy decreases rapidly. The process is called as the transient
state. After that, the mobility of flows keeps at the quasi-steady state.
Compared with a baffle plate, when adopting a row of baffles, the sand
and the baffle strike more violently, resulting in a temporary increase
in the speed of the sand and a crossover phenomenon. For the test
conducted on ceramsite, the crossover phenomenon is not obvious,
indicating that the velocity of materials with larger particle sizes will be
less affected after collision with the baffle. Therefore, the runout dis-
FIG. 6. The variations of final runout distance with time for different parts of the tance has a positive correlation with the slope angle and the particle
sand sliding body and installation distances. size of the sliding body.
collisions between the inner parts cause the speed of the sand to caused by landslides depend largely on the velocity of landslides,6,7 but
decrease rapidly. This also leads to the conclusion that if the quality of the mechanism of controlling the velocity is not fully understood.
the baffle is good enough, the baffle will block the faster debris flow Johannesson et al.54 built an empirical relationship based on snow ava-
more efficiently. The same rule can be found with two rows of baffles lanche mitigation, which illustrates that the velocity reduction arising
(Sand_45 _2 or 5 cm). from obstacles is proportional to the degree of blockage. Furthermore,
By comparing the two figures, a conclusion can be drawn that the Johannesson’s results reveal that the frontal velocity reduction is
with a smaller slope angle, the time required for the landslide body to influenced by the arrangement of baffles. Taking these findings into
decrease to zero velocity is longer. This is because when the slope angle consideration, the influence of different rows of baffles is explained in
is small, the speed of the landslide body hitting the baffle will not be the following part.
very large, the energy dissipation is not large, resulting in that there is Figure 15 shows the relationship between the time and the
still some remaining energy for taxiing. Therefore, it is also necessary frontal velocity for varying rows of baffles. In the absence of baffles,
to consider the slope angle in the prevention project. If the angle is the maximum frontal velocity of sand can reach up to 1.32 m=s. In
small, more baffles are required to be set; of course, the relationship the case of one row baffles, the maximum frontal velocity of the
between the baffle’s configuration and the slope angle still needs to be landslide can reach up to 1.06 m=s. In the case of two rows baffles,
further explored. the frontal velocity of the sliding body can decrease to 0.91 m=s.
By comparison, the maximum speed of ceramsite is higher than Compared to velocity in the absence of baffles, adopting up to a
sand at the same installation distance and angle. As the number of baf- row and two rows baffles leads to 20 % and 31% reduction in frontal
fles increased from one to two, the maximum speed of ceramsite velocity, respectively.
decreased by about 14.17% at a landslide angle of 45 , while the maxi- A series of laboratory model experiments were conducted to
mum speed of sand decreased by about 17.86%. The types of hazards explore the debris–baffle interaction mechanisms and energy
29
J. Fei, Y. Jie, X. Sun, and X. Chen, “Experimental investigation on granular 43
Y. Okura, H. Kitahara, H. Ochiai, T. Sammori, and A. Kawanami, “Landslide
flow past baffle piles and numerical simulation using a mu(i)-rheology-based fluidization process by flume experiments,” Eng. Geol. 66, 65–78 (2002).
approach,” Powder Technol. 359, 36–46 (2020). 44
P. Lu, T. Hou, X. Yang, M. Hao, and J. Zhou, “Physical modeling test for
30
S. P. Pudasaini, “A general two-phase debris flow model,” J. Geophys. Res.- entrainment effect of landslides and the related mechanism discussion,” Chin.
Earth Surf. 117, F03010 (2012). J. Rock Mech. Eng. 35, 1225–1232 (2016).
31
S. P. Pudasaini and M. Mergili, “A multi-phase mass flow model,” J. Geophys. 45
I. L. Singer, “Friction and energy-dissipation at the atomic-scale—A review,”
Res.-Earth Surf. 124, 2920–2942, https://doi.org/10.1029/2019JF005204 (2019). J. Vac. Sci. Technol. A 12, 2605–2616 (1994).
32
S. P. Pudasaini, S.-S. Hsiau, Y. Wang, and K. Hutter, “Velocity measurements 46
Y. Wu, D. Liu, and M. Li, “Landslide model experiment for energy dissipation
in dry granular avalanches using particle image velocimetry technique and law in sliding and impact processes,” Yanshilixue Yu Gongcheng Xuebao/
comparison with theoretical predictions,” Phys. Fluids 17, 093301 (2005). Chin. J. Rock Mech. Eng. 30, 693–701 (2011).
33
S. P. Pudasaini, “Some exact solutions for debris and avalanche flows,” Phys. 47
H. Q. Yang, S. G. Xing, Q. Wang, and Z. Li, “Model test on the entrainment
Fluids 23, 043301 (2011). phenomenon and energy conversion mechanism of flow-like landslides,” Eng.
34
N. Martin, I. R. Ionescu, A. Mangeney, F. Bouchut, and M. Farin, “Continuum Geol. 239, 119–125 (2018).
viscoplastic simulation of a granular column collapse on large slopes: Mu(i) 48
S. P. Pudasaini, B. Domnik, and S. A. J. E. Miller, “Multi-scale coupling strategy
rheology and lateral wall effects,” Phys. Fluids 29, 013301 (2017). for fully two-dimensional and depth-averaged models for granular flows,”
35
C.-H. Lee, Y. M. Low, and Y.-M. Chiew, “Multi-dimensional rheology-based Geophys. Res. Abstr. 15, EGU2013–7283 (2013).
two-phase model for sediment transport and applications to sheet flow and 49
C. Li, M. Wang, and K. Liu, “A decadal evolution of landslides and debris flows
pipeline scour,” Phys. Fluids 28, 053305 (2016). after the Wenchuan earthquake,” Geomorphology 323, 1–12 (2018).
36
O. Hungr, S. Leroueil, and L. Picarelli, “The Varnes classification of landslide 50
T. de Haas, L. Braat, J. R. F. W. Leuven, I. R. Lokhorst, and M. G. Kleinhans,
types, an update,” Landslides 11, 167–194 (2014). “Effects of debris flow composition on runout, depositional mechanisms, and
37
L. Olivares and L. Picarelli, “Modelling of flowslides behaviour for risk mitiga- deposit morphology in laboratory experiments,” J. Geophys. Res.-Earth Surf.
tion,” Physical Modelling in Geotechnics (General Report, Hong Kong, 2006), 120, 1949–1972, https://doi.org/10.1002/2015JF003525 (2015).
Vol. 1, pp. 99–113. 51
G. Devoli, F. V. De Blasio, A. Elverhi, and K. Heg, “Statistical analysis of
38
G. Wang and K. Sassa, “Factors affecting rainfall-induced flowslides in labora- landslide events in Central America and their run-out distance,” Geotech. Geol.
tory flume tests,” Geotechnique 51, 587–599 (2001). Eng. 27, 23–42 (2008).
39
R. M. Iverson, “The physics of debris flows,” Rev. Geophys. 35, 245–296, 52
Y. Okura, H. Kitahara, A. Kawanami, and U. Kurokawa, “Topography and vol-
https://doi.org/10.1029/97RG00426 (1997). ume effects on travel distance of surface failure,” Eng. Geol. 67, 243–254
40
R. M. Iverson and J. W. Vallance, “New views of granular mass flows,” (2003).
Geology 29, 115–118 (2001). 53
K. L. Johnson, Contact Mechanics (Cambridge University Press, Cambridge, 1985).
41
P. Si, H. Shi, and X. Yu, “Development of a mathematical model for submarine 54
T. J ohannesson, P. Gauer, M. Barbolini, U. Domaas, T. Faug, K.
granular flows,” Phys. Fluids 30, 083302 (2018). Hakonard ottir, C. Harbitz, D. Issler, F. Naaim-Bouvet, M. Naaim, and L.
42
D. Mancarella and O. Hungr, “Analysis of run-up of granular avalanches Rammer, The Design of Avalanche Protection Dams, Recent Practical and
against steep, adverse slopes and protective barriers,” Can. Geotech. J. 47, Theoretical Developments (International Symposium on Mitigative Measures
827–841 (2010). against Snow Avalanches, Egilsstadir, Iceland, 2008).