Riverbank stability assessment under flooding conditions in the Red River of Hanoi, Vietnam

Computers and Geotechnics 61 (2014) 178–189 Contents lists available at ScienceDirect Computers and Geotechnics journal homepage: www.elsevier.com/locate/compgeo Riverbank stability assessment under flooding conditions in the Red River of Hanoi, Vietnam Thi Toan Duong a,⇑, Hideo Komine a, Minh Duc Do b, Satoshi Murakami a a b Department of Urban and Civil Engineering, Ibaraki University, Japan Department of Geotechnics, VNU University of Science, Vietnam National University, Hanoi, Viet Nam a r t i c l e i n f o Article history: Received 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 stability 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 10À6 m/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 10À6 m/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 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 significantly influence the positive pore-water pressures [7–12], transient 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 contributes 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 fluctuations 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 drawdown, the FOS decreases significantly due to loss of the confining ⇑ Corresponding author Tel.: +81 08046300981 E-mail address: duongtoan109@gmail.com (T.T Duong) http://dx.doi.org/10.1016/j.compgeo.2014.05.016 0266-352X/Ó 2014 Elsevier Ltd All rights reserved 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 variation 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 investigators in the literature [11,12], which shows that the FOS is not affected by the drawdown rate when the hydraulic conductivity is low (10À8 m/s) However, when hydraulic conductivity is relatively high (>10À6 m/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 process 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 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 179 Fig Red River location in the Hanoi area Fig Location of the NT1 site on the Red River bank Fig Location of the XC site on the Red River bank Red River in Hanoi, Vietnam were analysed by considering both homogeneous and non-homogeneous materials using finite element 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 180 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 bank sites have been facing significant damage risks due to river meandering activities of both of the flow channels (the main channel of the Red River and the sub-channel 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 (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 (NT2) on the bank site (see Fig 4) 2.1 Case study background Fig Houses built near the river bank in NT2 Study area Fig shows 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 in Fig 1) Three riverbank sites located along the eastern side of the river were studied; 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 and Ngoc Thuy 2) These Table summarises 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 5–7 present the configurations and distribution of the simulated riverbanks’ initial porewater 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 À35 kPa, À40 kPa, and À80 kPa for XC, NT1, and NT2, Table Bank soil layers and hydraulic conditions used in simulated models for riverbank stability analysis Location Bank surface elevation (H) (m) Bank slope (°) Xuan Canh 10 Ngoc Thuy 12 Ngoc Thuy 25 Soil layer and elevation of surface soil layer Hydraulic conditions 81 Layer 1: 10–(À1)m XC Layer 2: (À1)-down bed sand 78 Layer Layer Layer Layer Layer Layer Layer 12–10 m silt NT11 10–9.5 m sand NT12 9.5–6 m silt NT13 6–5.5 m sand NT12 5.5–1.5 m silt NT14 1.5–1 m sand NT12 1-down bed sand Layer 1: 25–0 m silt clay NT2 Layer 2: 0-down bed sand 68 1: 2: 3: 4: 5: 6: 7: Initial elevation water level (m) Rate of water level change (m/d) Rise rates: 0.3, 0.5, 0.8 and m/day Draw down rates: 0.3, 0.5, 0.8 and m/day Fig The riverbank configuration along with the initial pore-water pressures at XC T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 181 Fig The riverbank configuration along with the initial pore-water pressures at NT1 Fig The riverbank configuration along with the initial pore-water pressures at NT2 respectively The maximum negative pressure head values for XC, NT1, and NT2 were 3.5 m, m, and 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 m, m, and 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 1971, 1996, 2000, and 2008) that have occurred in the past 50 years are shown in Fig The RWL is typically as low as 2– m during the dry season and as high as 8–14 m during the flooding season The rates of the RWL changes were estimated using the ratio of the changes to the RWL increment (m) and the elapsed time (days); these rates ranged from 0.1 m/d to m/d This information with respect to the changes in RWL values and their rates were then used in the model simulation 2.2 Hydraulic conditions The database from the National Hydro-Meteorological Service [21] at the Hanoi station (K65) was used for this study The flooding season typically coincides with the rainy season lasting 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., Methods 3.1 Soil testing Undisturbed and disturbed soil samples were collected at shallow depths at the XC, NT1, and NT2 locations The water contents 182 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 15 Year 14 1971 1996 2000 2008 Elevation of river water level (m) 13 12 11 10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 Time (days) Fig The daily change in the river water level during the flood season from June to October and bulk densities were determined according to the ASTM standards (ASTM D 2216 and D 2937-00, respectively) The SWCC, saturated shear strength parameters and saturated 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 measured 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 program 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 Geoslope program (Geo-slope International Ltd.) [25–29] The procedure 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 initial RWL was established on the boundary of the riverbank To simulate a change to the RWL, a boundary condition was defined using a function of water head versus time in the transient model The 100 Percent finer by weight (%) 90 80 70 60 50 Soil NT2 NT11 NT13 NT14 NT12 XC Bed river sand 40 30 20 10 0.0001 0.001 0.01 0.1 10 100 Grain Size (mm) Fig Grain size distribution of the soils at the river bank of XC, NT1 and NT2 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 configuration 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 Table Soil properties used in riverbank stability analysis Parameter Xuan Canh Ngoc Thuy (NT1) XC NT11 NT12 NT13 NT14 NT2 Mean grain size, D50, (mm) Hydraulic conductivity (m/s) Effective cohesion (kPa) Internal friction angle (°) 0.040 2.24 Â 10À6 30 0.009 8.32 Â 10À8 10 30 0.150 1.35 Â 10À4 39.8 0.013 4.32 Â 10À7 32.41 0.056 2.19 Â 10À6 32.82 0.0017 1.04 Â 10À8 10 30 The parameters of suction curves Saturated volume water content (%) Air-entry value (kPa) a n m 45 10 16.59 3.535 0.476 44.5 32 53.12 2.03 0.44 43 9.5 10.78 19.99 0.641 30.05 27 35.76 4.19 0.834 43 20 28.1 6.8 0.564 57 50 61 1.72 0.38 a, n, m are fitting data calculated by the Fredlund and Xing [22] equations Ngoc Thuy 183 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 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 2.4 40 2.2 Soil NT2 NT11 NT13 NT14 NT12 XC 20 10 100 1000 Factor of Safety Volumetric Water Content (%) 60 1.8 1.6 1.4 Suction (kPa) 1.2 Fig 10 Soil–water characteristic curves Hydraulic conductivity (m/s) 0.001 0.0001 1E-005 1E-006 1E-007 1E-008 1E-009 1E-010 1E-011 1E-012 1E-013 1E-014 1E-015 1E-016 1E-017 1E-018 1E-019 0.8 0.6 12 10 River water level (RWL) drawdown (m) Fig 13 The relationship between the FOS and RWL during the drawdown processes at XC and NT1 Soil NT2 NT11 NT13 NT14 NT12 XC 2.4 10 100 1000 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 2.2 Suction (kPa) Fig 11 Hydraulic conductivity functions for the unsaturated soils Factor of Safety 1.8 2.6 2.4 2.2 1.6 1.4 1.2 Factor of Safety 1.8 0.8 1.6 0.6 12 1.4 15 18 21 24 27 30 Time (days) 1.2 Fig 14 The relationship between the FOS and the elapsed time during the drawdown processes at XC 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 0.8 0.6 10 12 14 River water level (RWL) rising up (m) Fig 12 The relationship between the FOS and RWL during the rise processes at XC and NT1 were used to represent typical building loads, and the house loading stress was determined according to the Vietnam Design Standard for Loading and Dynamics [30] The finite element method was used in SLOPE/W to calculate the FOS of the different loading scenarios Results 4.1 Soil bank properties of loading due to nearby houses was also analysed A typical singlefamily 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 The grain-size distribution curves and saturated soil properties determined using the soil samples collected from XC, NT1, and NT2 184 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 4.2 Riverbank stability corresponding to water level changes 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 1.8 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 in Figs 12 through 17 The riverbanks at XC and NT1 were analysed with no external stress and the riverbank at NT2 was analysed with variable loading stresses Factor of Safety 1.6 1.4 1.2 0.8 10 15 20 25 30 35 40 Time (days) Fig 15 The relationship between the FOS and the elapsed time during the drawdown processes at NT1 are presented in Figs through 11 and Table 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 10À6 m/s, except for the sand layer, whose value was approximately 10À4 m/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 11 and Table present the SWCC, unsaturated hydraulic conductivity function and fitting data, respectively 4.2.1 Riverbank stability without vertical loading stress The relationships between the FOS and RWL changes for XC and NT1 are summarised in Figs 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 values These banks failed when the FOS value dropped below m as the RWL decreased to m at XC and to 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 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 15 present 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 17 present 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 2.8 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 2.4 Factor of Safety 2.2 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 2.6 2.4 without Building 2.2 Factor of Safety 2.6 1.8 50kPa 1.6 1.4 100kPa without Building 1.8 50kPa 1.6 1.4 100kPa 1.2 1.2 200kPa 200kPa 1 300kPa 300kPa 0.8 0.8 0.6 0.6 12 16 River water level rising up (m) 20 24 10 12 14 16 18 Time (days) Fig 16 The relationship between the FOS and RWL and the relationship between the FOS and the elapsed time during the rise processes at NT2 20 185 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 2.8 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 2.4 without Building 2.2 Factor of Safety 2.4 without Building 2.2 1.8 50kPa 1.6 1.4 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 2.6 Factor of Safety 2.6 100kPa 1.8 1.6 50kPa 1.4 1.2 100kPa 1.2 200kPa 300kPa 200kPa 0.8 300kPa 0.8 0.6 0.6 0.4 24 20 16 12 10 15 20 25 30 35 Time (days) River water level drawdown (m) 20 18 16 14 12 10 10 15 20 25 30 35 40 45 50 -P=-35kPa -200 -180 -160 -140 -120 -100 -80 -60 -40 -20 Pressure Head (m) Pressure (kPa) 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 -P +P Volumetric Water Content (%) Hydraulic conductivity (m/s) Fig 18 Soil–water characteristic curve and the initial pore water pressure condition at XC 0.001 0.0001 1E-005 1E-006 1E-007 1E-008 1E-009 1E-010 1E-011 1E-012 1E-013 1E-014 1E-015 1E-016 1E-017 k=2.24 x 10-4 m/s at saturation k=2.24 x 10-5 m/s at saturation k=2.24 x 10-6 m/s at saturation 10 100 1000 Suction (kPa) Fig 19 Three cases of the variation of simulated hydraulic conductivity at XC sites 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 Â 10À6 m/s, which is equal to the natural hydraulic conductivity of the XC silt, 2.24 Â 10À5 m/s, and 2.24 Â 10À4 m/s RWL rise rates of 0.1 m/d and 0.3 m/d were used in these simulations Fig 18 presents 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 À35 kPa (Fig 6) Fig 19 shows the three simulated hydraulic conductivity functions for saturated permeability values of ks = 2.24 Â 10À6 m/s, 2.24 Â 10À5 m/s, and 2.24 Â 10À4 m/s The unsaturated hydraulic conductivities for the initial negative pore-water pressure of À35 kPa were estimated as 10À8 m/s, 10À7 m/s, and 10À6 m/s, respectively (Fig 19) Fig 20 presents 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 at a high hydraulic conductivity of 10À6 m/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 10À7 m/d and 10À8 m/d, the probability of bank failure was high when the RWL rose from m to m at a rate of 0.1 m/d The FOS subsequently increased after the RWL rose higher than m Above m, the soil had a lower hydraulic conductivity value than the rate of the RWL, so the groundwater table did not rise rapidly Confining pressure was a dominant factor contributing to an increase in the FOS in this scenario Figs 21 and 22 show the groundwater table’s responses to a RWL increase rate of 0.1 m/d for hydraulic conductivity values of 10À6 m/s and 10À8 m/s, respectively At a high hydraulic conductivity (i.e., 10À6 m/s, Fig 21), the groundwater table reached a high level and caused failure (the FOS dropped to less than in Fig 20, left) However, the groundwater table reached a lower level at a lower hydraulic conductivity (i.e., 10À8 m/s, Fig 22), corresponding to a higher FOS value These results elucidated the manner 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 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] 186 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 2.4 2.4 Simulating for XC water rising at 0.1m/d with ks=2.24 x 10-6m/s; kp =10-8m/s 2.2 ks=2.24 x 10-5m/s; kp =10-7m/s ks=2.24 x 10-4m/s; kp =10-6m/s 2 1.8 1.8 Factor of Safety Factor of Safety Simulating for XC water rising 0.3m/d with ks=2.24 x 10-6m/s; kp =10-8m/s ks=2.24 x 10-5m/s; kp =10-7m/s ks=2.24 x 10-4m/s; kp =10-6m/s 2.2 1.6 1.4 1.6 1.4 1.2 1.2 1 0.8 0.8 0.6 0.6 River water level rising up (m) 10 River water level rising up (m) 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 10À6 m/s and a RWL rate of 0.1 m/d Fig 22 Levels of groundwater change with a hydraulic conductivity of 10À8 m/s and a RWL rate of 0.1 m/d 10 187 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 Fig 23 The simulated path of the rapid water flow in the sand layer in the non-homogeneous bank, NT1 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 23 shows the seepage path with a vector diagram along the sand layer The riverbank failure occurred due to the rapid increase in the porewater pressure Fig 24 presents the results for the FOS calculated with assumed sand layer thicknesses of 0.5 m, 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 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 1.8 NT1: RWL rising at 0.3m/d with 1.5m sand layer 1m sand layer 0.5m sand layer 1.6 Factor of Safety The 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 inducing 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 10À6 m/s) The FOS was not reduced because there were no significant changes in the positive pore-water pressures The summarised results differ from the results of other investigators 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 discrepancies 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 rising RWL and increasing 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] 1.4 1.2 0.8 10 12 River water level rising up (m) Fig 24 The effects of sand layer thickness on the NT1 bank stability 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 groundwater 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 m/d Previous studies [11,12] have suggested that a decrease in the FOS is not affected by the drawdown rate when the hydraulic conductivity is low (10À8 m/s) Most of the soil materials in the present study had hydraulic conductivity 188 T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 This study described the possible mechanisms of riverbank failure and the various factors influencing soil properties (i.e., hydraulic conductivity), pore-water pressure, and confining pressure and the rate of RWL change However, the water shear stress or seepage velocity, which would contribute to the erosion of soil particles in the bank-toe, was not considered Those topics will require further investigation 1.2 Simulating for XC with k=0.1935m/d,water down 0.3m/d down 0.3m/d ks=2.24 x 10-6m/s; RWL -6 RWL down 0.5m/d kk=0.1935m/d, water down 0.5m/d s=2.24 x 10 m/s; -5m/s; RWL down 0.3m/d kk=1.935m/d, drawdown at 0.3m/d s=2.24 x 10 water -5m/s; RWL down 0.5m/d kk=1.935m/d, drawdown at 0.5m/d s=2.24 x 10 water k=19.35m/d; downdown at 0.3m/d m/s; RWL 0.3m/d ks=2.24 x 10-4water k=19.35m/d; downdown at 0.5m/d m/s; RWL 0.5m/d ks=2.24 x 10-4water 1.1 Factor of Safety Conclusions The work presented in this paper evaluated riverbank stability along the Red River flowing through the Hanoi area Three different locations on the riverbank, all of which have different conditions, were analysed using Geo-slope software and data that included the results of field soil investigations, monitored hydraulic conditions, and geotechnical soil properties The results obtained in this study elucidated the mechanism of riverbank failures associated with RWL fluctuations The following conclusions were derived 0.9 0.8 0.7 0.6 12 10 River water level drawdown (m) Fig 25 The effects of hydraulic conductivity on the XC bank stability during the drawdown process values higher than 10À8 m/s, meaning the determined FOS values were not significantly affected by the rates of the RWL changes To identify the critical hydraulic conductivity with which control the effects of the RWL change rate on the FOS value, three different saturated hydraulic conductivity values of ks = 2.24 Â 10À6 m/s, 2.24 Â 10À5 m/s, and 2.24 Â 10À4 m/s were analysed The soil bank was assumed to be fully saturated, with the initial groundwater table equal to the river level at 10 m The rates of drawdown were 0.3 m/d and 0.5 m/d, and the FOS curve decreased with decreasing hydraulic conductivity The FOS values were higher when the RWL was drawn down at a lower rate (0.3 m/d) and a higher hydraulic conductivity (ks = 2.24 Â 10À5 m/s, 2.24 Â 10À4 m/s) (Fig 25) Furthermore, the FOS had the same value for the drawdown rates of 0.3 m/d and 0.5 m/d at a lower hydraulic conductivity value (ks = 2.24 Â 10À6 m/s) Based on these results, it could be concluded that the FOS was significantly influenced by the drawdown rate in soil with a high hydraulic conductivity (>10À6 m/s) However, the FOS was not affected by the drawdown rate in soil with a low hydraulic conductivity (approximately 10À6 m/s) 5.4 Effect of surcharge loading on the riverbank At NT2, which was simulated to include the effects of loading stress, the FOS was affected by the confining pressure of the RWL without loading stress or with low loading stress (50 and 100 kPa) At a higher loading stress of 200 kPa, the FOS was controlled predominantly by the loading stress rather than other factors The NT2 bank site was stable for both RWL processes (i.e., increase and draw down) without the surcharge loading However, the probability of bank site failure increased when the loading stress was greater than 100 kPa under drawdown conditions For this reason, it would not be safe for people to live near the riverbank at NT2 because the houses have been built without any protection measures along the segment of the Red River flowing through the Hanoi area The results obtained in this study showed that building on natural soil banks without protective construction measures should be avoided within 50 m from the lateral bank (1) During a short-term flood event, the FOS of the study riverbank increased with increasing RWL values and decreased with decreasing RWL values Water drawdown was the primary process causing bank failure in the studied bank areas at XC, NT1, and NT2 In terms of the relationship between the FOS and elapsed time, the FOS exhibited higher values with higher rates of RWL increase and lower values with higher rates of RWL drawdown (2) The FOS was approximately the same for soils with a low hydraulic conductivity (10À6 m/s), the pore-water pressure and rate of RWL change were the primary factors affecting the FOS During the RWL rising process, the FOS decreased and bank failure occurred when the hydraulic conductivity was equal to or higher than the RWL increase rate During the drawdown process, the FOS was significantly affected by the drawdown rate and was higher when the river water level drawdown process occurred at a lower rate (3) A sand layer with a high hydraulic conductivity was regarded as a primary factor affecting the FOS value A sand layer with a thickness of less than m did not significantly affect the FOS at the NT1 bank site However, the FOS value decreased to less than 1, and riverbank failure occurred in the presence of a 1.5 m sand layer (4) Riverbank failure was observed at NT2 when a house with a load greater than 50 kPa was built at a distance of approximately 50 m from the riverbank These results indicated that increased surcharged loading (>50 kPa) weakened the riverbank and became the dominant factor controlling the FOS when the RWL changed These results also indicated that building homes or other structures on natural soil banks without protective construction measures should be avoided within 50 m of the lateral bank The modelling techniques presented in this paper could be used to ascertain the safe distance for construction along the Red River bank in the future Acknowledgments The authors appreciate our mentors Prof Nobuo Mimura and Prof Kazuya Yasuhara, who have greatly contributed to our academic improvement The first author also expresses sincere thanks to the 322 Project Scholarship for supporting his PhD studies This research was partly supported by the Environment Research and T.T Duong et al / Computers and Geotechnics 61 (2014) 178–189 Technology Development Fund (S-8) of the Ministry of Environment, Japan (FY2010-FY2014, refer to Fig A1) and a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology (FY2011–FY2013, refer to Fig A2) The third author would like to express appreciation for the support provided by the project sponsored by Vietnam National University, Hanoi coded QGTÐ.12.06 We also express our sincere gratitude to all the participants in these projects References [1] Langendoen EJ CONCEPTS: conservational channel evolution and pollutant transport system software manual Research Report No 16, USDA-ARS National Sedimentation; 2000 [2] Simon A A model of streambank stability incorporating hydraulic erosion and the effects of riparian vegetation In: Proceedings of the eighth federal interagency sedimentation conference (8th FISC), April 2–6, Reno, NV, USA; 2006 [3] Simon A, Pollen-Bankhead N, Thomas RE Development and application of a deterministic bank stability and toe erosion model for stream restoration Stream Restoration Dynamic Fluvial Syst: Sci Approaches Anal Tools Am Geophys Union Geophys Monogr Ser 2011;194:453–74 [4] Darby SE, Rinaldi M, Dapporto S Couple simulations of fluvial erosion and mass wasting for cohesive river banks J Geophys Res 2007;112(F03022):1–15 [5] Rinaldi M, Mengoni B, Luppi L, Darby SE, Mosselman E Numerical simulation of hydrodynamics and bank erosion in a river bend Water Resour Res 2008;44(W09428):1–17 [6] Luppi L, Rinaldi M, Teruggi LB, Darby SE, Nardi L Mornitoring and mumerical model of river-bank erosion processes: a case study along the Cecina River (central Italy) Earth Surf Proc Land 2008;34(4):530–46 [7] 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 Landforms 1999;24:1095–114 [8] Rinaldi M, Casagli N Stability of streambanks formed in partially saturated soils and effects of negative pore-water pressures: Seive River (Italy) Geomorphology 1999;26:253–77 [9] Rinaldi M, Casagli N, Dapporto S, Gargini A Monitoring and modelling of porewater pressure changes and river-bank stability during flow events Earth Surf Proc Land 2004;29:237–54 [10] Dapporto S, Rinaldi M, Casagli N Failure mechanisms and pore pressure condition analysis of Riverbank along the Arno River (Central Italy) Eng Geol 2001;61:221–42 [11] Nian T, Jiang J, Wan S, Luan M Strength reduction FE Analysis of the stability of bank slopes subjected to transient unsaturated seepage Electr J Geotech Eng 2011;16:165–77 [12] Berilgen MM Investigation of stability of slope under draw-down Comput Geotech 2007;34:81–91 189 [13] Fox GA, Wilson GV, Periketi RK, Cullum RF Sediment transport model for seepage erosion of streambank sediment J Hydrol Eng 2006;11(6):603–11 [14] Fox GA, Wilson GV, Simon A, Langendoen EJ, Akay O, Fuchs JW Measuring streambank erosion due to ground water seepage: correlation to bank porewater pressure, precipitation and stream stage Earth Surf Proc Land 2007;32(10):1558–73 [15] Chu-Agor ML, Fox GA, Cancienne RM, Wilson GV Seepage caused tension failures and erosion undercutting of hillslopes J Hydrol 2008;359(3– 4):247–59 [16] Chu-Agor ML, Wilson GV, Fox GA Numerical modeling of bank instability by seepage erosion undercutting of layered streambanks J Hydrol Eng 2008;13(12):1133–45 [17] Chu-Agor ML, Fox GA, Wilson GV Empirical sediment transport function predicting seepage erosion undercutting for cohesive bank failure prediction J Hydrol 2009;377:155–64 [18] Fox GA, Heeren DM, Wilson GV, Langendoen EJ, Fox AK, Chu-Agor ML Numerically predicting seepage gradient forces and erosion: Sensitivity to soil hydraulic properties J Hydrol 2010;389:354–62 [19] Thorne CR Processes and mechanisms of River Bank Erosion In: Hey RD, Bathurst JC, Thorne CR, editors Gravel bed rivers New York: Wiley, John Wiley & Sons; 1982 p 227–71 [20] Fredlund DG, Rahardjo H Soil mechanics for unsaturated soil Hoboken: Wiley Interscience, John Wiley & Sons; 1993 [21] Doan TT Report of Project: TC-DT/07-03-3 In: Assessment Geo-environment engineering and propose strategy for land used planning in dynamic area near Red River bank in Hanoi city Report submitted to Vietnam Ministry of Civil Engineering; 2006 [22] Fredlund DG, Xing A Equations for the soil water characteristic curve Can Geotech J 1994;31(3):521–32 [23] Van Genuchten MTh A closed-form equation for predicting the hydraulic conductivity of unsaturated soils Soil Sci Soc Am J 1980;44(5):892–8 [24] Vanapalli SK, Fredlund DG, Pufahl DE, Clifton AW Model for the prediction of shear strength with respect to soil suction Can Geotech J 1996;33:379–92 [25] GEO-SLOPE International Ltd Seepage Modeling with SEEP/W; 2007 Web: [26] GEO-SLOPE International Ltd Stress-deformation modeling with SIGMA/W; 2007 Web: [27] GEO-SLOPE International Ltd Stability modeling with SLOPE/W; 2007 Web: [28] Krahn J The 2001 R.M Hardy lecture: the limits of limit equilibrium analyses Can Geotech J 2002;40:643–60 [29] Aryal KP Slope stability evaluations by limit equilibrium and finite element methods Doctoral Thesis at Norwegian University of Science and Technology; 2006 ISBN 82-471-7882-6 [30] TCVN 2737-1995 Vietnam design standard for loading and dynamic Civil Engineering Publisher, Hanoi; 1996 [31] Schnellmann R, Busslinger M, Schneider HR, Rahardjo H Effect of rising water table in an unsaturated slope Eng Geol 2010;114:71–83 [32] Vandamme J, Zou Q Investigation of slope instability induced by seepage and erosion by a particle method Comput Geotech 2013;48:9–20  ... 178–189 179 Fig Red River location in the Hanoi area Fig Location of the NT1 site on the Red River bank Fig Location of the XC site on the Red River bank Red River in Hanoi, Vietnam were analysed... been facing significant damage risks due to river meandering activities of both of the flow channels (the main channel of the Red River and the sub-channel of the Duong River, see Fig 1) The soil... [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