Professional Documents
Culture Documents
Numerical Assessment For The Efficiencies of Check Dams - 2020 - Computers and
Numerical Assessment For The Efficiencies of Check Dams - 2020 - Computers and
Research Paper
A R T I C LE I N FO A B S T R A C T
Keywords: Quantitative assessment for the efficiencies of check dams is necessary for the purpose of reducing the risk of
Debris flow potential debris flow disasters. In this paper, an improved finite difference model is adopted to assess the effi-
Disaster mitigation ciency of the check dam in Dagou debris flow gully, Gansu Province, China. Unlike other numerical models, the
Check dam improved model can simultaneously consider bed entrainment and the impeding effect of check dams, so it can
Numerical simulation
reflect the interaction between check dams and debris flow dynamics. The results show that the existing 10 m
Finite difference method
check dam may only be sufficient to prevent a debris flow with less than 0.8 times the volume of the catastrophic
event occurred in this gully in 2013. Otherwise, a higher check dam should be constructed in order to prevent a
larger event. The location of a check dam has significant influence on the dynamics of the debris flow. Generally,
a check dam constructed in the upper part of the gully performs better in controlling bed entrainment and the
discharge. Comparing with the locations in downstream, the dam site located in upstream of this gully seems to
be a better choice in terms of both efficiency and engineering quantity.
1. Introduction Various debris flow preventions works, such as check dams, slit
dams, channel works, wire nets etc., have been devised and put into
Debris flows are one of the most dangerous types of mass move- practice [12,29,42]. Among them, check dams may be the most widely-
ments which occur frequently in established channels on earth surface used countermeasures in debris flow gullies [11,49]. On the other hand,
[34]. These flows are characterized by the two-phase nature (i.e., they the design of check dams and the efficiency assessment of them still rely
consist of poorly-sorted sediments saturated with water, and both solid heavily on empirical methods currently [4,44,61]. For instance, the
particles and fluid significantly influence the motion of the flows) and storage capacity of a dam, which determines dam height, is a critical
periodical occurrence, which distinguish they from dry granular flows index in check dam design. However, this index is usually estimated
and floods. The two-phase nature of debris flows tends to render them according to the maximum static capacity of a reservoir using empirical
the high mobility that dry rock avalanches cannot reach. According to equations. Such a design method ignores the fact that the motion of a
field data, a debris flow can have more than twice the mobility (the debris flow is a dynamic process, and thus it tends to give an inaccurate
ratio of the horizontal run-out distance L to the drop height H: L/H) of a estimation. Another question which remains largely unsolved is the
dry rock avalanche [30], which suggests that debris flows usually efficiencies of the already constructed check dams in thousands of
possess much longer run-out distances and larger impact areas than debris flow gullies. If the assessment of their efficiencies is conducted
other landslides. As a consequence, a large debris flow alone may result only after the occurrence of real debris flow events, the risk of in-
in hundreds of deaths, such as the debris flow events occurred in China adequate design can be high and the subsequent catastrophic con-
[51,63], Venezuela [26,65] and many other countries [1,18,19]. Ad- sequences will be unavoidable. As a result, debris flow disasters, which
ditionally, with the rapid expansion of population and urbanization in are caused by the failures of inadequately designed check dams, are
developing countries, a growing number of infrastructures and re- reported by literatures at times [12,64]. Therefore, it is necessary to
sidence buildings are constructed on old debris flow fans without fully develop reliable quantitative methods for the design and assessment of
evaluating the potential risks, which makes the current situation worse check dams. To realize this goal, numerical simulation can be a pro-
[15,69,70]. Under such a circumstance, constructing efficient debris mising choice.
flow prevention works becomes an urgent mission. Many numerical models have been developed for simulating flow-
⁎
Corresponding author.
E-mail addresses: wei.shen2@unibo.it (W. Shen), dcdgx08@chd.edu.cn (T. Li), dcdgx07@chd.edu.cn (P. Li).
https://doi.org/10.1016/j.compgeo.2020.103541
Received 26 November 2019; Received in revised form 24 February 2020; Accepted 9 March 2020
Available online 14 March 2020
0266-352X/ © 2020 Elsevier Ltd. All rights reserved.
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
like mass movements like debris flows. The existing models can be 2. Principles of the numerical model
classified into three categories according to the mechanical methods,
namely discrete models (DMs), continuum models (CMs) and fluid-solid 2.1. Governing equations
coupling models (FSCMs). DMs regard the debris mass as an assemblage
of solid particles, and adopt discrete element methods [5,39,59] to The governing equations adopted in our model are deduced in a
solve the governing equations of the particle system. These methods global Cartesian coordinate [55]. They can be written as
perform well in simulating dry granular flows, while they may not be
∂U ∂F ∂G
capable of capturing the two-phase nature of debris flows since they + + =S
∂t ∂x ∂y (1)
ignore the influence of fluid. By contrast, CMs view the mixture of
debris flows as a continuum medium or both the solid and liquid parts where U is the vector of conservative variables, F and G are the vectors
as continuum mediums, while FSCMs are generally the combinations of of transport fluxes in the x and y directions, respectively, and S is the
DMs and CMs [8,47,62,68]. Theoretically, since FSCMs can directly vector of source terms. Their expressions are given by
depict the two-phase nature of debris flows, they should be able to
provide more realistic simulation results. However, the computational ⎛h⎞
cost of FSCMs is still unaffordable when simulating a real debris flow U = ⎜Q x ⎟
⎜Q ⎟
events, which constrains their applications. Therefore, there are only a ⎝ y⎠ (2)
few preliminary studies adopting FSCMs to simulate rock avalanches
⎛ Qx ⎞
generated impulsive waves. Given the above reasons, perhaps CMs are
still the most widely-used models that can be put into practice in debris F = ⎜Qx 2 h + k x gh2 2⎟
⎜ Qx Qy h ⎟
flow simulation. The existing CMs normally adopt shallow water as- ⎝ ⎠ (3)
sumption to simplify the problem, so that the original three-dimen-
sional problem can be reduced to a two-dimensional problem which ⎛ Qy ⎞
takes less computational resources. A variety of depth-averaged models, G=⎜ Qx Qy h ⎟
both those based on single-phase mixture fluid theory ⎜Q 2 h + k gh2 2 ⎟⎠
⎝ y y (4)
[7,13,28,41,43,45,48,53–55] and two-phase fluid theory [50,60], have
been proposed through similar manners. They can be solved by various Er
⎛ ⎞
numerical methods, such as finite difference method, finite volume
⎜ (Ag 2
+ B ) htanα
2 −
τ b Ab hv x
⎟
method and smoothed particle hydrodynamics. Additionally, some re- S = ⎜ tan α + tan β + 1 m v x 2 + v y 2 + vz 2
⎟
cent models have taken into account the effects of i) debris flow mass ⎜ (Ag + B ) h tanβ τ b Ab hv y
⎟
variation due to entrainment/deposition processes and ii) bed curva- ⎜ tan2α + tan2β + 1 − m v x 2 + v y 2 + vz 2
⎟
⎝ ⎠ (5)
ture [23,31]. These factors play dominant role in determining the dy-
namics of some debris flows. For instance, entraining saturated material In the above equations, h is the thickness of the debris flow,
can enhance the mobility of a debris flow [56], while the centrifugal Qx = vxh and Qy = vyh are the mass fluxes in the x and y directions,
force produced by curved bed can increase the total basal normal force respectively, kx and ky are the lateral earth pressure coefficients in the x
and interstitial pressure on bed, especially in the case of fine-grained and y directions, respectively, which are calculated according to the
material flow [22,58]. These new advances are of significant im- strain rate [55], Er is the entrainment rate, A and B are the terms related
portance to the improvement of simulation accuracy. On the other to the static and centrifugal normal forces on the basal surface, re-
hand, the current models are mainly formulated to simulate debris spectively, tanα = ∂Z ∂x and tanβ = ∂Z ∂y are the slopes of the basal
flows propagating on natural grounds. This limits their application in surface in the x and y directions, respectively, Z is the basal elevation, g
the design and efficiency assessment of the check dams in debris flow is the gravitational acceleration, τb is the basal shear stress, Ab is the
gullies. Although some recent studies and commonly used commercial basal area of the control volume, vx, vy and vz are the depth-averaged
models (e.g. FLO 2D and RAMMS) have accounted for the effect of velocities in the x, y and z directions, respectively. The expressions of A,
artificial structures in their models [17,36,37,40,66], they did not B and Ab are given by
consider bed entrainment which is a critical factor when simulating the ∂k x h2 2 ∂k y h2 2
dynamic process of a debris flow. As have been illustrated by previous A=1+ tan α + tan β
∂x ∂y (6)
studies [25,27,32,33,56], ignoring bed entrainment may lead to wrong
simulation results and misinterpretation on the run-out characteristic of 2 Cy ⎛ vy ⎞2
Cx v
a debris flow, especially when most material of a debris flow comes B= ⎛ x ⎞ + ⎜ ⎟
cos α ⎝ cos α ⎠ cos β ⎝ cos β ⎠ (7)
from entrainment (e.g., run-off induced debris flows). To overcome
these drawbacks in the previous CMs, recently Shen et al. [57] devel-
Ab = Δx Δy tan2α + tan2β + 1 (8)
oped a modified single-phase depth-averaged model which can si-
multaneously consider bed entrainment and the constraint of check where Cx and Cy are the bed curvatures in the x and y directions, re-
dams. The purpose of this paper is to apply this model to the efficiency spectively, and Δx, Δy are the sizes of the control volume in the x and y
assessment of the check dams in a real debris flow gully. In this case directions, respectively.
study, we evaluate the influences of check dam location, height and In rapid granular masses flowing along mountain streams, normal
debris flow magnitude on prevention efficiencies. The outcome of this and shear stresses associated with friction and collision [3] (i.e. dis-
study can be helpful in improving current understandings toward the persive pressures) simultaneously occur within a basal shear layer
interaction between debris flows dynamics and the impeding effect of [2,21,38,67]. Some studies indicate that the granulometric properties
check dams, and the method adopted here provides a promising ap- (i.e. average grain diameter) of debris mass, which is also related to
proach for the quantitative design of check dams and their efficiency dispersive pressure, can greatly influence the run-out length of a debris
assessment. flow [6,21]. However, for simplicity, we do not consider the stresses
In the following parts, the basic principles of the modified model are caused by the dispersive pressures here. The commonly used Mohr-
briefly presented. Then the Dagou debris flow gully with check dam is Coulomb criterion is adopted to determine both the shear stresses in the
applied as a case study. debris flow (τb) and the erodible mass (τe)
τb = σ (1 − rub) tan φb + cb (9)
2
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
τe = σ (1 − rue ) tan φe + ce (10) velocity value (vucal) will be updated using the computational result of
governing equations (Eq. (1)). Additionally, the check dam height will
where σ is the total normal stress on basal surface, the subscripts b and e
be included in the elevation of the basal surface (Z) when calculating
refer to debris flow and erodible mass, respectively, ru is the ratio of
vucal.
pore pressure u to total normal stress σ, which is an empirical para-
meter, tanφ is the effective frictional coefficient, and c is the cohesion.
2.3. Numerical scheme
The entrainment rate Er is calculated according to the momentum
conservation of the entrained mass [24]
The numerical scheme is briefly introduced here, and detailed in-
∂Z τb − τe formation can be found in [55]. Rectangular computational cells are
Er = − =
∂t ρe vx 2 + vy 2 used to discretize the governing equations. The debris flow depth h is
(11)
discretized at the center of these cells, and the mass fluxes Qx and Qy are
where ρe is the density of the entrained mass. discretized at the boundaries of these cells. The third-order Adams
Substituting Eq. (9) and (10) into the above equation, the expression predictor-corrector method [55] is used to discretize the temporal terms
of Er can be rewritten as (i.e., the first terms on the left hand side of Eq. (1)), and first-order
upwind method is adopted to discretize the convective terms (i.e., the
σ (1 − rub) tan φb + cb − σ (1 − rue ) tan φe − ce
Er = second and third terms on the left hand side of Eq. (1)). Source terms
ρe vx 2 + vy 2 (12) (i.e., the right hand side of Eq. (1)) are discretized by central difference
method. Appling the above numerical scheme, the governing equations
In summary, the governing equations (Eq. (1)) of the adopted model
will be transformed into linear algebraic equations which can be solved
are fully closed by Eqs. (9), (10) and (11), and the three elements in
numerically.
vector U (i.e., h, Qx and Qy) are the basic variables need to be calcu-
lated. These equations are discretized and solved by an explicit finite
3. Introduction to the Dagou landslide induced debris flow
difference scheme adopted in Shen et al. (2018).
3
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
Fig. 2. Panoramas of the Dagou gully (a) before slope failure, (b) after slope failure but before constructing the check dam, and (c) after constructing check dam.
(These satellite images were downloaded from Google Earth).
slope consist of thick mudstone and a thin layer of loess (only a few Eleven groups of simulations are conducted to investigate the in-
meters) covering on the top. Therefore, the rheological property of this fluences of check dam location, height and debris flow volume on the
debris flow should be controlled by the rheology of the mudstone. The prevention efficiency. The detailed settings for these groups are given in
influence of loess can be neglected, so we will only consider the Table 1. The first group (V-ND) absent of check dams is regarded as the
rheology of mudstone in the following simulations. reference group representing the 2013 debris flow event. The initial
volume of the debris flow is the initial volume V of the detached
4. Simulation settings mudstone in the 2013 event. Bed entrainment is considered in every
group.
The digital elevation model of the study area is built up using our The parameters adopted in each simulation group are the same
field survey data. The coordinate adopted in simulation is given in (Table 2). The effective and internal friction angles, basal cohesion and
Fig. 5. The sizes of the computational cells are 10 m in both x and y bulk density of the debris mass are given based on the mechanical
directions. Four locations along the gully are selected as the possible parameters of the mudstone samples [55]. The bulk density of the
places for building a check dam. These locations D1-D4 represent up- erodible mass takes the same value as that of the debris mass for sim-
stream, upper and middle stream, middle and lower stream and plification. The effective friction angle and basal cohesion of the
downstream of the gully, respectively. D1 is the location of the existing erodible mass take much lower values than those of the debris mass to
check dam in this gully. Additionally, three cross-sections are chosen to enable the occurrence of entrainment. Pore pressure coefficients are
monitoring the discharges along the gully. The locations of the check empirically given and calibrated to make the simulation results agree
dams and discharge monitoring cross-sections are illustrated in Fig. 6. well with the field observations.
4
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
Fig. 3. Back scarp in the source zone of Dagou gully and the check dam in the gully. The check dam is located in the throat area of the gully between the source zone
and transportation zone.
5. Results and discussions reason is that the velocity of the debris flow at D2 is relatively high, and
thus more debris mass is likely to surpass the dam. In the other three
5.1. Influence of check dam location situations, the check dam reduces 30–40 m of the final run-out distance.
The amounts of debris mass entrapped by the check dam in these three
Fig. 7 is the isopach maps of the debris deposit under the effect of situations are quite close. However, since the gully in the downstream is
different check dam locations. It shows that except for group V-D2 wider and flatter, the length of check dam at D1 will be longer than the
(Fig. 7c) the final distributions of the debris deposit in the other three rest two situations.
groups are quite similar. Although in group V-D2 the average thickness The above analysis suggests that those places in the upper and
of the deposit in the accumulation zone reduces 2–3 m, the run-out middle stream where the gully is narrow and steep should not be chosen
distance and covering area are basically the same as those of without a as the dam site. In addition, the flat and wide parts in the downstream
check dam (Fig. 7a), which means that the check dam at D2 cannot of the gully are also not suitable dam sites, because the total en-
prevent a debris flow like the 2013 event. This is perhaps caused by two gineering quantity tends to be larger in these locations. In comparison,
reasons. First of all, the width of the gully at D2 seems to be narrower the dam site D1 in upstream seems to be a good choice in terms of both
than the other three positions, and the slope there is also steeper. prevention effect and engineering quantity. Nevertheless, it illustrates
Therefore, the capacity of reservoir there tends to be smaller. Another that the current check dam may still not be sufficient to prevent a debris
Fig. 4. Outcrop at the back scarp exposed after the slope failure in the source zone of Dagou gully. The landslide mainly consists of mudstone, and the average
thickness of the loess covering upon the mudstone is only approximately 5 m.
5
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
Fig. 5. Three-dimensional digital elevation model of the computational zone for simulating the Dagou landslide-induced debris flow and the coordinates in this zone.
flow as large as the 2013 event. The debris flow may still endanger the Table 1
infrastructures and properties near to the outlet of the gully. Simulation conditions in different groups.
Fig. 8 illustrates the performance of the check dam in controlling Group names Initial volume V (m3) Dam location Dam Height (m)
the discharges in the gully. It is obvious that a check dam can reduce
the discharge on the downstream side of it, while it generally does not V-ND 151,000 Absence of check –
affect the discharge on the upstream side far away from it (Fig. 8d). The dams
V-D1 151,000 D1 10
reducing effect of check dam on the discharge is more significant when V-D2 151,000 D2 10
the dam is constructed in upstream of the gully. The peak discharges at V-D3 151,000 D3 10
the three cross-sections are almost the same (about 10,000 m3/s) before V-D4 151,000 D4 10
constructing a check dam. These values slump to about 8000 m3/s V-D1-15 151,000 D1 15
V-D1-20 151,000 D1 20
when the check dam is located at D1. By contrast, the discharge de-
0.8V-D1 0.8 × 151,000 D1 10
creases in groups V-D2 and V-D3 are relatively slight. Another apparent 0.6V-D1 0.6 × 151,000 D1 10
tendency is that the propagation speed of the debris flow will decrease 0.8V-ND 0.8 × 151,000 – –
due to the impeding effect of a check dam. In addition, the debris flow 0.6V-ND 0.6 × 151,000 – –
will spread slower when the dam is located in the upper gully (Fig. 8a).
Fig. 6. Original thickness distribution of the Dagou landslide, the locations of the check dams and the discharge monitoring profiles in the simulations.
6
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
Table 2
Parameters for simulating the Dagou landslide-induced debris flow.
Parameters Pore pressure coefficient ru Effective friction angle φ Basal cohesion c Internal friction angle φi Bulk density ρ
– ° kPa ° g/cm3
Fig. 7. Simulated final thickness distributions of the Dagou landslide-induced debris flow in groups V-ND and V-D1 to V-D4.
7
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
Fig. 8. Influence of check dam location on the discharges at the three monitoring profiles in Dagou gully.
outlet of the gully (Fig. 10a), while a debris flow has 0.6 times this feature is that the peak average velocity of the debris flow decreases
initial volume is more likely to stop before reaching the outlet. The with the location of the dam from the downstream to the upstream of
current check dam can further reduce the run-out distance in both the gully. Additionally, the average velocity will decrease with the in-
cases. Therefore, it can be concluded that the current check dam in crease of check dam height. Similarly, a check dam can reduce the total
Dagou gully may be efficient enough to prevent a debris flow with entrainment volume (Fig. 12). The constrain effect of check dam on bed
about 0.8 times of the initial volume of the 2013 event. entrainment tends to be more obvious when it is constructed in the
upstream of the gully. The characteristic data in Table 3 can well
5.3. Influences of check dams on debris flow dynamics support the above analysis.
The average velocity and entrainment process curves of the debris 6. Conclusions
flow in different groups are used to analyze the influences of check
dams on the dynamics of this debris flow. The average velocity curves The prevention effect of the check dam in Dagou gully is evaluated
(Fig. 11) show that the average velocity of the debris flow will be sig- by a modified finite different model. It shows that the current 10 m dam
nificantly reduced due to the impact of the check dam. An apparent may be insufficient to prevent a debris flow event as that occurred in
Fig. 9. Simulated final thickness distributions of the debris flow deposit under the effect of the check dams at D1 with different heights.
8
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
Fig. 10. Simulated final thickness distributions of the debris flow deposit with different initial volumes under the effect of the check dam at D1.
this gully in 2013. However, if the initial volume of a new debris flow is flow. Generally, if the check dam is located in the upper part of the
no larger than 0.8 times the volume of the 2013 event, the current gully, it will perform better in controlling the peak discharge, total
check dam may perform efficiently enough. Otherwise, a 15 m or average velocity and bed entrainment extent. The dam location in up-
higher check dam should be constructed to stop a debris flow like the stream of this gully should be a suitable place in terms of both pre-
2013 event. Further analysis indicates that the location of the check vention effect and engineering quantity. The simulations also suggest
dam can significantly influence the dynamic characteristic of the debris that those places in the upper and middle stream of the gully with
Fig. 11. Simulated average velocity curves of the debris flow in groups V-ND, V-D1 to V-D4, V-D1-15, and V-D1-20.
9
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
Fig. 12. Simulated total volume curves of the debris flow in groups V-ND, V-D1 to V-D4, V-D1-15, and V-D1-20.
10
W. Shen, et al. Computers and Geotechnics 122 (2020) 103541
DA, et al. Catastrophic detachment and high-velocity long-runout flow of Kolka [46] Peng J, Fan Z, Wu D, Zhuang J, Dai F, Chen W, et al. Heavy rainfall triggered
Glacier, Caucasus Mountains, Russia in 2002. Geomorphology loess–mudstone landslide and subsequent debris flow in Tianshui, China. Eng Geol
2009;105(3):314–21. 2015;186:79–90.
[20] Federico F, Amoruso A. Impact between fluids and solids. Comparison between [47] Peng X, Yu P, Chen G, Xia M, Zhang Y. Development of a coupled DDA–SPH method
analytical and FEA results. Int J Impact Eng 2009;36(1):154–64. and its application to dynamic simulation of landslides involving solid-fluid inter-
[21] Federico F, Cesali C. An energy-based approach to predict debris flow mobility and action. Rock Mech Rock Eng 2019.
analyze empirical relationships. Can Geotech J 2015;52(12):2113–33. [48] Pirulli M, Pastor M. Numerical study on the entrainment of bed material into rapid
[22] Federico F, Cesali C. Coupled effects of pore-water pressure evolution, slope cur- landslides. Géotechnique 2012;62(11):959–72.
vature, and mass variation on the kinematics of rapidly sliding fine-grained mate- [49] Piton G, Carladous S, Recking A, Tacnet JM, Liébault F, Kuss D, et al. Why do we
rials. Int J Geomech 2017;17(10):04017088. build check dams in Alpine streams? An historical perspective from the French
[23] Federico F, Cesali C. Effects of granular collisions on the rapid coarse-grained experience. Earth Surf Proc Land 2017;42(1):91–108.
materials flow. Géotech Lett 2019;9(4):278–83. [50] Pudasaini SP. A general two-phase debris flow model. J Geophys Res Earth Surf
[24] Fraccarollo L, Capart H. Riemann wave description of erosional dam-break flows. J 2012;117(F3).
Fluid Mech 2002;461:183–228. [51] Ren D. The devastating Zhouqu storm-triggered debris flow of August 2010: likely
[25] Frank F, McArdell BW, Huggel C, Vieli A. The importance of entrainment and causes and possible trends in a future warming climate. J Geophys Res: Atmos
bulking on debris flow runout modeling: examples from the Swiss Alps. Nat Hazards 2014;119(7):3643–62.
Earth Syst Sci 2015;15(11):2569–83. [52] Rickenmann D. Empirical relationships for debris flows. Nat Hazards
[26] García-Martínez R, López JL. Debris flows of December 1999 in Venezuela. Debris- 1999;19(1):47–77.
flow hazards and related phenomena. Berlin, Heidelberg: Springer Berlin [53] Sassa K, Nagai O, Solidum R, Yamazaki Y, Ohta H. An integrated model simulating
Heidelberg; 2005. p. 519–38. the initiation and motion of earthquake and rain induced rapid landslides and its
[27] Hungr O, Evans SG. Entrainment of debris in rock avalanches: an analysis of a long application to the 2006 Leyte landslide. Landslides 2010;7(3):219–36.
run-out mechanism. GSA Bull 2004;116(9–10):1240–52. [54] Savage SB, Hutter K. The motion of a finite mass of granular material down a rough
[28] Hungr O, McDougall S. Two numerical models for landslide dynamic analysis. incline. J Fluid Mech 1989;199:177–215.
Comput Geosci 2009;35(5):978–92. [55] Shen W, Li T, Li P, Guo J. A modified finite difference model for the modeling of
[29] Ikeya H. Debris flow and its countermeasures in Japan. Bull Int Assoc Eng Geol – flowslides. Landslides 2018;15(8):1577–93.
Bulletin de l'Association Internationale de Géologie de l'Ingénieur [56] Shen W, Li T, Li P, Shen Y, Lei Y, Guo J. The influence of the bed entrainment-
1989;40(1):15–33. induced rheology and topography changes on the propagation of flow-like land-
[30] Iverson RM. The physics of debris flows. Rev Geophys 1997;35(3):245–96. slides: a numerical investigation. Bull Eng Geol Environ 2019;78(7):4771–85.
[31] Iverson RM, George DL, Logan M. Debris flow runup on vertical barriers and ad- [57] Shen W, Wang D, Qu H, Li T. The effect of check dams on the dynamic and bed
verse slopes. J Geophys Res Earth Surf 2016;121(12):2333–57. entrainment processes of debris flows. Landslides 2019;16(11):2201–17.
[32] Iverson RM, Ouyang C. Entrainment of bed material by Earth-surface mass flows: [58] Siviglia A, Cantelli A. Effect of bottom curvature on mudflow dynamics: theory and
review and reformulation of depth-integrated theory. Rev Geophys experiments. Water Resour Res 2005;41(11).
2015;53(1):27–58. [59] Song Y, Huang D, Zeng B. GPU-based parallel computation for discontinuous de-
[33] Iverson RM, Reid ME, Logan M, LaHusen RG, Godt JW, Griswold JP. Positive formation analysis (DDA) method and its application to modelling earthquake-in-
feedback and momentum growth during debris-flow entrainment of wet bed sedi- duced landslide. Comput Geotech 2017;86:80–94.
ment. Nature Geosci 2011;4:116. [60] Tai Y-C, Heß J, Wang Y. Modeling two-phase debris flows with grain-fluid se-
[34] Jakob M, Hungr O. Introduction. Debris-flow hazards and related phenomena. paration over rugged topography: application to the 2009 Hsiaolin event, Taiwan. J
Berlin, Heidelberg: Springer Berlin Heidelberg, 2005. p. 1–7. Geophys Res: Earth Surf 2019;124(2):305–33.
[35] Jeong S, Lee K. Analysis of the impact force of debris flows on a check dam by using [61] Takahashi T, Das DK. Debris flow: mechanics, prediction and countermeasures. CRC
a coupled Eulerian-Lagrangian (CEL) method. Comput Geotech 2019;116:103214. Press; 2014.
[36] Kattel P, Kafle J, Fischer J-T, Mergili M, Tuladhar BM, Pudasaini SP. Interaction of [62] Tan H, Chen S. A hybrid DEM-SPH model for deformable landslide and its generated
two-phase debris flow with obstacles. Eng Geol 2018;242:197–217. surge waves. Adv Water Resour 2017;108:256–76.
[37] Kwan JSH, Sze EHY, Lam C. Finite element analysis for rockfall and debris flow [63] Tang C, Rengers N, van Asch TWJ, Yang YH, Wang GF. Triggering conditions and
mitigation works. Can Geotech J 2018;1–26. depositional characteristics of a disastrous debris flow event in Zhouqu city, Gansu
[38] Lanzoni S, Gregoretti C, Stancanelli LM. Coarse-grained debris flow dynamics on Province, northwestern China. Nat Hazards Earth Syst Sci 2011;11(11):2903–12.
erodible beds. J Geophys Res Earth Surf 2017;122(3):592–614. [64] Wang GL. Lessons learned from protective measures associated with the 2010
[39] Law RPH, Choi CE, Ng CWW. Discrete-element investigation of influence of gran- Zhouqu debris flow disaster in China. Nat Hazards 2013;69(3):1835–47.
ular debris flow baffles on rigid barrier impact. Can Geotech J 2015;53(1):179–85. [65] Wieczorek GF, Larsen MC, Eaton LS, Morgan BA, Blair JL. Debris-flow and flooding
[40] Liu J, Nakatani K, Mizuyama T. Effect assessment of debris flow mitigation works hazards associated with the December 1999 storm in coastal Venezuela and stra-
based on numerical simulation by using Kanako 2D. Landslides 2013;10(2):161–73. tegies for mitigation. Open-File Report. – ed: US Geological Survey 2001.
[41] McDougall S, Hungr O. Dynamic modelling of entrainment in rapid landslides. Can [66] Xiong M, Meng X, Wang S, Guo P, Li Y, Chen G, et al. Effectiveness of debris flow
Geotech J 2005;42(5):1437–48. mitigation strategies in mountainous regions. Prog Phys Geogr: Earth Environ
[42] Mizuyama T. Structural countermeasures for debris flow disasters. Int J Eros 2016;40(6):768–93.
Control Eng 2008;1(2):38–43. [67] Zhang D, Foda MA. Internal wave—granular temperature interaction: an energy
[43] Ouyang C, He S, Xu Q, Luo Y, Zhang W. A MacCormack-TVD finite difference balance study on granular flow. Acta Mech 1999;136(3):155–70.
method to simulate the mass flow in mountainous terrain with variable computa- [68] Zhao T, Utili S, Crosta GB. Rockslide and impulse wave modelling in the vajont
tional domain. Comput Geosci 2013;52:1–10. reservoir by DEM-CFD analyses. Rock Mech Rock Eng 2016;49(6):2437–56.
[44] Pal D, Galelli S, Tang H, Ran Q. Toward improved design of check dam systems: A [69] Zhou NQ, Zhao S. Urbanization process and induced environmental geological
case study in the Loess Plateau, China. J Hydrol 2018;559:762–73. hazards in China. Nat Hazards 2013;67(2):797–810.
[45] Pastor M, Haddad B, Sorbino G, Cuomo S, Drempetic V. A depth-integrated, coupled [70] Ziegler AD, Cantarero SI, Wasson RJ, Srivastava P, Spalzin S, Chow WTL, et al. A
SPH model for flow-like landslides and related phenomena. Int J Numer Anal Meth clear and present danger: Ladakh's increasing vulnerability to flash floods and
Geomech 2009;33(2):143–72. debris flows. Hydrol Process 2016;30(22):4214–23.
11