Riverbank stability assessment under flooding conditions in the RedRiver of Hanoi, Vietnam Thi Toan Duonga,⇑, Hideo Kominea, Minh Duc Dob, Satoshi Murakamia a Department of Urban and Civi
Trang 1Riverbank stability assessment under flooding conditions in the Red
River of Hanoi, Vietnam
Thi Toan Duonga,⇑, Hideo Kominea, Minh Duc Dob, Satoshi Murakamia
a
Department of Urban and Civil Engineering, Ibaraki University, Japan
b
Department of Geotechnics, VNU University of Science, Vietnam National University, Hanoi, Viet Nam
Article history:
Received 2 August 2013
Received in revised form 25 May 2014
Accepted 25 May 2014
Keywords:
Riverbank stability
River water level change
Hydraulic conductivity
Loading surcharge
a b s t r a c t
The literature contains limited information on variations in the factors of safety (FOS) of riverbank stabil-ity associated with river water level (RWL) fluctuations This paper analyses a case study on the portion of the Red River flowing through Hanoi using the finite element method and extending the mechanics of saturated and unsaturated soils to understand how the riverbank’s FOS varies with RWL fluctuations The results show that hydrostatic force is one of the key parameters influencing the FOS when the soil’s hydraulic conductivity is less than 106m/s However, the pore-water pressure and rate of RWL change are the key parameters influencing the FOS when the hydraulic conductivity is greater than 106m/s The study also indicates that a surcharge of 50 kPa or higher significantly weakens the riverbank stability and influences the FOS when the RWL rises The construction of residential or other structures without taking special protection measures within 50 m of the lateral riverbank should be avoided for safety reasons
Ó 2014 Elsevier Ltd All rights reserved
1 Introduction
Riverbank failures are typically caused by the coupled effects of
gravitational forces and the soil erosion associated with river water
level (RWL) fluctuations[1–6] While gravitational forces
signifi-cantly influence the positive pore-water pressures [7–12],
tran-sient seepage flow contributes to tension cracks and erosion
undercutting[13–18]that lead to riverbank instability Four main
factors should be considered when analysing the riverbank failures
associated with RWL fluctuations: (i) seepage flow, which
contrib-utes to changes in the groundwater table or pore-water pressures;
(ii) the influence of suction on the unsaturated soil’s engineering
properties; (iii) hydrostatic forces (referred to as confining
pressure in this paper) acting on the riverbank; and (iv) water
shear stress, which causes bank-toe erosion
Several investigators have studied the effects of RWL
fluctua-tions and pore-water pressure on the factor of safety (FOS) values
for riverbanks in various Italian river systems[7–10] These studies
show that an increase in the confining pressure leads to an increase
in the FOS when the RWL rises In other words, the riverbanks are
often more stable when the RWL is relatively high During
draw-down, the FOS decreases significantly due to loss of the confining
pressure’s influence Changes in soil suction also have a significant influence on the soil properties in the unsaturated zone and the riverbank’s stability[7,19,20]
The influence of hydraulic conductivity variation (i.e., the vari-ation of hydraulic conductivity with respect to soil suction) on the riverbank stability has not been widely investigated using case study results However, the effects of drawdown on the riverbank stability of saturated soils have been analysed by some investiga-tors in the literature [11,12], which shows that the FOS is not affected by the drawdown rate when the hydraulic conductivity
is low (108m/s) However, when hydraulic conductivity is relatively high (>106m/s), the FOS is significantly influenced by the drawdown rate Some researchers have suggested that tension cracks and undercutting influence a riverbank’s stability
[13–15,17,18] Riverbank failures created for modelling studies have been attributed to seepage forces when the groundwater table in the riverbank remains high, even after the drawdown pro-cess is completed The mechanics of riverbank stability related to the RWL fluctuation have recently been developed and analysed
to study the effects of seepage and undercutting erosion [1–6] These studies show that the erosion of soil particles significantly contributes to triggering mass failure above an overhang Past studies have identified the primary mechanisms of riverbank failure related to RWL changes[8,13,15,18,19] However, these studies are limited considering to homogeneous soil materials In the present study, three riverbank sites along the
http://dx.doi.org/10.1016/j.compgeo.2014.05.016
0266-352X/Ó 2014 Elsevier Ltd All rights reserved.
⇑Corresponding author Tel.: +81 08046300981.
E-mail address: duongtoan109@gmail.com (T.T Duong).
Contents lists available atScienceDirect
Computers and Geotechnics
j o u r n a l h o m e p a g e : w w w e l s e v i e r c o m / l o c a t e / c o m p g e o
Trang 2Red River in Hanoi, Vietnam were analysed by considering both
homogeneous and non-homogeneous materials using finite
ele-ment analysis with the commercial software GeoSlope The primary
objectives of the study were as follows
(i) Investigate the effects of RWL fluctuations on the stability of
the Red River bank
(ii) Investigate the effects of loading due to the nearby houses
(iii) Discuss the mechanisms of riverbank failure associated with variations in the soil’s hydraulic conductivity and various factors influencing the riverbank’s stability, such as RWL fluctuations, pore water pressure, confining pressure, and the rate of RWL change
All of the work presented in this paper was conducted based on two assumptions: that (i) water flow has no effect on the scouring
of the bank toe and (ii) there is no rainfall infiltration into the surface
Fig 1 Red River location in the Hanoi area.
Fig 2 Location of the XC site on the Red River bank.
Fig 3 Location of the NT1 site on the Red River bank.
Trang 32 Study area
Fig 1shows a map of the Red River flowing through Hanoi The
river flows approximately 40 km through an urban area extending
from Thuong Cat to Van Phuc (k48–k85, shown inFig 1) Three
riv-erbank sites located along the eastern side of the river were
stud-ied; one bank site was located at the Xuan Canh commune (the
Xuan Canh bank site) and the other two bank sites were located
at the Ngoc Thuy commune (Ngoc Thuy 1 and Ngoc Thuy 2) These
bank sites have been facing significant damage risks due to river meandering activities of both of the flow channels (the main chan-nel of the Red River and the sub-chanchan-nel of the Duong River, see
Fig 1) The soil bank material at the Xuan Canh (XC) bank site is homogeneous, with uniform silt-sand (Fig 2) The Ngoc Thuy 1 (NT1) soil bank has a thin fine sand layer sandwiched between silt layers, and can be considered non-homogeneous (Fig 3) Several residential structures constructed close to the stretch from K63
to K65 on the riverbank area have been prone to stability problems This region was identified as Ngoc Thuy 2 (NT2) on the bank site (seeFig 4)
2.1 Case study background
Table 1summarises the geometry of the three bank sites (i.e., the surface elevation and slope), the soil layers, and the hydraulic properties, including the elevation of the initial RWL and the rates
of RWL change during a flood event.Figs 5–7present the configu-rations and distribution of the simulated riverbanks’ initial pore-water pressures at XC, NT1, and NT2, respectively The initial pore-water pressures (i.e., the maximum negative pressure head) were estimated from the soil–water characteristic curve (SWCC) Above the water table, the negative pressure decreased linearly
as the height increased up to the maximum negative pressure head corresponding to the typical height of the capillary fringe Above this capillary fringe, the negative pressure had a constant value of 35 kPa, 40 kPa, and 80 kPa for XC, NT1, and NT2,
Fig 4 Houses built near the river bank in NT2.
Table 1
Bank soil layers and hydraulic conditions used in simulated models for riverbank stability analysis.
Location Bank surface
elevation (H) (m)
Bank slope (°) Soil layer and elevation
of surface soil layer
Hydraulic conditions Initial elevation water level (m)
Rate of water level change (m/d)
Layer 2: (1)-down bed sand Ngoc Thuy 1 12 78 Layer 1: 12–10 m silt NT11 2 Rise rates: 0.3, 0.5, 0.8 and 1 m/day
Layer 2: 10–9.5 m sand NT12 Layer 3: 9.5–6 m silt NT13 Draw down rates: 0.3, 0.5, 0.8 and 1 m/day Layer 4: 6–5.5 m sand NT12
Layer 5: 5.5–1.5 m silt NT14 Layer 6: 1.5–1 m sand NT12 Layer 7: 1-down bed sand
Layer 2: 0-down bed sand
Trang 4respectively The maximum negative pressure head values for XC,
NT1, and NT2 were 3.5 m, 4 m, and 8 m, respectively At XC, NT1,
and NT2, the bank surface elevations (H) were 10 m, 12 m, and
25 m and the initial elevations of the RWL were 1 m, 2 m, and
5 m, respectively The elevations of the RWL were set to change
with the elapsed time The peak river-stage values were 10 m,
12 m, and 20 m at XC, NT1, and NT2, respectively
2.2 Hydraulic conditions
The database from the National Hydro-Meteorological Service
[21]at the Hanoi station (K65) was used for this study The
flood-ing season typically coincides with the rainy season lastflood-ing from
June to October During this period, the RWL goes through several
cycles of rises and falls The daily RWL changes during the flooding
seasons (1st June–30th October) of four excessive flood years (i.e.,
1971, 1996, 2000, and 2008) that have occurred in the past
50 years are shown inFig 8 The RWL is typically as low as 2–
6 m during the dry season and as high as 8–14 m during the flood-ing season The rates of the RWL changes were estimated usflood-ing the ratio of the changes to the RWL increment (m) and the elapsed time (days); these rates ranged from 0.1 m/d to 1 m/d This infor-mation with respect to the changes in RWL values and their rates were then used in the model simulation
3 Methods 3.1 Soil testing
Undisturbed and disturbed soil samples were collected at shal-low depths at the XC, NT1, and NT2 locations The water contents
Fig 6 The riverbank configuration along with the initial pore-water pressures at NT1.
Fig 7 The riverbank configuration along with the initial pore-water pressures at NT2.
Trang 5and bulk densities were determined according to the ASTM
stan-dards (ASTM D 2216 and D 2937-00, respectively) The SWCC,
sat-urated shear strength parameters and satsat-urated hydraulic
conductivity were determined using a pressure plate apparatus,
triaxial shear apparatus (JGS 0524: 2000), and the falling head
method (JIS A: 1218), respectively Reconstituted samples were
prepared at predetermined water contents and dry density values
to determine these soil properties All soil properties were
mea-sured in a laboratory at Ibaraki University, Japan
Fredlund and Xing’s[22]equation was used to determine the
fitting parameters of the SWCC, and the van Genuchten model
[23] was applied to determine the hydraulic conductivity of the
unsaturated soil using the hydraulic function in the SEEP/W
pro-gram The shear strength of the unsaturated soil was determined
by the Vanapalli model [24] using the saturated shear strength
parameters and the SWCC with the aid of the SLOPE/W program
3.2 Modelling of the seepage flow, loading surcharge, and riverbank
stability
The riverbank stability was analysed using the commercial
Geo-slope program (Geo-Geo-slope International Ltd.)[25–29] The
proce-dure for analysing the RWL fluctuation’s influence on the riverbank
stability included the following steps The riverbank configuration
was constructed using a finite element grid in the SEEP/W The
ini-tial RWL was established on the boundary of the riverbank To
sim-ulate a change to the RWL, a boundary condition was defined using
a function of water head versus time in the transient model The
simulated results were then used for a slope stability analysis using SLOPE/W Spencer’s method was applied to calculate the FOS in SLOPE/W, in which a constant inclination was assumed for inter-slice forces The FOS was computed to satisfy both the moment and force equilibrium conditions This model configura-tion was applied to analyse both the homogeneous soil bank (XC) and non-homogeneous bank (NT1)
For NT2, a coupled analysis was used in SIGMA/W to simulate the effects of the changes in pore-water pressure caused by RWL fluctuations, the stress of the river water, and the vertical loading associated with the houses built near the riverbank The influence
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Year 1971 1996 2000 2008
Time (days)
Fig 8 The daily change in the river water level during the flood season from June to October.
0 20 40 60 80 100
10 30 50 70 90
Soil NT11 NT13 NT14 NT12 XC Bed river sand
Grain Size (mm)
Fig 9 Grain size distribution of the soils at the river bank of XC, NT1 and NT2.
Table 2
Soil properties used in riverbank stability analysis.
Hydraulic conductivity (m/s) 2.24 10 6
8.32 10 8
1.35 10 4
4.32 10 7
2.19 10 6
1.04 10 8
The parameters of suction curves
Trang 6of loading due to nearby houses was also analysed A typical
single-family house in Hanoi is 1–5 floors high, with an area of 30–100 m2
per floor Thus, five different stresses of 0, 50, 100, 200, and 300 kPa
were used to represent typical building loads, and the house load-ing stress was determined accordload-ing to the Vietnam Design Stan-dard for Loading and Dynamics[30] The finite element method was used in SLOPE/W to calculate the FOS of the different loading scenarios
4 Results 4.1 Soil bank properties
The grain-size distribution curves and saturated soil properties determined using the soil samples collected from XC, NT1, and NT2
Suction (kPa)
0
20
40
60
Soil
NT2 NT11 NT13 NT14 NT12 XC
Fig 10 Soil–water characteristic curves.
1E-019
1E-018
1E-017
1E-016
1E-015
1E-014
1E-013
1E-012
1E-011
1E-010
1E-009
1E-008
1E-007
1E-006
1E-005
0.0001
0.001
Soil NT2 NT11 NT13 NT12 XC
Suction (kPa)
Fig 11 Hydraulic conductivity functions for the unsaturated soils.
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
Rising process
XC, RWL rising at 0.5m/d
XC, RWL rising at 1m/d NT1, RWL rising at 0.5m/d NT1, RWL rising at 1m/d
River water level (RWL) rising up (m)
Fig 12 The relationship between the FOS and RWL during the rise processes at XC
and NT1.
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4
2.6
Drawdown process
XC, RWL drawdown at 0.5m/d
XC, RWL drawdown at 1m/d NT1, RWL drawdown at 0.5m/d NT1, RWL drawdown at 1m/d
River water level (RWL) drawdown (m)
Fig 13 The relationship between the FOS and RWL during the drawdown processes at XC and NT1.
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
2.4
XC, Drawdown process RWL drawdown at 0.3m/d RWL drawdown at 0.5m/d RWL drawdown at 0.8m/d RWL drawdown at 1m/d
Time (days)
Fig 14 The relationship between the FOS and the elapsed time during the drawdown processes at XC.
Trang 7are presented inFigs 9through11andTable 2 The mean grain
size D50 ranged from 0.0017 to 0.04 mm for silt soil and 0.15–
0.18 mm for sand (Fig 9) The soil sample in NT2 had the highest
fines content, whereas that in XC had the lowest fines content
The fines content in NT1 decreased with the bank depth and the
presence of a sandwiched sand layer The grain size distribution
had a linear relationship with the saturated hydraulic conductivity,
which was lower in soil with higher fines content (Table 2) Most of
the soils had low hydraulic conductivity values close to or less than
106m/s, except for the sand layer, whose value was
approxi-mately 104m/s The effective cohesion (c0) was higher in soil with
higher fines content, but the friction angle (/0) was higher in soils
with higher sand content.Figs 10 and 11andTable 2present the
SWCC, unsaturated hydraulic conductivity function and fitting
data, respectively
4.2 Riverbank stability corresponding to water level changes
The riverbank stability corresponding to the RWL fluctuations is presented as (i) the relationship between the FOS and the elevation
of the RWL and (ii) the relationship between the FOS and the elapsed time during the RWL change and is summarised inFigs 12
through17 The riverbanks at XC and NT1 were analysed with no external stress and the riverbank at NT2 was analysed with vari-able loading stresses
4.2.1 Riverbank stability without vertical loading stress The relationships between the FOS and RWL changes for XC and NT1 are summarised inFigs 12 and 13 The FOS increased for the bank sites of both XC and NT1 with an increase in the RWL (Fig 12) Conversely, the FOS decreased with decreasing RWL val-ues These banks failed when the FOS value dropped below 1 m as the RWL decreased to 5 m at XC and to 4 m at NT1 (Fig 13) When considering the relationship between the FOS and RWL, the FOS did not change noticeably with the variation of the RWL The FOS had the same value for the water level change rates of 0.5 m/d and 1 m/d (Figs 12 and 13) In terms of the relationship between the FOS and elapsed time, the FOS was higher for a rising RWL and lower with a higher drawdown rate (Figs 14 and 15present the results for the drawdown process)
4.2.2 Riverbank stability with different vertical loading stresses The riverbank slope stability at NT2 was studied by applying five different loading stresses of 0, 50, 100, 200, and 300 kPa at a distance of 20–50 m.Figs 16 and 17present the effects of loading stress on the FOS during the RWL changes For a riverbank without
a loading stress, the FOS response to the RWL displayed the same trend as that shown for XC and NT1 The FOS increased at low stress values (i.e., 50 and 100 kPa), but decreased concomitantly
at a loading of more than 200 kPa (Fig 16) During the drawdown process, this bank failure occurred at a loading stress of 100 kPa (Fig 17)
4.3 Effects of hydraulic conductivity on riverbank stability Hydraulic conductivity is a key factor controlling the seepage flow in riverbank soil However, the materials in the three bank
0.8
1
1.2
1.4
1.6
1.8
2
NT1, Drawdown process RWL drawdown at 0.3m/d RWL drawdown at 0.5m/d RWL drawdown at 0.8m/d RWL drawdown at 1m/d
Time (days)
Fig 15 The relationship between the FOS and the elapsed time during the
drawdown processes at NT1.
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
Rising process Without Building , 1m/d Without Building, 0.5m/d 50kPa Building stress, 1m/d 50kPa Building stress, 0.5m/d 100kPa Building stress, 1m/d 100kPa Building stress, 0.5m/d 200kPa Building stress, 1m/d 200kPa Building stress, 0.5m/d 300kPa Building stress, 1m/d 300kPa Building stress, 0.5m/d
River water level rising up (m)
300kPa 200kPa 100kPa
without Building
50kPa
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
2.8
Rising process Without Building , 1m/d Without Building, 0.5m/d 50kPa Building stress, 1m/d 50kPa Building stress, 0.5m/d 100kPa Building stress, 1m/d 100kPa Building stress, 0.5m/d 200kPa Building stress, 1m/d 200kPa Building stress, 0.5m/d 300kPa Building stress, 1m/d 300kPa Building stress, 0.5m/d
Time (days)
300kPa 200kPa 100kPa without Building
50kPa
Trang 8sites on the Red River have low hydraulic conductivity To evaluate
the effects of hydraulic conductivity on bank stability when the
RWL changes, three values of saturated hydraulic conductivity
(ks) were assumed in the XC area: ks= 2.24 106m/s, which is
equal to the natural hydraulic conductivity of the XC silt,
2.24 105m/s, and 2.24 104m/s RWL rise rates of 0.1 m/d
and 0.3 m/d were used in these simulations.Fig 18presents the
SWCC and initial pore pressure distribution The volumetric water
content of a soil bank near the surface was 28%, and the negative pressure was determined to be 35 kPa (Fig 6) Fig 19 shows the three simulated hydraulic conductivity functions for saturated permeability values of ks= 2.24 106m/s, 2.24 105m/s, and 2.24 104m/s The unsaturated hydraulic conductivities for the initial negative pore-water pressure of 35 kPa were estimated
as 108m/s, 107m/s, and 106m/s, respectively (Fig 19)
Fig 20presents analytical results for XC with the simulated hydraulic conductivity during the RWL rise at rates of 0.1 m/d and 0.3 m/d The FOS dropped below 1 at a high hydraulic conduc-tivity of 106m/s and a low RWL rate of 0.1 m/d, indicating that bank failure had occurred (Fig 20) The FOS also decreased at an RWL rate of 0.3 m/d and RWL rise of 1–7 m For lower hydraulic conductivity values of 107m/d and 108m/d, the probability of bank failure was high when the RWL rose from 1 m to 4 m at a rate
of 0.1 m/d The FOS subsequently increased after the RWL rose higher than 4 m Above 4 m, the soil had a lower hydraulic conduc-tivity value than the rate of the RWL, so the groundwater table did not rise rapidly Confining pressure was a dominant factor contrib-uting to an increase in the FOS in this scenario
Figs 21 and 22show the groundwater table’s responses to a RWL increase rate of 0.1 m/d for hydraulic conductivity values of
106m/s and 108m/s, respectively At a high hydraulic conductiv-ity (i.e., 106m/s,Fig 21), the groundwater table reached a high level and caused failure (the FOS dropped to less than 1 in
Fig 20, left) However, the groundwater table reached a lower level
at a lower hydraulic conductivity (i.e., 108m/s,Fig 22), corre-sponding to a higher FOS value These results elucidated the man-ner in which hydraulic conductivity, pore-water pressure, and confining pressure affected the stability of the riverbanks Bank failure occurred when the hydraulic conductivity was equal to or greater than the rate of the RWL increase
5 Discussion 5.1 Effects of the RWL increase
The FOS increased with increasing RWL at the XC, NT1, and NT2 riverbanks when the stress was lower than 100 kPa These results are consistent with those obtained from earlier studies[7–12,19]
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
Drawdown process Without Building , 1m/d Without Building, 0.5m/d 50kPa Building stress, 1m/d 50kPa Building stress, 0.5m/d 100kPa Building stress, 1m/d 100kPa Building stress, 0.5m/d 200kPa Building stress, 1m/d 200kPa Building stress, 0.5m/d 300kPa Building stress, 1m/d 300kPa Building stress, 0.5m/d
River water level drawdown (m)
50kPa
300kPa 200kPa 100kPa without Building
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
2.8
Drawdown process Without Building , 1m/d Without Building, 0.5m/d 50kPa Building stress, 1m/d 50kPa Building stress, 0.5m/d 100kPa Building stress, 1m/d 100kPa Building stress, 0.5m/d 200kPa Building stress, 1m/d 200kPa Building stress, 0.5m/d 300kPa Building stress, 1m/d 300kPa Building stress, 0.5m/d
Time (days)
50kPa
300kPa 200kPa 100kPa without Building
Fig 17 The relationship between the FOS and RWL, and the relationship between the FOS and the elapsed time during the drawdown processes at NT2.
10 15 20 25 30 35 40 45 50
0
-20
-40
-60
-80
-100
-120
-140
-160
-180
-200
0 2 4 6 8 10 12 14 16 18 20
Volumetric Water Content (%)
+P
-P
Fig 18 Soil–water characteristic curve and the initial pore water pressure
condition at XC.
1E-017
1E-016
1E-015
1E-014
1E-013
1E-012
1E-011
1E-010
1E-009
1E-008
1E-007
1E-006
1E-005
0.0001
0.001
Suction (kPa)
Fig 19 Three cases of the variation of simulated hydraulic conductivity at XC.
Trang 90 1 2 3 4 5 6 7 8 9 10
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
Simulating for XC water rising at 0.1m/d with
River water level rising up (m)
k k k
s=2.24 x 10- 6m/s; kp =10-8m/s s=2.24 x 10- 5m/s; kp =10-7m/s s=2.24 x 10- 4m/s; kp =10-6m/s
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
2.4 Simulating for XC water rising 0.3m/d with
River water level rising up (m)
ks ks ks
=2.24 x 10- 6m/s; kp =10-8m/s
=2.24 x 10- 5
m/s; kp =10-7m/s
=2.24 x 10- 4m/s; kp =10-6m/s
Fig 20 The effect of the variation of simulated hydraulic conductivity on the XC bank stability.
Fig 21 Levels of groundwater change with a hydraulic conductivity of 10 6 m/s and a RWL rate of 0.1 m/d.
8
Trang 10The FOS increased because of the high confining pressure induced
on the riverbank by the rising RWL The confining pressure, or
water force, applied both vertical and horizontal pressures[27]
and the increased confining pressure was the primary factor
induc-ing increased resistance forces against bank failure Moreover, the
results obtained for the XC and NT1 sites in particular showed an
increase in the FOS values with increasing RWL values because
the saturated hydraulic conductivity was low (equal to or less than
106m/s) The FOS was not reduced because there were no
signif-icant changes in the positive pore-water pressures
The summarised results differ from the results of other
investi-gators reported in[13–18,31,32] It is important, however, to note
that these studies were based on seepage models that simulated
the RWL to obtain a groundwater table increase and FOS value
The FOS values calculated for the opposite side of the river were
affected by seepage flow and high pore-water pressures The FOS
values also decreased with increasing RWL because of the high
positive pore-water pressure and soil particle erosion caused by
seepage forces
The current model ignored the effects of water shear stress, so
the FOS changes modelled with increasing RWL showed some
dis-crepancies from the studies reported in the literature[1–6], where
riverbank geometries have been changed and redrawn based on
changes in the erosion distance from the toe of the bank The
FOS values presented in these studies therefore decrease with
ris-ing RWL and increasris-ing erosion distance The differences between
the results of the present study and previous studies are due to the
deformation of the riverbank associated with bank toe erosion,
which has also been asserted in the previous studies[6,15]
5.2 Effect of water level increases with different sand layer thicknesses
on NT1
The soil layers at the NT1 bank site were not homogeneous due
to an interstitial sand layer between the silt layers.Fig 23shows
the seepage path with a vector diagram along the sand layer The
riverbank failure occurred due to the rapid increase in the
pore-water pressure.Fig 24presents the results for the FOS calculated
with assumed sand layer thicknesses of 0.5 m, 1 m, and 1.5 m at
the NT1 bank site No apparent effect on the FOS was observed
when the sand layer had a thickness of less than 1 m, but riverbank
failure occurred with a 1.5 m sand layer
5.3 Effect of water level changes during drawdown
In contrast to the effects of increasing the RWL with soil having
a relatively low hydraulic conductivity, the FOS decreased during
the drawdown process, and similar modelling result trends have been reported in the literature[7,9,11,12] Bank failure occurred before the RWL drawdown occurred to its initial level The reason for the bank failure was attributed not only to the reduction in the confining pressure or the high positive pore pressure but also to the reduction in the soil suction The soil suction increased after drawdown, but did not recover to its previous values A decrease
in soil suction monitored from other case studies that supports the present modelling results is available in the literature [7] Moreover, the RWL drawdown occurred quickly when the ground-water table was high, contributing to a continuous decrease in the FOS until the groundwater table reached equilibrium with the RWL During this time, transient seepage toward the river often caused tension cracking and soil particle erosion[13–18] During the drawdown process, riverbank failure occurred at both the XC and NT1 bank sites with drawdown rates of 0.3 m/d, 0.5 m/d, 0.8 m/d, and 1 m/d Previous studies [11,12] have sug-gested that a decrease in the FOS is not affected by the drawdown rate when the hydraulic conductivity is low (108m/s) Most of the soil materials in the present study had hydraulic conductivity
Fig 23 The simulated path of the rapid water flow in the sand layer in the non-homogeneous bank, NT1.
0.8 1 1.2 1.4 1.6
1.8 NT1: RWL rising at 0.3m/d with
1.5m sand layer 1m sand layer 0.5m sand layer
River water level rising up (m)
Fig 24 The effects of sand layer thickness on the NT1 bank stability.