You are on page 1of 13

Technical Note

Landslides C. W. W. Ng I U. Majeed I C. E. Choi I W. A. R. K. De Silva


DOI 10.1007/s10346-021-01631-7
Received: 25 August 2020
Accepted: 19 January 2021 New impact equation using barrier Froude number
© Springer-Verlag GmbH Germany
part of Springer Nature 2021
for the design of dual rigid barriers against debris flows

Abstract In the design of multiple rigid barriers, the height of the shed light on how the spacing between barriers should depend on
first barrier governs the impact dynamics of debris flow on the more than just the final volume retained, Hákonardóttir et al.
next barrier in a channel. However, current design approaches (2003a) conducted a series of experiments that modelled dry
neglect the height of the first barrier, and no specific guideline is granular flow impacting against avalanche mounds. They reported
given on the design of the impact load on the second barrier. In that the spacing between rows of obstacles depends on the over-
this study, a new impact equation that explicitly considers the flow trajectory, which is dependent on the height of the first row of
effects of the height of the first barrier on the impact dynamics is obstacles. Furthermore, Faug et al. (2002) reported that the height
proposed. This is achieved by adopting the barrier Froude number of an obstacle also influences the size of the dead zone that forms
Frb, which is the ratio of inertia to barrier potential. Thereby, the behind an obstacle. They proposed a dimensionless number called
new impact equation accounts for the static load as a function of the barrier Froude number Frb, (more details discussed later)
the first barrier height. The equation is evaluated using physical which showed a strong correlation with dead zone length. The
experiments carried out in a 5-m-long flume. The experiments Frb was also used to characterise the impact dynamics relative to
modelled dry sand and water flows impacting dual rigid barriers. the barrier height when overflow occurs (Faug 2015). Furthermore,
These two idealised flow types represent extreme cases of frictional Zanuthigh and Lamberti (2006) showed an increase in the total
and viscous flows, which exhibit entirely different impact mecha- load on the barrier with the size of the dead zone. Despite the
nisms. A comparison of the experimental results from this study importance of the height of a barrier, current guidelines (VanDine
shows that the proposed impact model using the barrier Froude 1996; Kwan 2012) do not explicitly consider the influence of barrier
number provides a reasonably conservative estimate for the nor- height for overflow conditions. For engineering design, debris
malised impact force on the first rigid barrier with overflow. flows are often idealised as an equivalent fluid (Hungr et al.
Furthermore, a bilinear design envelop for the impact force 1984; Zanuttigh and Lamberti 2006) to enable continuum mechan-
exerted on the second barrier is proposed based on the barrier ics to be adopted. The most common continuum-based ap-
Froude number of the first barrier. proaches to estimate the impact force exerted by debris flows are
the hydrostatic (Scotton and Deganutti 1997) and hydrodynamic
Keywords Landslides . Debris flow . Multiple barriers . Barrier (Kwan 2012) approaches as shown in Fig. 1. Both approaches
height . Impact assume that a flow exerts a distributed load on a barrier over the
initial flow depth before impact ho and channel width w. The
Introduction hydrostatic approach is given as follows (Armanini and Scotton
Flow-like landslides such as debris flow surge downslope rapidly 1993; Scotton and Deganutti 1997):
(Hungr et al. 2014) and often result in loss of lives and infrastruc-
1
ture in densely populated mountainous areas (Koo et al. 2018; F peak ¼ kρgh2o w ð2Þ
Froude and Petley 2018; Cui et al. 2019). Countermeasures such 2
as flexible barriers (Wendeler 2016; Koo et al. 2017b) and rigid
barriers (Song et al. 2017; Ng et al. 2017) are commonly installed in
the channel to arrest debris flows. Mitigation of large volume where Fpeak is the maximum impact force; k is the empirical static
debris flows often requires installing more than a single barrier pressure coefficient; ρ is the bulk density; and g is the acceleration
along the predicted flow path. Barriers in series are ideal for due to earth’s gravity; The empirical coefficient k is reported to
progressively arresting debris and distributing its loading among range from 9 to 11 by Armanini (1997) and 3 to 30 by Scheidl et al.
barriers (Osti and Egashira 2008; Takahashi 2014; Kwan et al. 2015; (2013). Rather high k values are reported in literature to account
Ng et al. 2018). Such an approach enables the height and, therefore, for dynamic loading in Eq. 2, which is explicitly static. The hydro-
the size of each barrier to be reduced. Existing guidelines only dynamic approach is as follows (Hungr et al. 1984):
consider the barrier height when calculating the final volume that
F peak ¼ αρv2 ho w ð3Þ
a barrier can retain (CGS 2004; NILIM 2007). Based on the final
retained volume, the recommended minimum spacing Lmin be-
tween barriers can be calculated (VanDine 1996) as follows:
where α is an empirical hydrodynamic coefficient, and v is the
Lmin > 2l >
H
ð1Þ flow velocity before impact. The hydrodynamic coefficient α ac-
tanθ−tanψ counts for the simplifications and assumptions made in Eq. 3. For
example, the momentum flux is constant, the contact area is
assumed to remain unchanged, and the formula is explicitly dy-
where H is the barrier height; θ is the channel inclination; l is the namic. Equation 3 can be rearranged for α, which is given as α =
length of potential scour; and Ψ is the deposition angle, which for Fpeak/ρv2how. Theoretically, an α of unity implies that the linear
granular flow can be taken as the angle of repose (Hungr 2008). To impact momentum is not conserved or the impact scenario is

Landslides
Technical Note
inelastic. An α of two indicates that the linear impact momentum
is conserved or the impact scenario is elastic. However, α has been
reported to vary considerably, ranging from 0.2 to 18 in the liter-
ature (Scheidl et al. 2013; Cui et al. 2015). The broad range of α
reported in the literature is mainly because of Eq. 3, which is
explicitly dynamic and only captures the effects of static loading
ho
through larger values of α. Based on the above considerations, Eq.
2 should be equal to Eq. 3. However, experimental data reveal that
flowing sediments impacting a barrier have both static and dy-
namic components (Armanini et al. 2020). Therefore, a general- Rigid barrier
ised impact model is needed to account for the influence of flow (a)
velocity and the barrier height to improve estimates of the impact
force when overflow occurs.
Depending on the proportion of sediments to fluid, the impact
dynamics of a debris flow can be dominated by frictional or
viscous stresses. Choi et al. (2015a) highlighted that for the range
of Froude numbers Fr (more details discussed later) that engineers
mitigate (Faug 2015), the impact mechanism between frictional
and viscous flows differs significantly. Therefore, a robust impact
ho
equation is needed to capture the impact dynamics from these two
extreme flow types. Furthermore, flows enriched with large and
hard inclusions have also been reported to compound the chal- Rigid barrier
lenges in predicting the impact force. More specifically, large and (b)
hard inclusions exert sharp concentrated impulses that Eqs. 2 and
3 cannot capture (Hübl et al. 2017; Song et al. 2018b). To capture Fig. 1 Comparison of approaches to estimate impact force, a hydrostatic and b
concentrated loading, Hertz contact mechanics is often used in hydrodynamic
design guidelines (VanDine 1996; SWCB 2005; Kwan 2012). The
Hertz equation is given as follows:
and μb of 0.2. The barrier is made of reinforced concrete with
F b ¼ K c na1:5 ð4Þ typical EB of 25×109 N/m2 and μB of 0.2. Based on the assumptions
mentioned above, the Hertz equation can be simplified as follows
(Kwan 2012):
where Fb is the force due to boulder impact and Kc is the load
reduction factor. The parameters n and a are defined as follows: F b ¼ K c 4000vb 1:2 r 2b ð9Þ

4r 0:5 The Fb using above equation has units of kN. The recommend-
n¼ b
ð5Þ
3πðkb þ kB Þ ed values of Kc vary in the literature from 0.1 and 0.5 (Hungr et al.
1984; SWCB 2005; Hübl et al. 2017). Kwan (2012), in particular,
recommended a combination of Eqs. 3 (α = 2.5) and 9 (Kc = 0.1) to
 0:4 estimate the impact force exerted by boulder enriched debris
5mb v2b flows.
a¼ ð6Þ
4n In this study, a new impact equation is proposed to account for
the static load as a function of the barrier height when overflow
occurs. To evaluate the impact equation, physical flume experi-
ments using dry sand and water were used to model idealised
1−μ2b frictional and viscous flows impacting dual rigid barriers. Further-
kb ¼ ð7Þ
πEb more, a design envelope that considers the barrier Froude number
of the first barrier is recommended for predicting the impact load
on the second barrier in a dual barrier system.

1−μ2B
kB ¼ ð8Þ Flume modelling
πEB A laboratory-scale flume that is 5 m long, 0.2 m wide, and 0.5 m
high was used in this study to carry out the experiments (Fig. 2).
The storage container of flume has a capacity of 0.06 m3. The
where rb, mb, vb, Eb, and μb, are the radius, mass, velocity, Young’s source material is retained behind a gate, which is controlled by
modulus, and Poisson ratio of the boulder, respectively. Similarly, using a hydraulic actuator. The first rigid barrier is installed
EB, and μB is Young’s modulus and Poisson ratio of the barrier, perpendicularly to the flume bed at an inclined distance of 0.8 m
respectively. The boulder is assumed spherical (i.e. mb=4/3 πprb3) from the gate and the second barrier is installed at an inclined
and made of granite with density 2650 kg/m3, Eb of 50×109 N/m2 distance of 1.0 m downstream from the first barrier. Barriers in this

Landslides
study are idealised to have a fixed boundary condition, as is the with horizontal lines separated by 50 mm. Additional camera
case with most flow-barrier interaction studies (Zanuttigh and (model no: GoPro Hero 5) with a sampling rate of 200 fps and
Lamberti 2006; Song et al. 2017; Armanini et al. 2020). However, a resolution of 1280 × 720 was installed above the channel bed to
in reality, boundary conditions vary depending on the type of capture the movement of the flow front. Frontal flow velocities
barriers and foundation. Rigid barriers embedded in soil (Kwan were estimated by recording the time flow front took to pass
2012) resist the impact force via mobilising the soil shear strength marked lines. The frontal velocity is assumed to be uniform
due to sliding, whereas for some cases barrier self-weight counters throughout the flow depth following the shallow water assump-
the impact force (Lam et al. 2018). Furthermore, flexible barriers tion. Both flow velocity and flow height change with time at a
have poles and anchors that provide lateral resistance to the specific location. However, for simplicity, the flow depth and
barrier. In addition, flexible barriers attenuate the impact energy velocities are assumed constant at barrier location following
via deformation (Ashwood and Hungr 2016; Song et al. 2018a). The the design recommendation in Hong Kong (Kwan 2012; Kwan
aforementioned barrier types in the field would mobilise the shear and Koo 2015). Assuming frontal velocity and maximum flow
resistance in the soil upon impact. By assuming a fixed boundary depth for the estimation of impact force is robust and conserva-
condition, this study models a conservative impact scenario. tive (Kwan 2012). In this study, frontal velocity and maximum
flow depth at the barrier location obtained from the control test
is used to estimate the flow Froude number and theoretical force.
Flow characterisation
The Froude number Fr governs the dynamics of open channel flow Test programme
(Hübl et al. 2009; Armanini et al. 2011) and is defined as a ratio of Water and dry Toyoura sand (d50 ~ 0.2 mm) were used to model
the inertial to gravitational forces: two extreme flow types, specifically dry sand (frictional) and
water (viscous) flows. It is acknowledged that the two idealised
v
Fr ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð10Þ flow types in this study neglect the influence of fluid-solid inter-
gcosθho action, which governs the dynamics of two-phase debris flows in
the field. In such flows, pore pressure generation regulates the
The numerator of the Froude number (Eq. 10) is the inertial macroscopic flow dynamics (Iverson 1997). However, two-phase
component of the flow. As the velocity increases, so does the debris flows modelled at smaller scales cannot capture appropri-
inertial force. Inertial flows are defined as when Fr >> 1 ate timescales for pore pressure dissipation and the relative
(Takahashi 2014; Saingier et al. 2016). The length scale in Eq. 10 contributions from viscous and inertial stresses that are general-
ho can be substituted with the height of the barrier H to give the ly observed in the field (Iverson 2015). It is worthwhile to point
barrier Froude number (Faug et al. 2002; Hákonardóttir et al. out that Iverson (2004) stressed that scale effects are less impor-
2003b; Faug 2015): tant for dry granular flows and water (Ng et al. 2013; Choi et al.
v 2015b; Ng et al. 2017). Therefore, to minimise scaling conflicts,
Fr b ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð11Þ idealised flows were adopted in this study. For water flows, flume
gcosθH
inclinations of 0°, 5°, 10°, and 15° were used. For dry sand flows,
flume inclinations of 26°, 35°, and 45° were used. For each test
The barrier Froude number is the ratio of the inertial force of configuration with a different flow type and flume inclination, a
the flow before impact to the barrier potential, which is the separate test was carried out without a barrier installed in the
gravitational force that the flow needs to overcome to climb over flume to measure the maximum flow depth and frontal velocity
the barrier. The Frb provides an indication of whether overflow at the location where the barrier would be installed. By doing so,
will occur for a given barrier height. the Fr of the flows before impact could be characterised. To
obtain Froude numbers with similarity to dynamic conditions
Instrumentation observed in the field, the location of the first barrier was deter-
Figure 2 shows a schematic of the physical model setup and the mined to be installed at an inclined distance of 0.8 m down-
instrumentation layout. Two laser displacement sensors (model stream from the gate. By using the selected channel inclinations,
no.: Wenglor YT44MTGV) were installed 100 mm upstream of Froude numbers ranging from 1 and 5 were obtained at the first
each model barriers to measure the flow depth at the centreline barrier location. This Froude number range is typical of debris
of the flume. Additionally, load cells (model no.: Kyowa Lux-B- flows in Hong Kong where 50% of slopes have inclination greater
500N (first barrier); Omega LC101-500 (second barrier)) were than 25° (Ng et al. 2012). The influence of the height of the first
sandwiched between an acrylic load plate and a reaction frame barrier was examined. These heights were selected based on open
mounted to the flume to measure the impact force exerted by the channel experiments. Barrier Froude numbers from 0.8 to 3.5
model flows. A high-speed camera (model no.: Mikrotron EoSens were modelled. These barrier Froude numbers are typical of
mini2), with a frame rate of 200 frames per second (fps) and a barrier heights (2 to 15 m) observed in the field (Lo 2000). For
resolution of 1696 × 1440, was mounted on the side of the flume the impact tests conducted with barriers, a second barrier with a
to capture the overflow and landing kinematics through the constant height of 500 mm was installed at an inclined distance
transparent side wall of the flume. Another camera (model no: of 1 m from the first barrier. A barrier spacing of 1 m ensures that
GoPro Hero 5) with a sampling rate of 200 fps and a resolution of overflow from the first barrier lands on the flume before
1280 × 720 was installed at the side of the flume to capture the impacting the second barrier (Kwan et al. 2015; Ng et al. 2018)
impact kinematics on the second barrier. For control tests with- for each dual rigid barrier test. The overflow distances were
out barriers installed in the flume, the flume base was marked estimated using the trajectory equation proposed by Kwan

Landslides
Technical Note

Fig. 2 Schematic of laboratory flume setup with instrumentation

et al. (2015), which is given as: allowed. A summary of the test programme and open flume
" sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi# measurements is given in Table 1.
v2m 2gH
xi ¼ tanθ þ tan2 θ þ 2 ð12Þ
g vm Test setup and modelling procedure
For each dual rigid barrier test, the first and second model barriers
were initially installed in the flume. The gate was closed before the
where xi is the maximum landing distance and vm is the
source material was prepared into the storage container. For dry
horizontal component of launch velocity. The estimation of the
sand flows, a total mass of 100 kg was used. For water flows, a total
overflow distance requires the information of launch velocity and
mass of 30 kg was used for Frb greater than unity. To develop a Frb
launch angle at the barrier crest. Kwan et al. (2015) in their
< 1 for water flow, a smaller mass of 15 kg was used for some
derivation of overflow model assumed that the flow launches
experiments (Table 1). The flume was gradually inclined to the
horizontally from the barrier crest once the barrier is filled. Al-
appropriate inclination. The gate was lifted by activating the pneu-
though velocity attenuation occurs during the impact process
matic actuator. The source material was allowed to flow down the
(Koo et al. 2017a), using the frontal velocity v to estimate the
flume and impact the dual rigid barriers.
overflow distance xi would provide a conservative estimate of the
overflow distance. Adopting Eq. 12 enables the effects of barrier
New impact model
spacing on the impact load on the second barrier to be studied
The term impact in this study is defined as the process that starts
rather than solely the retained volume as proposed by VanDine
when the flow front impacts the barrier and ends when the flow
(1996). In their approach, they do not consider the impact dynam-
comes to rest (Song et al. 2018a). The peak impact force Fpeak is
ics between barriers. In this study, some test configurations have
conventionally calculated using continuum approaches, which
barrier spacing that is greater, and some less, than that recom-
idealised the impact process is either explicitly hydrostatic (Eq.
mended by VanDine (1996). However, all of the configurations
2) or hydrodynamic (Eq. 3). However, such an idealisation requires
satisfy Eq. 12. Such an approach allows us to compare the two
empirical coefficients to compensate for static or dynamic effects
methods and optimise the impact load on the second barrier. The
implicitly. Thus, recommended empirical coefficients are generally
height of the second barrier was selected so that overflow was not
inconsistent in the literature and design guidelines (Kwan 2012;
Vagnon 2020). To explicitly capture the effects of static and

Landslides
Table 1 Test programme and corresponding dimensionless numbers
Test ID Material Upstream barrier Downstream Flume Velocity, v (m/s) Flow depth, ho (m) Condition Lmin Froude Barrier Froude number,
height, H1 (mm) barrier height, H2 (mm) inclination, I (°) < L satisfied? number, Fr (-) Frb (-)

W_C_I0* Water - - 0 1.5 0.04 - 2.5 -

W_C_I0 0 2.0 0.05 2.9

W_C_I05 5 2.3 0.05 3.4

W_C_I10 10 2.4 0.05 3.5

W_C_I15 15 2.9 0.06 4.1

S_C_I26 Sand 26 1.2 0.09 1.4

S_C_I35 35 1.5 0.10 1.7

S_C_I45 45 2.0 0.11 2.3

W_H10_I0 Water 100 500 0 2.0 0.05 - 2.9 2.0

W_H10_I05 5 2.3 0.05 No 3.4 2.3

W_H10_I10 10 2.4 0.05 3.5 2.4

W_H10_I15 15 2.9 0.06 4.1 3.0

W_H18_I0 180 0 2.0 0.05 - 2.9 1.5

W_H18_I05 5 2.3 0.05 Yes 3.4 1.7

W_H18_I10 10 2.4 0.05 3.5 1.8

W_H18_I15 15 2.9 0.06 4.1 2.3

W_H26_I0* 260 0 1.5 0.04 - 2.5 0.9

W_H26_I0 0 2.0 0.05 - 2.9 1.3

W_H26_I05 5 2.3 0.05 Yes 3.4 1.4

W_H26_I10 10 2.4 0.05 3.5 1.5

W_H26_I15 15 2.9 0.06 4.1 1.9

S_H10_I26 Sand 100 26 1.2 0.09 - 1.4 1.3

S_H10_I35 35 1.5 0.10 No 1.7 1.7

S_H10_I45 45 2.0 0.11 Yes 2.3 2.4

S_H18_I26 180 26 1.2 0.09 - 1.4 1.0

S_H18_I35 35 1.5 0.10 No 1.7 1.2

S_H18_I45 45 2.0 0.11 Yes 2.3 1.8

S_H26_I26 260 26 1.2 0.09 - 1.4 0.8

S_H26_I35 35 1.5 0.10 No 1.7 1.0

S_H26_I45 45 2.0 0.11 Yes 2.3 1.5

Note: Lmin is estimated by using deposit height in front of second barrier

Landslides
*Represents tests with 15 kg water
Technical Note
dynamic loading, the peak force Fpeak induced on a rigid barrier than unity for both flow types (Table 1). The impact kinematics of
can be taken as the sum of Eqs. 2 and 3 as follows: dry sand (Fig. 3a) and water (Fig. 3b) flows with barrier Froude
number Frb less than unity were compared. The flow direction is
kρgho 2 w from right to left. The time just before the flow impacts the barrier
F peak ¼ þ αρv2 ho w ð13Þ
2 is taken as zero. Figure 3a shows that as dry sand impacts the
barrier, a wedge-like dead zone forms. The dead zone acts as a
By normalising Fpeak in Eq. 13 with 0:5kρh2o w and rearranging, ramp for the subsequent flow to climb up along the barrier face
Eq. 13 becomes: (Fig. 3a, t = 0.5 s). Subsequent flow rides over the deposited
material. During this interaction, energy loss is partly due to the
F peak αv2
2 ¼1þ ð14Þ conversion of kinetic to potential energy during the deposition
0:5kρgho w 0:5kgho
process and partly due to frictional loss between the deposited
material and flow overriding the deposited material (Koo et al.
Simplifying Eq. 14 gives: 2017a). As deposition takes place and the dead zone increases in
F peak 2α 2 size, the remaining flow does not have enough energy to further
¼1þ Fr ð15Þ climb up the dead zone, and a granular bore occurs (Faug et al.
F static k
2008). The bore propagates back in the upstream direction (Fig. 3a,
t = 1 s). The remaining flow is deposited upstream of the barrier,
Furthermore, to account for the influence of barrier height, Eq.
and only a small amount of overflow is observed. Similarly, for
15 can be written as:
water flows with Frb less than unity, no overflow occurs. However,
F peak 2α 2 H in contrast to dry sand, water flow exhibits a jet-like run-up
¼1þ Fr b ð16Þ mechanism. Figure 3b shows water flow impacting against a bar-
F static k ho
rier and a jet-like run-up mechanism whereby water jumps verti-
cally along the barrier height (Fig. 3b, t = 0.25 s). Once the water
where F static ¼ 0:5kρgh2o w. Equation 16 describes the general
jumps to its maximum height, it rolls back towards the flume bed
form of the impact model in terms of Frb, which accounts for the
and no overflow is observed (Fig. 3b, t = 0.50 s). A comparison
barrier height on the impact force when overflow occurs. By
between dry sand and water flow shows that for Frb less than unity,
assuming that the linear momentum is not conserved (Ng et al.
no overflow occurs. Therefore, designers can use the Frb to assess
2019b), α is unity. Besides, the earth pressure coefficient k can have
whether overflow or overspill will occur. This is especially impor-
values greater than unity for frictional flows due to flow being in
tant for congested cities where the consequence of overspill can be
the passive state during impact (Faug 2020). Whereas for two-
serious. To mitigate overspill, deflectors can be installed (Choi
phase debris flows that are fluidised due to sustained pore pres-
et al. 2016).
sure in the flow, it can be assumed to have isotropic stress distri-
bution, thus k = 1 (Kwan 2012; Ashwood and Hungr 2016; Iverson
Observed impact kinematics for flow with Frb >1
et al. 2016). Therefore, Eq. 16 can be rewritten as:
Figure 4 shows a comparison between dry sand and water flows
F peak H impacting a rigid barrier. The barrier Froude number Frb for dry
¼ 1 þ 2 Fr 2b ð17Þ
F static ho sand and water flow is 2.4 and 3.0, respectively (Table 1). The
barrier height for both water and dry sand flows is 100 mm. The
In Eq. 17, Frb also accounts for the drag force exerted on flume inclination for the dry sand and water flows is 45° and 15°,
barriers when overflow occurs. The applicability of Eq. 17 is eval- respectively. The flow direction is from right to left for each
uated using the physical experimental results from this study and snapshot. Figure 4a shows dry sand impacting against a barrier.
with other data published in the literature. In a similar manner as that observed for dry sand in Fig. 3a, a dead
zone forms (Fig. 4a, t = 0.25 s). The dead zone increases in size as
Interpretation of results the deposited material propagates back upslope as a bore. Subse-
quent sand flow layers on top of the dead zone until overflow are
Observed impact kinematics for flow with Frb <1 observed (Fig. 4a, t = 0.50 s). Overflow occurs when the barrier
Figure 3 shows the images captured from the high-speed camera potential is insufficient to arrest the kinetic energy of the flow (i.e.
and the corresponding particle image velocimetry (PIV) analysis Frb >1). Overflow launches from the crest of the barrier and lands
(White et al. 2003). The frontal velocities deduced from PIV on the flume before flowing into the second barrier. The trajectory
analysis are up to 30% lower compared to the velocities estimated equation proposed by Kwan et al. (2015) based on the assumption
using high-speed camera. The discrepancy is attributed to the poor that overflow is a point mass can be used to estimate the landing
texture of images due to the reflection of the lighting off the sand distance. The landing distance based on the trajectory equation is
grains and water in water flows (Take 2015; Chen et al. 2017). 0.9 m. By contrast, the actual landing distance is 0.6 m and the
Therefore, the velocity fields generated from PIV analysis only second barrier is 1.0 m downstream from the first barrier. Further-
reveals the impact kinematics in this study qualitatively. For dry more, dry sand deposits in front of the second barrier and a
sand, the barrier height and flume inclination were 260 mm and granular bore propagate back upstream towards the first barrier.
26°, respectively. For water flows, a barrier height of 260 mm and a The final deposited profile (Fig. 4a, t = 6.5 s) shows that a spacing
flume inclination of 0° were used. The barrier height and flume of 1 m between barriers is inadequate as the dead zone of the
inclination were adequate to obtain a barrier Froude number less second barrier extends upstream and interacts with the first bar-
rier. This means the volume retention capacity of the second

Landslides
t = 0.00 s Flow direction t = 0.00
s
Rigid barrier

t = 0.50 s t = 0.25 s

Dead zone

t = 1.00 s Granular bore t = 0.50 s

Dead zone

t = 1.50 s Granular bore t = 0.75 s

Dead zone

(a) (b)
Fig. 3 Interaction mechanism of rigid barrier with Frb < 1 for a dry sand; b water

barrier is compromised by the close position of the first barrier. Similar to Fig. 3b, water flow exhibits a jet-like run-up (Fig. 4b, t
Therefore, to ensure that the flow lands on the channel bed before = 0.25 s). Part of the jet rolls back towards the flume, while the rest
impacting the second barrier, the barrier spacing L should be of the jet overspills the barrier (Fig. 4b. t = 0.5 s). Overspill
greater than the overflow distance xi (Kwan et al. 2015) and greater eventually becomes overflow, which lands in between the two
than the Lmin reported by VanDine (1996). When Lmin > L > xi, barriers at an inclined distance of about 0.5 m from the first
overflow lands on the channel before the second barrier, but the barrier. The calculated landing distance using the trajectory equa-
volume retention capacity of the second barrier is underutilised tion (Eq. 12) is 0.74 m. Similar to dry sand flow in Fig. 4a, the final
because of the close proximity of the second barrier to the first deposition profile shows that the spacing between barriers is
barrier. In contrast, when xi > L > Lmin, the retention capacity of inadequate. The minimum specified spacing recommended by
the second barrier is fully utilised but overflow may directly Eq. 1 is Lmin ≈ 1.2 m, which is larger than the actual barrier spacing
impact the second barrier or even overtop the second barrier of 1.0 m adopted in the experiments.
without landing on the channel first. Therefore, for a conservative A comparison of the observed kinematics between flows with
design, barrier spacing should be greater than both, xi and Lmin. Frb higher than and lower than unity reveals that the Frb should be

Landslides
Technical Note
Flow direction t = 0.00 s

Rigid barrier

t = 0.25 s

t = 0.50 s
Dead zone

t = 1.50 s
Dead zone

t = 6.50 s t = 3.00 s

15°±1°
15°

(a) (b)
Fig. 4 Interaction mechanism of rigid barriers with Frb > 1 for a dry sand; b water

used to assess whether overspill and overflow will occur. More- of the comparison between measured and calculated normalised
over, for a conservative spacing design, barrier spacing should be peak force. All the reported data in Fig. 5a falls within a wide range
larger than both xi and Lmin. of Froude numbers from 0 to 9. Froude numbers differ according
to geomorphological settings involved. For example, in Hong
Verification of the proposed impact model Kong, debris flows travel on steep terrain of up to 40° with high
Figure 5a shows the effects of Froude number Fr and flow type on velocity over short travel distance but with relatively small volume
the normalised impact force Fpeak/Fstatic, which is the peak impact in the order of about 500 m3 or less. In contrast to areas like Hong
force normalised by the theoretical static force Fstatic given by Kong with steep mountains, the Froude conditions tend to be
0.5kρgho2w (Armanini and Scotton 1993; Armanini 2009) acting lower (Fr < 3) in the Alps, that have gentle inclinations and larger
on the first rigid barrier. The measured normalised impact force flow volumes (Hübl et al. 2009). Two theoretical bounding refer-
resulting from dry granular flow (this study; Zanuttigh and ence lines from Eq. 15 are shown. The first line adopts k = 1 (no
Lamberti 2006; Caccamo et al. 2012), water (this study), and two- internal shear strength) and α = 1 (inelastic collision). In Eq. 15
phase flow experiment conducted in a 28-m-long flume (Liu 2019) when k = 1 (no internal shear strength) and α = 1.5, the equation
is compared with the theoretical normalised peak force (Eq. 15). provides an upper bound for the measured data from this study
The data from 28-m-long flume is shown to increase the reliability and from other experimental results reported in the literature.

Landslides
This upper bound equation can be used to provide conservative sand that is about 14 mm diameter, while the large particle diam-
design impact loads on a single rigid barrier. It is worthwhile to eter was varied as 224, 492, and 874 mm in three separate tests.
mention that the sand flows in this study exhibited lower impact Song et al. (2018b) also reported mono-disperse granular flows
forces compared to the water flows in this study. This may be with particle diameters of 67, 224, 492 and 874 mm. Table 2
attributed to a higher degree of bulk compressibility and internal summarises the experimental data from Ng et al. (2019a) and Song
energy dissipation in dry granular flows leading to higher energy et al. (2018b). The proposed upper bound equation for normalised
dissipation during the impact process (Choi et al. 2015a). Similarly, debris flow impact force Fpeak/Fstatic on a single rigid barrier, using
the dry granular flows reported in the literature (Zanuttigh and Eq. 15 and adopting k = 1 (no internal shear strength) and α = 1.5
Lamberti 2006; Caccamo et al. 2012) for higher Froude numbers (Fig. 4a), is shown in Fig. 6. The proposed equation provides
(Fr > 4) also tend to exhibit lower impact force compared to two- conservative estimates of the normalised impact forces for all the
phase debris flows (Liu 2019). mono- and bi-disperse flows reported by Song et al. (2018b) except
The different impact mechanisms between dry granular flow for the 2 cases including the largest prototype particle diameter of
and water flow necessitate the use of different flow height ratios. 874 mm. In contrast, all of the bi-disperse bouldery flows in Ng
For the tests using dry granular flow, a progressive pileup mech- et al. (2019a) have 874 mm diameter particles. Thus, the normal-
anism (Song et al. 2017) was observed. Thus, the peak impact load ised impact forces for these data points are not captured by the
occurs when the barrier has been filled to its crest and overflow upper bound equation. Therefore, a comparison of these measured
induces a drag load on the barrier. For dry granular flow, a flow and estimated normalised impact force values show that the dis-
height ratio H/ho of unity is appropriate for characterising the crete nature of flows in the presence of large particles tends to
peak impact load using Eq. 17. For tests using water flows, a generate sharp impulses leading to peak impact loads (see Fig. 4 of
vertical jet impact mechanism occurs (Armanini et al. 2020). Thus, Song et al. 2018b and Fig. 5 of Ng et al. 2019a) that cannot be
the peak load occurs from the impact of the flow front before the captured using the proposed equation as it is based on continuum
barrier is filled to its crest. To characterise the peak impact load for mechanics.
water, the flow height ratio H/ho should be greater than unity. The As for the estimation of the impact force exerted by bouldery
flow height ratio may even depend on the ascent of the vertical jet. flows on a rigid barrier, different approaches have been proposed
However, due to the turbulent impact kinematics for water flows, in the literature (SWCB 2005; Kwan 2012). The measured impact
the flow depth correlating to the peak impact load could not be loads reported by Song et al. (2018b) and Ng et al. (2019a), for
easily characterised. Therefore, only the data for dry granular flows with large boulders with a diameter δL of 874 mm, are
flows with overflow is shown in Fig. 5b. The barrier Froude num- underestimated by Eq. 15 in Fig. 6. The impact load from these
ber is shown on abscissa, while the normalised impact force Fpeak/ flows is replotted on Fig. 7, where the ordinate shows the normal-
Fstatic acting on the first rigid barrier is shown on the ordinate. The ised peak force Fpeak/Fstatic, and the abscissa shows the Froude
impact force estimated using Eq. 17 with H/ho = 1 is also compared number. The calculated impact force using the approach proposed
with the measured results for the dry granular flow tests. A com- by Kwan (2012) is based on the superposition of Eq. 3 using α = 2.5
parison of measured impact force with the predicted force reveals and Eq. 9 using Kc = 0.1. According to SWCB (2005), the design
that Eq. 17 predicts the trend reasonably for all the reported dry load for hard inclusions is calculated by using Eq. 9 with a load
granular flows. However, some data points are underestimated by reduction factor Kc of 0.2 and 0.5. A comparison of the measured
Eq. 17. The underestimation is probably because complexities such and calculated peak loads shows that the approach by Kwan (2012)
as stress state during impact (Faug 2020), basal friction may not always yield a conservative estimate of the peak load.
(Ahmadipur et al. 2019), and discontinuities such as jump in the Whereas, the approach proposed by SWCB (2005) with a Kc of 0.5
flow-barrier interaction are not captured with simple model. provides an upper bound estimate of the load for all flows
Therefore, an upper bound equation for the normalised impact compared.
load 6 + 2Frb2 may be used to obtain a conservative estimate of the
impact load acting on the first rigid barrier. Influence of first barrier on second barrier impact force
To further analyse the applicability of the proposed impact Whether overflow occurs or not depends on the remaining kinetic
model in Fig. 5a, for bouldery flows, the measured normalised energy of flow after the debris has reached the crest of the barrier.
impact force Fpeak/Fstatic induced by bouldery flows on a single Figure 8 shows the influence of the barrier Froude number of the
rigid barrier without overflow from literature is compared. Figure 6 first barrier Frb_1 (addressed as Frb in previous sections) on the
shows the effects of Froude number Fr on the impact force nor- impact force exerted on the second barrier. The abscissa shows
malised by the theoretical static load Fpeak/Fstatic for bouldery Frb_1 and ordinate shows the relative impact load between the first
flows. The reported experimental data are from centrifuge tests and second barrier F2/F1. The measured impact load for dry sand
conducted at 22.4g. The flow reported by Ng et al. (2019a) consists and water flows in this study is compared. F2/F1 increases with
of a bi-disperse mixture of two different particle diameters of glass Frb_1 for both dry sand and water flows. The increase in F2/F1 with
spheres. The mixture consists of 70% small particles δS and 30% increasing Frb_1 is due to the sediments having less energy dissi-
large particles δL by volume. The diameter of the large particles is pation when flow overtops the barrier because of the lower barrier
874 mm, while the diameter of the small particles was varied as 67, potential relative to flow inertia (Frb_1 > 1). However, when Frb_1 <
224, and 492 mm in three separate tests. The particle diameters are 1, almost no overflow occurs (Fig. 3). This means that no impact
presented in prototype scale. Similarly, a mixture with 70% small load is measured on the second barrier. Furthermore, when Frb_1
particles and 30% large particles was used by Song et al. (2018b) in becomes greater than 1.6 for dry sand and 2.5 for water flows, the
their study of the impact behaviour of bi-disperse flows. However, measured load on the second barrier becomes even higher than
in their flows, the finer particles consisted of Leighton Buzzard that of the first barrier. This indicates that designing the multiple-

Landslides
Technical Note
120 120
Eqn. 15

Normalised impact force: Fpeak/Fstatic


Normalis ed impact force Fpeak/Fstatic

100 100

80 80
Eqn. 17

60 60

40 40

20 20

0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5
Froude number Fr = /(gho)0.5 Barrier Froude number Frb = /(gH)0.5
Water (this work)
Sand (this work) Sand (this work)
Dry granular (Caccamo et al. 2012) Dry granular (Caccamo et al. 2012)
Dry granular (Zanuttigh and Lamberti 2006) Calculated using Eqn. 17; H/ho = 1
Series14
Two-phase (Liu 2019) Calculated using Fpeak/Fstatic = 6+2Frb2
Fp/Fs
Calculated using Eqn.
Calculated using Eqn.

(a) (b)
Fig. 5 Comparison of the influence of a Froude number Fr and b barrier Froude Frb number on normalised impact force (Fpeak/Fstatic) acting on the first barrier

barrier system with the same barrier height is not an effective These relationships can be given as F2/F1 ≈ 1.25Frb_1 for dry sand
solution. Therefore, designs should be optimised by using differ- and F2/F1 ≈ 0.85Frb_1 for water flows. In principle, debris flow,
ent barrier heights. The ratio F2/F1 follows a linear trend with Frb_1. which is a mixture of solid and fluid, should lie in somewhere
between these two idealised flow types.
140 Figure 9 shows the effects of the barrier Froude number of the
first barrier Frb_1 on the load induced on the second barrier
L= 874 mm relative to the theoretical static force F2/Fstatic. The velocity and
120
Normalised impact force: Fpeak /Fstatic

flow depth at the location of the second barrier is obtained from


L= 874 mm control tests without barriers installed in the flume. The F2/Fstatic
100 Eqn. 15 generally increases with Frb_1 for both water and dry sand flows.
Evidently, for similar Frb_1, the F2/Fstatic for dry sand flow is less
80 than that of water flow. In fact, the peak normalised impact forces
L= 874 mm
for water are up to three times higher than dry sand flow for the
60 same first barrier Froude number because of the different flow
properties and impact mechanisms. Dry sand flow generally re-
40 L= 874 mm L= 492 mm sults in a pileup mechanism (Choi et al. 2015a; Koo et al. 2017a),
L= 224 mm while water flow results in a vertical jet mechanism (Choi et al.
L= 874 mm 2015a). Higher bulk compressibility and internal shear strength of
20 L= 224 mm
dry sand flow compared to water promote energy dissipation via
L= 67 mm
L= 492 mm shearing among grains during impact resulting in a pileup. On the
0 contrary, water flows are nearly incompressible and have lower
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
internal dissipation due to low viscosity that leads to a vertical jet-
Froude number Fr = /(gho)0.5
like run-up at impact (Choi et al. 2015a). F2/Fstatic is close to zero
Measured (bouldery flows; Ng et al. 2019b) when Frb_1 <1 because no overflow occurs when Frb_1 <1 (Fig. 3).
Measured (bouldery flows; Song et al. 2018b) Therefore, no load is recorded on the second barrier. Based on the
Calculated using Eqn. 15; measured results of the two extreme flow types, bilinear design
envelopes are suggested. Designers may use these envelopes to
Fig. 6 Comparison of normalised impact force (Fpeak/Fstatic) acting on the first estimate the impact force on the second barrier by knowing the
barrier for bouldery flows barrier Froude number of the first barrier. The distinction of a
separate design envelope for water flow and dry sand flow

Landslides
Table 2 Flow parameters of bouldery flows reported in the literature (dimensions in prototype)
Reference Small particle, δS Large particle, δL Velocity, Flow depth, Froude number, Flow type
(mm) (mm) ν (m/s) ho (m) Fr (-)
Ng et al. 67 874 12.7 1.21 3.87 Bi-disperse
(2019a)
224 874 9.6 1.79 2.41
492 874 8.6 2.42 1.85
Song et al. 14 224 11.4 0.59 5.00
(2018b)
14 492 9.1 0.66 3.75
14 874 8.7 0.87 3.13
- 67 14.2 1.03 4.69 Mono-disperse
- 224 21.9 1.19 6.74
- 492 9.4 1.34 2.72
- 874 9.1 0.87 3.27

represents the different physical interaction mechanisms in the (a) Dynamic impact coefficient of α = 1.5 and static pressure
two flow types. For water flows, the F2/Fstatic = 20Frb_1 – 1.0 when coefficient of k = 1 are recommended for providing conser-
Frb_1 <2.2 and F2/Fstatic = 30 when Frb_1 >2.2. For dry sand flows, vative impact forces exerted by dry sand, water, and two-
the F2/Fstatic = 8.2Frb_1– 6.9 when Frb_1 <1.85 and F2/Fstatic = 9 when phase debris flows on the first rigid barrier of a dual rigid
Frb_1 >1.85. barrier system, but cannot capture the impact load induced
by large inclusions.
Conclusions (b) Based on the observed impact kinematics, the barrier Froude
In this study, a new impact equation that considers the static load number is an appropriate dimensionless number to discern
as a function of the barrier height by using the barrier Froude whether overflow or overspill will occur. However, for
number Frb is proposed. This newly proposed equation is verified characterising the impact dynamics, only dry granular flow
by physical flume experiments carried out to investigate the influ- can be bounded for the first barrier using Fpeak/Fstatic = 6 +
ence of the barrier height of the first barrier on the impact dy- 2Frb2 when H/ho = 1. The impact dynamics using the barrier
namics on the second barrier. Two idealised flow types, dry sand Froude number could not be easily characterised for water
and water, were used to represent extreme cases of frictional and flows since H/ho could not be readily assessed from the
viscous flows. Key findings may be drawn as follows: kinematics for a turbulent vertical jet mechanism.

250
kc+alpha
Eqn. 3 and 9 (Kc = 0.1; = 2.5; Kwan 2012)
2.0
Measured second barrier force / measured

Eqn. in
F(kc) 9 kN
(Kc = 0.2; SWCB 2005) Water (this work)
Normalised impact force: Fpeak/Fstatic

Eqn. in
F(kc) 9 kN
(Kc = 0.5; SWCB 2005)
Sand (this work)
Measured
Measured
200 1.6 F 2/F1 ==11
FD/FU
first barrier force: F2/F1

150 1.2

100 0.8 F2/F1 = 1.25Frb _1 - 1.2

F2/F1 = 0.85Frb _1 - 1.0


50 0.4

0 0.0
1 2 3 4 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2
Froude number Fr = /(gho)0.5 First barrier Froude number Frb_1

Fig. 7 Comparison of measured and calculated impact forces of bouldery flows Fig. 8 Influence of first barrier Froude number Frb_1 on measured impact force at
with 874 mm diameter particles (dimensions in prototype) the second barrier

Landslides
Technical Note
35 Armanini A (2009) Discussion on: Experimental analysis of the impact of dry avalanches
Water (this work) F2/Fstatic = 30 on structures and implication for debris flow (Zanuttigh Lamberti). J Hydraul Res
Measured second barrier force / theoretical second
for Frb_1 > 2.2 47:381–383
30 F2/Fstatic = 20Frb_1 - 1.0
Armanini A, Scotton P (1993) On the dynamic impact of a debris flow on structures. In:
for 0.75 ≤ Frb_1 ≤ 2.2
Proceedings of XXV IAHR Congress Tokyo (Tech Sess B III), pp 203–210
25 Sand (this work) Armanini A, Larcher M, Odorizzi M (2011) Dynamic impact of a debris flow front against
a vertical wall. In: Genevois R, Hamilton DL, Prestininzi A (eds) Padua ItalyIn
F2/Fstatic = 8.2Frb_1 - 6.9
barrier force: F2/Fstatic

proceeding of 5th international conference on debris-flow hazards Mitigation me-


for 0.75 ≤ Frb_1 ≤ 1.85
20 chanics prediction and assessment. Casa Editrice Università La Sapienza, Rome Italy,
pp 1041–1049
Armanini A, Rossi G, Larcher M (2020) Dynamic impact of a water and sediments surge
15 against a rigid wall. J Hydraul Res 58:314–325
Ashwood W, Hungr O (2016) Estimating total resisting force in flexible barrier impacted
F2/Fstatic = 9 by a granular avalanche using physical and numerical modeling. Can Geotech J
10 for Frb_1 > 1.85
53(10):1700–1717
Caccamo P, Chanut B, Faug T, Bellot H, Bouvet FN (2012) Small-scale tests to investigate
5 the dynamics of finite-sized dry granular avalanches and forces on a wall-like
obstacle. Granul Matter 14:577–587
CGS (China Geological Survey) (2004) Design code for debris flow disaster mitigation
0 measures (DZ/T0239-2004). CGS, Beijing (in Chinese)
0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 Chen Z, Li K, Omidvar M, Iskander M (2017) Guidelines of DIC in geotechnical
First barrier Froude number Frb_1 engineering research. Int J Phys Model Geotech 17(1):13–22
Choi CE, Au-Yeung SCH, Ng CWW, Song D (2015a) Flume investigation of landslide
Fig. 9 Proposed design envelop for the second barrier force in the presence of first granular debris and water run-up mechanisms. Géotech Lett 5(1):28–32
Choi CE, Ng CWW, Au-Yeung SCH, Goodwin GR (2015b) Froude characteristics of both
barrier
dense granular and water flows in flume modelling. Landslides 12(6):1197–1206
Choi CE, Ng CWW, Goodwin GR, Liu LHD, Cheung WW (2016) Flume investigation of the
influence of rigid barrier deflector angle on dry granular overflow mechanisms. Can
Geotech J 53(10):1751–1759
(c) The impact force on the second barrier increases linearly with Cui P, Zeng C, Lei Y (2015) Experimental analysis on the impact force of viscous debris
an increasing barrier Froude number Frb_1 for the first barri- flow. Earth Surf Process Landf 40:1644–1655
er. When Frb_1 >1, the barrier potential is less than the flow Cui Y, Cheng D, Choi CE, Jin W, Lei Y, Kargel JS (2019) The cost of rapid and haphazard
kinetic energy. This implies that barriers in dual rigid barrier urbanisation: lessons learned from the Freetown landslide disaster. Landslides
systems should not be designed to have the same barrier 16(6):1167–1176
Faug T (2015) Depth-averaged analytic solutions for free-surface granular flows
height; rather they should be designed based on the Frb_1.
impacting rigid walls down inclines. Phys Rev E 92:062310
(d) Based on the measured results of frictional and viscous flows, Faug T (2020) Impact force of granular flows on walls normal to the bottom: slow versus fast
bilinear design envelopes for the impact load on the second impact dynamics. Can Geotech J 58:114–124. https://doi.org/10.1139/cgj-2019-0399
barrier based on the Frb_1 of the first rigid barrier are pro- Faug T, Lachamp P, Naaim M (2002) Experimental investigation on steady granular flows
posed. Engineers may leverage the proposed envelopes to interacting with an obstacle down an inclined channel: study of the dead zone
upstream from the obstacle Application to interaction between dense snow ava-
estimate the design force on the second barrier. The proposed
lanches and defence structures. Nat Hazards Earth Syst Sci 2:187–191
envelopes are for viscous flows, the F2/Fstatic = 20Frb_1 – 1.0 Faug T, Chanut B, Naaim M, Perrin B (2008) Avalanches overflowing a dam: dead zone
when Frb_1 <2.2 and F2/Fstatic = 30 when Frb_1 >2.2; and for granular bore and run-out shortening. Ann Glaciol 49:77–82
frictional flows, the F2/Fstatic = 8.2Frb_1 – 6.9 when Frb_1 <1.85 Froude MJ, Petley DN (2018) Global fatal landslide occurrence from 2004 to 2016. Nat
and F2/Fstatic = 9 when Frb_1 >1.85.> Hazards Earth Syst Sci 18:2161–2181
Hákonardóttir KM, Hogg AJ, Batey J (2003a) Flying avalanches. Geophys Res Lett
30(23):2191
Hákonardóttir KM, Hogg AJ, Jóhannesson T, Kern M, Tiefenbacher F (2003b) Large-scale
avalanche braking mound and catching dam experiments with snow: a study of the
Funding airborne jet. Surv Geophys 24:543–554
The work described in this paper is supported by a grant from the Hübl J, Suda J, Proske D, Kaitna R, Scheidl C (2009) Debris flow impact estimation. In: In
Research Grants Council of the Hong Kong Special Administrative The Proceedings of the 11th International Symposium on Water Management and
Region, China (Project Nos. AoE/E-603/18, T22-603/15N, 16212618, Hydraulic Engineering Ohrid Macedonia, pp 1–5
Hübl J, Nagll G, Suda J, Rudolf-Miklau F (2017) Standardised stress model for design of
16209717, and 16210219). The authors also received financial
torrential barriers under impact by debris flow (according to Austrian standard
sponsorship from the National Natural Science Foundation of regulation 24801). Int J Eros Control Eng 10(1):47–55
China (51709052) and the financial support from Higher Education Hungr O (2008) Simplified models of spreading flow of dry granular material. Can
Commission of Pakistan. Author W. A. R. K. De Silva received the Geotech J 45(8):1156–1168
support of Hong Kong PhD Fellowship Scheme (HKPFS) provided Hungr O, Morgan GC, Kellerhals R (1984) Quantitative analysis of debris torrent hazards
for design of remedial measures. Can Geotech J 21(4):663–677
by the RGC of HKSAR.
Hungr O, Leroueil S, Picarelli L (2014) The Varnes classification of landslide types, an
update. Landslides 11(2):167–194
Iverson RM (1997) The physics of debris flows. Rev Geophys 35(3):245–296
References Iverson RM (2015) Scaling and design of landslide and debris-flow experiments.
Geomorphology 244:9–20
Ahmadipur A, Qiu T, Sheikh B (2019) Investigation of basal friction effects on impact Iverson RM, Logan M, Denlinger RP (2004) Granular avalanches across irregular three-
force from granular sliding mass to a rigid obstruction. Landslides 16:1089–1105 dimensional terrain: 2 Experimental tests. J Geophys Res Earth Surf 109(F1):e085
Armanini A (1997) On the dynamic impact of debris flows. In: Recent Developments on Iverson RM, George DL, Logan M (2016) Debris flow run-up on vertical barriers and
Debris Flows Lecture notes in Earth Sciences, vol 64. Springer Verlag, Berlin, pp 208–226 adverse slopes. J Geophys Res Earth Surf 121(12):2333–2357

Landslides
Koo RCH, Kwan JSH, Ng CWW, Lam C, Choi CE, Song D, Pun WK (2017a) Velocity Saingier G, Deboeuf S, Lagrée PY (2016) On the front shape of an inertial granular flow
attenuation of debris flows and a new momentum-based load model for rigid down a rough incline. Phys Fluids 28(5):053302
barriers. Landslides 14:617–629 Scheidl C, Chiari M, Kaitna R, Mulleger M, Krawtschuk A, Zimmermann T, Proske D (2013)
Koo RCH, Kwan JSH, Lam C, Ng CWW, Yiu J, Choi CE, Ng AKL, Ho KKS, Pun WK (2017b) Analysing debris-flow impact models based on a small scale modelling approach. Surv
Dynamic response of flexible rockfall barriers under different loading geometries. Geophys 34:121–140
Landslides 14(3):905–916 Scotton P, Deganutti AM (1997) Phreatic line and dynamic impact in laboratory debris
Koo RCH, Kwan JSH, Lam C, Goodwin GR, Choi CE, Ng CWW, Pun WK (2018) Back- flow experiments. In: In Proceeding of 1st International Conference on Debris Flow
analysis of geophysical flows using three-dimensional runout model. Can Geotech J Hazards Mit Mech Predic & Assess San Francisco USA
55(8):1081–1094 Song D, Ng CWW, Choi CE, Zhou GGD, Kwan JSH, Koo RCH (2017) Influence of debris
Kwan JSH (2012) Supplementary technical guidance on design of rigid debris-resisting flow solid fraction on rigid barrier impact. Can Geotech J 54(10):1421–1434
barriers. GEO Report No. 270, Geotechnical Engineering Office, Civil Engineering and Song D, Choi CE, Ng CWW, Zhou GGD (2018a) Geophysical flows impacting a flexible
Development Department, Hong Kong SAR Government barrier: effect of solid-fluid interaction. Landslides 15(1):99–110
Kwan JSH, Koo RCH (2015) Preliminary back analysis of open hillside landslide impacting Song D, Choi CE, Zhou GGD, Kwan JSH, Sze HY (2018b) Impulse load characteristics of
on a flexible rockfall barrier at Jordan Valley. Report No. 308, Geotechnical Engineer- bouldery debris flow impact. Géotech Lett 8(2):111–117
ing Office, Civil Engineering and Development Department, Hong Kong SAR SWCB (2005) Manual of Soil and Water Conservation. Soil and Water Conservation
Government Bureau, Taiwan 692 p (in Chinese)
Kwan JSH, Koo RCH, Ng CWW (2015) Landslide mobility analysis for design of multiple Takahashi T (2014) Debris flow: mechanics prediction and countermeasures, 2nd edn.
debris-resisting barriers. Can Geotech J 52(9):1345–1359 Taylor & Francis Group, London UK
Lam C, Yong ACY, Kwan JSH, Lam NTK (2018) Overturning stability of L-shaped rigid Take WA (2015) Thirty-Sixth Canadian Geotechnical Colloquium: advances in visualisation
barriers subjected to rockfall impacts. Landslides 15(7):1347–1357 of geotechnical processes through digital image correlation. Can Geotech J
Liu H (2019) Impact mechanisms of debris flow against multiple rigid barriers with basal 52(9):1199–1220. https://doi.org/10.1139/cgj-2014-0080
clearance. Dissertation. Hong Kong University of Science and Technology, Hong Kong Vagnon F (2020) Design of active debris flow mitigation measures: a comprehensive
SAR analysis of existing impact models. Landslides 17(2):313–333
Lo DOK (2000) Review of natural terrain landslide debris-resisting barrier design. GEO VanDine DF (1996) Debris flow control structures for forest engineering. Ministry of
Report No. 104, Civil Engineering and Development Department, Hong Kong SAR Forests Research Program Working Paper 22/1996, Vancouver Canada, Government of
Government the Province of British Columbia
Ng CWW, Choi CE, Kwan JSH, Shiu HYK, Ho KKS, Koo RCH (2012) Flume modelling of Wendeler C (2016) Debris-flow protection systems for mountain torrents. In: Steffen K
debris flow resisting baffles. In: In proceeding of One-Day Seminar on Natural Terrain (ed) WSL Berichte Swiss Federal Institute for Forest Snow and Landscape Research
Hazard Mitigation Measures Hong Kong 16 October 2012 WSL CH-8903. Birmensdorf Swiss Federal Institute for Forest Snow and Landscape
Ng CWW, Choi CE, Law RP (2013) Longitudinal spreading of granular flow in trapezoidal Research WSL
channels. Geomorphology 194:84–93 White DJ, Take WA, Bolton MD (2003) Soil deformation measurement using particle
Ng CWW, Choi CE, Liu LHD, Wang Y, Song D, Yang N (2017) Influence of particle size on image velocimetry (PIV) and photogrammetry. Géotechnique 53(7):619–631
the mechanism of dry granular run-up on a rigid barrier. Géotech Lett 7(1):79–89 Zanuttigh B, Lamberti A (2006) Experimental analysis of the impact of dry avalanches on
Ng CWW, Choi CE, Koo RCH, Goodwin GR, Song D, Kwan JSH (2018) Dry granular flow structures and implication for debris flows. J Hydraul Res 47(4):381–383
interaction with dual-barrier systems. Géotechnique 68(5):386–399
Ng CWW, Choi CE, Cheung DKH, Cui Y (2019a) Effects of dynamic fragmentation on the
C. W. W. Ng : U. Majeed : W. A. R. K. De Silva
impact force exerted on rigid barrier: centrifuge modelling. Can Geotech J
56(9):1215–1224
Department of Civil and Environmental Engineering,
Ng CWW, Choi CE, Majeed U, Poudyal S, De Silva WARK (2019b) Fundamental framework
The Hong Kong University of Science and Technology,
to design multiple rigid barriers for resisting debris flows. In: In proceeding of the
Hong Kong, SAR, China
16th Asian Regional Conference on Soil Mechanics and Geotechnical Engineering
October 14–18 Taipei Taiwan
C. E. Choi ())
NILIM (2007) Manual of Technical Standard for Establishing Sabo Master Plan for Debris
Department of Civil Engineering,
Flow and Driftwood Technical Note of NILIM No 364. Natural Institute for Land and
University of Hong Kong,
Infrastructure Management Ministry of Land Infrastructure and Transport Japan 18 p
Hong Kong, SAR, China
(in Japanese)
Email: cechoi@hku.hk
Osti R, Egashira S (2008) Method to improve the mitigative effectiveness of series of
check dams against debris flows. Hydrol Process 22:4986–4996

Landslides

You might also like