Professional Documents
Culture Documents
1
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
2
5 Department of Geotechnical Engineering, College of Civil Engineering, Tongji University, Shanghai, 20009, China
8 * Corresponding author.
9 Address: Geo-disaster Prevention Laboratory, West 2-1110, 744 Motooka, Nishi-ku, Fukuoka, 819-0395, Japan;
12
13 Abstract:
14 This paper analyzes the effects of complex geometries on three-dimensional (3D) slope stability using an
15 elastoplastic finite difference method (FDM) with strength reduction technique. A series of special 3D slopes with various
16 geometric configurations, including curving slope surface, turning corners, turning arcs, and turning forms, are presented
17 in terms of factor of safety, shear slip surface and deformed mesh. More than 180 cases with various geometries for
18 different slope gradient (90º, 45º and 26.57º) under different boundary conditions (smooth-smooth, rough-smooth and
19 rough-rough) are calculated and discussed in detail. Many interesting results are obtained and some of them appear to be
20 surprising. These results can be directly used for offering suggests for landslide hazard preparedness or safe and economic
22 Key words: three-dimensional slope stability, strength reduction technique, finite difference method, curving slope surface,
24 Introduction
25 Slope stability analysis remains one of the important problems in geotechnical engineering. In the framework of slope
1
Page 2 of 44
26 stability analysis, the vast majority of analyses are performed in two dimensions (2D) under the assumption of plane strain
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
27 conditions, in which the value of strain component perpendicular to the plane of interest is zero. The plane strain assumption
28 is valid if one dimension is very large in comparison with the other two. It also requires that no curvature or corner exists in
29 geometry of slope and no curvature exists in the shape of failure surface in the direction perpendicular to the plane of
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
30 interest, i.e., failure surface is the same in any cross section (Azizian and Popescu 2006). However, the vast majority of
31 slopes, both natural and constructed, exhibit a complex configuration and three-dimensional (3D) state, whereas slopes
32 satisfying the assumption of plain strain are seldom encountered. Hence, a 3D slope stability analysis is required to obtain
34 In recent years, there were various developments in the 3D slope stability analysis. These methods can fall into three
35 general categories: (1) limit equilibrium method (LEM), (2) limit analysis method (LAM), and (3) stress-strain method.
36 Each of these types of analysis has strengths and weaknesses. These methods are briefly retrospected as follow:
37 The most common approach to 3D slope stability analysis is still the LEM, which is usually a direct extension of the
38 various 2D slices methods to 3D columns methods, for example, the ordinary, modified Bishop, Morgenstern and Price,
39 Spencer, and Janbu methods (Hungr 1987; Hungr et al. 1989; Zhang 1988; Lam and Fredlund 1993; Jiang and Yamagami
40 2004; Razdolsky 2009), and numerous improvements to these methods (Hovland 1977; Chen and Chameau 1983; Duncan
41 1996; Stark and Eid 1998; Huang and Tsai 2000; Huang et al. 2002; Chang M 2002; Cheng et al. 2005; Silvestri 2006;
42 Cheng and Yip 2007; Zheng 2012). The 3D LEM is simple and fast. Unfortunately, the 3D LEM involves various
43 assumptions and is not readily modified to account for realistic boundary conditions in the third dimension. In addition,
44 another limitation of the 3D LEM is the difficulty of locating the critical general 3D failure surface in both shape and
46 Several 3D limit analysis models for slope stability analysis were also presented in literatures (Michalowski 1989;
47 2010; Chen et al. 2001a; 2001b; Farzaneh and Askari 2003; Chen et al. 2003a; 2005; Li et al. 2009; 2010). The limit
48 analysis method is used to find solutions to soil stability problems based on the lower and upper bound theorems in the
2
Page 3 of 44
49 theory of plasticity. In 3D slope stability analysis, the upper-bound method is widely popular (Donald and Chen 1997).
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
50 Optimization technique such as the linear or nonlinear programming was used to determine the critical FoS and
51 corresponding failure surface. The most important advantage of the upper-bound method is that the solution is approached
52 by the energy-work balance equation in which only one unknown, the factor of safety (FoS), is involved (Chen et al.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
53 2006). This point contributes greatly to the success of developing a theoretically more rigorous and numerically tractable
54 3D method. However, the construction of the 3D failure mechanism for limit analysis is difficult for a complicated slope,
56 Stress-strain method is another type of approach used for 3D slope stability analysis. In this method, a numerical
57 simulation is used to compute the displacement field to simulate the stress-strain behavior of the slope. Despite the fact that
58 this method has a large computational burden, it can simultaneously provide the FoS and the critical slip surface, together
59 with the stress, deformation, and progressive shear failure of the slope. With the developments of computer technology and
60 simulation approach in recent decades, this method is becoming increasingly popular and has been widely and successfully
61 used in 3D slope stability analysis. In addition, 3D stress-strain method can exactly reflect the slope dimensions, boundary
62 conditions, and a realistically complex geometric configuration in all three directions (Nian et al. 2012). The most
63 commonly applied stress-strain methods can be divided into two main categories: continuum approach and discontinuum
64 approach. The behavior of the slope studied here may be realistically reproduced under the continuum assumption.
65 Commonly used continuum approaches for slope stability are the finite element method (FEM) and the finite difference
66 method (FDM). The FEM solves a weak (variational) form of the governing equation on an unstructured mesh, whereas the
67 FDM solves a strong form, often on a structure grid (Zienkiewicz and Tayor 1989; Gharti et al. 2012). Compared to the
68 FEM, the FDM has a main advantage that the discretization and the solution procedure are relatively simple. Moreover, the
69 FEM or FEM combined with other method has been widely applied to 3D slope stability analysis (Ugai and Leshchinsky
70 1995; Cai et al. 1998; Chen et al. 2003a; Griffiths and Marquez 2007; Stolle and Guo 2008; Li et al. 2009; 2010; Stianson et
71 al. 2011; Nian et al 2012; Gharti et al. 2012), whereas the FDM has been relatively less (Chugh 2003; Wei et al. 2009).
3
Page 4 of 44
72 Based on this observation, an elastoplastic 3D FDM with strength reduction technique is used here to investigate the
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
73 stability of 3D slopes.
74 Existing 3D slope stability analysis, however, mainly emphasized 3D effects and boundary conditions, whereas the
75 effect of the complex geometries on 3D slope stability is rarely studied. In geotechnical practice, it’s clearly more likely that
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
76 the geometry of 3D slope will be complex. For example, nature slope often has a curving surface, turning corner often
77 appeared in excavation engineering (concave) and masonry engineering (convex), and turning arc often appeared in
78 open-pit mining engineering and road engineering in mountainous area. The effect of complex geometries on 3D slope
79 stability is still not very clear but it is important in slope engineering, especially in safe and economic design of
80 infrastructures. Cheng et al. (2007) and Wei et al. (2009) involved a vertical cut slope with 90º convex-turning corner in
81 their papers of comparative analysis among different 3D slope stability analysis methods. Farzaneh et al. (2008) analysed
82 convex slopes in plan-view based on the upper-bound theorem of the limit analysis approach. Michalowski and Drescher
83 (2009) presented an excavation using 3D limit analysis. Nian et al. (2012) studied several special 3D slopes (90º and 135º
84 convex turning corners slopes, a 180º convex turning arc slope and a 90º concave vertical cut) and draw some meaningful
85 conclusions. In general, existing studies related to slope geometries are scattered. With the aim of systematical study the
86 effects of complex geometries on 3D slope stability, an elastoplastic 3D FDM with strength reduction technique is used in
87 here to investigate the stability of 3D slopes with various complex geometries, including i) curving slope surface; ii)
88 convex- and concave-turning corners; iii) convex- and concave-turning arc; and iv) turning forms. All these geometries are
89 used for three examples that have different slope gradient (90º, 45º and 26.57º) under various boundary conditions
91 Three-
Three-dimensional FDM with strength reduction technique
93 The finite difference method code, FLAC3D (Fast Lagrangian Analysis of Continua in 3-Dimension) (Itasca 2007), is
94 used to carry out the computation. Three-dimensional analysis of elastic, perfectly plastic soils with a Mohr-Coulomb
4
Page 5 of 44
95 yield criterion combined with association rule (dilation angle is equal to friction angle) is performed, and no tensile
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
97 The convergence criterion for FLAC3D is the nodal unbalanced force, the sum of forces acting on a node from its
98 neighbouring elements (Dawson et al. 1999). If a node is in equilibrium, these forces should sum to zero. For this study,
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
99 the maximal unbalanced force of all nodes is normalized by the gravitational body force acting on that node. A simulation
100 is considered to have converged when the normalized unbalanced force of every node in the mesh is less than 10-5.
102 A generally accepted definition of FoS is that the ratio of the available strength of soil material to that required to
103 maintain equilibrium. Based on this definition, the current approach uses a shear strength reduction technique in which
c tanϕ
105 [1] c ' = , ϕ ' = arctan
SRF SRF
106 are used here, where c and φ are the actual values of soil material, cohesion and friction angle, respectively, c' and φ' are
107 the relevant parameters required to maintain the limit equilibrium, and SRF is a strength reduction factor. To obtain the
108 correct FoS, it is essential to trace the SRF gradually until the reduced strength parameters c' and φ' bring the slope to a
109 limit equilibrium state. In this process, the SRF is assumed to apply equally to both c and φ. When the slope reaches the
110 limit state, the FoS = SRF. The strength reduction technique is identical to those used in the LEM, LAM. This definition
111 used by Zienkiewicz et al. (1975), Matsui and San (1992), Ugai and Leshchinsky (1995), Duncan (1996), Griffiths and
112 Lane (1999), Dawson et al. (1999), Griffiths and Marquez (2007), Cheng et al. (2007), Zheng et al. (2009), Wei et al.
113 (2009) and others. The calculation of FoS is automatically carried out by the FLAC3D program.
115 For a 3D slope with any complex shape, boundary surfaces can be generally fall into four kinds: (1) bottom-surface;
116 (2) front-surface; (3) back-surface; and (4) side-surface (see Fig. 1). Usually, all movements are restrained in the bottom
117 plane of a model (i.e., ux, y, z=0, where u is the displacement); front-surface and back-surface are restrained in the normal
5
Page 6 of 44
118 direction, but free to move in the dip direction and strike direction. Different side-surface condition embodies different
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
119 boundary conditions. As most literatures about 3D FEM or FDM slope stability analysis (Chugh 2003; Griffiths and
120 Marquez 2007; Nian et al. 2012), the side-surface conditions can be generally classified into three types:
121 1. Smooth-smooth (SS) boundary condition: All side-surfaces are restrained in the normal direction, but free to move in
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
123 2. Rough-smooth (RS) boundary condition: Half of side-surfaces are restrained in the normal direction, but free to
124 move in the dip direction and strike direction; the other half of side-surfaces are restrained in all three normal, dip,
126 3. Rough-rough (RR) boundary condition: All side-surfaces are restrained in all three normal, dip, and strike directions.
127 All three kinds of boundary condition are often used in the real example or numerical simulation. In particular, RR
128 boundary condition can be used to represent a rigid contact with no possibility of movement. SS boundary condition is
129 used to represent a contact with a rigid, smooth abutment that can provide a reactive thrust, but no in-plane shear restraint.
130 And RS boundary condition usually implies a symmetric analysis of a RR boundary condition, and therefore only half of
131 the actual slope is analyzed for simplification. A useful technical instruction for practical slope engineering of the three
132 boundary conditions was illustrated by Nian et al. (2012). For details of the applicable condition, see the source reference.
133 Validations
Validations
135 Three typical slope examples with different slope gradients are analysed in this paper. The cross sections of three
136 examples are shown in Fig. 2. These slopes are inclined at angles of 90º, 45º and 26.57º to the horizontal, respectively.
137 Examples described here are homogeneous as the main objective of the current study is the effects of complex geometries
138 on 3D slope stability. Former two examples are a typical vertical cut and a simple slope with ratio of 1V:1H (V: vertical;
139 H: horizontal), respectively. They are taken from Wei et al. (2009). Example 3 is a gentle slope with ratio of 1V:2H. This
140 model was originally used in 2D by Fredlund and Krahn (1977) and later extended to 3D by Zhang (1988). Since then,
6
Page 7 of 44
141 various investigators used it as a benchmark for validating their methods (e.g. Lam and Fredlund 1993; Huang and Tsai
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
142 2000; Chen et al. 2001b; Chen et al. 2003b; Griffiths and Marquez 2007; Nian et al. 2012; Gharti et al. 2012). Table 1 lists
143 material properties of three examples. It should be noted that in the absence of data for Young’s modulus E, Poisson’s
144 ratio ν and dilation angle ψ from source literatures, they are estimated and given nominal values (e.g. E=10MPa, ν=0.25,
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
145 ψ =φ) as they have little influence on the predicted FoS in slope stability analysis (Dawson et al. 1999).
146 For verification of the present method, 3D analyses for three examples are presented for comparison with results
147 obtained by various investigators using other 3D slope stability techniques. To make adequate comparisons with the
148 existing literatures possible, the boundary condition and the extended width in the third dimension are as far as possible
149 the same with those given in source literatures. The boundary conditions for example 1 and example 2 are SS, therefore
150 the extended width of the slope have no influence on the FoS because of symmetry. For example 3, the extended width is
151 60m under RR boundary condition, but the actual mesh extended width varied by half this amount also because of
152 symmetry. The results from the FDM analysis in the present study, together with results obtained by several other
153 investigators who considered the same slope using different methods, are summarized in Table 2. We can found that the
154 FoS for three examples are in good agreement with the other solutions. The shear strain contour and the deformed mesh
155 for example 3 are shown in Fig. 3. It can be found that the failure mechanism is similar to that obtained by 3D LEM
158 The results from FDM with strength reduction technique are sensitive to the design of the mesh (Shukha and Baker
159 2003). In order to choose appropriate mesh model, two popular types of mesh form, hexahedra and tetrahedra, and three
160 mesh sizes, coarse, optimized and fine, are covered and validated for example 1. Six cases indicated in Table 3 are run to
161 illustrate the sensitivity of results. From the Table 3, it can be seen that all cases can give a reasonable slip surface but
162 wide-ranging FoS. For hexahedral models (case 1, case 2 and case 3), the obtained FoS are 1.72, 1.61 and 1.57 for coarse,
163 optimized and fine meshes, respectively. The same change trend (FoS = 2.35, 1.92 and 1.89 for case 4, 5 and 6,
7
Page 8 of 44
164 respectively) can be found for tetrahedron models but the values of FoS are much larger than those obtained from
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
165 hexahedral models with the same mesh size. In addition, mesh parameters including numbers of nodes and elements,
166 elapsed times are listed in Table 3. The elapsed time for tetrahedron models are much longer than those spend in
168 With either type of element, the fewer the number of nodes, the lower the accuracy and the less elapsed time. This
169 point is obvious and comparable to those of other reputable researches. It’s the main reason of difference in elapsed time
171 The reasons for significant different results between two element forms (hexahedron and tetrahedron) are: i)
172 Hexahedron can generally deform in a lower strain energy state, thus making them more accurate than tetrahedron in
173 numerous situations, ii) Tetrahedral model can produce a larger discretization error in the process of finite difference
174 method than hexahedral model as the grid discretization of the former is less uniform than that of the latter, and iii) The
175 numbers of nodes and elements for hexahedrons are less than those for tetrahedrons under the same element scale
176 situation, and these numbers have directly relationships with elapsed time and memory consumption. Thus, the mesh form
177 selection criterion is: for simple geometries, or for applications in which it is possible to build a mesh ‘by hand,’ the
178 hexahedral element (called ‘brick’ in FLAC3D) is the better choice; nevertheless, tetrahedral element is selected for those
179 cases in which the 3D model cannot be decomposed into an assembly of bricks and the number (or the percent) of
181 Therefore, hexahedral mesh form is chosen in present study and the mesh size is fine as far as possible as long as the
182 elapsed time is acceptable. Otherwise, optimized mesh size is selected to reduce the unacceptable computing time.
184 In 2D slope stability analysis, the plane strain assumption requires no curvature exist in geometry of slope, say, slope
185 surface is a plane in the direction perpendicular to the plane of interest. However, many natural and constructed slopes
186 exhibit a curvature in the third dimension on slope surface and therefore 2D slope stability analysis is inadaptable. On the
8
Page 9 of 44
187 other hand, existing 3D slope stability analysis rarely emphasized this issue. 3D slopes with curving slope surfaces are
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
188 often appear in practical slope engineering, e.g. natural mountain slope. In order to investigate the effect of curving slope
189 surface on stability, the FoS and potential slip surface of 3D slopes with various curvatures are presented in following. It
190 should be noted that the curving slope surface studied in here is curvature in the third dimension, the direction
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
191 perpendicular to the plane of interest, while curvature in the plane of interest can analysed using 2D method and not be
193 3D slopes with different degree of convex and concave curvatures for three examples are considered in the present
194 analysis. An instance with example 2, 3D slopes with different convex and concave curvatures are shown in Fig. 4. The
195 degree of curvature is measured using the ratio of maximal convex or concave length L to half of slope width W/2,
196 Rcur=L/(W/2) (see Fig. 4(a) (b)). For each curvature case, two types of constraints, RR and SS boundary conditions, are
197 incorporated into the present analysis. The FoS of the 3D slope corresponding to various curvatures and constraints are
198 calculated and listed in Table 4. Differences of the FoS of slopes with and without curvature, DF = (FoS-FoScommon) /
200 It can be seen from Table 4 that i) the boundary condition has a significant influence on the FoS for any case with a
201 given curving slope surface; ii) for SS boundary condition, both convex and concave curvature slopes are more stable
202 than common slope but the effect is insignificant (less than 3%); iii) for RR boundary condition, the FoS of three
203 examples increases with the change curvature from the greatest degree of convex to the greatest degree of concave and
204 the effect of curvature is significant (larger than 10%); iv) slope gradient has obviously influence on the effect of
205 curvature, steep slope has more significant influence than gentle slopes except concave curving cases under SS boundary
206 condition.
207 In order to further investigate the effect of curvature on 3D slope stability under different boundary conditions, shear
208 strain contours and deformed meshes for three representative curvature slopes of example 2 under two boundary
209 conditions are shown in Figs. 5 and 6. It can be found that the impact of curving slope surface on potential slip surface is
9
Page 10 of 44
210 obvious under SS boundary condition. The slip surface seems has the same shape with the slope surface configuration.
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
211 For SS boundary condition, the curving slip surface has the larger contact area between the potential slip mass and the
212 stable mass than that of common 3D slip surface. This is exactly why the FoS for 3D slopes with curving slop surface is
213 larger than that given by a common 3D slopes under SS boundary condition. In the other hand, RR boundary condition
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
214 can change the 3D slope’s failure surface and the volume of failure mass. And the volume of convex slope is larger than
215 that of concave one. As we know, under the same control condition, the smaller the volume of failure mass, the larger the
216 FoS. That’s why the RR boundary condition makes the FoS of a convex slope have an insignificant improvement (see Fig.
217 6(a) (d)) whereas that of a concave slope have a remarkable improvement (see Fig. 6(c) (f)). This could also explain why
218 the FoS increases with the change curving slope surface from the greatest degree of convex to the greatest degree of
219 concave under RR boundary condition. 3D slopes with other gradients have the similar results.
221 The 3D slopes with turning corners often appear in practical engineering, e.g. excavation engineering and open pit
222 mining. These slopes (see Fig. 7(a) ~ (e)) can be considered as: two common 3D slopes parallelly closed to each other at
223 first and then one of them rotates about the midpoint of slope surface sideline. Different rotation direction (left or right)
224 and rotation angle forms different 3D turning corners slopes (see Fig. 7(f) ~ (j)). In this study, 3D slopes with 90º, 135º
225 and 180º convex- and concave-turning corners for three examples under three boundary conditions are considered.
226 Specially, 3D slope with 180º turning corner actually is the common 3D slope (see Fig. 7(c) (h)). The FoS corresponding
227 to various turning corners under three boundary conditions are calculated and listed in Table 5. Difference of the FoS
228 between turning corner slopes and common slope are also shown in Table 5.
229 From Table 5, we can found that: i) for convex-turning corners, the FoS for a vertical cut slope decreases as the slope
230 turning corner decreases from 180º to 90º while entirely opposite result can be found for 1V:1H and 1V:2H slopes
231 regardless of the boundary conditions; ii) for concave-turning corners, the FoS for three examples increases as the slope
232 turning corner increases from 180º to 90º under three boundary conditions; iii) the influence of concave-turning corners
10
Page 11 of 44
233 on the FoS is larger than that from convex-turning corners (about twice); iv) the FoS will increase as the boundary
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
234 constraints for a given slope corner (from SS to RS then RR boundary condition) but the increment from SS to RS
235 boundary condition is small. Another interesting result is the FoS for some 3D vertical cut convex-turning slopes is
236 smaller than that of a common 3D slope which is almost equivalent to the 2D plane strain problem under SS boundary
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
237 condition, e.g. the FoS of 90º convex-turning corner is 1.53 which is smaller than that of common slope, 1.57. This result
238 refutes the prevailing opinion that 2D analysis yields conservative results compared to 3D analysis, i.e. FoS2D<FoS3D.
239 To further investigate the variations in stability under various turning corners and slope gradients, the shear strain
240 contours and deformed meshes for three examples under RR boundary condition are shown in Figs. 8 and 9. Comparing
241 the shear strain contours and the deformed meshes of convex-turning corner cases of three slope gradients (Figs. 8(a) ~ (i)
242 and 9(a) ~ (i)), we can found that the convex-turning corners act to some extent as a constraint to resist potential sliding
243 for 1V:1H and 1V:2H slopes whereas an entirely opposite effect act on the vertical cut slope. This is exactly why the FoS
244 for a vertical cut slope with convex-turning corners is smaller than that for the common 3D slope whereas the FoS of
245 slopes with gradients of 1V:1H and 1V:2H are larger than those of the common slopes. On the other hand, from the shear
246 strain contours and the deformed meshes of concave-turning corner cases (Figs. 8(g) ~ (o) and 9(g) ~ (o)), the
247 concave-turning corner shows an ability to resist potential sliding for all three examples. In addition, both the convex
248 corners and the concave corners, the effects of turning become more noticeable when the turning angle is smaller.
249 Because the change trend of stability situation under different boundary conditions is similar, plots for SS and RS
252 In mountainous areas, turning arc slopes are often appears especially in road engineering or open pit mining. Fig. 10
253 shows the formation mode of convex-turning arc slopes (Fig. 10(a)) and concave-turning arc slopes (Fig. 10(b)). Turning
254 arc slope can be considered as: the base plane (Fig. 10) rotates about the coplanar vertical line OL with a vertical distance
255 of R to midpoint of slope surface. Clockwise and anticlockwise rotation direction form the convex- and concave-turning
11
Page 12 of 44
256 arc respectively and rotation angle α notes the degree of turning arc. The FoS corresponding to various turning arcs under
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
257 three boundary conditions are listed in Table 6. It should be declared that results of some cases can obtained from other
258 ones because of symmetry. In fact, all 3D turning arc slopes under various boundary conditions are identified as slope
259 with different degree of turning arc under RR boundary condition because of symmetry, e.g. a slope with 45º turning arc
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
260 under RS boundary condition can be identified as 90º turning arc under RR boundary condition, another example, a slope
261 with random turning angle under SS boundary condition can be identified as ∞º turning arc under RR boundary condition.
262 It can be found from Table 6: i) for convex turning arc, the FoS of steep slope is smaller than that of common slope
263 with the same width cases with negative DF in Table 6 while opposite result can be found for gentle slope cases with
264 positive DF in Table 6, i.e., influence of convex turning arc on 3D slope FoS has relationship with the slope gradient; ii)
265 for concave turning arc, the FoS is larger than that of common slope with the equivalent width regardless of the slope
266 gradient and the boundary condition; iii) the influence of concave turning arcs on the FoS is larger than that from
267 convex-turning corners; iv) for a given slope turning arc, the FoS will increase as the boundary constraints change from
268 SS to RS then RR boundary condition); v) the FoS decreases as the increases of angle of turning arc from 45º to 180º
269 under RS and RR boundary conditions whereas invariant under SS boundary condition with the reason of symmetry. The
270 same with convex-turning corner described in previous section, some convex-turning arc slope have the smaller FoS than
271 common slope under SS boundary condition. These convex-turning arc slopes also provide special cases of FoS3D<FoS2D.
272 To further investigate the variations in potential slip surface under various turning arcs and boundary conditions, the
273 shear strain contours and deformed meshes for example 3 are shown in Figs. 11 and 12. Because the influence of slope
274 gradient on change trend of the FoS has the similar impact with the turning corner slopes that was presented in previous
275 section, Fig. 11 emphasizes the effect of turning angle of arc on the 3D slopes stability. Plots for vertical cut and 1V:1H
276 slope are omitted owing to space reasons. It can be seen from Figs. 11(b) ~ (d), (g) ~ (i), (l) ~ (n) and 12(b) ~ (d), (g) ~ (i),
277 (l) ~ (n) that the shear strain contours and deformed meshes under RS boundary condition have not been extended to the
278 end of the rough boundary. This is because the 3D turning arc slope is somewhat different from the common 3D slope or
12
Page 13 of 44
279 3D turning corner slopes discussed in previous sections. Comparing the shear strain contours and deformed meshes of
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
280 convex-turning arc slopes and common slopes (Figs. 11(a) ~ (j) and 12(a) ~ (j)), we can found that the convex-turning
281 arcs act to some extent as a constraint to resist potential sliding for 1V:2H slopes but the influence is small. More
282 noticeable influence of the concave-turning arcs can be found from Figs. 11(f) ~ (o) and 12(f) ~ (o).
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
284 The effects of turning corner and turning arc on the 3D slope stability have been sufficiently analyzed as described
285 above. However, in practical slope engineering, turning slope is not only the single form (turning corner or turning arc)
286 but also coupled with each other or combined with other forms. In the following discussion, a series of 1V:1H 3D 90
287 corner slopes with different turning forms are investigated. Fig. 13 shows the outlines of these special slopes. The
288 calculated results for the FoS obtained using FDM are listed in Table 7. Difference of the FoS between turning corner
290 From Table 7, we can see that: i) for convex-turning slopes, turning forms with certain arc supply almost the same
291 FoS under three boundary conditions, i.e., turning arc has insignificant influence on the FoS (less than 2%); ii) for
292 concave-turning slopes, the FoS of turning slopes increases as the arc increases regardless boundary conditions and the
294 Figs. 14 and 15 show the shear strain contours and deformed meshes for the slopes with different turning forms
295 under SS boundary condition. It can be seen from Figs. 14(a) ~ (d) and 15(a) ~ (d) that failure mode of slopes with convex
296 turning form is entire failure. Different failure mode for slopes with concave turning in which the concave turning act to
297 some extent as a constraint to resist potential sliding can be found from Figs. 14(e) ~ (h) and 15(e) ~ (h) and the effect of
298 constraint is larger when the arc increases. This is exactly why the FoS of concave turning slopes is increases as the arc
299 increases.
300 Discussions
301 More than 180 cases with various geometries for different slope gradient under different boundary conditions were
13
Page 14 of 44
302 calculated. The results from these cases demonstrate that the FoS and the critical slip surface for a 3D slope with different
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
303 geometries (curvature, turning corner, turning arc and combined turning form), slope angle and boundary conditions are
304 different.
305 In fact, the effects of any geometry on 3D slope stability are reflected by two aspects: change area of sliding surface
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
306 and change volume of failure mass, which are relative to common 3D slope under the same boundary condition. The
307 variation in FoS depends on these changes. For example, for a concave slope with different curvature characteristics in
308 Table 4, the volume of failure mass for any of the three examples under RR boundary condition is far smaller than that of
309 the convex slope. That’s why the significant difference for the computed FoS of concave slope is far greater than that of
310 the convex slope. In addition, from the Fig. 6(d), the larger degree of the convex seems to induce the smaller the area of
311 sliding surface under RR boundary condition. It’s the reason of decrease in FoS with the increase of convex degree.
312 For most investigation cases, either convex- or concave-shape 3D slope including turning corner, turning arc and
313 coupled turning form, the computed FoS is larger than that of common 3D slope. The most important reason is these
314 turnings provide some extra constraint effects and improve the stability of the 3D slope. Some especial results, however,
315 were also observed, as would be the cases of convex-shaped steep slopes with turning corner or turning arc (see the cases
316 with negative DF shown in Table 5 and Table 6). Nian et al. (2012) has investigated the convex-shaped vertical slope with
317 90º and 135º turning corner, observed the similar result and attribute to two points: unique failure model (pyramid-shaped)
318 and free-surface effects. The authors agree with them. These two points, however, are limited to the cases of turning
319 corner slope but not suitable to explain the cases of turning arc slope. For convex-shaped steep slope with turning corner
320 or turning arc, more failure mass or smaller sliding surface than common 3D slope is the reason of reduction in FoS. In
321 this point, an analytical method can be used to drive the critical slope angle of turning slope (both corner and arc) that
322 makes the 3D FoS smaller than 2D FoS. This analytical method should be explored in the following research.
323 The boundary condition has a significant influence on the FoS and sliding surface. The physical reason for the
324 influence can also be explained by the changes of the area of sliding surface and the volume of failure mass. For a
14
Page 15 of 44
325 common 3D slope, the change of boundary condition from SS to RR can be considered to decrease the volume of failure
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
326 mass and increase the area of sliding surface. Therefore, the RR boundary condition can increase the FoS. The improving
327 degree, however, depends on the width of 3D slope: the wider, the smaller influence is. For 3D slopes with various
328 geometries, the boundary condition has the similar effects. For example, the turning slopes studied in here (Figs. 7 and 10
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
329 and Tables 5 and 6) are about twice as width as the curvature slopes (Fig. 4 and Table 4), therefore, the difference in FoS
330 under two boundary conditions for curvature slope is larger than that for turning slope.
331 Conclusions
332 A three-dimensional finite difference method with strength reduction technique was carried out to study the effect of
333 complex geometries on slope stability in terms of the factor of safety, shear strain contour and deformed mesh. First,
334 validations for some examples were carried out and the numerical results obtained clearly demonstrated that the 3D
335 strength-reduction FDM presented in this paper is reasonable and feasible. Then, the FoS as a function of various
336 geometries, including curvature feature of slope surface in third dimension, turning corner, turning arc and turning form,
337 was studied in detail. Three slope gradients (90º, 45º, 26.57º) and three boundary conditions (SS, RS and RR) were fully
338 considered. Many interesting results were obtained and some of them appeared to be surprising, and the authors have
339 carried out a detailed investigation to confirm their validity. From the findings of this study, the following conclusions can
340 be drawn:
341 1. Effect of curving surface: Both convex- and concave-curving slope surface are enhance the stable of 3D slope
342 than common slope surface under SS boundary condition but the difference in FoS is less than 3%. Under RR
343 boundary condition, the FoS increases with the change curving slope surface from the greatest degree of convex
344 to the greatest degree of concave and the difference is up to 11%. In general, the effect of curving slope surface
345 on steep slope shows more significant influence than that of gentle slope.
346 2. Effect of turning corner: The convex-turning corners act to some extent as a constraint to resist potential sliding
347 for gentle slope (increase the FoS up to 7.5%) whereas an entirely opposite effect acts on steep slope (decrease
15
Page 16 of 44
348 the FoS up to 7%). The concave-turning corners show a significant constraint effects on the corner for any slope
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
349 gradient, and therefore a considerable higher FoS (up to 20%) is obtained than for a common 3D slope. In
350 particular, the smaller the turning corner angle (both convex- and concave-turning), the more significant effect
352 3. Effect of turning arc: The convex-turning arc can improve the stability of gentle slope whereas opposite effect
353 for steep slope. The concave-turning arc can improve the slope stability regardless the slope gradient. Both the
354 convex- and concave-turning arc, the FoS decreases as the increases of angle of turning arc from 45º to 180º
355 under RS and RR boundary conditions whereas invariant under SS boundary condition with the reason of
356 symmetry. In the other hand, the difference in FoS between turning arc slope and common slope with the
358 4. Effect of turning form: For the convex-turning slopes, turning forms with certain arc supply almost the same FoS
359 under three boundary conditions, i.e., turning arc has insignificant influence on the FoS (less than 2%). For
360 concave-turning slopes, the FoS of turning slopes increases as the increases of arc regardless boundary
362 Acknowledgements
363 This study has received financial support from the Global Environment Research Found of Japan (S-8) and from
364 Grants-in-Aid for Scientific Research (Scientific Research (B), 22310113, G. Chen) from Japan Society for the Promotion
365 of Science. The financial support is gratefully acknowledged. The authors thank Mr. H, Zhang for his assistant in
366 checking grammar at proof stage. The authors also thank the Associate Editor and two anonymous reviewers for their
368 References
369 Azizian, a., and Popescu, R. 2006. Three-dimensional seismic analysis of submarine slopes. Soil Dynamics and Earthquake
370 Engineering, 26(9): 870-887. doi:10.1016/j.soildyn.2005.10.008.
371 Cai, F., Ugai, K., Wakai, A., and Li, Q. 1998. Effects of horizontal drains on slope stability under rainfall by three-dimensional
372 finite element analysis. Computers and Geotechnics, 23(4): 255-275. doi:10.1016/S0266-352X(98)00021-4.
373 Chang, M. 2002. A 3d slope stability analysis method assuming parallel lines of intersection and differential straining of block
374 contacts. Canadian Geotechnical Journal, 39(4): 799-811. doi:10.1139/t02-020.
375 Chen, J., Yin, J.-H., and Lee, C.F. 2003a. Upper bound limit analysis of slope stability using rigid finite elements and nonlinear
16
Page 17 of 44
384 Chen, Z., Wang, J., Wang, Y., Yin, J.-H., and Haberfiel, C. 2001b. A three-dimensional slope stability analysis method using the
385 upper bound theorem part ii: Numerical approaches, applications and extensions. International Journal of Rock
386 Mechanics and Mining Sciences, 38: 379-397. doi:10.1016/S1365-1609(01)00013-2.
387 Chen, Z., Mi, H., Zhang, F., and Wang, X. 2003b. A simplified method for 3d slope stability analysis. Canadian Geotechnical
388 Journal, 40(3): 675-683. doi:10.1139/T03-002.
389 Chen, Z.Y., Yin, J.H., and Wang, Y.J. 2006. The three‐dimensional slope stability analysis recent advances and a forward look.
390 In Advances in Earth Structures: Research to Practice.
391 Cheng, Y.M., Lansivaara, T., and Wei, W.B. 2007. Two-dimensional slope stability analysis by limit equilibrium and strength
392 reduction methods. Computers and Geotechnics, 34(3): 137-150. doi:10.1016/j.compgeo.2006.10.011.
393 Cheng, Y.M., Liu, H.T., Wei, W.B., and Au, S.K. 2005. Location of critical three-dimensional non-spherical failure surface by
394 nurbs functions and ellipsoid with applications to highway slopes. Computers and Geotechnics, 32(6): 387-399.
395 doi:10.1016/j.compgeo.2005.07.004.
396 Cheng, Y.M., and Yip, C.J. 2007. Three-dimensional asymmetrical slope stability analysis extension of bishop’s, janbu’s, and
397 morgenstern–price’s techniques. Journal of Geotechnical and Geoenvironmental Engineering, 133(12): 1544-1555.
398 doi:10.1061//asce/1090-0241/2007/133:12/1544.
399 Chugh, A.K. 2003. On the boundary conditions in slope stability analysis. International Journal for Numerical and Analyrucal
400 Methods in Geomechanics, 27: 905-926. doi:10.1002/nag.305.
401 Dawson, E.M., Roth, W.H., and Drescher, A. 1999. Slope stability analysis by strength reduction. Géotechnique, 49(6): 835 –840.
402 doi:10.1680/geot.1999.49.6.835.
403 Donald, I.B., and Chen, Z. 1997. Slope stability analysis by the upper bound approach: Fundamentals and methods. Canadian
404 Geotechnical Journal, 34: 853-862. doi:10.1139/t97-061.
405 Duncan, J.M. 1996. State of the art: Limit equilibrium and finite-element analysis of slopes. Journal of Geotechnical Engineering,
406 122(7): 577-596. doi:10.1061/(ASCE)0733-9410(1996)122:7(577).
407 Farzaneh, O., and Askari, F. 2003. Three-dimensional analysis of nonhomogeneous slopes. Journal of Geotechnical and
408 Geoenvironmental Engineering, 129(2): 137-145. doi:10.1061/(ASCE)1090-0241(2003)129:2(137).
409 Farzaneh, O., Askari, F., and Ganjian, N. 2008. Three-dimensional stability analysis of convex slopes in plan view. Journal of
410 Geotechnical and Geoenvironmental Engineering, 134(8): 1192-1200.
411 doi:10.1061//(ASCE)/1090-0241/2008/134:8/1192.
412 Fredlund, D.G., and Krahn, J. 1977. Comparison of slope stability methods of analysis. Canadian Geotechnical Journal, 14(3):
413 429-439. doi:10.1139/t77-045.
414 Gharti, H.N., Komatitsch, D., Oye, V., Martin , R., and Tromp, J. 2012. Application of an elastoplastic spectral-element method
415 to 3d slope stability analysis. International Journal for Numerical Methods in Engineering, 91(1): 1-26.
416 doi:10.1002/nme.3374.
417 Griffiths, D.V., and Lane, A.P. 1999. Slope stability analysis by finite element. Géotechnique, 49(3): 387-403.
418 doi:10.1680/geot.1999.49.3.387.
419 Griffiths, D.V., and Marquez, R.M. 2007. Three-dimensional slope stability analysis by elasto-plastic finite elements.
420 Géotechnique, 57(6): 537-546. doi:10.1680/geot.2007.57.6.537.
421 Hovland, H.J. 1977. Three-dimensional slope stability analysis method. American Society of Civil Engineers, Journal of the
422 Geotechnical Engineering Division, 103(9): 971-986.
17
Page 18 of 44
423 Huang, C.-C., and Tsai, C.-C. 2000. New method for 3d and asymmetrical slope stability analysis. Journal of Geotechnical and
424 Geoenvironmental Engineering, 126(10): 917-927. doi:10.1061//(ASCE)1090-0241(2000)126_10(917).
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
425 Huang, C.-C., Tsai, C.-C., and Chen, Y.-H. 2002. Generalized method for three-dimensional slope stability analysis. Journal of
426 Geotechnical and Geoenvironmental Engineering, 128(10): 836-848. doi:10.1061/(ASCE)1090-0241/2002/128:10/836.
427 Hungr, O. 1987. An extension of bishop's simplified method of slope stability analysis to three dimensions. Géotechnique, 37(1):
428 113-117. doi:10.1680/geot.1987.37.1.113.
429 Hungr, O., Salgado, F.M., and Byrne, P.M. 1989. Evaluation of a three-dimensional method of slope stability analysis. Canadian
430 Geotechnical Journal, 26(4): 679-686. doi:10.1139/t89-079.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
431 Itasca 2007. Fast lagrangian analysis of continua in 3-dimension (flac3d v3.1). Itasca Consulting Group, Minnesota.
432 Jiang, J.C., and Yamagami, T. 2004. Three-dimensional slope stability analysis using an extended spencer method. Soils and
433 Foundations, 44(4): 127-135. doi:10.3208/sandf.44.4_127.
434 Lam, L., and Fredlund, D.G. 1993. A general limit equilibrium model for three-dimensional slope stability analysis. Canadian
435 Geotechnical Journal, 30(6): 905-919. doi:10.1139/t93-089.
436 Li, A.J., Merifield, R.S., and Lyamin, A.V. 2009. Limit analysis solutions for three dimensional undrained slopes. Computers and
437 Geotechnics, 36(8): 1330-1351. doi:10.1016/j.compgeo.2009.06.002.
438 Li, A.J., Merifield, R.S., and Lyamin, A.V. 2010. Three-dimensional stability charts for slopes based on limit analysis methods.
439 Canadian Geotechnical Journal, 47(12): 1316-1334. doi:10.1139/t10-030.
440 Matsui, T., and San, K.C. 1992. Finite element slope stability analysis by shear strength reduction technique. Soils and
441 Foundations, 32: 59-70. doi:10.3208/sandf1972.32.59.
442 Michalowski, R.L. 1989. Three-dimensional analysis of locally loaded slopes. Géotechnique, 39(1): 27-38.
443 doi:10.1680/geot.1989.39.1.27
444 Michalowski, R.L. 2010. Limit analysis and stability charts for 3d slope failures. Journal of Geotechnical and Geoenvironmental
445 Engineering, 136(4): 583-593. doi: 10.1061/(ASCE)GT.1943-5606.0000251.
446 Michalowski, R.L., and Drescher, A. 2009. Three-dimensional stability of slopes and excavations. Géotechnique, 59(10): 839-850.
447 doi:10.1680/geot.8.P.136.
448 Nian, T.K., Huang, R.Q., Wan, S.S., and Chen, G.Q. 2012. Three-dimensional strength-reduction finite element analysis of slopes:
449 Geometric effects. Canadian Geotechnical Journal, 49(5): 574-588. doi:10.1139/t2012-014.
450 Razdolsky, A.G. 2009. Slope stability analysis based on the direct comparison of driving forces and resisting forces. International
451 Journal for Numerical and Analytical Methods in Geomechanics, 33(8): 1123-1134. doi:10.1002/nag.761.
452 Shukha, R., and R., B. 2003. Mesh geometry effects on slope stability calculation by flac strength reduction method - linear and
453 non-linear failure criteria. In 3rd international conference on FLAC and numerical modeling in geomechanics, Ontaria,
454 Canada, pp. 109-116.
455 Silvestri, V. 2006. A three-dimensional slope stability problem in clay. Canadian Geotechnical Journal, 43(2): 224-228.
456 doi:10.1139/t06-001.
457 Stark, T.D., and Eid, H.T. 1998. Performance of three-dimensional slope stability methods in practice. Journal of Geotechnical
458 and Geoenvironmental Engineering, 124(11): 1049-1060. doi:10.1061/(ASCE)1090-0241(1998)124:11(1049).
459 Stianson, J.R., Fredlund, D.G., and Chan, D. 2011. Three-dimensional slope stability based on stresses from a stress-deformation
460 analysis. Canadian Geotechnical Journal, 48(6): 891-904. doi:10.1139/t11-006.
461 Stolle, D., and Guo, P. 2008. Limit equilibrium slope stability analysis using rigid finite elements. Canadian Geotechnical Journal,
462 45(5): 653-662. doi:10.1139/t08-010.
463 Ugai, K., and Leshchinsky, D. 1995. Three-dimensional limit equilibrium and finite element analyses: A comparison of results.
464 Soils and Foundations, 35(4): 1-7. doi: doi:10.3208/sandf.35.4_1.
465 Wei, W.B., Cheng, Y.M., and Li, L. 2009. Three-dimensional slope failure analysis by the strength reduction and limit
466 equilibrium methods. Computers and Geotechnics, 36(1-2): 70-80. doi:10.1016/j.compgeo.2008.03.003.
467 Zhang, X. 1988. Three-dimensional stability analysis of concave slopes in plan view. Journal of Geotechnical Engineering, 114(6):
468 658-671. doi: 10.1061/(ASCE)0733-9410(1988)114:6(658).
469 Zheng, H. 2012. A three-dimensional rigorous method for stability analysis of landslides. Engineering Geology.
18
Page 19 of 44
470 doi:10.1016/j.enggeo.2012.06.010.
471 Zheng, Y.R., Tang, X., Zhao, S., Deng, C., and Lei, W. 2009. Strength reduction and step-loading finite element approaches in
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
472 geotechnical engineering. Journal of Rock Mechanics and Geotechnical Engineering, 1(1): 21-30.
473 doi:10.3724/SP.J.1235.2009.00021.
474 Zienkiewicz, O.C., Humpheson, C., and Lewis, R.W. 1975. Associated and non-associated visco-plasticity and plasticity in soil
475 mechanics. Géotechnique, 25(4): 671-689. doi:10.1680/geot.1975.25.4.671.
476 Zienkiewicz, O.C., and Tayor, R.L. 1989. The finite element method, vol.1, 4th edition. McGraw-Hill, London, New York.
477
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
19
Page 20 of 44
480 Table 2. Comparison of FoS from various investigators for three examples.
481 Table 3. Comparison of different models for example 1 under smooth-smooth boundary condition.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
482 Table. 4. Stability of 3D slopes with different curvature for three examples under different boundary conditions.
483 Table. 5. Stability of 3D slopes with different turning corners of three examples under different boundary conditions.
484 Table. 6. Stability of 3D slopes with different turning arcs of three examples under different boundary conditions.
485 Table. 7. Stability of 3D slopes with different turning forms under different boundary conditions.
20
Page 21 of 44
490 Fig. 4. Top views and FDM models of 3D slopes with convex and concave curving slope surface. (a) Top view of 3D
491 slope with a convex surface; (b) top view of 3D slope with a concave surface; (c) FDM model of 3D slope with a
492 convex surface; (d) FDM model of 3D slope with a concave surface.
493 Fig. 5. Shear strain contours of example 2 with different curving slope surface under two boundary conditions (Rcur=0.5):
494 (a) convex surface, (b) common, and (c) concave surface under smooth-smooth boundary condition; (d) convex
495 surface, (e) common, and (f) concave surface under rough-rough boundary condition.
496 Fig. 6. Deformed models of example 2 with different curving slope surface under two boundary conditions (Rcur=0.5): (a)
497 convex surface, (b) common, and (c) concave surface under smooth-smooth boundary condition; (d) convex surface,
498 (e) common, and (f) concave surface under rough-rough boundary condition.
499 Fig. 7. Top views with formation mode and outlines of 3D slopes with different turning corners. Top views of (a) convex
500 90º, (b) convex 135º, (c) common 180º, (d) concave 135º, (e) concave 90º; outlines of (f) convex 90º, (g) convex
501 135º, (h) common 180º, (i) concave 135º, (j) concave 90º. Little blue and little green note two slopes, dark gray
502 means intersection of two slopes because of rotation, and little gray means new generated zone.
503 Fig. 8. Shear strain contours with factors of safety of three examples with different turning corners under rough-rough
505 Fig. 9. Deformed meshes of three examples with different turning corners under rough-rough boundary condition.
506 Fig. 10. Formation mode of 3D slopes with different turning arcs.
507 Fig. 11. Shear strain contours with factors of safety of example 3 with different turning arcs.
508 Fig. 12. Deformed meshes with factors of safety of example 3 with different turning arcs.
21
Page 22 of 44
509 Fig. 13. Outlines of 90º turning slopes with different turning form.
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
510 Fig. 14. Shear strain contours with factors of safety of slopes with different turning forms under smooth-smooth boundary
511 condition.
512 Fig. 15. Deformed meshes with factors of safety of slopes with different turning forms under smooth-smooth boundary
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
513 condition.
22
Page 23 of 44
Table 3. Comparison of different models for example 1 under smooth-smooth boundary condition.
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Table. 4. Stability of 3D slopes with different curvature for three examples under different boundary conditions.
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Table. 5. Stability of 3D slopes with different turning corners of three examples under different boundary conditions.
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Factor of safety
Example Turning corner Smooth-smooth (SS) Rough-smooth (RS) Rough-rough (RR)
FoS DF, % FoS DF, % FoS DF, %
90º 1.53 -2.55 1.53 -3.77 1.55 -6.63
Convex
135º 1.56 -0.64 1.57 -1.26 1.61 -3.01
Example 1 Common 180º 1.57 1.59 1.66
135º 1.65 5.10 1.66 4.40 1.78 7.23
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
Concave
90º 1.65 5.10 1.66 4.40 1.82 9.64
90º 1.89 3.85 1.93 3.21 2.08 6.12
Convex
135º 1.84 1.10 1.88 0.53 1.97 0.51
Example 2 Common 180º 1.82 1.87 1.96
135º 1.94 6.60 1.96 4.81 2.13 8.67
Concave
90º 2.05 12.64 2.06 10.16 2.37 20.92
90º 2.20 4.76 2.22 3.74 2.37 7.24
Convex
135º 2.15 2.38 2.17 1.40 2.25 1.81
Example 3 Common 180º 2.10 2.14 2.21
135º 2.19 4.29 2.22 3.74 2.33 5.43
Concave
90º 2.26 7.62 2.27 6.08 2.46 11.31
Note: Half of width of 3D slope, W, are 10 m, 9 m and 30 m for example 1, 2 and 3, respectively (see
Fig. 7).
Page 28 of 44
Table. 6. Stability of 3D slopes with different turning arcs of three examples under different boundary conditions.
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Factor of safety
Example Turning arc Smooth-smooth (SS) Rough-smooth (RS) Rough-rough (RR)
FoS DF, % FoS DF, % FoS DF, %
45º 1.69 -6.63 1.93 -6.31
Convex 90º 1.62 -2.99 1.63 -4.68 1.69 -6.63
180º 1.62 -3.57 1.63 -4.68
Example 1
Table. 7. Stability of 3D slopes with different turning forms under different boundary conditions.
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Factor of safety
Turning
Smooth-smooth (SS) Rough-smooth (RS) Rough-rough (RR)
form
FoS DF, % FoS DF, % FoS DF, %
a 1.84 1.10 1.86 1.64 1.90 1.60
b 1.85 1.60 1.86 1.64 1.90 1.60
Convex
c 1.84 1.10 1.86 1.64 1.90 1.60
d 1.84 1.10 1.86 1.64 1.89 1.07
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
Fig. 4. Top views and FDM models of 3D slopes with convex and concave curving slope surface. (a) Top view
of 3D slope with a convex surface; (b) top view of 3D slope with a concave surface; (c) FDM model of 3D
slope with a convex surface; (d) FDM model of 3D slope with a concave surface.
111x68mm (300 x 300 DPI)
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record. Page 34 of 44
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
Fig. 5. Shear strain contours of example 2 with different curving slope surface under two boundary
conditions (Rcur=0.5): (a) convex surface, (b) common, and (c) concave surface under smooth-smooth
boundary condition; (d) convex surface, (e) common, and (f) concave surface under rough-rough boundary
condition.
82x37mm (300 x 300 DPI)
Page 35 of 44
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
Fig. 6. Deformed models of example 2 with different curving slope surface under two boundary conditions
(Rcur=0.5): (a) convex surface, (b) common, and (c) concave surface under smooth-smooth boundary
condition; (d) convex surface, (e) common, and (f) concave surface under rough-rough boundary condition.
88x42mm (300 x 300 DPI)
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record. Page 36 of 44
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
Fig. 7. Top views with formation mode and outlines of 3D slopes with different turning corners. Top views of
(a) convex 90º, (b) convex 135º, (c) common 180º, (d) concave 135º, (e) concave 90º; outlines of (f)
convex 90º, (g) convex 135º, (h) common 180º, (i) concave 135º, (j) concave 90º. Little blue and little
green note two slopes, dark gray means intersection of two slopes because of rotation, and little gray means
new generated zone.
77x33mm (300 x 300 DPI)
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record. Page 37 of 44
condition.
249x341mm (300 x 300 DPI)
Fig. 9. Deformed meshes of three examples with different turning corners under rough-rough boundary
Page 38 of 44
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/19/13
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record. Page 39 of 44
boundary condition.
76x32mm (300 x 300 DPI)
Fig. 15. Deformed meshes with factors of safety of slopes with different turning forms under smooth-smooth
Page 44 of 44