water

Article
Riverbank Stability Assessment under River Water
Level Changes and Hydraulic Erosion
Toan Duong Thi * and Duc Do Minh
VNU University of Science, Vietnam National University, Hanoi, Hanoi 100000, Vietnam; ducdm@vnu.edu.vn
* Correspondence: duongtoan@hus.edu.vn; Tel.: +84-093-454-3261




Received: 5 October 2019; Accepted: 7 December 2019; Published: 10 December 2019


Abstract: The dominant mechanism of riverbank cantilever failure is soil erosion of the bank toe and
near bank zone. This paper demonstrates that the shape of the riverbank cantilever failure depends
on the properties of the soil and the fluctuation of the river water level (RWL). With a stable RWL,
a riverbank with higher resistance force leads to failure with larger and deeper overhang erosion
width. When RWL rises, a less cohesive soil bank will be eroded over a larger width and riverbank
failure will occur earlier. With a low rate of rising RWL, riverbank failure may happen in a type of
mass failure. With a high rate of rising RWL, a riverbank will fail in a type of overhang riverbank
failure, with the soil erosion rate being the main affected factor.

Keywords: water shear stress; soil erosion; riverbank cantilever failure

1. Introduction
Riverbank failure occurs by multiple factors, especially during a rainy season, such as the
fluctuation of river water level (RWL), groundwater, pore water pressure, soil strength (soil suction
and shear strength), and soil erosion. The two main mechanics of riverbank failure mentioned in most
previous research are mass failure and cantilever failure [1]. The role of the related factors in mass
failure mechanics have been studied quite clearly in [2–9], which found that the relationship between
the soil hydraulic and the rate of RWL change is the main factor controlling these factors as well as the
riverbank stability. While the cantilever failure mechanic is more complex, it often includes not only
change in pore water pressure and soil strength, but also the process of soil erosion in the toe of the
riverbank. To build an understanding of cantilever failure mechanics, a couple of the factors in the
mass failure and soil erosion will be analyzed in this paper.
Soil erosion is one of the main processes having a significant impact on riverbank stability and has
often been mentioned by hydraulic dynamic erosion [10–33], and seepage erosion [34–43]. Riverbank
failure deals with the soil fluvial erosion known as the cantilever type, which is a combination of
undercutting erosion and mass collapse of the upper part of the riverbank [10,17,18,20,22]. Soil erosion
occurs when the water shear stress caused by water flow around the interface of water and soil is
higher than the critical shear stress [44]. Shear stress is the pressure caused by water flow around the
interface of water and soil. In the rainy season, the river water level increases and flows with a high
shear stress, expanding the soil erosion width and creating an overhanging riverbank shape.
To understand the mechanism of the cantilever riverbank failure, previous research has attempted
to combine the processes of soil erosion and change of riverbank geometry into riverbank stability
analysis. The Department of Agriculture’s Agricultural Research Service developed the Bank-Stability
and Toe-Erosion Model (BSTEM), which can be used to estimate the erosion of banks and bank-toe
materials [10–16,22]. The GeoSlope program has been used to aid understanding of the mechanisms of
riverbank failure coupled with soil bank erosion and riverbank groundwater seepage [5,17,21]. Other
models have also been suggested for understanding and analysis of overhanging riverbank failure.

Water 2019, 11, 2598; doi:10.3390/w11122598 www.mdpi.com/journal/water

Water 2019, 11, 2598 2 of 20

Yong et al. (2014) [22] used the extensively verified and validated 2D depth-averaged mobile-bed
model SRH-2D and the BSTEM to predict channel adjustment and platform development. This was
considered the first study to accomplish a complete coupling of a multilayer cohesive bank model and
a multidimensional mobile-bed model. Patsinghasanee et al. (2015, 2017, 2018) [24,25,30] developed
a coupled model implementing a triple-grid approach from/to a Cartesian grid cell, consisting of a
coarse one-dimensional (1D) grid for the flow field in the lateral direction; a fine 1D grid for sediment
transport, and bed deformation in the lateral direction; and a 2D grid for cantilever failure in the
vertical and lateral directions. Abderrezzak (2016) [28] focused on the simulation of non-cohesive,
non-uniform bank material, implemented a simple bank failure in SISYPHE and the sediment transport
module of TELEMAC-MASCARET. From those researches it can be seen that fluvial erosion has
significant impact on riverbank stability. In addition, the mechanics of cantilever riverbank failure
give rise to complex problems, including instantaneous processes such as soil erosion, change of bank
geometry, groundwater level, and unsaturated soil properties. These processes all require numerical
analysis, for which previous models are not adequate. Analysis of the cantilever riverbank requires
further development.
Regarding the factors affecting cantilever riverbank failure, the physical properties of the soil,
particularly grain size and density, are the main factors affecting both the fluvial erosion rate and
riverbank stability [19,20,23–25,30]. Samadi et al. (2011) [19] performed the first experiment involving
the creation of an overhanging riverbank block to understand the failure mechanism of overhanging
blocks. The experimental results were compared with the results obtained from a numerically simulated
model [20]. By analyzing riverbank soils with different densities, Samadi et al. (2011, 2013) [19,20]
concluded that with increasing soil density, the undermining depth at which failure occurs will increase,
as will the width ratio of upper to lower overhanging blocks in failure conditions. Yu et al. (2015) [23]
showed different mechanics of riverbank failure owing to different soil materials. For cohesive soil,
as the sediment-moving incipient velocity is much greater than that of non-cohesive soil and its bed
erosion is weaker, bank failure occurs deeper, near the bank toe. The non-cohesive bank tends to
collapse near the water surface. Patsinghasanee et al. (2017, 2018) [24,25] recommended that the
physical and erosion properties of soil should continue to be investigated in further studies.
From an overview of the previous research focused on soil erosion and riverbank failure by water
flow stress, it may appear that soil properties are important factors impacting on soil erosion and
riverbank failure [19,20,23–25,30,43–56]. However, few studies have focused on the effects of soil
properties and erosion on riverbank cantilever failure. Moreover, analysis of riverbank erosion should
include the effects of the river water level (RWL), which changes during the flooding season due to
heavy rains and sudden recharge from upstream recharge. In most previous studies, the RWL has
been held constant, and the pore pressure and groundwater were not accounted for. The primary
objectives in this paper as follows. (i) Analysis of the effect of soil erosion on riverbank failure. Three
different kinds of bank soils were analyzed in cases both with erosion and without soil undercutting
erosion. (ii) Discussion of the mechanics of riverbank failure, including the effect of factors such as soil
property and river water level change. The discussion of these and previous results will improve our
understanding of riverbank failure mechanics.

2. Methods
Figure 1 shows the framework of the present study, which includes two main parts, namely,
a laboratory soil experiment and simulation models in the GeoSlope program. Laboratory soil
properties were soil erosion, saturated soil properties (i.e., physical properties, shear strength, and
hydraulic conductivity), and unsaturated soil property (soil suction).
Water 2019, 11, 2598 3 of 20
Water 2019, 11, x FOR PEER REVIEW 3 of 20

I. FIELD INVESTIGATION AND SOIL SAMPLE COLLECTION

Soil erosion test Saturated soil Unsaturated soil

properties properties

III. GEOSLOPE ANALYSIS

Case 1—Couple riverbank stability Case 2—Riverbank stability

analysis with RWL changes and analysis with RWL changes
soil erosion

Step 1: Build bank geometry, soil

Step 1: Build bank geometry, soil
layers and the hydraulic boundary
layers and the hydraulic boundary

Step 2: Analysis riverbank seepage Step 2: Analysis riverbank seepage

and stability with the initial condition and stability with the initial condition

Step 3: Rise river water level from (Li)

to (Li+1) Step 3: Rise river water level from (Li)
Rise RWL to (Lf)

Step 4: Calculate the shear stress;

determine soil erosion from Step 4: Calculate the FOS with
experiment graph; calculate erosion RWL rises from (Li) to (Lf)
distance

Step 5: Redraw the new bank line

with loss soil bank by change erosion
distance
Results
- The effects of RWL and soil
erosion to the riverbank stability
FOS > 1 Step 6: Calculate the FOS
- Discussion on mechanics of the
cantilever riverbank failure

The new bank line is

FOS < 1
the failure surface

FigureFigure
1. The 1. The research
research framework.
framework.

The physical
The physical properties,
properties, shear shear strength,
strength, and hydraulic
and hydraulic conductivity
conductivity were determined
were determined following
following
the standards
the standards of ASTM
of ASTM D 2216,DD 2216, D 2937-00,
2937-00, JGS 0524:
JGS 0524: 2000,JIS
2000, and and
A: JIS A: respectively.
1218, 1218, respectively. Soil suction
Soil suction
was determined
was determined using using a pressure
a pressure plate apparatus,
plate apparatus, as shown
as shown in [57].inA[57]. A description
description of theof the equipment
equipment
and method
and method used toused to determine
determine soilvelocity
soil erosion erosion velocity is
is provided provided
below. below. used
This research This the
research used the
GeoSlope
GeoSlope
program program
to analyze to analyze
riverbank riverbank
stability stability
in two cases. In in
thetwo
firstcases.
case,In the first case,
riverbank riverbank
stability stability was
was analyzed
analyzed
coupled coupled
with soil erosionwith
and soil erosion
change andwater
of river change of river
level. water
In this, level.soil
all three In this, all three
laboratory soil laboratory
experiment
experiment groups (soil erosion, and saturated and unsaturated soil properties) were used in the
analysis process. In the second case, riverbank stability was analyzed with only river water level
Water 2019, 11, 2598 4 of 20

groups (soil erosion, and saturated and unsaturated soil properties) were used in the analysis process.
In the second case, riverbank stability was analyzed with only river water level change, and without
soil bank erosion. The details of the soil erosion test and analysis model in the GeoSlope program are
described below.

2.1. Determining Soil Erosion Rate in Laboratory

Soil erosion, which in the bank toe is caused by river water flow, was determined in an open
pipe in the laboratory. This test was based on the erosion function apparatus invented by Briaud et
al. (1999, 2001a, 2001b, 2004, 2008) [43,53–56]. The equipment includes an open rectangular pipe 2 m
Water 2019, 11, x FOR PEER REVIEW 5 of 20
long, 0.2 m wide, and 0.1 m high (Figure 2). The water flowing through the pipe is supplied by a water
tank. The slope of (5)
Equations theand
water(6) pipe and
can be theto
used height of the
estimate thewater tank
friction can be
factor changed
without to control
having to usethe
thewater
Moody
velocity.
chart [58,62], although multiple iterations may be required to obtain convergence. In the this
A cylindrical mold attached to the bottom of the water pipe has a diameter of 7.6 cm; curve
is used to compact
relating soil erosionthe rates
soil sample. Thestress,
and water protrusion of soilshear
the critical in this mold(τcan
stress be manipulated by hand
c) is defined as the stress value
control in thesoil
at which bottom
beginsof the soil column.
to erode. The
Finally, theexperimental model measures
results are presented by thethe time needed
erosion curve asto erode
showna in
certain soil4,height
Figure whichbyprovides
water flowing through a simulated
the relationship between theopening channel.rate
soil erosion Theand
soil erosion
the waterrateflow
(mm/h)
shear
is thus defined
stress. as the ratio of the eroded height (mm) to the elapsed time (h).

Figure 2. Configuration of soil erosion test.

Figure 2. Configuration of soil erosion test.

In this analysis, a soil sample with predetermined water content and dry density was mounted
in the sample cylinder. During the test, the core soil was made to protrude in intervals of 1 mm
higher than the bottom of the water pipe (Figure 3). The velocity of the water flow was set at a
predetermined rate (m/s) and continued flowing until 1 mm of soil was completely eroded. The time it
took to erode 1 mm was recorded, and this represents the duration ∆t needed to erode the height of
the soil ∆h (mm). The results of this analysis represent the relationship between the erosion rate and
shear stress (Figure 4), in which the erosion rate and shear stress are calculated using the following
equations. The soil erosion rate (ε) is the ratio of the height of the eroded soil (∆h) to the elapsed time
(∆t) [53–56,58–62]:
Figure 3. Soil sample protruding 1 mm before the start of water flow.
ε = ∆h/∆t (1)
Water 2019, 11, 2598 5 of 20
Figure 2. Configuration of soil erosion test.
Figure 2. Configuration of soil erosion test.

Figure 3. Soil sample protruding 1 mm before the start of water flow.

Figure3.3.Soil
Figure Soilsample
sampleprotruding
protruding11mm
mmbefore
beforethe
thestart
startofofwater
waterflow.
flow.

Figure 4. Result curve: relationship between soil erosion rate and water shear stress.

Figure
The shear stress4.from
Resultthe
curve: relationship
water flow can between soil erosion
be calculated usingrate
theand waterchart:
Moody shear stress.
Figure 4. Result curve: relationship between soil erosion rate and water shear stress.
2.2. Riverbank Erosion Simulation ρ f V2
2.2. Riverbank Erosion Simulation τ= (2)
8
2.2.1. The Boundary Shear Stress Causing Soil Erosion
in2.2.1.
which The Boundary
τ is the shearShear
stressStress
(N/mCausing
2 or Pa), Soil
ρ is Erosion
the density of water, V is the average water velocity,
and f is a friction coefficient. Theoretically, f is obtained from the Moody diagram using the relative
roughness (roughness height/conduit diameter) and the Reynolds number, which is calculated as:

VD
Re = (3)
υ
where D is the hydraulic diameter of the pipe and υ is the kinematic viscosity of water. The hydraulic
diameter is calculated as:
2ab
D= (4)
(a + b)
The open pipe is assumed to be smooth; a represents the width, and b represents the height of
water flow in the open pipe. For turbulent flow in smooth conduits, the following approximations of
the friction coefficient and the Reynolds number dependence can be used. For Re > 3000,

1 p
p = 2.0 log(Re f ) − 0.8 (5)
f

When Re < 105 , the above equation can be approximated as:

0.316
f = (6)
R1/4
e

Equations (5) and (6) can be used to estimate the friction factor without having to use the Moody
chart [58,62], although multiple iterations may be required to obtain convergence. In the curve relating
soil erosion rates and water stress, the critical shear stress (τc ) is defined as the stress value at which
Water 2019, 11, 2598 6 of 20

soil begins to erode. Finally, the results are presented by the erosion curve as shown in Figure 4, which
provides the relationship between the soil erosion rate and the water flow shear stress.

2.2. Riverbank Erosion Simulation

2.2.1. The Boundary Shear Stress Causing Soil Erosion

To assess the erosion of soil at the lateral regions of the riverbank and the bank toe prior to
riverbank collapse, the river flow shear stress profile must be determined. The basic method for
calculating the shear stress boundary is described using the following equation (Chow, 1959) [59]:

A
τo = ρw sin θ · = ρw sin θ · R (7)
P
In this equation, the shear stress is calculated based on the hydraulic radius (R), which is
determined by dividing the cross-sectional area of the flow (A) by the wetted perimeter (P), in which
sinθ is the slope of the river. The shear stress also can be calculated based on the mean flow depth
(Hickin, 2004) [60], where the following equation is used:

τo = ρw sin θ(d) (8)

where d is the shear depth of the water, which represents the distance from the RWL surface to a
calculated point. This equation is applied with the assumption that the flow is steady and uniform.
During a flood event, the RWL continually increases, which induces changes in water shear stress
with increasing shear depth. Therefore, it is necessary to use a shear stress profile to simulate likely
changes to the lateral geometry and depth of the riverbank. Equation (9) shows the distribution of
shear stress from the riverbed to the water surface, where D is the current water depth at boundary, y
is the elevation above the boundary of a considered section across the channel, and the shear depth
d is equal to (D − y). The shear stress boundary (τb ) at the interface between the river channel and
riverbank can be determined using Equation (10) [60]:

τ y = ρw sin θ(d) = ρw sin θ(D − y) (9)

τb = 0.75ρw sin θ(D − y) (10)

where τy represents the shear stress acting across the bottom of the block of fluid (i.e., it represents the
downslope component of the weight of fluid in the block) at a height y above the riverbed. At the
riverbed (i.e., y = 0), the shear stress is greatest and is referred to as τo ; this represents the force per
unit area acting on the bed that is available to move sediment. τb represents the shear stress near the
riverbank and is used to simulate the water stress-induced erosion of soil at the riverbank.

2.2.2. Soil Erosion and Riverbank Erosion Width

When the boundary shear stress (τb ) at the riverbank is higher than the critical shear stress (τc ),
which is determined by the erosion test, riverbank soil erosion occurs. Based on the erosion curve
determined by the erosion test, the erosion rate responding to the shear stress distribution at the
riverbank will be defined. The soil erosion width (∆E) caused by the increase of the RWL and the time
elapsed while submerged (∆t) is calculated by the following equation:

∆E = ε∆t (11)

where the ε is the soil erosion rate responding to the specific shear stress and determined by the erosion
curve. The value of the soil erosion rate and erosion width will be used in the analysis model of
riverbank stability presented in Section 2.3. The applied idea for calculating erosion width in analysis
riverbank stability here is the same as for models used in [10–18].
Water 2019, 11, 2598 7 of 20

2.3. Model Analysis of Riverbank Stability with Seepage and Soil Erosion
This model
Water analyzes
2019, 11, riverbank
x FOR PEER REVIEW stability under the joint effects of seepage and soil erosion with 7 of 20
changing RWL. The concept of this model is modified from the analysis procedure performed using
GeoSlope in previous
stress and porestudies
water [17,18,37–39]. In general,
pressure to simulate RWL the changes
procedure andused to model
water stressbank-toe
(Step 1).erosion
Analysis of
and the changing
seepage andbank line response
stability is presented in Case
to the initial1conditions
of Figure 1,(Step
and 2).
is outlined as follows: Set up the
initial riverbank configuration
The following in thedifferential
governing SEEP/W module in GeoSlope
equation program to seepage
for two-dimensional analyze wascoupled
usedshear
in SEEP/W
stress and pore water
to determine pressure
the changestoinsimulate RWL changes
pore pressure andcalculate
state and water stress (Step 1). soil
unsaturated Analysis of seepage
properties [44]:
and stability response to the initial conditions (Step 2).
 ∂  for ∂two-dimensional
∂  ∂H wequation Hw  ∂θ seepage∂was ( )
− y in SEEP/W
H wused
 kx
The following governing differential  +  k y  + Q = ＝ mwγ w (12)
to determine the changes in pore ∂x pressure  ∂yand
∂x state ∂y  unsaturated
 calculate ∂t soil properties∂t [44]:
where kx (resp., k∂ ∂Hhydraulic
∂ conductivity
∂Hw in ∂θ ∂(H(resp.,
! !
y) is the w the x-direction w − y) y-direction), Q is the applied
kx + ky +Q = = mw γw (12)
hydrodynamic boundary, θ is the volumetric water content, and ∂t
∂x ∂x ∂y ∂y ∂t t is time, Hw is the total hydraulic
where head. While
kx (resp., ky ) is the hydraulic conductivity in the x-direction (resp., y-direction), Q is the applied
hydrodynamic boundary, θ is the volumetric water content, and t is time, Hw is the total hydraulic
∂θ = mw ∂u w = mwγ w ∂( H w − y)
head. While (13)
∂θ = mw ∂uw = mw γw ∂(Hu ww−=yγ)wu(H w –γy).
w = w (Hw − y). (13)

mw is the
mw slope
is the of the of
slope suction curve curve
the suction and depends on the on
and depends type theoftype
soil,ofγwsoil,
is the
γw unit
is theweight of water
unit weight and and
of water
y is they elevation, (uw )
is the elevation, is pore water pressure.
(uw) is pore water pressure.
Here, the bank is first built using the initial geometry of the bank line (li ) and the initial RWL (Li )
Here, the bank is first built using the initial geometry of the bank line (li) and the initial RWL (Li)
defined at the time ti (Figure 5a) and, then, the hydraulic boundary as the RWL changes from Li to
defined at the time ti (Figure 5a) and, then, the hydraulic boundary as the RWL changes from Li to
Li+1 , which corresponds to moving in time from ti to ti+1 in (Step 3). The simulation is then run to
Li+1, which corresponds to moving in time from ti to ti+1 in (Step 3). The simulation is then run to model
model transient seepage. During flood events, the RWL is simulated as rising at a fixed rate. Bank
transient seepage. During flood events, the RWL is simulated as rising at a fixed rate. Bank erosion
erosion occurs in submerged areas below the surface of the RWL, based on its water shear stress (Step
occurs in submerged areas below the surface of the RWL, based on its water shear stress (Step 4). The
4). The change in the riverbank line, which represents the surface between the soil bank and the water
change in the riverbank line, which represents the surface between the soil bank and the water river
river channel, is designated as the lateral eroded distance or the erosion width (∆Ei+1 ) (Figure 5b),
channel, is designated as the lateral eroded distance or the erosion width (ΔEi+1) (Figure 5b), which is
which is the distance the soil is eroded in a specific elapsed time, ∆ti+1 , determined using the following
the distance the soil is eroded in a specific elapsed time,Δti+1 , determined using the following
equation:
equation:
∆Ei+1 = ε∆ti+1 ; ∆ti+1 = ti+1 − ti (14)
ΔE i +1 = εΔt i +1 ; Δt i +1 = t i +1 − t i (14)

li

Li

(a) Initial geometry of riverbank line (li) and initial river water level (RWL) (Li)

Figure 5. Cont.
 Water 2019, 11, 2598 8 of 20
Water 2019, 11, x FOR PEER REVIEW 8 of 20

∆Ei+1: erosion width at ti+1

ε : soil erosion rate

li: initial bank

li+1: bank at ti+1

∆Ei+1 = ε ∆ti+1

(b) RWL rises from Li to Li+1 after Δti+1 (days)

Ln

Li

(c) RWL rises to Ln after Δtn (days)

(d) Riverbank line after failure

Figure 5.5.
Figure Examples ofof
Examples changes inin
changes riverbank geometry
riverbank due
geometry toto
due hydraulic erosion.
hydraulic erosion.

InInEquation
Equation(12), (12),ε εisisthe
thesoil
soilerosion
erosionrate,
rate,andandΔt∆tisisthe
thetotal
totalelapsed
elapsedtime
time(in (indays).
days).TheThesoilsoil
erosion
erosionrate rate(ε)(ε)isisdetermined
determinedbybyerosion
erosiontests
tests(as
(asininSection
Section2.1).
2.1).AtAt
the end
the endofof
this
this time
time period,
period,(i.e.,
(i.e.,
atatti+1ti+1
), the bank
), the bank line has
line has already
alreadybeen changed
been changedbybywater watererosion
erosionby distanceΔE
byaadistance ∆Ei+1i+1
, therefore,
, therefore,a new
a new
bank
bankline line(li+1) is) is
(li+1 manually
manually drawn
drawn byby
modifying
modifying thethe point inin
point the
thelateral bank
lateral bankunder
underthe theriver
riverwater
water
level (Step 5 and Figure 5b). The modify function in the GeoSlope program is used for redrawing the
riverbank.
Water 2019, 11, 2598 9 of 20

level (Step 5 and Figure 5b). The modify function in the GeoSlope program is used for redrawing
Water 2019, 11, x FOR PEER REVIEW 9 of 20
the riverbank.
Lastly, the factor of safety (FOS) is calculated by SLOPE/W in the GeoSlope program with the
Lastly, the factor of safety (FOS) is calculated by SLOPE/W in the GeoSlope program with the
current high river water level, pore water and new bank at which soil under the river water level
current high river water level, pore water and new bank at which soil under the river water level was
was eroded (Step 6). In SLOPE/W, it is impossible to simulate the slope of the overhang shape, so the
eroded (Step 6). In SLOPE/W, it is impossible to simulate the slope of the overhang shape, so the area
area between the initial bank line (li ) and the new bank line (li+1 ), was simulated as a low density
between the initial bank line (li) and the new bank line (li+1), was simulated as a low density material
material with properties similar to those of water. In the finite element modes of SEEP/W, the elements
with properties similar to those of water. In the finite element modes of SEEP/W, the elements are
are required to be connected at their corners by nodes that are not representative of an undercutting
required to be connected at their corners by nodes that are not representative of an undercutting
process, where an element tends to break away from the adjacent element, and the geometry of the
process, where an element tends to break away from the adjacent element, and the geometry of the
riverbank with an overhang cannot be simulated in SLOPE/W [17,18]. To overcome this, the change in
riverbank with an overhang cannot be simulated in SLOPE/W [17,18]. To overcome this, the change
geometry is produced by changing the material properties, and the eroded soil area is filled with very
in geometry is produced by changing the material properties, and the eroded soil area is filled with
low-strength soil.
very low-strength soil.
The obtained FOS is used in further analysis: (1) If the FOS is higher than 1.0, the analysis process
The obtained FOS is used in further analysis: (1) If the FOS is higher than 1.0, the analysis process
returns to Step 3, and RWL will continue to rise to level Ln . In this case, the riverbank geometry is
returns to Step 3, and RWL will continue to rise to level Ln. In this case, the riverbank geometry is
continually updated and redrawn by the new lines induced by the development of the erosion width
continually updated and redrawn by the new lines induced by the development of the erosion width
below the current RWL, ∆En (Figure 5c); then recalculate the FOS. (2) If the obtained FOS is less than
below the current RWL, ΔEn (Figure 5c); then recalculate the FOS. (2) If the obtained FOS is less than
1.0, the riverbank fails. The new bank line is updated by both the new erosion width and the failed slip
1.0, the riverbank fails. The new bank line is updated by both the new erosion width and the failed
surface (Figure 5d), then the FOS is recalculated with the new bank line. At that time, the FOS of new
slip surface (Figure 5d), then the FOS is recalculated with the new bank line. At that time, the FOS of
bank will higher than 1, and the analysis process returns to Step 3 if the RWL is continually rising.
new bank will higher than 1, and the analysis process returns to Step 3 if the RWL is continually
By building this framework (Figure 1), both processes of water level and soil erosion will be
rising.
involved. Thus, the issues limiting previous research, such as the process of seepage, change of
By building this framework (Figure 1), both processes of water level and soil erosion will be
groundwater level, pore pressure, and unsaturated soil properties, are addressed in this study.
involved. Thus, the issues limiting previous research, such as the process of seepage, change of
The process for determining the FOS in SLOPE/W using the limit equilibrium method (LE method).
groundwater level, pore pressure, and unsaturated soil properties, are addressed in this study.
Calculations for the FOS are performed by dividing the riverbank block into vertical slices. The forces
The process for determining the FOS in
acting on a slice are shown in Figure 6 for a composite slip surface [45].

Figure 6.
Figure 6. Forces
Forces acting
acting on
on aa slice of aa sliding
slice of sliding mass
mass with
with aa composite
composite slip
slip surface
surface [45].
[45].

In Figure 6, W is the total weight of a slice of width b and height h; N is the total normal force on
the base of the
the slice;
slice; SSmm is the force mobilized
mobilized on
on the
the base
base ofof each
each slice;
slice; EELL and
and EERR are the horizontal
horizontal
forces at
interslice normal forces at the
the left
left and
and right
right sides
sides of
of the
the slice,
slice, respectively;
respectively;XXLL and
and X XRR are the vertical
interslice shear forces at the left and right sides of the slice, respectively; D is an external point load;
kW is the horizontal seismic load; R is the radius for a circular slip surface; α is the angle between the
tangent to the center of the base of each slice and the horizontal; β is the sloping distance across the
base of a slice; ω is the angle of the point load from the horizontal; f is the perpendicular offset of the
normal force from the center of rotation; x is the horizontal distance from the centerline of each slice
to the center of rotation; h is the vertical distance from the center of the base of each slice to the
uppermost line in the geometry; e is the vertical distance from the centroid of each slice to the center
 Water 2019, 11, 2598 10 of 20

tangent
Water 2019, to
11, the center
x FOR PEER of the base
REVIEW of each slice and the horizontal; β is the sloping distance across 10 of the
20
base of a slice; ω is the angle of the point load from the horizontal; f is the perpendicular offset of the
ofnormal force
rotation; d from
is thethe center of rotation;
perpendicular x is the
distance fromhorizontal
a pointdistance
load to from the centerline
the center of each
of rotation; a isslice
theto
perpendicular distanceh from
the center of rotation; is thethe
vertical distance
resultant from water
external the center
forceofto
thethe
base of each
center slice to the
of rotation; anduppermost
F is the
line in the
resultant geometry;
external e isforce
water the vertical distance
FL,R is the fromexternal
resultant the centroid
waterofforce
each at
slice
lefttoand
the right
centersides
of rotation;
of failured is
the perpendicular distance from a point load to the center of rotation; a is the perpendicular distance
block.
from ThetheFOS
resultant
can beexternal
computed water
basedforce
ontomoment
the center of rotation;
equilibrium (Fmand
) andF is the resultant
horizontal forceexternal water
equilibrium
(Fforce FL,R factors
f). These is the resultant
of safetyexternal
may vary water force at left
depending andpercentage
on the right sides (λ)
of failure
of the block.
force function used in
The FOS canInbethis
the computation. computed
study, thebased
FOSonwasmoment
computedequilibrium
by using (Fmthe
) and horizontal
Spencer method,forcebecause
equilibrium
this
(Ff ). These
method factors
satisfied of safety
both moment mayequilibrium
vary depending on the percentage
and horizontal (λ) of the force
force equilibrium. function
As shown inused
Figurein the
7,
computation.
FOS In thispoint
is the crossover study, ofthe
theFOS
momentwas computed by using theFOS
and force equilibrium Spencer method,
values. because this
The following method
equations
ofsatisfied
statics in both moment
solving equilibrium
for the FOS by Fandm andhorizontal
Ff: force equilibrium. As shown in Figure 7, FOS is the
crossover point of the moment and force equilibrium FOS values. The following equations of statics in
solving for the FOS by Fm and Ff :  tan φ b  tan φ b  
 c′β R +  N − u w β − u a β 1 −
tan φb ′ tan φ ′
 R tan φ ′
 P      (15)
FOS = Fm = tan ϕ

tan ϕb

c0 βR + N − uw β tan ϕ0 − ua β 1 − tan ϕ0 R tan ϕ0
FOS = Fm = PWx ±  P
Fa P − Nf
(15)
Wx ± Fa− Nf
 P 0  tan ϕφ
tan b b  tan ϕtan b 
φtanϕ0 cos α 

 b

 c ′βc β
cos α
cos +
α 
+ NN−−uuwwββ tan ϕ 0 − −
uauβ β
1 
− 1 −
a  tan ϕ 0 
  tan φ ′ cos α 
FOS = F f =  P tan φ ′ P  tan φ ′   (16)
(16)
FOS = F f = N sin α ± F
 N sin α ±  F

FactorofofSafety
Figure7.7.Factor
Figure Safety(FOS)
(FOS)versus
versuslambda
lambdaplot.
plot.

3.3.Study
StudyArea
AreaConditions
Conditions
Figure88shows
Figure showsthe
thestudy
studyarea
areaofofthe
theRed
RedRiver
Riverbank,
bank,part
partofofthe
theRed
RedRiver
Riverdelta
deltaininHanoi
Hanoicity
cityinin
the north of Vietnam. Ngoc Thuy (NT1), Xuan Canh (XC) and Hai Boi (HB)
the north of Vietnam. Ngoc Thuy (NT1), Xuan Canh (XC) and Hai Boi (HB) are natural bank areasare natural bank areas and
have
and been
have strongly
been eroded
strongly by water
eroded by water flowflow
in the bending
in the area.
bending TheThe
area. riverbank in the
riverbank Xuan
in the Canh
Xuan area
Canh
was selected for simulating in the analysis model. This model includes the effects of
area was selected for simulating in the analysis model. This model includes the effects of both seepageboth seepage and
soilsoil
and erosion by water
erosion shear
by water stress
shear nearnear
stress the riverbank. Three
the riverbank. broad
Three categories
broad of data
categories are required
of data to run
are required
these simulation models: (1) bank geometry and stratigraphy data, (2) hydraulic
to run these simulation models: (1) bank geometry and stratigraphy data, (2) hydraulic boundary, boundary, and (3)
geotechnical data.
and (3) geotechnical data.
Water 2019, 11, 2598 11 of 20
Water 2019, 11, x FOR PEER REVIEW 11 of 20

Figure 8. The study area.

Figure 8. The study area.

The geometric
The geometric behaviors
behaviors of ofriverbanks
riverbanksand andthethehydraulic
hydraulicboundary
boundary are are
listed in Table
listed 1. This
in Table 1.
riverbank has a bank surface elevation of 10 m, with a bank slope of 81°. Hydraulic
◦
This riverbank has a bank surface elevation of 10 m, with a bank slope of 81 . Hydraulic dynamics dynamics include
the initial
include the RWL, the initial
initial RWL, pore pore
the initial pressure, the the
pressure, change
changein in
RWL RWLatatdifferent
differentrates
rates during rising and
during rising and
drawdown, and the distribution of shear stress near the bank, which causes soil
drawdown, and the distribution of shear stress near the bank, which causes soil erosion. The initial erosion. The initial
RWL,set
RWL, setatat00m,
m,isisthe
theelevation
elevationof ofRWL
RWLbefore
beforethe
therainy
rainyseason.
season.TheTheinitial
initialpore
porewater
waterpressure
pressureisis
defined as the maximum negative pressure head at 7 m so that the negative
defined as the maximum negative pressure head at 7 m so that the negative pressure at the pressure at the bank
bank
surface is −50 kPa (Figure 9). A simulated flood event with a rate of RWL change
surface is −50 kPa (Figure 9). A simulated flood event with a rate of RWL change of 1 m/d for both of 1 m/d for both
riseand
rise anddrawdown
drawdownisisused usedfor
forthese
thesemodels.
models. These
Thesesame
samedata
dataare
areused
usedto todetermine
determinethe thewater
watershear
shear
stress profile.
stress profile.

Table1.1.Bank
Table Banksoil
soillayers
layersand
andhydraulic
hydraulicconditions
conditionsused
usedin
insimulations
simulationsof
ofriverbank
riverbankstability.
stability.

Riverbank
RiverbankProperties
Properties Unit Unit Descriptions
Descriptions
Bank surface elevation (H) (m) 10
Bank surface elevation (H) (m) 10
BankBank
slopeslope ( o)
◦
( ) 8181
Soil layer and elevation of Layer 1: from
Layer 1: from1010mmtoto(−1)
(−1)m
msoil
soil bank
bank
(m) (m)
Soil layer and elevation of surface soil layer
surface soil layer Layer 2: from
Layer 2: from(−1)
(−1)mmdown
downbed
bed sand
sand
InitialInitial elevation
elevation of RWL
of RWL (m) (m) 00
Initial pore pressure
Initial pore pressure (m) 7m
(Maximum negative head) (m) 7m
(Maximum negative head)
Rate of water level change (m/d) 1
Rate of water level
River change
slope (m/d) (m/m) 1
0.0001
River slope (m/m) 0.0001
 Water 2019, 11, 2598 12 of 20
Water 2019, 11, x FOR PEER REVIEW 12 of 20

Figure 9. 9.
Figure Configuration of of
Configuration riverbank.
riverbank.

ToTosimulate
simulateriverbank
riverbankstability
stabilityand
anderosion,
erosion,three
threedifferent
different soil
soil materials
materials (Silt I-01 (NT03), Silt
Silt I-03
I-
03(HB21),
(HB21),andandSiltSilt
II-03 (XC))
II-03 willwill
(XC)) be set
beup
setinup
theinsame
the riverbank configuration
same riverbank as Xuanas
configuration Canh
Xuan riverbank.
Canh
Table 2 shows
riverbank. Tablethe properties
2 shows of these three
the properties soils.three
of these The soils.
stability
Theofstability
the riverbank was then analyzed
of the riverbank was thenin
the following
analyzed in the two scenarios:
following two Case 1—The
scenarios: soil1—The
Case bank is soil
erodedbankbased on thebased
is eroded distribution
on the of shear stress
distribution
ofwith
sheardepth.
stress In thisdepth.
with case, the soilcase,
In this bankthe
is eroded,
soil bankand then theand
is eroded, bank geometry
then the bankisgeometry
updated is asupdated
described
asin Section 2.3.
described Case 2—There
in Section is no soil is
2.3. Case 2—There erosion,
no soil and bankand
erosion, geometry remains remains
bank geometry constantconstant
as the RWLas
changes
the at a rateat
RWL changes ofa1 rate
m/d.ofFigure
1 m/d.1 Figure
shows different
1 shows steps
differentin both
stepsanalysis
in bothprocesses
analysis with erosion
processes withand
without
erosion anderosion.
without erosion.
Comparing the results of these two scenarios allows the stability of the riverbank to be evaluated
Table 2. Soil properties used in riverbank stability analysis.
using the calculated FOS. Cases of erosion of the bed sand layer were not considered; riverbank
stability was only assessed by tracking changes in the Siltsilt bank layer.Silt I-03
I-01 Silt II-03
Soil Properties
(NT 03) (HB21) (XC)
Sand Table 2. Soil properties
0.25–0.075 mm used in riverbank
5 stability analysis.
15 35
Silt 0.075–0.005 mm 70 I-01
Silt Silt70
I-03 55
Silt II-03
Clay <0.005 mm
Soil Properties 25 03)
(NT 15
(HB21) (XC)10
Saturated volumetric water content (%) 30 42 45
Sand 0.25–0.075 mm 5 15 35
Air-entry value (kPa) 27 20 10
Silt 0.075–0.005 mm 70 70 55
Hydraulic conductivity (m/s) 4.32 × 10−7 7.39 × 10−7 2.24 × 10−6
Clay
Cohesion force (kPa) <0.005 mm 7.525 15
5.0 10 5.0
Saturated volumetric
Internal friction anglewater
(◦ ) content (%) 32 30 42
32 45 30
Air-entry value shear stresses. (kPa)
Critical erosion (kPa) 1.527 20
1.1 10 1.0
Hydraulic conductivity (m/s) 4.32 × 10−7 7.39 × 10−7 2.24 × 10−6
Cohesion force (kPa) 7.5 5.0 5.0
Comparing the results
Internal friction angle of these two scenarios(o) allows 32the stability of
32 the riverbank 30to be evaluated
using the calculated
Critical erosionFOS.
shearCases of erosion of(kPa)
stresses. the bed sand1.5layer were not1.1considered; riverbank
1.0 stability
was only assessed by tracking changes in the silt bank layer.
4. Results and Discussion
4. Results and Discussion
4.1. Effects of Soil Erosion to Riverbank Stability
4.1. Effects of Soil Erosion to Riverbank Stability
Figure 10 shows the soil erosion curves for three different soils of Silt I-01 (NT03), Silt I-03
Figure 10 shows the soil erosion curves for three different soils of Silt I-01 (NT03), Silt I-03 (HB21),
(HB21), and Silt II-03 (XC), which have increasing sand contents of 5%, 15% and 36%, respectively.
and Silt II-03 (XC), which have increasing sand contents of 5%, 15% and 36%, respectively. The critical
The critical shear stresses, defined as the shear stress, are small and are nearly consistent between all
shear stresses, defined as the shear stress, are small and are nearly consistent between all soils of these
soils of these groups; however, this value increases significantly with increasing fine grain size
groups; however, this value increases significantly with increasing fine grain size content.
content.
Figure 11 (left) shows the calculated shear stress ranging from the initial RWL at 0 m to the peak
RWL at 10 m. The shear stress is calculated by Equation (10), with D = 10 m, and y changing from 0
to 10 m. The shear stress increases with the decrease of y and reaches a maximum in the bank toe at
0 m, thus demonstrating that the bank toe is often the most heavily eroded region of the riverbank.
 m on Day 1 to 3 m on Day 4; during this time, the soil was insignificantly eroded, and the bank
Water geometry
2019, 11, x FOR
did PEER REVIEW
not change 13 of 20
substantially. The bank line was updated after the last day of the first period
(on the fourth day). The stability of the riverbank after this first period was determined by calculating
Fromthe this
FOSshear
when stress profile,
the RWL wasthe at responding
3 m, and theerosion rates
bank line wasfor Silt I-01,
eroded by threeSilt I-03,
daysand Silt II-03
of flow. Aftersoils
this
are determined
first period,by thethe soil
next erosion
stage rate curve
was divided intoinaFigure 10.day-long
series of Then theintervals.
soil erosion width interval
A one-day for Silt I-01,
was
SiltWater
I-03,
chosenand11,
2019, 2598
Silt
becauseII-03
theriverbanks
riverbank may are calculated
have failedusing Equation
after only (13).
a short timeFigure
of soil11 (right)
bank shows
erosion. 13
the end
At the of 20
soil
erosion width
of each responding
day, the bank lineto the
wasshear stress
updated andof the
these
newthree
FOSsoils. The soil with higher erosion rate has
was calculated.
a larger erosion width; these results may have a great effect on the riverbank stability. These are the
900
results of the shear stress and erosion width corresponding to the RWL rising 10m. Analysis of the
800
difference of RWL used the same method and process.

Soil erosion rate (mm/h)

To simulate changes in the geometry of the riverbank due to bank soil erosion, this simulation
700

model was run over multiple 600 periods. During the first period, from Days 1–4, the RWL rose from 0
m on Day 1 to 3 m on Day 4;500during this time, the soil was insignificantly eroded, and the bank
geometry did not change substantially. The bank line was updated after the last day of the first period
400
(on the fourth day). The stability of the riverbank after this first period was determined by calculating
300
the FOS when the RWL was at 3 m, and the bank line was eroded by three days of flow. After this
first period, the next stage was 200 divided into a series of day-long intervals.
Erosion rate of A one-day interval was
Silt II-03
chosen because the riverbank 100 may have failed after only a short time of Silt soil I-03 bank erosion. At the end
Silt I-01
of each day, the bank line was updated 0
and the new FOS was calculated.
0 5 10 15 20 25 30 35 40 45 50 55
900 Water shear stress (Pa)

Figure10. 800
10.Erosion
Erosionrates
ratesof
ofSilt
Silt I-01,
I-01, Silt
Silt I-03
I-03 and
and Silt
SiltII-03
II-03atataadensity
densityofof1515
kN/m 3. 3
Figure kN/m .
Soil erosion rate (mm/h)

700
Figure10 11 (left) shows the calculated shear stress ranging from the initial RWL at 0 m 10 to the peak
600 Distance of soil erosion
= 10 m, and y changing
The elevation above the boundary (y, m)

RWL at 10 9m. The shear stress is calculated by Equation (10), withSiltDII-03 9 from 0

The elevation above the boundary (y, m)

500 Silt I-03
to 10 m. The8 shear stress increases
τb = 0.75ρw sinθwith
(D− y) the decrease of y and reaches Silt I-01
a maximum in the 8 bank toe at

0 m, thus demonstrating
7 400that the bank toe is often the most heavily eroded region of the 7 riverbank.
From this shear
6
stress profile,
300
the responding erosion rates for Silt I-01, Silt I-03, and Silt II-03
6 soils are
Eroded area
determined5 by the soil erosion rate curve in Figure 10. Then the soil erosion width for Silt 5I-01, Silt I-03,
200 and occupied by water
Erosion rate of
and Silt II-03
4
riverbanks are calculated using Equation (13). Figure 11 (right) shows the4 soil erosion
Silt II-03
100 Silt I-03
width responding to the shear stress of these three soils. The soil with higher erosion rate3 has a larger
3 Silt I-01
erosion width; these results 0 may have a great effect on the riverbank stability. These are the results of
2 2
the shear stress and erosion0 width 5 10 15
corresponding20 25 30
to the 35
RWL 40
rising45
10m.50 55
Analysis of the difference of
1 Water shear stress (Pa) 1
RWL used the same method and process.
0
Figure
0 10. Erosion rates of Silt I-01, Silt I-03 and7Silt II-03
6
at5 a density
4 3
of 152 kN/m
1
3.
0
0 1 2 3 4 5 6 7 8
Shear stress (Pa) Eroded distance (m)

10 10
Distance
Figure 11. Shear stress boundary near riverbank (left),ofand
soil erosion
eroded distance (right).
The elevation above the boundary (y, m)

9 Silt II-03 9
The elevation above the boundary (y, m)

Silt I-03
Figures 12 and 13τb =show
0.75ρw sinθ(D− y)of the FOS corresponding to
results 8
and soil8 erosion of RWL
Silt I-01 increasing

the three
7 soils. Figure 12 includes the change of RWL, and the result of FOS with both change 7 of RWL
and soil
6 erosion, and the results of FOS with only change of RWL for soil Silt II-03. In the case
6 without
soil erosion, the change of FOS has the same trend as the Eroded area of RWL, the FOS increases with
changes 5
5 and occupied by water
4 4

3 3

2 2

1 1

0 0
0 1 2 3 4 5 6 7 8 7 6 5 4 3 2 1 0

Shear stress (Pa) Eroded distance (m)

Figure 11. Shear stress boundary near riverbank (left), and eroded distance (right).
Figure 11. Shear stress boundary near riverbank (left), and eroded distance (right).

Figures 12 andchanges
To simulate 13 showin results of the FOS
the geometry of corresponding to RWL
the riverbank due increasing
to bank and soil
soil erosion, thiserosion of
simulation
the three soils. Figure 12 includes the change of RWL, and the result of FOS with both change of RWL
model was run over multiple periods. During the first period, from Days 1–4, the RWL rose from 0 m
and soil erosion, and the results of FOS with only change of RWL for soil Silt II-03. In the case without
on Day 1 to 3 m on Day 4; during this time, the soil was insignificantly eroded, and the bank geometry
soil erosion, the change of FOS has the same trend as the changes of RWL, the FOS increases with
did not change substantially. The bank line was updated after the last day of the first period (on the
fourth day). The stability of the riverbank after this first period was determined by calculating the FOS
when the RWL was at 3 m, and the bank line was eroded by three days of flow. After this first period,
 Water 2019, 11, 2598 14 of 20

the next stage was divided into a series of day-long intervals. A one-day interval was chosen because
the riverbank may have failed after only a short time of soil bank erosion. At the end of each day, the
bank line was updated and the new FOS was calculated.
Figures 12 and 13 show results of the FOS corresponding to RWL increasing and soil erosion of
the three soils. Figure 12 includes the change of RWL, and the result of FOS with both change of RWL
and soil erosion, and the results of FOS with only change of RWL for soil Silt II-03. In the case without
Water 2019, 11, x FOR PEER REVIEW 14 of 20
soil2019,
Water erosion, thePEER
11, x FOR change of FOS has the same trend as the changes of RWL, the FOS increases
REVIEW 14 ofwith
20
increasing RWL, and FOS decreases with decreasing RWL. These trends happen for all three soils.
increasing
increasing RWL, and FOS
FOS decreases with
with decreasing RWL. These trends happen for all three soils.
However,RWL, andchanges
the FOS decreases
with a different decreasing RWL.
trend in the These
case trends
of both RWLhappen
changesfor allsoil
and three soils. as
erosion
However,
However, the FOS
FOS changes with
with aa different trend
trend in the
the case
case of
of both
both RWL
RWL changes
changes and
and soil
soil erosion
shown inthe Figure changes
13 and outlined different
in the following indescription. erosion
as
as shown
shown in in Figure
Figure 1313 and
and outlined
outlined in
in the
the following
following description.
description.
6 12
6 River water level (1m/d) 12
River
with water level (1m/d)
erosion
5.5 10
5.5 with erosion
without erosion 10
5 without erosion
5 8
8
4.5
4.5 6

River
6

Riverwater
Safety

4
FactorofofSafety

4
4
3.5 4

waterlevel
3.5
2
3 2
Factor

level(m)
3
2.5 0
2.5 0

(m)
2 -2
2 -2
1.5
1.5 -4
-4
1
1 -6
-6
0.5
0.5
0 2 4 6 8 10 12 14 16 18 20 22
0 2 4 6 8 10 12 14 16 18 20 22
Time (days)
Time (days)
Figure 12. 12.
Evolution of bank
Evolution stability
of bank for for
stability Silt II-03.
Figure
Figure 12. Evolution of bank stability for SiltSilt II-03.
II-03.
4.5 12
4.5 River water level (1m/d) 12
River water level (1m/d)
Silt I-01 (erosion) 10
4 Silt I-01 (erosion) 10
4 Silt I-03 (erosion)
Silt I-03 (erosion) 8
3.5
Silt II-01 (erosion) 8
3.5
Silt II-01 (erosion)
6
River

6
Riverwater
Safety

3
FactorofofSafety

3 4
4
waterlevel

2.5
2.5 2
2
Factor

level(m)

2 0
2 0
(m)

1.5 -2
1.5 -2

-4
1 -4
1
-6
-6
0.5
0.5
0 2 4 6 8 10 12 14 16 18 20 22
0 2 4 6 8 10 12 14 16 18 20 22
Time (days)
Time (days)
Figure 13. Comparison of bank stability with erosion for three soils.
Figure 13.13.
Figure Comparison of bank
Comparison stability
of bank with
stability erosion
with forfor
erosion three soils.
three soils.
In the
In In
thethefirst
first period, 0–3 m, in which the shear stress was small, the FOS increased slightly. The
firstperiod,
period,0–3 0–3m, m,in
inwhich
which the the shear stress
stress waswassmall,
small,the theFOSFOSincreased
increased slightly.
slightly. The TheFOS
FOS
FOS had
had already
already begun
begun to
to decrease
decrease slightly
slightly during
during the last two days of
of the first period when the
had already begun to decrease slightly during thethelastlast
twotwo daysdays
of the the
firstfirst
periodperiod
when when the
the water
water
water shear
shear stress
stress exceeded
exceeded the
the critical
critical shear stress. Following the first
first period, the FOS
FOS continued to
shear stress exceeded the critical shearshear
stress.stress. Following
Following the firsttheperiod, period,
the FOS thecontinuedcontinued to
to decrease.
decrease.
decrease. The
The bank
bank line above
lineRWLabove RWL remained
RWL remained the
the same at all steps until the FOS exceeded 1, but the
The bank line above remained the same atsame
all stepsat all steps
until theuntil
FOSthe FOS exceeded
exceeded 1, but the 1, but
bank theline
bank
bank line under
linethe
under the RWL
thewas
RWL was
was updated
updated after
after this
this rise inin RWL to
to reflect the
the cumulative erosion width.
under RWL updated after this rise inrise
RWL RWL
to reflect reflect
the cumulativecumulativeerosionerosion
width. width.
In the
In
In the
the riverbank
riverbank of
of Silt
Silt II-03
II-03 soil,
soil, the FOS decreased to 0.916
0.916 whenwhen thethe RWL rose to 5 m after six days.
riverbank of Silt II-03 soil, thetheFOSFOSdecreased
decreasedto to0.916 when RWL
RWL rose rosetoto5 5mm after sixsix
after days.days.
This
This bank
bank collapsed
collapsed one
one day earlier than the riverbank of Silt I-03, in which the FOS value decreased
This bank collapsed oneday dayearlier
earlierthanthanthe
theriverbank
riverbankofofSilt SiltI-03,
I-03,ininwhich
which thethe FOS
FOS value
value decreased to
to
to 0.972
0.972 at at 66 m.
m. The
The riverbank
riverbank of of Silt
Silt I-01
I-01 collapsed
collapsed last
last when
when the the RWL
RWL roserose toto 10
10 mm after
after 1111 days,
days, and
and
the FOS dropped to 0.88. Generally, the FOS decreased to a value of
the FOS dropped to 0.88. Generally, the FOS decreased to a value of less than 1 first in the Silt II-03 less than 1 first in the Silt II-03
riverbank,
riverbank, then then in in the
the Silt
Silt I-03
I-03 riverbank
riverbank and and finally
finally inin the
the Silt
Silt I-01
I-01 riverbank
riverbank (Figure
(Figure 13).13).
After
After riverbank failure, the bank line was modified based on the failure surface, as
riverbank failure, the bank line was modified based on the failure surface, as the
the riverbank
riverbank
was
was assumed to have fallen into the river channel along the failure surface. Consequently, the
assumed to have fallen into the river channel along the failure surface. Consequently, the new
new
bank line above the current RWL represented the failure surface, and the bank line under the current
Water 2019, 11, 2598 15 of 20

0.972 at 6 m. The riverbank of Silt I-01 collapsed last when the RWL rose to 10 m after 11 days, and
the FOS dropped to 0.88. Generally, the FOS decreased to a value of less than 1 first in the Silt II-03
riverbank, then in the Silt I-03 riverbank and finally in the Silt I-01 riverbank (Figure 13).
After riverbank failure, the bank line was modified based on the failure surface, as the riverbank
was assumed to have fallen into the river channel along the failure surface. Consequently, the new
bank line above the current RWL represented the failure surface, and the bank line under the current
RWL represented the bank line updated to reflect the occurrence of soil erosion. The new bank line was
updated, and the pore pressure and RWL were inherited from the previous period, producing a new
bank with higher stability. Therefore, the FOS suddenly increased after these bank failures, and the
Water 2019, 11, x FOR PEER REVIEW 15 of 20
Water 2019, 11,
maximum x FOR
FOS PEER were
values REVIEW obtained in this step during the simulated flood event. The maximum 15 of 20
value
valuewas wasobtained
obtainedusing
usingthetheupdated
updatedbank bankgeometry
geometryresulting
resultingfromfromfailure
failurebybyerosion,
erosion,ininwhich
whichthe the
value wasFOS
resulting obtained
values using
were the updated
3.98 for Silt bank
I-01, geometry
1.56 for resulting
Silt I-03, from
and failure
1.42 for byII-03.
Silt erosion, in whichthis
Following the
resulting FOS values were 3.98 for Silt I-01, 1.56 for Silt I-03, and 1.42 for Silt II-03. Following this
resulting
point, FOS values were 3.98 for the
Silt I-01, 1.56 for Silt I-03, and 1.42 for Silt II-03. decreasing
Following this
point,asasthetheRWL
RWLcontinued
continuedtotorise,
rise, thebank
banksoil
soilwaswaseroded
erodedatata ahigher
higherwidth,
width,thus
thus decreasingthe the
point,
FOS as the RWL continued to rise, the bank soil was eroded at a higher width, thus decreasing the
values.
FOS values.
FOSFigures
values. 14–16 show the evolution of riverbank stability with erosion at the bank toe for the three
Figures 14–16 show the evolution of riverbank stability with erosion at the bank toe for the three
types Figures 14–16 show the evolutionThereof riverbank stability withstability
erosion at the bank toe for the three
types ofmodeled
of modeledsoil soilbank
bankmaterials.
materials. Thereare arethree
threeriverbank
riverbank stabilitystates:states:(1)
(1)atatthe
theend
endofofthe
the
types
first of modeled soil bank materials. There are three riverbank stability states: (1) at the end ofand
the
firststate, after
state, thethe
after RWLRWL rose from
rose from0 m0tom3tom;3 (2)
m;when
(2) when the FOS dropped
the FOS to a value
dropped of lessofthan
to a value less1than 1
firstoriginal
the state, after
bankthe RWL rose
collapsed; andfrom 0 mthe
(3) after to 3bank
m; (2)
linewhen the FOS and
was updated dropped to a value
the riverbank of less than 1
and the original bank collapsed; and (3) after the bank line was updated and thestate recovered.
riverbank state
and the
These original
figures bankdetails
provide collapsed;
of theand (3) after
hydraulic andthegeometric
bank lineconditions
was updated and the riverbank state
recovered. These figures provide details of the hydraulic and geometriccorresponding to the phases
conditions corresponding to
recovered.
discussed These
above. figures
Changes provide
in RWL details
are shownof the hydraulic
on theare and
riverside, geometric conditions corresponding to
the phases discussed above. Changes in RWL shown and changes
on the in the and
riverside, groundwater
changes table
in the
the phases
(dashed bluediscussed
line) above. Changes in RWL on arethe shown on the riverside, and changes inwasthe
groundwater tableand soil erosion
(dashed blue line)areand
shown
soil erosion bank
are shownside. The
on thesoil
bankarea under
side. Thethe soilRWL
area under
groundwater
eroded andwas table (dashed
removed, blue line) and soil erosion are shown onriver the bank side. The soil area under
the RWL erodedafter
and which
removed, this after
entire area
which was
this occupied
entire areabywas water.
occupied by river water.
the RWL was eroded and removed, after which this entire area was occupied by river water.

Figure 14.Simulated
Simulated Silt I-01 riverbank after rise in RWL 3tom,
3 10
m, m10(before
m (before bank update), and 10
Figure 14. SimulatedSilt
Figure14. SiltI-01
I-01riverbank
riverbankafter
afterrise
riseininRWL
RWLtoto 3 m, 10 m bank
(before update),
bank andand
update), 10 m
10
m (after
(after bankbank update).
update).
m (after bank update).

Figure 15. Simulated Silt I-03 riverbank after rise in RWL to 3, 6 and 8 m.
Figure 15. Simulated Silt I-03 riverbank after rise in RWL to 3, 6 and 8 m.
Figure 15. Simulated Silt I-03 riverbank after rise in RWL to 3, 6 and 8 m.
Water 2019, 11, 2598 16 of 20
Figure 15. Simulated Silt I-03 riverbank after rise in RWL to 3, 6 and 8 m.

Figure 16. Simulated Silt II-03 riverbank after rise in RWL to 3, 5 and 6 m.

The soil erosion and riverbank failure process depend on the erosional behavior of the bank soil.
Three types of soil were modeled; in order of increasing erosion rate, these are Silt I-01, Silt I-03, and
Silt II-03, which have critical shear stresses of 1.49, 1.09, and 1.00 Pa, respectively. Critical shear stresses
do not vary substantially among these soils, but their erosion rates vary quite substantially at higher
shear stresses. Less cohesive soils with higher erosion rates become unstable earlier. This can be seen
in the recorded times to failure determined in the simulations, which are six days for Silt II-03, seven
days for Silt I-03, and 11 days for Silt I-01. The less cohesive soil bank (Silt II-03) fails at lower RWL
(6 m) and more rapidly (six days) than the more cohesive soil banks (Silt I-03 and Silt I-01). Moreover,
different soils produce different riverbank failure shapes. When riverbank failure occurs, the overhang
shape is greater in the less cohesive soil (Silt II-03) than in the more cohesive soil (Silt I-01) because of
the higher width of soil erosion at the toe of the riverbank, then the riverbank in the less cohesive soil
(Silt II-03) fails earlier than that in cohesive soil.
The effects of soil properties on the process of riverbank failure have previously been discussed
in [19,20,22–25,30], which simulated this process at the same river water level. The riverbank in
non-cohesive soil banks (mainly those comprising sandy soil) tends to fail near the water surface,
whereas cohesive bank failure occurs near the bank toe, where the water velocity is much higher [23].
A series of simulations modeling the full processes of riverbank failure, depicting the regions both
under and above the undercutting overhang were performed in [19,20]. These results demonstrate
that soils with higher soil density have higher resistance force and thus higher undermining failure
depths at which the riverbank failure occurs. These previous studies clearly illustrated the effects
of soil properties on riverbank failure in cases without changes in the RWL. In this study, modeled
simulations were performed using variations in RWL and soils. The Silt I-01 bank failed at an RWL
of 10 m and a maximum erosion width of 0.55 m; the Silt I-03 bank failed at an RWL of 6 m and a
maximum erosion width of 0.72 m; and the Silt II-03 bank failed at an RWL of 5 m and a maximum
erosion width of 1.2 m. During flood events in which RWL increases at the same rate, the less cohesive
soil bank will be eroded in a larger toe cantilever shape, and riverbank failure will occur earlier than the
more cohesive soil banks. The results obtained in this study include the effects of soil properties and
RWL conditions and show that both soil bank properties and RWL can greatly affect riverbank stability.
The upper riverbank overhang failure may occur as a type of topping failure, as described in [19,20].

4.2. Complexity of Factors in Riverbank Stability Analysis

In the scenario analyzed without soil erosion, the results demonstrate that FOS increased during
RWL rise and decreased during drawdown. The studies analyzing the type of mass failure [2–9]
illustrated the effects of factors, such as confining pressure, pore pressure, and unsaturated soil
properties on riverbank stability. In particular, Duong et al., (2014) [6] clarified cases of riverbanks
controlled by soil hydraulic conductivity. In cases of low hydraulic conductivity (<10−6 m/s) and
those in which soil erosion does not occur in the toe of the riverbank, the confining pressure is the
Water 2019, 11, 2598 17 of 20

main factor affecting the FOS value. However, when riverbank soil has high hydraulic conductivity
(>10−6 m/s), increasing the RWL will cause the pore water pressure to also increase. In soils with higher
hydraulic conductivity, pore water pressure and the rate of water level change are the primary factors
affecting the FOS value. Those results may contrast with results obtained in this study, as well as those
obtained from studies analyzing the effects of soil erosion and RWL on riverbank stability [17,18,21].
As the rates of RWL change increase, water stress increases more rapidly, and the soil is more eroded.
Because both higher water shear stress and higher rates of soil erosion produce larger soil erosion
widths, the bank failures are caused more rapidly. In this study, RWL change was defined as 1 m/day
and erosion in the toe of bank was accounted for. Riverbank failure occurred after the RWL rose by 5
m, 6 m, and 10 m and after six days, seven days, and 11 days for the soil banks of Silt II-03, Silt I-03,
Silt I-01, respectively. However, without accounting for soil bank erosion, the riverbank of Silt II-03
only failed when its hydraulic conductivity was higher than the increase in the RWL [6]. Combining
these results with those of previous studies, the understanding of the mechanisms of riverbank failure
should cover all factors, such as the rate of RWL fluctuation, seepage and soil erosion. With low rates of
RWL change (especially when they are equal to or smaller than the hydraulic conductivity of the soil),
riverbank failure may occur as a mass failure even without toe-bank soil erosion due to increasing pore
pressure by seepage process. With high rates of RWL change as well as high water stress, toe-bank soil
erosion will occur, and the riverbank fails as an overhang riverbank failure. In the latter case, the soil
properties and erosion rate are the main factors causing riverbank failure, and the confining pressure
continually supports the riverbank when the RWL rises. The pore water pressure and soil suction may
not significantly affect riverbank stability because the groundwater in the riverbank area is not greatly
increased by the increase of the RWL.

5. Conclusions
This paper simulated the processes of riverbank failure using three different types of bank soil and
by modeling both mass failure and undercutting erosion. By combining the results of these simulations
with those of previous studies, the following conclusions can be reached.
Soil erosion is closely related to the content of fine-grained particles. The higher the fine content,
the lower the erosion rate. The silt soils found along the Red River bank have a very high erosion rate.
Their critical shear stress is small and is nearly constant at 1–1.5 Pa for a soil density of 15 kN/m3 .
During RWL rise, the soil bank erosion width and overhang shape develops due to increased
shear stress. In the same configuration of the riverbank and changing RWL, the bank has a higher soil
erosion rate, and riverbank failure occurs within a shorter time and at a lower RWL. In the three cases
of Silt I-01, Silt I-03, and Silt II-03, which are listed in order of increasing erosion rate, riverbank failure
occurs after 10 days, seven days and six days, respectively. The less cohesive soil bank (Silt II-03) fails
at a lower RWL (6 m) and in a shorter amount of time (six days) than the more cohesive soil banks (Silt
I-03 and Silt I-01).
The shape of the riverbank cantilever failure depends on the soil properties and RWL. Without
changing the RWL, riverbanks with higher resistance force (i.e., higher cohesive force and soil density)
experience failure with larger and deeper overhang erosion. In flood events in which RWL increases at
the same rate, less cohesive soil banks will be eroded in larger toe overhang shapes than more cohesive
soil banks, and riverbank failure will occur earlier.
In cases with a low RWL rising rate (especially when equal to or smaller than soil hydraulic
conductivity), riverbank failure may occur as a mass failure. In cases of both high RWL rising rate and
high water stress, toe-bank soil erosion will occur, and the riverbank will undergo overhang riverbank
failure. The soil properties and erosion rate are the main factors causing riverbank failure.

Author Contributions: Conceptualization, T.D.T. and D.D.M.; methodology, T.D.T. and D.D.M.; software, T.D.T.;
validation, T.D.T. and D.D.M.; formal analysis, T.D.T.; investigation, T.D.T. and D.D.M.; resources, T.D.T.; data
curation, T.D.T. and D.D.M.; writing—original draft preparation, T.D.T.; writing—review and editing, T.D.T. and
D.D.M.; visualization, T.D.T.; supervision, D.D.M.; project administration, T.D.T.; funding acquisition, T.D.T.
Water 2019, 11, 2598 18 of 20

Funding: This research was funded by the project Code 105.08-2015.24, which was sponsored by Nafosted,
Ministry of Science and Technology, Vietnam.
Acknowledgments: This paper was completed with the support of the project Code 105.08-2015.24, which was
sponsored by Nafosted, Ministry of Science and Technology, Vietnam. The authors express their sincere gratitude
for this support.
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. Watson, A.; Basher, L. Stream bank erosion: A review of processes of bank failure, measurement and
assessment techniques, and modelling approaches. In A Report Prepared for Stakeholders of the Motueka
Integrated Catchment Management Programme and the Raglan Fine Sediment Study; Landcare Research: Hamilton,
New Zealand, 2006.
2. Casagli, N.; Rinaldi, M.; Gargini, A.; Curini, A. Pore water pressure and streambank stability: Results from a
monitoring site on the Seive River, Italy. Earth Surf. Process. Landf. 1990, 24, 1095–1114. [CrossRef]
3. Dapporto, S.; Rinaldi, M.; Casagli, N. Failure mechanisms and pore pressure condition analysis of Riverbank
along the Arno River (Central Italy). Eng. Geol. 2001, 61, 221–242. [CrossRef]
4. Rinaldi, M.; Casagli, N.; Dapporto, S.; Gargini, A. Monitoring and modeling of pore water pressure changes
and river-bank stability during flow events. Earth Surf. Process. Landf. 2004, 29, 237–254. [CrossRef]
5. Rinaldi, M.; Mengoni, B.; Luppi, L.; Darby, S.E.; Mosselman, E. Numerical simulation of hydrodynamics and
bank erosion in a river bend. Water Resour. Res. 2008, 44, 1–17. [CrossRef]
6. Duong, T.T.; Komine, H.; Do, M.D.; Murakami, S. Riverbank stability assessment under flooding conditions
in the Red River of Hanoi, Vietnam. Comput. Geotech. 2014, 61, 178–189. [CrossRef]
7. Hui, L. Yangtze Riverbank Slope Stability by Water Level Change in the Three Gorges Reservoir. Electron. J.
Geotech. Eng. 2015, 20, 1847–1856.
8. Liang, C.; Jaksa, M.; Ostendorf, B.; Kuo, Y. Influence of river level fluctuations and climate on riverbank
stability. Comput. Geotech. 2015, 63, 83–98. [CrossRef]
9. Chen, C.-H.; Hsieh, T.-Y.; Yang, J.-C. Investigating effect of water level variation and surface tension crack on
riverbank stability. J. Hydro-Environ. Res. 2017, 15, 41–53. [CrossRef]
10. Langendoen, E.J. Concepts: Conservational Channel Evolution and Pollutant Transport System Software Manual;
Research Report No. 16; USDA-ARS National Sedimentation: Oxford, MS, USA, 2000.
11. Simon, A.; Collison, A.J.C. Quantifying the mechanical and hydrological effects of riparian vegetation on
stream-bank stability. Earth Surf. Process. Landf. 2002, 27, 527–546. [CrossRef]
12. Simon, A.; Curini, A.; Darby, S.; Langendoen, E. Stream-bank mechanics and the role of bank and nearbank
processes in incised channels. In Incised River Channels; Darby, S., Simon, A., Eds.; John Wiley and Sons:
New York, NY, USA, 1999; pp. 123–152.
13. Simon, A.; Curini, A.; Darby, S.E.; Langendoen, E.J. Bank and near-bank processes in an incised channel.
Geomorphology 2000, 35, 193–217. [CrossRef]
14. Simon, A.; Pollen, N.; Langendoen, E.J. Influence of two woody riparian species on critical conditions
for streambank stability: Upper Truckee River, California. J. Am. Water Resour. Assoc. 2006, 42, 99–113.
[CrossRef]
15. Simon, A.; Pollen-Bankhead, N.; Thomas, R.E. Development and application of a deterministic bank stability
and toe erosion model for stream restoration. In Stream Restoration in Dynamic Fluvial Systems: Scientific
Approaches, Analyses and Tools; Geophysical Monograph Series; American Geophysical Union: Washington,
DC, USA, 2011; Volume 194, pp. 453–474.
16. Simon, A.; Pollen-Bankhead, N.; Mahacek, V.; Langendoen, E.J. Quantifying reductions of mass failure
frequency and sediment loadings from streambanks using toe protection and other means: Lake Tahoe,
United States. J. Am. Water Resour. Assoc. 2009, 45, 170–186. [CrossRef]
17. Darby, S.E.; Rinaldi, M.; Dapporto, S. Coupling simulations of fluvial erosion and mass wasting for cohesive
river banks. J. Geophys. Res. 2007, 112, 1–15. [CrossRef]
18. Luppi, L.; Rinaldi, M.; Teruggi, L.B.; Darby, S.E.; Nardi, L. Monitoring and numerical model of river-bank
erosion processes: A case study along the Cecina River (central Italy). Earth Surf. Process. Landf. 2008, 34,
530–546. [CrossRef]
Water 2019, 11, 2598 19 of 20

19. Samadi, A.; Davoudi, M.H.; Amiri-Tokaldany, E. Experimental Study of Cantilever Failure in the Upper Part
of Cohesive Riverbanks. Res. J. Environ. Sci. 2011, 5, 444–460. [CrossRef]
20. Samadi, A.; Amiri-Tokaldany, E.; Davoudi, M.H.; Darby, S.E. Experimental and numerical investigation of
the stability of overhanging riverbanks. Geomorphology 2013, 184, 1–19. [CrossRef]
21. Bo, N.I.; Wu, S.Q.; Tan, Y.F.; Xie, X.H.; Jun, Y.A.; Yan, Z.M.; Geng, Y.Q. Coupling effect of seepage flow and
river flow on bank failure. J. Hydrodyn. 2011, 23, 834–840.
22. Yong, G.L.; Robert, E.; Thomas, Y.O.; Simon, A.; Blair, P.; Greimann, K.W. Modeling of multilayer cohesive
bank erosion with a coupled bank stability and mobile-bed mode. Geomorphology 2014, 243, 116–129.
23. Yu, M.-H.; Wei, H.-Y.; Wu, S.-B. Experimental study on the bank erosion and interaction with near-bank bed
evolution due to fluvial hydraulic force. Int. J. Sediment Res. 2015, 30, 81–89. [CrossRef]
24. Patsinghasanee, S.; Kimura, I.; Shimizu, Y. Experimental and Numerical Study on Overhanging Failure Of
River Bank. J. Jpn. Soc. Civ. Eng. Ser. B1 Hydraul. Eng. 2015, 71, 127–132. [CrossRef]
25. Patsinghasanee, S.; Kimura, I.; Shimizu, Y.; Nabi, M. Cantilever failure investigations for cohesive riverbanks.
In Proceedings of the Institution of Civil Engineers-Water Management; Thomas Telford Ltd.: London, UK, 2015;
Volume 2, pp. 93–108.
26. Kimiaghalam, N.; Goharrokhi, M.; Clark, S.P.; Ahmari, H. A comprehensive fluvial geomorphology study of
riverbank erosion on the Red River in Winnipeg, Manitoba. Can. J. Hydrol. 2015, 529, 1488–1498. [CrossRef]
27. Daly, E.R.; Miller, R.B.; Fox, G.A. Modeling streambank erosion and failure along protected and unprotected
composite streambanks. Adv. Water Resour. 2015, 81, 114–127. [CrossRef]
28. El Kadi Abderrezzak, K.; Die Moran, A.; Tassi, P.; Ata, R.; Hervouet, J.-M. Modelling river bank erosion
using a 2D depth-averaged numerical model of flow and non-cohesive, non-uniform sediment transport.
Adv. Water Resour. 2016, 93, 75–88. [CrossRef]
29. Klavon, K.; Fox, G.; Guertault, L.; Langendoen, E.; Enlow, H.; Miller, R.; Khanal, A. Evaluating a process-based
model for use in streambank stabilization: Insights on the Bank Stability and Toe Erosion Model (BSTEM).
Earth Surf. Process. Landf. 2017, 42, 191–213. [CrossRef]
30. Patsinghasanee, S.; Kimura, I.; Shimizu, Y.; Nabi, M.; Chub-Uppakarn, T. Coupled studies of fluvial erosion
and cantilever failure for cohesive riverbanks: Case studies in the experimental flumes and U-Tapao River.
J. Hydro-Environ. Res. 2017, 16, 13–26. [CrossRef]
31. Jugie, M.; Gob, F.; Virmoux, C.; Brunstein, D.; Tamisier, V.; Le Coeur, C.; Grancher, D. Characterizing
and quantifying the discontinuous bank erosion of a small low energy river using Structure-from-Motion
Photogrammetry and erosion pins. J. Hydrol. 2018, 563, 418–434. [CrossRef]
32. Patsinghasanee, S.; Kimura, I.; Shimizu, Y.; Nabi, M. Experiments and modelling of cantilever failures for
cohesive riverbanks. J. Hydraul. Res. 2018, 56, 76–95. [CrossRef]
33. Dutta, S.; Karmakar, T. Erosion Processes in Composite Riverbanks: Experiments and Modelling. Int. J. Res.
Eng. Technol. 2015, 4, 85–92.
34. Cancienne, R.M.; Fox, G.A.; Simon, A. Influence of seepage undercutting on the stability of root-reinforced
streambanks. Earth Surf. Process. Landf. 2008, 33, 1769–1786. [CrossRef]
35. Fox, G.A.; Wilson, G.V.; Periketi, R.K.; Cullum, R.F. Sediment transport model for seepage erosion of
streambank sediment. J. Hydrol. Eng. 2006, 11, 603–611. [CrossRef]
36. Fox, G.A.; Wilson, G.V.; Simon, A.; Langendoen, E.J.; Akay, O.; Fuchs, J.W. Measuring streambank erosion
due to ground water seepage: Correlation to bank pore water pressure, precipitation and stream stage.
Earth Surf. Process. Landf. 2007, 32, 1558–1573. [CrossRef]
37. Chu-Agor, M.L.; Fox, G.A.; Cancienne, R.M.; Wilson, G.V. Seepage caused tension failures and erosion
undercutting of hill slopes. J. Hydrol. 2008, 359, 247–259. [CrossRef]
38. Chu-Agor, M.L.; Fox, G.A.; Wilson, G.V. Empirical sediment transport function predicting seepage erosion
undercutting for cohesive bank failure prediction. J. Hydrol. 2009, 377, 155–164. [CrossRef]
39. Chu-Agor, M.L.; Wilson, G.V.; Fox, G.A. Numerical modeling of bank instability by seepage erosion
undercutting of layered streambanks. J. Hydrol. Eng. 2008, 13, 1133–1145. [CrossRef]
40. Fox, G.A.; Heeren, D.M.; Wilson, G.V.; Langendoen, E.J.; Fox, A.K.; Chu-Agor, M.L. Numerically predicting
seepage gradient forces and erosion: Sensitivity to soil hydraulic properties. J. Hydrol. 2010, 389, 354–362.
[CrossRef]
41. Masoodi, A.; Majdzadeh Tabatabai, M.; Noorzad, A.; Samadi, A. Effects of soil physico-chemical properties on
stream bank erosion induced by seepage in northeastern. Iran Hydrol. Sci. J. 2017, 62, 2597–2613. [CrossRef]
Water 2019, 11, 2598 20 of 20

42. Masoodi, A.; Noorzad, A.; Majdzadeh Tabatabai, M.R.; Samadi, A. Application of short-range
photogrammetry for monitoring seepage erosion of riverbank by laboratory experiments. J. Hydrol.
2018, 558, 380–391. [CrossRef]
43. Briaud, J.L. Case histories in soil and rock erosion: Woodrow Wilson Bridge, Brazos River Meander,
Normandy Cliffs, and New Orleans Levees. J. Geotech. Geoinviron. Eng. ASCE 2008, 134, 1424–1447.
[CrossRef]
44. GEO-SLOPE International Ltd. Seepage Modeling with SEEP/W. Available online: http://www.geo-slope.com
(accessed on 9 December 2019).
45. GEO-SLOPE International Ltd. Stability Modeling with SLOPE/W. Available online: http://www.geo-slope.
com (accessed on 9 December 2019).
46. Jepsen, R.; Roberts, J.; Lick, W. Effects of Bulk Density on Sediment Erosion Rates Water. Air Soil Pollut. 1997,
99, 21–31. [CrossRef]
47. Roberts, J.; Jepsen, R.; Gotthard, D.; Lick, W. Effects of particle size and bulk density on erosion of quartz
particles. J. Hydraul. Eng. ASCE 1998, 24, 1261–1267. [CrossRef]
48. Lick, W.; McNeil, J. Effects of sediment bulk properties on erosion rates. Sci. Total Environ. 2001, 266, 41–48.
[CrossRef]
49. Couper, P. Effects of silt—clay content on the susceptibility of river banks to subaerial erosion. Geomorphology
2003, 56, 95–108. [CrossRef]
50. Ahmad, M.F.; Dong, P.; Mamat, M.; Wan Nik, W.B.; Mohd, M.H. The Critical Shear Stresses for Sand and
Mud Mixture. Appl. Math. Sci. 2011, 5, 53–71.
51. Wang, Y.-C. Effects of Physical Properties and Rheological Characteristics on Critical Shear Stress of Fine
Sediments. Ph.D. Thesis, Georgia Institute of Technology, Atlanta, GA, USA, 2013.
52. Guo, Y.; Yang, Y.; Yu, X. Influence of particle shape on the erodibility of non-cohesive soil: Insights from
coupled CFD–DEM simulations. Particuology 2018, 39, 12–24. [CrossRef]
53. Briaud, J.L.; Ting, F.; Chen, H.C.; Gudavalli, R.; Perugu, S.; Wei, G. SRICOS: Prediction of scour rate in
cohesive soils at bridge piers. J. Geotech. Geoenviron. Eng. ASCE 1999, 125, 237–246. [CrossRef]
54. Briaud, J.L.; Ting, F.; Chen, H.C.; Cao, Y.; Han, S.W.; Kwak, K.W. Erosion function apparatus for scour rate
predictions. J. Geotech. Geoenviron. Eng. ASCE 2001, 127, 105–113. [CrossRef]
55. Briaud, J.L.; Chen, H.C.; Kwak, K.W.; Han, S.W.; Ting, F. Multifood and multilayer method for scour rate
prediction at bridge piers. J. Geotech. Geoenviron. Eng. ASCE 2001, 127, 114–125. [CrossRef]
56. Briaud, J.L.; Chen, H.C.; Li, Y.; Nurtjahyo, P.; Wang, J. SRICOS-EFA method for complex piers in fine-grained
soils. J. Geotech. Geoenviron. Eng. ASCE 2004, 130, 1180–1191. [CrossRef]
57. Duong, T.T. Assessment of Riverbank Stability—The Perspectives of Unsaturated Soils and Erosion Function.
Ph.D. Thesis, Ibaraki University, Mito, Japan, 2014.
58. Julien, P.Y. River Mechanics; Cambridge University Press: New York, NY, USA, 2002.
59. Chow, V.T. Open-Channel Hydraulic; McGraw-Hill: New York, NY, USA, 1959.
60. Hickin, E.J. River Hydraulics and Channel Form; Wiley: Chichester, UK, 2004.
61. Goharrokhi, M. Effect of Hydraulic Shear Stress on the Banks of the Red River. Master’s Thesis, The University
of Manitoba, Winnipeg, MB, Canada, 2015.
62. Thomas, J.M. Erodibility Testing of Cohesion Soil. Master’s Thesis, Auburn University, Auburn, Alabama,
2009.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).