0 оценок0% нашли этот документ полезным (0 голосов)

12 просмотров61 страницаUntitled 2

Sep 29, 2018

© © All Rights Reserved

DOCX, PDF, TXT или читайте онлайн в Scribd

Untitled 2

© All Rights Reserved

0 оценок0% нашли этот документ полезным (0 голосов)

12 просмотров61 страницаUntitled 2

© All Rights Reserved

Вы находитесь на странице: 1из 61

Circular failure

present a series of slope stability charts for circu-

Although this book is concerned primarily with the

lar failure. These charts enable the user to carry

stability of rock slopes containing well-defined

out a rapid check on the factor of safety of a

sets of discontinuities, it is also necessary to

slope, or upon the sensitivity of the factor of

design cuts in weak materials such as highly

safety to changes in ground water conditions,

weathered or closely fractured rock, and rock fills.

slope angle and material strength properties.

In such materials, failure occurs along a sur-face

These charts should only be used for the analysis

that approaches a circular shape (Figure 8.1),

of circular failure in slope materials that are

and this chapter is devoted to a discussion on the

homo-genous and where the conditions apply

stability analysis of these materials.

that were assumed in deriving the charts (see

In a review of the historical development of

Section 8.4). More comprehensive methods of

slope stability theories, Golder (1972) traced the

analysis are presented in Section 8.6. These

subject back almost 300 years. Much of the

methods can be used, for example, where the

development of circular failure analysis meth-ods

material properties vary within the slope, or

was carried out in the 1950s and 1960s, and

where part of the slide surface is at a soil/rock

these techniques have since been used to

interface and the shape of the slide surface

prepare computer programs that have the ver-

differs significantly from a simple circular arc.

satility to accommodate a wide range of geo-

This chapter primarily addresses the stability

logic, geometric, ground water and external

of slopes in two dimensions, and assumes that

loading conditions. This chapter discusses the

the slope can be modeled as a unit slice

principles of the theoretical work, and demon-

through an infinitely long slope, under plane-

strates their application in design charts and in

strain condi-tions. Section 8.6.5 discusses

the results of computer analyses. During the past

three-dimensional circular failure analysis, and

half century, a vast body of literature on the

Section 10.3.1 dis-cusses the influence of the

subject of circular failure has accumulated, and

radius of curvature of the slope on stability.

no attempt will be made to summarize the

material in this chapter. Standard soil mechan-ics

text books such as those by Taylor (1937),

8.2 Conditions for circular failure

Terzaghi (1943) and Lambe and Whitman (1969),

and methods of analysis

and papers by Skempton (1948), Bishop (1955),

Janbu (1954), Morgenstern and Price (1965), In the previous chapters, it has been assumed

Nonveiller (1965), Peck (1967), Spencer (1967, that the failure of rock slopes is controlled by

1969) and Duncan (1996) all contain excellent geological features such as bedding planes and

discussions on the stability of soil slopes. joints that divide the rock into a discontinuous

Circular failure 177

failure in highly

weathered, granitic

rock (on Highway 1,

near Devil’s Slide,

Pacifica, California).

mass. Under these conditions, one or more of the basalts, will also tend to fail in this manner. It is

discontinuities normally defines the slide sur-face. appropriate to design slopes in these materials

However, in the case of a closely fractured or on the assumption that a circular failure

highly weathered rock, a strongly defined process will develop.

structural pattern no longer exists, and the slide

surface is free to find the line of least resistance

8.2.1 Shape of slide surface

through the slope. Observations of slope failures

in these materials suggest that this slide surface The actual shape of the “circular” slide surface is

generally takes the form of a circle, and most influenced by the geological conditions in the

stability theories are based upon this observa- slope. For example, in a homogenous weak or

tion. Figure 8.1 shows a typical circular failure in weathered rock mass, or a rock fill, the failure is

a highly weathered rock slope above a high-way. likely to form as a shallow, large radius sur-face

The conditions under which circular failure will extending from a tension crack close behind the

occur arise when the individual particles in a soil crest to the toe of the slope (Figure 8.2(a)). This

or rock mass are very small compared with the contrasts with failures in high cohesion, low

size of the slope. Hence, broken rock in a fill will friction materials such as clays where the surface

tend to behave as a “soil” and fail in a circular may be deeper with a smaller radius that may exit

mode when the slope dimensions are beyond the toe of the slope. Figure 8.2(b) shows

substantially greater than the dimensions of the an example of conditions in which the shape of

rock fragments. Similarly, soil consisting of sand, the slide surface is modified by the slope geology.

silt and smaller particle sizes will exhibit circular Here the circular surface in the upper, weathered

slide surfaces, even in slopes only a few meters rock is truncated by the shallow dipping, stronger

in height. Highly altered and weathered rocks, as rock near the base. Stability analyses of both

well as rock with closely spaced, randomly ori- types of surface can be carried out using circular

ented discontinuities such as some rapidly cooled failure methods, although for the latter case it is

178 Circular failure

(a)

R

Vertical

slice

E

i –1

i –1

h

i –1

Wi b

i

h iE Ai

i

S = (ci + i tan i ) Ai

FS

Circular sliding i Ai

surface Forces acting on slice, i

(b)

Non-circular

sliding surface

Figure 8.2 The shape of

typical sliding surfaces:

(a) large radius circular

surface in homogeneous,

weak material, with the

detail of forces on slice;

(b) non-circular surface in

weak, surficial material with

stronger rock at base.

8.2.2 Stability analysis procedure

shape of the surface to be defined.

For each combination of slope parameters The stability analysis of circular failure is carried

there will be a slide surface for which the factor out using the limit equilibrium procedure similar to

of safety is a minimum—this is usually termed that described in earlier chapters for plane and

the “critical surface.” The procedure to find the wedge failures. This procedure involves compar-

critical surface is to run a large number of ana- ing the available shear strength along the sliding

lyses in which the center co-ordinates and the surface with the force required to maintain the

radius of the circle are varied until the surface slope in equilibrium.

with the lowest factor of safety is found. This is The application of this procedure to circu-lar

an essential part of circular slope stability failures involves division of the slope into a series

analysis. of slices that are usually vertical, but may

Circular failure 179

be inclined to coincide with certain geological and rearranging this equation, we have

features. The base of each slice is inclined at angle

ψb and has an area A. In the simplest case, the c + σ tan φ

τ (8.2)

forces acting on the base of each slice are the shear e= FS

resistance S due to the shear strength of the rock

(cohesion c; friction angle φ), and forces E (dip The method of solution for the factor of safety

angle ψ; height h above base) acting on the sides of is to use an iterative process in which an initial

the slice (see detail Figure 8.2(a)). estimate is made for FS, and this is refined

The analysis procedure is to consider equilib- with each iteration.

rium conditions slice by slice, and if a condition of The influence of various normal stress distri-

equilibrium is satisfied for each slice, then it is butions upon the factor of safety of soil slopes

also satisfied for the entire sliding mass. The has been examined by Frohlich (1955) who found

number of equations of equilibrium avail-able that a lower bound for all factors of safety that

depends on the number of slices N, and the satisfy statics is given by the assumption that the

number of equilibrium conditions that are used. normal stress is concentrated at a single point on

The number of equations available is 2N if only the slide surface. Similarly, the upper bound is

force equilibrium is satisfied, and 3N if both force obtained by assuming that the normal load is

and moment equilibrium are satis-fied. If only concentrated at the two ends of the slide surface.

force equilibrium is satisfied, the number of The unreal nature of these stress distributions

unknowns is (3N − 1), while, if both force and is of no consequence since the object of the

moment equilibria are satisfied, the number of exer-cise, up to this point, is simply to determine

unknowns is (5N − 2). Usually between 10 and 40 the extremes between which the actual factor of

slices are required to realist-ically model the safety of the slope must lie. In an example con-

slope, and therefore, the number of unknowns sidered by Lambe and Whitman (1969), the

exceeds the number of equations. The excess of upper and lower bounds for the factor of safety of

unknowns over equations is (N − 1) for force a particular slope corresponded to 1.62 and 1.27,

equilibrium analysis, and (2N − 2) for analyses respectively. Analysis of the same problem by

that satisfy all conditions of equilib-rium. Thus, Bishop’s simplified method of slices gives a factor

the analyses are statically indeterm-inate and of safety of 1.30, which suggests that the actual

assumptions are required to make up the factor of safety may lie reasonably close to the

imbalance between equations and unknowns lower bound solution.

(Duncan, 1996). Further evidence that the lower bound solution

The various limit equilibrium analysis pro- is also a meaningful practical solution is provided

cedures either make assumptions to make up the by an examination of the analysis that assumed

balance between known and unknowns, or they the slide surface has the form of a logarithmic

do not satisfy all the conditions of equilib-rium. spiral (Spencer, 1969). In this case, the factor of

For example, the Spencer Method assumes that safety is independent of the normal stress distri-

the inclination of the side forces is the same for bution, and the upper and lower bounds coincide.

every slice, while the Fellenius and Bishop Taylor (1937) compared the results from a num-

methods do not satisfy all conditions of ber of logarithmic spiral analyses with results of

equilibrium. 1

lower bound solutions and found that the dif-

The factor of safety of the circular failure based

ference is negligible. Based on this comparison,

on limit equilibrium analysis is defined as

FS = 1 The lower bound solution discussed in this chapter is usu-

shear stress required for equilibrium on slip surface(τe) ally known as the Friction Circle Method and was used by

(8.1) Taylor (1937) for the derivation of his stability charts.

180 Circular failure

Taylor concluded that the lower bound solution (e) The locations of the tension crack and of

provides a value of the factor of safety that is the slide surface are such that the factor

sufficiently accurate for most practical of safety of the slope is a minimum for the

problems involving simple circular failure of slope geometry and ground water

homogeneous slopes. conditions considered.

The basic principles of these methods of (f) Ground water conditions vary from a dry

ana-lyses are discussed in Section 8.6. slope to a fully saturated slope under

heavy recharge; these conditions are

8.3 Derivation of circular failure charts defined in Figure 8.4.

(g) Circular failure charts are optimized for a

This section describes the use of a series of charts 3

rock mass density of 18.9 kN/m .

that can be used to determine rapidly the factor of

Densities higher than this give high factors

safety of circular failures. These charts have been of safety, densities lower than this give

developed by running many thousands of circular low factors of safety. Detailed circular

analyses from which a number of dimensionless analysis may be required for slopes in

parameters were derived that relate the factor of which the mate-rial density is significantly

safety to the material unit weight, friction angle and 3

different from 18.9 kN/m .

cohesion, and the slope height and face angle. It

has been found that these charts give a reliable The charts presented in this chapter corres-

estimate for the factor of safety, provided that the pond to the lower bound solution for the factor of

conditions in the slope meet the assumptions used safety, obtained by assuming that the normal

in developing the charts. In fact, the accuracy in load is concentrated on a single point on the slide

calculating the factor of safety from the charts is surface. These charts differ from those published

usually greater than the accuracy in determining the by Taylor in that they include the influence of a

shear strength of the rock mass. critical tension crack and of ground water.

Use of the stability charts presented in this

chapter requires that the conditions in the

8.3.1 Ground water flow assumptions

slope meet the following assumptions:

In order to calculate the forces due to water

(a) The material forming the slope is homogen- pres-sures acting on the slide surface and in

eous, with uniform shear strength properties the tension crack, it is necessary to assume a

along the slide surface. set of ground water flow patterns that coincide

(b) The shear strength τ of the material is char- as closely as pos-sible with conditions that are

acterized by cohesion: c and a friction believed to exist in the field.

angle φ, that are related by the equation τ In the analysis of rock slope failures discussed in

= c + σ tan φ (see Section 1.4). Chapters 6, 7 and 9, it is assumed that most of the

(c) Failure occurs on a circular slide surface, water flow takes place in discontinuities in the rock

2 and that the rock itself is practically imper-meable.

which passes through the toe of the slope.

(d) A vertical tension crack occurs in the In the case of slopes in soil or waste rock, the

upper surface or in the face of the slope. permeability of the mass of material is gener-ally

several orders of magnitude higher than that of

2 Terzaghi (1943: 170), shows that the toe failure assumed for intact rock and, hence, a general flow pattern will

this analysis gives the lowest factor of safety provided that φ develop in the material behind the slope.

◦

> 5 . The φ = 0 analysis, involving failure below the toe of Figure 5.10(a) shows that, within the rock mass,

the slope through the base material has been discussed by the equipotentials are approximately perpendi-cular

Skempton (1948) and by Bishop and Bjerrum (1960) and is

applicable to failures which occur during or after the rapid to the phreatic surface. Consequently, the flow lines

construction of a slope. will be approximately parallel to the

Circular failure 181

x

(a)

Tension

crack

Face

Phreatic surface

H

Assumed equipotentials

Sliding surface

water flow patterns used in

(b) Surface recharge due to heavy rain circular failure analysis of slopes

in weak and closely fractured

Tension crack rock: (a) ground water flow

pattern under steady state

Sliding surface drawdown conditions where the

H phreatic surface coincides with

the ground surface at a distance

Assumed equipotentials x behind the toe of the slope.

The distance x is measured in

multiples of the slope height H ;

Assumed flow lines (b) ground water flow pattern in

a saturated slope subjected to

surface recharge by heavy rain.

phreatic surface for the condition of steady-state Figure 8.4 shows five ground water

drawdown. Figure 8.3 shows that this approx- conditions ranging from fully drained to

imation has been used for the analysis of the water saturated, based on the models shown in

pressure distribution in a slope under condi-tions of Figure 8.3. For con-ditions 2, 3 and 4, the

normal drawdown. Note that the phreatic surface is position of the ground water table is defined by

assumed to coincide with the ground surface at a the ratio x/H . These five ground water

distance x, measured in multiples of the slope conditions are used in conjunc-tion with the

height, behind the toe of the slope. This may circular failure charts discussed in Section 8.4.

correspond to the position of a surface water source,

or be the point where the phreatic surface is judged

to intersect the ground surface. 8.3.2 Production of circular failure charts

The phreatic surface itself has been obtained, The circular failure charts presented in this

for the range of the slope angles and values of x chapter were produced by running a search

considered, by solution of the equations proposed routine to find the most critical combination of

by L. Casagrande (1934), and discussed in the slide surface and tension crack for each of a wide

textbook by Taylor (1937). For the case of a sat- range of slope geometries and ground water con-

urated slope subjected to heavy surface ditions. Provision was made for the tension crack

recharge, the equipotentials and the associated to be located in either the upper surface, or in the

flow lines used in the stability analysis are based face of the slope. Detailed checks were carried

upon the work of Han (1972). This work involved out in the region surrounding the toe of the slope

the use of an electrical resistance analogue where the curvature of the equipotentials results

method to study ground water flow patterns in in local flow which differs from that illustrated in

slopes comprised of isotropic materials. Figure 8.3(a).

182 Circular failure

behind toe of slope

behind toe of slope

behind toe of slope

5

Figure 8.4 Ground water flow

models used with circular

Saturated slope subjected to

failure analysis

heavy surface recharge charts—Figures 8.6–8.10.

The charts are numbered 1–5 (Figures 8.6– the slope and choose the chart which

8.10) to correspond with the ground water con- is closest to these conditions, using

ditions defined in Figure 8.4. Figure 8.4.

Step 2: Select rock strength parameters appli-

8.3.3 Use of the circular failure charts cable to the material forming the slope.

Step 3: Calculate the value of the dimensionless

In order to use the charts to determine the factor

ratio c/(γ H tan φ) and find this value on

of safety of a slope, the steps outlined here and

the outer circular scale of the chart.

shown in Figure 8.5 should be followed. Step 4: Follow the radial line from the value

found in step 3 to its intersection with

Step 1: Decide upon the ground water con- the curve which corresponds to the slope

ditions which are believed to exist in angle.

Circular failure 183

φ/FS or c/(γ H FS), depending upon

which is more convenient, and

calculate the factor of safety.

c

Consider the following example: tan

H tan

◦ 3

A 15.2-m high cut with a face angle of 40 is FS

to be excavated in overburden soil with a dens-

3

ity γ = 15.7 kN/m , a cohesion of 38 kPa and a 4

◦

friction angle of 30 . Find the factor of safety of 4

the slope, assuming that there is a surface

water source 61 m behind the toe of the slope.

The ground water conditions indicate the use c

of chart number 3 (61/15.2 ∼ 4). The value of H FS

Figure 8.5 Sequence of steps involved in using circular

◦

value of tan φ/FS, for a 40 slope, is 0.32. failure charts to find the factor of safety of a slope.

Hence, the factor of safety of the slope of 1.80.

2.0 .04 .05

.06 .07

.08

.09

1.8 .10

.11

.12

.13

1.6 .14

.15 c

.16

.17 H tan

.18

1.4 .19

.20

1.2 .25

tan

FS .30

1.0 90

.35

.40

0.8

Slope angle .45

80 .50

0.6 .60

70 .70

60 .80

0.4 50 .90

40 1.0

30 1.5

0.2 20 2.0

10

4.0

0 8

0 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26 .28 .30 .32 .34

c

H FS

184 Circular failure

2.0 .05 .06 .07

.08

.09

1.8 .10

.11

.12

.13

1.6 .14

.15 c

.16

.17 H tan

.18

1.4 .19

.20

1.2 .25

tan

FS .30

1.0 90

.35

.40

0.8

Slope angle .45

.50

80

0.6 60

70 .70

60 .80

0.4 50 .90

40 1.0

30

20 1.5

0.2 10 2.0

4.0

0 8

0 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26 .28 .30 .32 .34

c

H FS

Figure 8.7 Circular failure chart number 2—ground water condition 2 (Figure 8.5).

were determined for each slope analyzed.

these charts, they are ideal for checking the sens-

These locations are presented, in the form of

itivity of the factor of safety of a slope to a wide

charts, in Figures 8.11 and 8.12.

range of conditions. For example, if the cohesion

It was found that, once ground water is present

were to be halved to 20 kPa and the ground water

in the slope, the locations of the critical circle and

pressure increased to that represented by chart

the tension crack are not particularly sensitive to

number 2, the factor of safety drops to 1.28.

the position of the phreatic surface and hence

only one case, that for chart number 3, has been

plot-ted. It will be noted that the location of the

8.4 Location of critical slide

critical circle center given in Figure 8.12 differs

surface and tension crack signific-antly from that for the drained slope

During the production of the circular failure plotted in Figure 8.11.

charts presented in this chapter, the locations These charts are useful for the construction of

drawings of potential slides and for estimating the

of both the critical slide surface and the critical

friction angle when back-analyzing existing

tension crack for limiting equilibrium (FS = 1)

Circular failure 185

0 .01 .02

2.0 .03 .04 .05

.06 .07

.08 .09

1.8 .10

.11

.12

.13

1.6 .14

.15 c

.16

.17 H tan

.18

1.4 .19

.20

1.2

.25

tan

FS 90

.30

1.0

.35

Slope angle .40

0.8

.45

80 .50

0.6

70 .60

60 .70

50 .80

0.4 40 .90

30 1.0

20

1.5

0.2 2.0

4.0

0 8

0 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26 .28 .30 .32 .34

c

H FS

Figure 8.8 Circular failure chart number 3—ground water condition 3 (Figure 8.4).

circular slides. They also provide a start in 8.5.1 Example 1—China clay pit slope

locating the critical slide surface when carrying

Ley (1972) investigated the stability of a China

out more sophisticated circular failure analysis.

clay pit slope which was considered to be

As an example of the application of these

charts, consider the case of a drained slope potentially unstable, and that a circular failure

◦ was the likely type of instability. The slope pro-

having a face angle of 30 in a soil with a friction

file is illustrated in Figure 8.14 and the input

◦

angle of 20 . Figure 8.11 shows that the critical data used for the analysis is included in this

slide circle center is located at X = 0.2H and Y = figure. The material, a heavily kaolinized

1.85H and that the critical tension crack is at a granite, was tested in direct shear to determine

distance b = 0.1H behind the crest of the slope. the friction angle and cohesion.

These dimensions are shown in Figure 8.13.

Two piezometers in the slope and a known

water source some distance behind the slope

8.5 Examples of circular failure analysis enabled an estimate to be made of the pos-

ition of the phreatic surface as shown in Figure

The following two examples illustrate the use of 8.14. From Figure 8.14, chart number 2

the circular failure charts for the study of the corresponds most closely to these ground

stability of slope in highly weathered rock. water conditions.

186 Circular failure

2.0 .04

.05 .06

.07

.08

.09

1.8 .10

.11

.12

.13

1.6 .14

.15 c

.16

.17 H tan

.18

1.4 .19

.20

1.2 .25

tan 90

FS .30

1.0

.35

80 .45

.50

0.6 70 .60

60 .70

50 .80

0.4 40 .90

1.0

1.5

0.2 2.0

4.0

0 8

0 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26 .28 .30 .32 .34

c

H FS

Figure 8.9 Circular failure chart number 4—ground water condition 4 (Figure 8.4).

From the information given in Figure 8.14, was required to check whether the cut would be

the value of the ratio c/(γ H tan φ) = 0.0056, stable. The slope was in weathered and altered

and the corresponding value of tan φ/FS from material, and failure, if it occurred, would be a

chart number 2, is 0.76. Hence, the factor of circular type. Insufficient time was available for

safety of the slope is 1.01. A number of trial ground water levels to be accurately established

calculations using Janbu’s method (Janbu et or for shear tests to be carried out. The stability

al., 1956) for the critical slide circle shown in analysis was carried out as follows:

Figure 8.14, found a factor of safety of 1.03. For the condition of limiting equilibrium, FS = 1

These factors of safety indicated that the and tan φ/FS = tan φ. By reversing the pro-cedure

stabil-ity of the slope was inadequate under outlined in Figure 8.5, a range of friction angles

the assumed conditions, and steps were taken were used to find the values of the ratio c/(γ H tan

to deal with the problem. ◦

φ) for a face angle of 42 . The value of the cohesion

c which is mobilized at failure, for a given friction

8.5.2 Example 2—highway slope angle, can then be calculated. This analysis was

carried out for dry slopes using chart number 1 (line

◦ B, Figure 8.15), and for saturated slopes using chart

A highway plan called for a cut at an angle of 42 .

The total height of the cut would be 61 m and it number 5 (line A, Figure 8.15).

Circular failure 187

.04

2.0 .05 .06

.07

.08

.09

1.8 .10

.11

.12

.13

1.6 .14

.15

.16

.17 c

.18

1.4 H tan

.19

.20

1.2

.25

tan

FS .30

1.0

.35

Slope angle

.40

0.8

80 .45

.50

70

0.6 .65

60

50 .70

40 .80

0.4 30 .90

1.0

20

1.5

0.2 10 2.0

4.0

0 8

0 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26 .28 c .30 .32 .34

H FS

ratio c/(γ H tan φ) for a flatter slope of 30 , in the

and cohesions that would be mobilized at failure. ◦

same way as it was found for the 42 slope. The

The shaded circle (D) included in Figure 8.15 dashed line (C) in Figure 8.15 indicates the shear

indicated the range of shear strengths that were strength, which is mobilized in a dry slope with a

considered probable for the material under ◦

face angle of 30 . Since the mobilized shear

consideration, based upon the data presented in strength C is less than the available shear

Figure 4.21. This figure shows that the available strength D, the dry slope is likely to be stable.

shear strength may not be adequate to maintain

stability in this cut, particularly when the cut is 8.6 Detailed stability analysis

saturated. Consequently, the face angle could be

of circular failures

reduced, or ground water conditions investigated

to establish actual ground water pressures and The circular failure charts presented earlier in this

the feasibility of drainage. chapter are based upon the assumption that the

The effect of reducing the slope angle can be material forming the slope has uniform proper-

checked very quickly by finding the value of the ties throughout the slope, and that failure occurs

188 Circular failure

+X

critical circle

0.3 = 10°

Ratio b/H

b

Y 0.2

Tension = 20 °

= 30°

H crack 0.1

= 40°

Drained slope 0

0 10 20 30 40 50 60 70 80 90

Failure through Slope face angle (°)

toe of slope Location of critical tension crack position

Distance X

−3H −2H −H 0 H 2H 3H

4H 4H

20˚ Friction angle

=30° = = 10°10°

angle

3H 20 3H

30

Distance Y

Slope

=

40 40°

=

2H 50 50°

2H

60°

70°

80°

H H

0 0

−3H −2H −H 0 H 2H 3H

Location of center of critical circle for failure through toe

Figure 8.11 Location of critical sliding surface and critical tension crack for drained slopes.

(1955) and the Janbu’s modified method of slices

the slope. When these conditions are not satis-fied,

(1954) are given in Figures 8.16 and 8.17 respect-

it is necessary to use one of the methods of slices

ively. Bishop’s method assumes a circular slide

published by Bishop (1955), Janbu (1954),

surface and that the side forces are horizontal; the

Nonveiller (1965), Spencer (1967), Morgenstern and

analysis satisfies vertical forces and overall moment

Price (1965) or Sarma (1979). This section

describes in detail the simplified Bishop and Janbu equilibrium. The Janbu method allows a slide

methods of stability analysis for circular failure. surface of any shape, and assumes the side forces

are horizontal and equal on all slices; the analysis

satisfies vertical force equilibrium. As pointed out by

8.6.1 Bishop’s and Janbu’s method of slices Nonveiller (1965), Janbu’s method gives reasonable

factors of safety when applied to shallow slide

The slope and slide surface geometries, and the surfaces (which are typ-ical in rock with an angle of

equations for the determination of the factor of ◦

safety by the Bishop’s simplified method of slices friction in excess of 30 and rockfill), but it is

seriously in error

Circular failure 189

+X

0.4

Location of center of = 10°

critical circle

0.3

Ratio b/H

b

= 20°

Y 0.2

Ground water = 30°

surface

H 0.1

Tension crack = 40° = 50°

Slope with ground water = 60°

0

(chart number 3) 0 10 20 30 40 50 60 70 80 90 Angle of slope face (°)

Failure through

toe of slope Location of critical tension crack position

Distance X

−3H −2H −H 0 H 2H 3H

4H

10°

3H

Friction angle

20

20° =

Distance Y

30

=

2H Slope angle 30°

=10

40°

70° 60° 50°

80°

H

=

60

=

50

40

=

0

−3H −2H −H 0 H 2H 3H

Location of center of critical circle for failure through toe

Figure 8.12 Location of critical sliding surface and critical tension crack for slopes with ground water present.

X = 0.2 H

b = 0.1 H

Y = 1.85 H

Figure 8.13 Location of critical slide surface and critical tension crack for a drained slope at an

◦ ◦

angle of 30 in a material with a friction angle of 20 .

190 Circular failure

Unit weight = 21.5 kN/m3

Friction angle = 37

Cohesion c = 6.9 kPa

Measured water

level 31 76.8 m

Critical failure

Figure 8.14 Slope profile

circle for Janbu

of China clay pit slope

analysis

considered in Example 1.

A—Saturated 42 slope the charts given in Figures 8.11 and 8.12 can

B—Dry 42 slope be used to estimate the center of the circle with

250 C—Dry 30 slope

D—Probable shear strength range for

the lowest factor of safety. In the Janbu

material in which slope is cut analysis, the slide surface may be defined by

200 (see Figure 4.21). known structural features or weak zones within

A the rock or soil mass, or it may be estimated in

c (kPa)

150

In either case, the slide surface assumed for

Cohesio

B of safety, and a series of analyses are required

n

100 D

with variations on this pos-ition to find the

C surface with the lowest factor of safety.

50 Step 2: Slice parameters. The sliding mass

assumed in step 1 is divided into a number of

0 slices. Generally, a minimum of five slices should

0 10 20 30 40 50 be used for simple cases. For complex slope pro-

Friction angle () files, or where there are different materials in the

rock or soil mass, a larger number of slices may

Figure 8.15 Comparison between shear be required in order to define adequately the

strength mobilized and shear strength

available for slope considered in Example 2. prob-lem. The parameters which have to be

defined for each slice are as given here:

• base angle ψb;

in materials with low friction angles. • the weight of each slice W is given by the

The procedures for using Bishop’s and product of the vertical height h, the unit

Janbu’s methods of slices are very similar and weight γr of the rock or soil and the width of

it is con-venient to discuss them together.

the slice x : W = (h γr x); and

• uplift water pressure U on the base of each

Step 1: Slope and slide surface geometry. The slice is given by the product of the height hw

geometry of the slope is defined by the actual or to the phreatic surface, the unit weight γw of

the designed profile as seen in a vertical section water and the width of the slice x, that is,

through the slope. In the case of a circular failure, U = (hw γw x).

Circular failure 191

Tension crack

Ground water

surface

R

Y Z

H

½ wz2 z/3

Typical slice

of slope

x

Factor of safety:

x /(1 + Y/FS) (8.3)

FS = Z + Q

where

X = [c + ( rh – whw) tan ] (x /cos b) (8.4)

Y = tan b tan (8.5)

h

Z = rh x sin b (8.6)

hw

Slice weight Q = ½ wz 2 ( /R ) (8.7)

= rhx

b Figure 8.16 Bishop’s

simplified method of slices for

The following conditions must be satisfied for each slice:

the analysis of non-circular

rh – whw – c (tan b /FS) failure in slopes cut into

(1) = (8.8) materials in which failure is

1 + Y/FS

defined by the Mohr–Coulomb

(2) cos b (1 + Y/FS) > 0.2 (8.9) failure criterion.

Step 3: Shear strength parameters. The shear defined by non-linear failure criterion as dis-

strength acting on the base of each slice is cussed in Section 4.5, it is necessary to

required for the stability calculation. In the case of determ-ine the cohesion and friction angle for

a uni-form material in which the failure criterion is each slice at the effective normal stress for that

assumed to be that of Mohr–Coulomb (equa-tion slice (see Figure 4.23).

(1.1) in Section 1.4), the shear strength Step 4: Factor of safety iteration. When the

parameters c and φ will be the same on the base slice and shear strength parameters have been

of each slice. When the slope is cut in a number defined, the values of X, Y and Z are calculated

of materials, the shear strength para-meters for for each slice. The water force Q is added to Z,

each slice must be chosen according to the

the sum of the components of the weight of each

material in which it lies. When the shear strengths

slice acting parallel to the slide surface. An initial

of the materials forming the slope are

estimate of FS = 1.00 for the factor of safety is

192 Circular failure

Tension crack

Ground water

surface

Typical slice

z

H ½ wz2 z/3

Failure through

toe of slope

x

Factor of safety:

f0 X/(1 + Y/FS)

FS = (8.10)

Z + Q

where

X = [c + ( rh – whw) tan ] (1 + tan2 b) x (8.11)

Y = tan b tan (8.12)

Z = rh x tan b (8.13)

h

Q = ½ wz 2 (8.14) Figure 8.17 Janbu’s

hw modified method of

Slice weight

= r hx slices for the analysis

Note: angle b is negative when sliding uphill

of non-circular failure

in slopes cut into

Approximate correction factor f0 materials in which

b

f0 = 1 + K(d /L – 1.4(d /L)2) (8.15) for c = 0; K = failure is defined by

0.31 the Mohr–Coulomb

c > 0, > 0; K = 0.50 failure criterion.

used, and a new factor of safety is calculated from of each slice is always positive. If this con-

equations (8.3) and (8.10) given in Figures 8.16 and dition is not met for any slice, the inclusion of a

8.17, respectively. If the difference between the tension crack into the analysis should be

calculated and the assumed factors of safety is considered. If it is impossible to satisfy this

greater than 0.001, the calculated factor of safety is condition by readjustment of the ground water

used as a second estimate of FS for a new factor of conditions or the introduction of a tension

safety calculation. This process is repeated until the crack, the analysis as presented in Figure 8.16

difference between successive factors of safety is should be abandoned and a more elaborate

less than 0.001. For both the Bishop and the Janbu form of analysis, to be described later, should

methods, approximately seven iteration cycles will be adopted.

be required to achieve this result for most slope and Condition 2 in Figure 8.16 was suggested by

slide surface geometries. Whitman and Bailey (1967) and it ensures that

Step 5: Conditions and corrections. Figure 8.16 the analysis is not invalidated by conditions which

lists two conditions (equations (8.8) and (8.9)) can sometimes occur near the toe of a slope in

that must be satisfied for each slice in the Bishop which a deep slide surface has been assumed. If

analysis. The first condition ensures that the this condition is not satisfied by all slices, the

effective normal stress on the base slice dimensions should be changed and, if this

Circular failure 193

fails to resolve the problem, the analysis this procedure until the difference between

should be abandoned. successive factors of safety is less than 0.001.

Figure 8.17 gives a correction factor f0, which

Generally, about ten iterations will be required

is used in calculating the factor of safety by

means of the Janbu method. This factor allows to achieve the required accuracy in the

for inter-slice forces resulting from the shape of calculated factor of safety.

the slide surface assumed in the Janbu analysis.

The equation for f0 given in Figure 8.17 has been 8.6.3 Example of Bishop’s and

derived by Hoek and Bray (1981) from the curves Janbu’s methods of analysis

published in Janbu (1954).

A slope is to be excavated in blocky sandstone

with very closely spaced and persistent discon-

8.6.2 Use of non-linear failure criterion tinuities. The slope will consist of three, 15-m

in Bishop stability analysis high benches with two 8-m wide berms, the

primary function of which are to collect surface

When the material in which the slope is cut obeys runoff and control erosion (Figure 8.19). The

the Hoek–Brown non-linear failure criterion dis- ◦

bench faces will be at 75 to the horizontal, and

cussed in Section 4.5, the Bishop’s simplified

the slope above the crest of the cut will be at

method of slices as outlined in Figure 8.18 can be ◦

used to calculate the factor of safety. The an angle of 45 . The assumed position of the

following procedure is used, once the slice para- water table is shown on the figure. It is required

meters have been defined as described earlier for to find the factor of safety of the overall slope,

assuming that a circular type stability analysis

the Bishop and Janbu analyses:

is appropriate for these conditions.

The shear strength of the jointed rock mass is

1 Calculate the effective normal stress σ act-

based on the Hoek–Brown strength criterion, as

ing on the base of each slice by means of

discussed in Section 4.5, which defines the strength

the Fellenius equation (equation (8.17) on

as a curved envelope. The cohesion and friction

Figure 8.18).

angle for this criterion are calculated using the pro-

2 Using these values of σ , calculate tan φ

gram ROCLAB 1.004 (RocScience, 2002a), for

and c for each slice from equations (4.24)

which the input parameters are as follows:

and (4.25).

3 Substitute these values of tan φ and c into the • Very poor quality rock mass, GSI = 20;

factor of safety equation in order to obtain the

• Uniaxial compressive strength of intact rock

first estimate of the factor of safety. (from point load testing) ≈ 150 MPa;

4 Use this estimate of FS to calculate a new value

• Rock material constant, mi = 15;

of σ on the base of each slice, using the Bishop 3

• Unit weight of rock mass, γr = 0.025 MN/m ;

equation (equation (8.18) on Figure 8.18). 3

• Unit weight of water, γw = 0.00981 MN/m ;

5 On the basis of these new values of σ ,

• For careful blasting used in excavation, dis-

calcu-late new values for tan φ and c.

turbance factor D = 0.7; and

6 Check that conditions defined by equa-tions • Average slice height = 24 m (this height

(8.8) and (8.9) on Figure 8.16 are sat-isfied together with the rock mass unit weight

for each slice. defines the average vertical stress on the

7 Calculate a new factor of safety for the new sliding surface).

values of tan φ and c.

8 If the difference between the first and second Using these parameters, ROCLAB calculates, at the

factors of safety is greater than 0.001, return appropriate vertical stress level, a best fit line to the

to step 4 and repeat the analysis, using the curved strength envelope to define a friction angle

second factor of safety as input. Repeat ◦

of 43 and a cohesion of 0.145 MPa. This is

194 Circular failure

X Center of rotation

b Tension crack

Ground water

surface

R

Y z

H 2 z/3

½ wz

Typical slice

Failure through toe

of slope

x Factor of safety:

FS = (8.16)

rh x sin b + ½ wz 2 /R

where

2

= rh cos b – whw (Fellenius solution) (8.17)

and

h

rh – whw – (ci tan b/FS)

hw rh = (Bishop solution) (8.18)

1+ (tan i tan b/FS)

a –1

6am b (s + m b 3n)

= sin

–1

a –1 (4.24)

b 2(1 + a)(2 + a) + 6 am (s + m )

b b 3n

a –1

ci [(1 + 2a)s + ( 1– a)m b 3n] ( + mb 3n)

c = a –1 (4.25)

(1 + a)(2 + a) – 1+ [6am (s + m s) ][(1 + a)(2 + a)]

b b 3n

where = = /

3n 3 max ci

(1) > 0, where is calculated by Bishop’s method

(2) cos b [1 + (tan b tan i )/FS] > 0.2

Figure 8.18 Bishop’s simplified method of slices for the analysis of circular failure in slope in material

in which strength is defined by non-linear criterion given in Section 4.5.

the equivalent Mohr–Coulomb shear strength. Table 8.1 shows the input parameters for the

In addition, the program calculates the Bishop and Janbu stability analyses assuming a

instantan-eous friction angle and cohesion linear shear strength, and for the Janbu analysis

corresponding to the effective normal stress on assuming a non-linear shear strength. The slope is

the base of each slice. divided into eight slices, and for each slice the base

Circular failure 195

angle, weight, pore pressure and width are meas- the slices, and the corresponding instantaneous

ured. The program SLIDE (RocScience, 2002b) is friction angle and cohesion are also calculated,

then used to calculate the factors of safety for the as shown in the last three columns of Table 8.1.

linear and non-linear shear strengths. In the case The calculated factors of safety estimates are

of the non-linear strength analysis, the val-ues of

Bishop simplified method of slices for = 1.39

the effective normal stress on the base of

Mohr–Coulomb shear strength

Janbu modified method of slices for = 1.26

Mohr–Coulomb shear strength

Critical center for = 45

Bishop simplified method of slices for = 1.39

non-linear shear strength

Note that the stability analysis of this slope

using numerical analysis methods is shown in

Section 10.4.1.

5m

0 10 20 analysis computer programs

2

Scale - m

The circular failure charts discussed in Section 8.3

3

provide a rapid means of carrying out stability

4 n/R = 0.401 analyses, but are limited to simple conditions as

5

7 6 d /L = 0.117 illustrated in the examples. More complex ana-lyses

8 can be carried out using the Bishop and Janbu

methods discussed in Section 8.6.3, and the

Figure 8.19 Section of sandstone slope showing

water table, slice boundaries, tension crack and the purpose of providing details on the procedures is to

expected circular sliding surface. show the principles of the analyses.

Table 8.1 Calculated shear strength values of slices using Mohr–Coulomb and Hoek–Brown failure criteria

values for Bishop values for Bishop analysis

and Janbu analyses

Slice Angle Slice Pore Slice Friction Cohesion, Base Inst. Inst.

member of slice height pressure, width angle, c (MPa) effective friction cohesion,

base, (MN) γw h w (m) normal angle, ci (MPa)

(degrees) (MPa) (degrees) stress, φi

σn (degrees)

(MPa)

2 29 1.597 0.047 6.1 43 0.145 0.169 53.8 0.068

3 34 2.603 0.071 6.1 43 0.145 0.270 49.8 0.096

4 39 2.635 0.087 6.1 43 0.145 0.261 51.3 0.085

5 44 3.501 0.095 6.1 43 0.145 0.323 49.0 0.104

6 50 3.914 0.087 6.1 43 0.145 0.324 48.6 0.107

7 57 3.592 0.047 6.1 43 0.145 0.240 50.3 0.092

8 65 2.677 0 6.1 43 0.145 0.109 54.5 0.064

196 Circular failure

There are, of course, computer programs slope excavated in sandstone, shale and

available to carry out stability analyses of siltstone above a proposed highway. For these

slopes, where the circular failure charts are not conditions the shape of the failure surfaces are

applicable. Important features included in influenced by the position and thickness of the

these programs, which allows them to be used beds of the weaker material.

for a wide range of conditions, are as follows:

8.6.5 Three-dimensional circular

• Slope face can include benches and a

failure analysis

variety of slope angles;

• Boundaries between the materials can be posi- The program XSTABL examines the stability of

tioned to define layers of varying thickness and a unit width slice of the slope, which is a two-

inclination, or inclusions of any shape; dimensional analysis that ignores any shear

• Shear strength of the materials can be stresses on the sides of the slice (this is the

defined in terms of Mohr–Coulomb or same principle that is used in the plane fail-ure

Hoek–Brown criterion; analysis described in Chapter 6). While two-

• Ground water pressures can be defined on dimensional procedures have found to be a

single or multiple water tables, or as reliable method of analysis, there may be cir-

specified pressure distributions; cumstances where three-dimensional analysis

• External loads, in any direction within the is required to define the slide surface and

plane of the slope cross-section, can be slope geometry more precisely. One program

posi-tioned at their correct location on the that provides a three-dimensional analysis is

slope. Such loads can include bridge and CLARA (Hungr, 1987), which divides the

building foundations and bolting forces; sliding mass into columns, rather than slices as

• Earthquake acceleration which is applied used in the two-dimensional mode. Figure 8.21

as a horizontal force in order to carry out shows an example of the CLARA analysis for a

pseudo-static stability analysis; partially saturated slope in which the water

• The shape and position of the slide surface table is below the bottom of the tension crack.

can be defined as a circular arc or straight

line segments;

8.6.6 Numerical slope stability analysis

• A search routine finds the slide surface with

the minimum factor of safety; This chapter has been concerned solely with

• Deterministic and probabilistic analysis the limit equilibrium method of analysis in

meth-ods that calculate the factor of safety which the factor of safety is defined by the ratio

and probability of failure, respectively. The of the resisting to the displacing forces on the

prob-abilistic analysis requires that the slide surface.

design para-meters be defined as An alternative method of analysis is to exam-

distributions rather than single values; ine the stresses and strains within the slope as a

• Error messages which identify negative means of assessing stability conditions. If the

stresses along the slide surface; and slope is close to failure, then a zone of high strain

• Drawing of slope showing slope geometry, will develop within the slope with a shape that will

material boundaries, ground water table(s) be approximately coincident with the circu-lar

and slide surface(s). slide surface. If the shear strength properties are

progressively reduced, there will be a sudden

One program that contains all these functions increase in the movement along the shear zone

is the program XSTABL (Sharma, 1991). An indicating that the slope is on the point of failure.

example of the output (partial) produced by The approximate factor of safety of the slope can

XSTABL is shown in Figure 8.20 for a benched be calculated from the ratio of the actual shear

Circular failure 197

Surface of minimum Janbu Overburden

FS = 1.101

Sandstone

995

Shale

905

Y-axis (m)

Siltstone and

sandstone

815

Shale

Sandstone

725 Shale

Siltstone and sandstone

835

0 90 180 270 360 450 540 630 720

X-axis (m)

circular failure analysis

Statement

Surcharge on

crest of slope ◦

A 22-m high rock cut with a face angle of 60 has

Retaining wall been excavated in a massive, very weak vol-

canic tuff. A tension crack has opened behind the

crest and it is likely that the slope is on the point

Failure surface of failure, that is, the factor of safety is approx-

imately 1.0. The friction angle of the material is

m ◦ 3

10 5 0 estimated to be 30 , its density is 25 kN/m , and

the position of the water table is shown on the

sketch of the slope (Figure 8.22). The rock con-

tains no continuous joints dipping out of the face,

Figure 8.21 Three-dimensional stability analysis of and the most likely type of failure mode is circular

a slope incorporating a retaining wall and a

failure.

surcharge at the crest.

Required

strength of the rock to the shear strength at

which the sudden movement occurred. This (a) Carry out a back analysis of the failure to

method of stability analysis is described in determine the limiting value of the

more detail in Section 10.3. cohesion when the factor of safety is 1.0.

198 Circular failure

◦

analysis. When φ = 30 and FS = 1.0,

60

tan φ/FS = 0.58

Ground water table 22 m The intersection of this value for tan φ/FS

◦

and the curve for a slope angle of 60 gives

Estimated slide surface

c

γ H FS = 0.086

(b) b = 2.9 m

∴ c = 0.086 × 25 × 22 × 1.0

= 47.3 kPa

X = –7.7 m

(b) If the slope were completely drained, circular

Y = 22 m failure chart 1 could be used for analysis.

c 47.3

=

γ H tan φ 25 × 22 × tan(30)

= 0.15

Figure 8.22 Slope geometry for Example The intersection of this inclined line with the

Problem 8.1: (a) slope geometry for circular ◦

curved line for a slope angle of 60 gives

failure with ground water table corresponding to

circular failure chart number 3; (b) position of tan φ

=

critical slide surface and critical tension crack. 0.52 FS

∴ FS = tan 30

(b) Using the strength parameters calculated in 0.52

(a), determine the factor of safety for a com-

= 1.11

pletely drained slope. Would drainage of the

slope be a feasible method of stabilization?

This factor of safety is less than that usually

(c) Using the ground water level shown in accepted for a temporary slope, that is, FS =

Figure 8.22 and the strength parameters 1.2, so draining the slope would not be an

cal-culated in (a), calculate the reduction effective means of stabilization.

in slope height, that is, amount of ◦

(c) When FS = 1.3 and φ = 30 , then

unloading of the slope crest required to

tan φ/FS = 0.44.

increase the factor of safety to 1.3.

On circular failure chart 3, the inter-

(d) For the slope geometry and ground water

section of this horizontal line with the curved

level shown in Figure 8.22, find the coordin-

◦

ates of the center of the critical circle and line for a slope angle of 60 gives

the position of the critical tension crack. c

γ H FS = 0.11

47.3

Solution

∴ H = 25

× ×

1.3 0.11

(a) The ground water level shown in Figure 8.22

corresponds to ground water condition 3 in = 13.2 m

Table 8.4 in the manual, so circular failure This shows that the slope height must be

reduced by 8.8 m to increase the factor of

Circular failure 199

safety of 1.3 would only be achieved if the

ground water level dropped by an amount = 22 m, that is, 22 m above the toe

equivalent to the unloading.

(d) The critical circle and critical tension crack The location of the tension crack behind

for a slope with ground water present are the crest is

located using the graphs in Figure 8.12.

◦

For a slope angle of 60 and a friction b/H = 0.13

◦

angle of 30 , the coordinates of the center

of the circle are: b = 2.9 m

= −7.7 m, that is, 7.7 m 8.22(b).

horizontally beyond the toe

Chapter 9

Toppling failure

of topples.

The failure modes discussed in the three

Papers concerning field studies of toppling fail-

previous chapters all relate to sliding of a rock or

ures include de Freitas and Waters (1973) who

soil mass along an existing or induced sliding

discuss slopes in Britain, and Wyllie (1980) who

surface. This chapter discusses a different failure

demonstrates stabilization measures for toppling

mode—that of toppling, which involves rotation of

failures related to railway operations.

columns or blocks of rock about a fixed base.

Most of the discussion that follows in this

Similar to the plane and wedge failures, the

chapter is based on a paper by Goodman and

stability analysis of toppling failures involves, first,

Bray (1976) in which a formal mathematical

carrying out a kinematic analysis of the structural

solu-tion to a simple toppling problem is

geology to identify potential toppling conditions,

shown. This solution, which is reproduced

and then, if this condition exists, performing a

here, represents a basis for designing rock

stability analysis specific to toppling failures.

slopes in which top-pling is present, and has

One of the earliest references to toppling fail-ures

been further developed into a more general

is by Muller (1968) who suggested that block

design tool (Zanbak, 1983; Adhikary et al.,

rotation or toppling may have been a contribut-ory

1997; Bobet, 1999; Sageseta et al., 2001).

factor in the failure of the north face of the Vaiont

slide (Figure 9.1). Hofmann (1972) car-ried out a

number of model studies under Muller’s direction to 9.2 Types of toppling failure

investigate block rotation. Similar model studies Goodman and Bray (1976) have described a

carried out by Ashby (1971), Soto (1974) and Whyte number of different types of toppling failures that

(1973), while Cundall (1971), Byrne (1974) and may be encountered in the field, and each is dis-

Hammett (1974) who incor-porated rotational failure cussed briefly on the following pages. The

modes into computer analysis of rock mass import-ance of distinguishing between types of

behavior. Figure 9.2 shows a computer model of a toppling is that there are two distinct methods of

toppling failure in which the solid blocks are fixed stability analysis for toppling failures as described

and the open blocks are free to move. When the in the following pages—block and flexural

fixed blocks at the face are removed, the tallest toppling— and it is necessary to use the

columns of blocks topple because their center of appropriate analysis in design.

gravity lies outside the base. The model illustrates a

typical feature of toppling failures in which the

9.2.1 Block toppling

tension cracks are wider at the top than at the base.

This con-dition, which can best be observed when As illustrated in Figure 9.3(a), block toppling

looking occurs when, in strong rock, individual columns

Toppling failure 201

Figure 9.1 Suggested toppling mechanism of the north face of Vaiont slide (Muller, 1968).

Figure 9.2 Computer generated model of toppling failure; solid blocks are fixed in space while open

blocks are free to move (Cundall, 1971).

are formed by a set of discontinuities dipping steeply of rock, separated by well developed, steeply

into the face, and a second set of widely spaced dipping discontinuities, breaking in flexure as

orthogonal joints defines the column height. The they bend forward. Typical geological condi-

short columns forming the toe of the slope are tions in which this type of failure may occur are

pushed forward by the loads from the longer thinly bedded shale and slate in which ortho-

overturning columns behind, and this slid-ing of the gonal jointing is not well developed. Generally,

toe allows further toppling to develop higher up the the basal plane of a flexural topple is not as

slope. The base of the failure generally consists of a well defined as a block topple.

stepped surface rising from one cross joint to the Sliding, excavation or erosion of the toe of the

next. Typical geological con-ditions in which this slope allows the toppling process to start and it

type of failure may occur are bedded sandstone and retrogresses back into the rock mass with the

columnar basalt in which orthogonal jointing is well formation of deep tension cracks that become

developed. narrower with depth. The lower portion of the

slope is covered with disordered fallen blocks and

it is sometimes difficult to recognize a toppling

9.2.2 Flexural toppling

failure from the bottom of the slope. Detailed

The process of flexural toppling is illustrated in examination of toppling slopes shows that the

Figure 9.3(b) that shows continuous columns outward movement of each cantilevered column

202 Toppling failure

(a) (b)

(c)

Figure 9.3 Common classes of toppling failures: (a) block toppling of columns of rock containing widely

spaced orthogonal joints; (b) flexural toppling of slabs of rock dipping steeply into face; (c) block flexure

toppling characterized by pseudo-continuous flexure of long columns through accumulated motions

along numerous cross-joints (Goodman and Bray 1976).

produces an interlayer slip and a portion of the 9.2.4 Secondary toppling modes

upper surface of each plane is exposed in a

Figure 9.4 illustrates a number of possible

series of back facing, or obsequent scarps,

secondary toppling mechanisms suggested by

such as those illustrated in Figure 9.3(a).

Goodman and Bray. In general, these failures are

initiated by some undercutting of the toe of the

9.2.3 Block-flexure toppling slope, either by natural agencies such as scour

As illustrated in Figure 9.3(c), block-flexure or weathering, or by human activities. In all

toppling is characterized by pseudo-continuous cases, the primary failure mode involves sliding

flexure along long columns that are divided by or physical breakdown of the rock, and toppling is

numerous cross joints. Instead of the flexural induced in the upper part of the slope as a result

failure of continuous columns resulting in flex- of this primary failure (Figure 9.4(a) and (b)).

ural toppling, toppling of columns in this case Figure 9.4(c) illustrates a common occurrence of

results from accumulated displacements on the toppling failure in horizontally bedded sand-stone

cross-joints. Because of the large number of and shale formations. The shale is usu-ally

small movements in this type of topple, there significantly weaker and more susceptible to

are fewer tension cracks than in flexural weathering than the sandstone, while the sand-

toppling, and fewer edge-to-face contacts and stone often contains vertical stress relief joints. As

voids than in block toppling. the shale weathers, it undermines support for the

Toppling failure 203

(a) (b)

2050 Circular sliding

surfaces

2000

Toppling

Elevation (m)

1950

at pit crest

1900

1850 Sandstone

Talus

1800

Fault

1750

Siltstone

Conglomerate

Figure 9.4 Secondary toppling modes: (a) toppling at head of slide; (b) toppling at toe of slide with shear

movement of upper slope (Goodman and Bray, 1976); (c) toppling of columns in strong upper material due

to weathering of underlying weak material; (d) toppling at pit crest resulting in circular failure of upper slope

(Wyllie and Munn, 1978).

sandstone and columns of sandstone, with 30 m occurred on the slope above the pit, result-

their dimensions defined by the spacing of the ing in cracks opening in the crest of the mountain

vertical joints, topple from the face. At some that were several meters wide and up to 9 m

locations the overhangs can be as wide as 5 deep. Continuous movement monitoring was

m, and failures of substantial volumes of rock used to allow mining to proceed under the moving

occur with little warning. slope, and finally the slope was stabilized by

The example of the slide base toppling mode back-filling the pit (Wyllie and Munn, 1979).

shown in Figure 9.4(d) is the failure of a pit slope A further example of the toppling mechan-

in a coal mine where the beds at the crest of the ism is illustrated in Figure 9.5 (Sjöberg, 2000).

◦

pit dipped at 70 into the face, and their strike was In open pit mines where the depth of the slope

parallel to the face. Mining of the pit slope at an progressively increases, minor toppling move-

◦ ment may eventually develop into a substantial

angle of 50 initiated a toppling failure at the crest

of the pit, which in turn resulted in a circular failure. Careful monitoring of the movement,

failure that extended to a height of 230 m above and recognition of the toppling mechanism, can

the base of the topple. Detailed monitoring of the be used to anticipate when hazardous

slope showed that a total movement of about conditions are developing.

204 Toppling failure

I II III

Crest

Elastic Joint slip fully

rebound developed

Joint slip (exaggerated

Joints displacements)

Stress

redistribution Toe

New mining

step

IV V VI

Compression and

bending of columns

Movement on

slide surface

at base of rotation propagated to crest starting from toe

Figure 9.5 Failure stages for large-scale toppling failure in a slope (Sjöberg, 2000).

The potential for toppling can be assessed

ψp < φp (Stable) (9.1)

from two kinematic tests described in this

section. These tests examine first the shape of

but will topple when the center of gravity of the

the block, and second the relationship between

block lies outside the base, that is, when

the dip of the planes forming the slabs and the

face angle. It is emphasized that these two

x/y < tan ψp (Topple) (9.2)

tests are useful for identifying potential toppling

conditions, but the tests cannot be used alone For example, for a 3 m wide block on a base

as a method of stability analysis. ◦

plane dipping at 10 , toppling will occur if the

height exceeds 17 m.

9.3.1 Block shape test

9.3.2 Inter-layer slip test

The basic mechanics of the stability of a block on

a plane are illustrated in Figure 9.6(a) (see also A requirement for toppling to occur in the mech-

Figure 1.10). This diagram shows the conditions anisms shown in Figures 9.3 and 9.5 is shear

that differentiate stable, sliding or toppling blocks displacement on the face-to-face contacts on the

with height y and width x on a plane dipping at an

top and bottom faces of the blocks. Sliding on

angle ψp. If the friction angle between the base these faces will occur if the following conditions

of the block and the plane is φp, then the block are met (Figure 9.6(b)). The state of stress close

will be stable against sliding when the dip of the to the slope face is uniaxial with the direction of

base plane is less than the friction angle, the normal stress σ aligned parallel to the slope

Toppling failure 205

(a) (b)

x f

d

y

f

(90 – f)

d

d

f d

d

(180 – f – d)

Figure 9.6 Kinematic conditions for flexural slip preceding toppling: (a) block height/width test for toppling;

(b) directions of stress and slip directions in rock slope; (c) condition for interlayer slip; (d) kinematic

test defined on lower hemisphere stereographic projection.

face. When the layers slip past each other, σ the field shows that instability is possible where

must be inclined at an angle φd with the normal the dip direction of the planes forming sides of

◦

to the layers, where φd is the friction angle of the blocks, αd is within about 10 of the dip

the sides of the blocks. If ψf is the dip of slope direction of the slope face αf , or

face and ψd is the dip of the planes forming the

sides of the blocks, then the condition for ◦

interlayer slip is given by (Figure 9.6(c)): |(αf − αd)| < 10 (9.5)

The two conditions defining kinematic stability of

topples given by equations (9.4) and (9.5) can be

(180 − ψf − ψd) ≥ (90 − φd) (9.3)

depicted on the stereonet (Figure 9.6(d)). On the

or stereonet, toppling is possible for planes for

which the poles lie within the shaded area,

provided also that the base friction properties and

ψd ≥ (90 − ψf ) + φd (9.4) shape of the blocks meet the conditions given by

9.3.3 Block alignment test equations (9.1) and (9.2), respectively.

9.4 Limit equilibrium analysis of

the planes forming the blocks should strike

toppling on a stepped base

approximately parallel to the slope face so that

each layer is free to topple with little constraint The method of toppling analysis described in this

from adjacent layers. Observations of topples in section utilizes the same principles of limiting

206 Toppling failure

a2 b

x s

n p

yn

H ( f – p) d

Stable

b

f 2

a Figure 9.7 Model for

1 Topple

1 limiting equilibrium

Slide analysis of toppling on a

stepped base (Goodman

and Bray, 1976).

equilibrium that have been used throughout this a pseudo-static force acting on each block

book. While this method of analysis is limited to a (see Section 6.5.4), water forces can act on

few simple cases of toppling failure, it provides a the base and sides of each block, and loads

basic understanding of the factors that are produced by bridge foundations can be added

important in toppling, and allows stabilization to any specified block (Wyllie, 1999).

options to be evaluated. The stability analysis As an alternative to the detailed analysis

involves an iterative process in which the dimen- described in this section, Zanbak (1983)

sions of all the blocks and the forces acting on developed a series of design charts that can

them are calculated, and then stability of each is be used to identify unstable toppling slopes,

examined, starting at the uppermost block. Each and to estimate the support force required for

block will either be stable, toppling or sliding, and limiting equilibrium.

the overall slope is considered unstable if the

lowermost block is either sliding or toppling. A 9.4.1 Block geometry

basic requirement of this analysis is that the fric-

tion angle on the base of each block is greater The first step in toppling analysis is to calculate

the dimensions of each block. Consider the

than the dip angle of the base so that sliding on regu-lar system of blocks shown in Figure 9.7

the base plane does not occur in the absence of in which the blocks are rectangular with width x

any external force acting on the block (see and height yn. The dip of the base of the

equation (9.1)).

blocks is ψp and the dip of the orthogonal

The limit equilibrium method of analysis is planes forming the faces of the blocks is

ideally suited to incorporating external forces

ψd(ψd = 90 − ψp). The slope height is H , and

acting on the slope to simulate a wide variety of

actual conditions that may exist in the field. For the face is excavated at angle ψf while the

example, if the lower block or blocks are upper slope above the crest is at angle ψs.

unstable, then tensioned anchors with a specified

tensile strength and plunge can be installed in Angle of base plane (ψb). The base of the top-

these blocks to prevent movement. Also, ground pling blocks is a stepped surface with an overall

motion due to earthquakes can be simulated by dip of ψb (Figure 9.7). Note that there is no

Toppling failure 207

= x (1 – cos )

Upslope x

Dilatancy

Downslope

p Figure 9.8 Dilatancy of toppling

blocks with base plane

movement, coincident with normal to dip of

blocks (Zanbak, 1983).

meter ψb. However, it is necessary to use an

appropriate value for ψb in the analysis because this ◦ ◦

has a significant effect on the stability of the slope. ψb ≈ (ψp + 10 ) to (ψp + 30 ) (9.6)

That is, as the base angle become flatter, the

lengths of the blocks increase and there is a greater

It is considered that an appropriate stability

tendency of the taller blocks to topple resulting in

analysis procedure for situations where the

decreased stability of the slope. If the base angle is value of ψb is unknown is to carry out a

coincident with the base of the blocks (i.e. ψb = ψp), sensitivity analysis within the range given by

then the geometry of top-pling requires dilatancy δ

equation (9.6) and find the value that gives the

of the blocks along the base plane and shearing on

least stable condition.

the faces of the blocks (Figure 9.8). However, if the Based on the slope geometry shown in

Figure 9.7, the number of blocks n making up

base is stepped (i.e. ψb > ψp), then each block can the system is given by

topple without dilatancy, provided there is

displacement on the face-to-face contacts (Figure

9.7). It is expected that more energy is required to n H cosec(ψ ) cot(ψb) − cot(ψf ) sin(ψ )

dilate the rock mass than to develop shear along = x b + sin(ψb − ψf ) s

existing dis-continuities, and so a stepped base is (9.7)

more likely than a planar base. Examination of base

friction, centrifugal and numerical models (Goodman

and Bray, 1976; Pritchard and Savigny, 1990, 1991; The blocks are numbered from the toe of the

Adhikary et al., 1997) show that base planes tend to slope upwards, with the lowest block being 1 and

be stepped, and the approximate dip angle is the upper block being n. In this idealized model,

the height yn of the nth block in a position below

208 Toppling failure

the crest of the slope is Pn are Mn and Ln on the upper and lower

faces respectively of the block, and are given

by the following.

yn = n (a1 − b) (9.8) If the nth block is below the slope crest, then

while above the crest

Mn = yn (9.13)

yn = yn−1 − a2 − b (9.9)

Ln = yn − a1 (9.14)

The three constants a1, a2 and b that are defined

by the block and slope geometry and are given by If the nth block is the crest block, then

a

1 = x tan(ψf − ψp) (9.10) Mn = yn − a2 (9.15)

a2 = x tan(ψp − ψs) (9.11)

Ln = yn − a1 (9.16)

b = x tan(ψb − ψp) (9.12)

If the nth block is above the slope crest, then

9.4.2 Block stability

Figure 9.7 shows the stability of a system of Mn = yn − a2 (9.17)

blocks subject to toppling, in which it is

possible to distinguish three separate groups Ln = yn (9.18)

of blocks according to their mode of behavior:

For an irregular array of blocks, yn, Ln and Mn

(a) A set of stable blocks in the upper part of can be determined graphically.

the slope, where the friction angle of the When sliding and toppling occurs, frictional

base of the blocks is greater than the dip forces are generated on the bases and sides of

of this plane (i.e. φp > ψp), and the height the blocks. In many geological environments, the

is limited so the center of gravity lies friction angles on these two surfaces are likely to

be different. For example, in a steeply dipping

inside the base (y/ x < cot ψp). sedimentary sequence comprising sandstone

(b) An intermediate set of toppling blocks where beds separated by thin seams of shale, the shale

the center of gravity lies outside the base. will form the sides of the blocks, while joints in

(c) A set of blocks in the toe region, which are the sandstone will form the bases of the blocks.

pushed by the toppling blocks above. For these conditions, the friction angle of the

Depending on the slope and block geometries,

sides of the blocks (φd) will be lower than friction

the toe blocks may be stable, topple or slide.

angle on the bases (φp). These two friction

Figure 9.9 demonstrates the terms used to define angles can be incorporated into the limit

the dimensions of the blocks, and the posi-tion equilibrium analysis as follows.

and direction of all the forces acting on the blocks For limiting friction on the sides of the block:

during both toppling and sliding. Figure 9.9(a)

shows a typical block (n) with the normal and

shear forces developed on the base (Rn, Sn), Qn = Pn tan φd (9.19)

and on the interfaces with adjacent blocks (Pn, Q

n−1 = Pn−1 tan φd (9.20)

Qn, Pn−1, Qn−1). When the block is one of the

toppling set, the points of application of all forces By resolving perpendicular and parallel to the

are known, as shown in Figure 9.9(b). The points base of a block with weight Wn, the normal and

of application of the normal forces shear forces acting on the base of block n are,

Toppling failure 209

(a) (b)

Qn Pn tan d

x

Pn Pn

Q

n–1

Mn

P yn

n–1

n

W

Ln n

P

n–1

p

S tan b n Pn – 1 tan d

Rn

kn

Rn

(c) Sn

Qn

Pn

n Wn

T Rn tan b

P Q

n–1 n–1

L Rn

1

Figure 9.9 Limiting equilibrium conditions for toppling and sliding of nth block: (a) forces acting on nth block;

(b) toppling of nth block; (c) sliding of nth block (Goodman and Bray, 1976).

the sliding set (Figure 9.9(c)),

Sn = Rn tan φp (9.24)

Sn = Wn sin ψp + (Pn − Pn−1) (9.22)

However, the magnitudes of the forces Qn−1,

Considering rotational equilibrium, it is found Pn−1 and Rn applied to the sides and base of the

that the force Pn−1 that is just sufficient to

block, and their points of application Ln and Kn,

prevent toppling has the value are unknown. Although the problem is

indeterminate, the force Pn−1 required to pre-

Pn−1,t = [Pn(Mn − x tan φd) + (Wn/2) vent sliding of block n can be determined if it is

assumed that Qn−1 = (tan φd · Pn−1). Then the

(yn sin ψp − x cos ψp)]/Ln (9.23) shear force just sufficient to prevent sliding has

210 Toppling failure

relatively short block that does not satisfy

P P Wn(cos ψp tan φp − sin ψp) equation (9.2) for toppling, may still topple

(1 − tan φp tan φd) if the moment applied by the thrust force on

n−1,s = n−

(9.25) the upper face is great enough to satisfy

the condition stated in (v) above. If the

9.4.3 Calculation procedure for toppling condi-tion Pn−1,t > Pn−1,s is met for all

stability of a system of blocks blocks, then toppling extends down to

block 1 and sliding does not occur.

The calculation procedure for examining toppling (vii) Eventually a block may be reached for

stability of a slope comprising a system of blocks which Pn−1,s > Pn−1,t . This establishes

dipping steeply into the faces is as follows:

block n2, and for this and all lower blocks,

the critical state is one of sliding. The

(i) The dimensions of each block and the stability of the sliding blocks is checked

number of blocks are defined using using equation (9.24), with the block being

equations (9.7)–(9.12). unstable if (Sn = Rn tan φb). If block 1 is

(ii) Values for the friction angles on the sides stable against both sliding and toppling (i.e.

and base of the blocks (φd and φp) are P0 < 0), then the overall slope is

assigned based on laboratory testing, or considered to be stable. If block 1 either

inspection. The friction angle on the base topples or slides (i.e. P0 > 0), then the

should be greater than the dip of the overall slope is considered to be unstable.

base to prevent sliding (i.e. φp > ψp).

(iii) Starting with the top block, equation (9.2)

is used to identify if toppling will occur, 9.4.4 Cable force required to stabilize a slope

that is, when y/ x > cot ψp. For the upper

toppling block, equations (9.23) and If the calculation process described in Section

(9.25) are used to calculate the lateral 9.4.3 shows that block 1 is unstable, then a

forces required to prevent toppling and tensioned cable can be installed through this

sliding, respectively. block and anchored in stable rock beneath the

(iv) Let n1 be the uppermost block of the zone of toppling to prevent movement. The

toppling set. design parameters for anchoring are the bolt

(v) Starting with block n1, determine the lat-eral tension, the plunge of the anchor and its

forces Pn−1,t required to prevent top-pling, position on block 1 (Figure 9.9(c)).

and Pn−1,s to prevent sliding. If Pn−1,t > Suppose that an anchor is installed at a

Pn−1,s, the block is on the point of toppling plunge angle ψT through block 1 at a distance

and Pn−1 is set equal to Pn−1,t , or if Pn−1,s L1 above its base. The anchor tension required

> Pn−1,t , the block is on the point of sliding to prevent toppling of block 1 is

and Pn−1 is set equal to Pn−1,s.

In addition, a check is made that there W1/2(y1 sin ψp − x cos ψp) + P1(y1 − x tan φd )

Tt = L1 cos(ψp + ψT )

is a normal force R on the base of the

block, and that sliding does not occur on (9.26)

the base, that is while the required anchor tension to prevent

sliding of block 1 is

Rn > 0 and (|Sn| > Rn tan φp)

P1(1 − tan φp tan φd) − W1(tan φp cos ψp − sin ψp)

(vi) The next lower block (n1 − 1) and all the Ts = tan φp sin(ψp + ψT ) + cos(ψp + ψT )

Toppling failure 211

When the force T is applied to block 1, the overturns by a small amount, there are edge-to-

normal and shear force on the base of the face contacts between the blocks, and the friction

block are, respectively, required to prevent further rotation increases.

Hence, a slope just at limiting equilibrium is meta-

R1 = P1 tan φd + T sin(ψp + ψT) + W1 cos ψp stable. However, rotation equal to 2(ψb − ψp) will

(9.28) convert the edge-to-face contacts along the sides

S1 = P1 − T cos(ψp + ψT) + W1 sin ψp of the columns into continuous face contacts and

the friction angle required to prevent further

(9.29)

rotation will drop sharply, possibly even below

that required for initial equilibrium. The choice of

The stability analysis for a slope with a tensioned factor of safety, therefore, depends on whether or

anchor in block 1 is identical to that described in not some deformation can be tolerated.

Section 9.4.3, apart from the calculations relating

The restoration of continuous face-to-face

to block 1. The required tension is the greater of

contact of toppled columns of rock is probably an

Tt and Ts defined by equations (9.26) and (9.27). important arrest mechanism in large-scale top-

pling failures. In many cases in the field, large

9.4.5 Factor of safety for limiting equilibrium surface displacements and tension crack forma-

analysis of toppling failures tion can be observed and yet the volumes of rock

that fall from the face are small.

For both reinforced and unreinforced slopes, the

factor of safety can be calculated by finding the

friction angle for limiting equilibrium. The pro-

cedure is first to carry out the limiting equilibrium 9.4.6 Example of limit equilibrium

stability analysis as described in Section 9.4.3 analysis of toppling

using the estimated values for the friction angles.

If block 1 is unstable, then one or both of the fric- The following is an example of the application

tion angles are increased by increments until the of the Goodman and Bray limit equilibrium ana-

value of P0 is very small. Conversely, if block 1 is lysis to calculate the factor of safety and

stable, then the friction angles are reduced until required bolting force of the toppling failure

illustrated in Figure 9.10(a).

P0 is very small. These values of the friction A rock face 92.5 m high (H ) is cut at an angle

angles are those required for limiting equilibrium. ◦

of 56.6 (ψf ) in a layered rock mass dipping at

The limiting equilibrium friction angles are ◦ ◦

termed the required friction angles, while the 60 into the face (ψd = 60 ); the width of each

block is 10 m ( x). The angle of the slope above

actual friction angles of the block surfaces are ◦

termed the available friction angles. The factor of the crest of the cut is 4 (ψs), and the base of the

safety for toppling can be defined by dividing the blocks is stepped 1 m at every block (atn (1/10) =

◦ ◦

tangent of the friction angle believed to apply 5.7 , and ψb = (5.7+ψp) = 35.7 ). Based on this

geometry, there are 16 blocks formed between

to the rock layers (tan φavailable), by the tangent the toe and crest of the slope (equation 9.7);

of the friction angle required for equilibrium block 10 is at the crest. Using equations (9.10)–

(tan φ ). (9.12), the constants are a1 = 5.0 m, a2 = 5.2 m

required

tan φ and b = 1.0 m. These constants are used to

available

calcu-late the height yn of each block, and the

FS = height to width ratio yn/ x as shown on the table

tan φrequired (9.30) in Figure 9.10(b).

The actual factor of safety of a toppling slope The friction angles on the faces and bases of

depends on the details of the geometry of the top- the blocks are equal and have a value of

pling blocks. Figure 9.7 shows that once a column ◦

38.15 (φavailable). The unit weight of the rock is

212 Toppling failure

(a) 5.8

56.6

4

30

16

15

13 14

11 12

10

9

8

m

7

S

92.5

6 n

5

R

4 n

3

2

1 T

(b)

n y P P

n yn /x Mn Ln n·t n·s Pn Rn Sn Sn /Rn Mode

16 4.0 0.4 0 0 0 866 500 0.577

15 10.0 1.0 0 0 0 2165 1250 0.577 STABLE

14 16.0 1.6 0 0 0 3463 2000 0.577 ________

13 22.0 2.2 17 22 0 0 0 4533.4 2457.5 0.542

12 28.0 2.8 23 28 292.5 –2588.7 292.5 5643.3 2966.8 0.526 T

11 34.0 3.4 29 34 825.7 –3003.2 825.7 6787.6 3520.0 0.519 O

10 35

40.0 4.0 35 1556.0 –3175.0 1556.0 7662.1 3729.3 0.487 P

9 36

36.0 3.6 31 2826.7 –3150.8 2826.7 6933.8 3404.6 0.491 P

8

32.0 3.2 32 27 3922.1 –1409.4 3922.1 6399.8 3327.3 0.520 L

7

28.0 2.8 28 23 4594.8 156.8 4594.8 5872.0 3257.8 0.555 I

6

5 24.0 2.4 24 19 4837.0 1300.1 4837.0 5352.9 3199.5 0.598 N

4 20.0 2.0 20 15 4637.5 2013.0 4637.5 4848.1 3159.4 0.652 G

3 16.0 1.6 16 11 3978.1 2284.1 3978.1 4369.4 3152.5 0.722

2 12.0 1.2 12 7 2825.6 2095.4 2825.6 3707.3 2912.1 0.7855 ________

1 8.0 0.8 8 3 1103.1 1413.5 1413.5 2471.4 1941.3 0.7855 SLIDING

4.0 0.4 4 – –1485.1 472.2 472.2 1237.1 971.8 0.7855

(c) Block

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

0

1

2

S

(M

N)

3 n

rce

4

Fo

6

R

n

7

Figure 9.10 Limited equilibrium analysis of a toppling slope: (a) slope geometry; (b) table listing block

dimensions, calculated forces and stability mode; (c) distribution of normal (R) and shear (S) forces on

base of blocks (Goodman and Bray, 1976).

3 ◦

25 kN/m . It is assumed that the slope is dry, of the blocks is 38.15 and the dip of the base is

and that there are no external forces acting. ◦

30 , the upper blocks are stable against sliding.

The stability analysis is started by examining Equation (9.2) is then used to assess the toppling

the toppling/sliding mode of each block, starting mode. Since cot ψp = 1.73, blocks 16, 15 and 14

at the crest. Since the friction angle on the base are stable, because for each the ratio yn/ x is

Toppling failure 213

less than 1.73. That is, these three blocks are short defined by Rn > 0 and |Sn| < Rn tan φp are

and their center of gravity lies inside the base. satisfied everywhere.

For block 13, the ratio yn/ x has the value 2.2,

which is greater than 1.73 and the block topples. 9.4.7 Application of external

Therefore, P13 is equal to 0 and P12 is calcu- forces to toppling slopes

lated as the greater of P12,t and P12,s given by

There may be circumstances where there are

equations (9.23) and (9.25) respectively. This

calculation procedure is used to examine the external forces acting on the slope and it is

stability of each block in turn progressing down necessary to investigate their effect on stability.

the slope. As shown in the table of forces in Examples of external forces include water forces

Figure 9.10(b), Pn−1,t is the larger of the two acting on the sides and bases of the blocks, earth-

quake ground motion simulated as a horizontal

forces until a value of n = 3, whereupon Pn−1,s force acting on each block (see Section 6.5.4), and

is larger. Thus blocks 4 to 13 constitute the

point loads produced by bridge piers located on a

poten-tial toppling zone, and blocks 1 to 3

constitute a sliding zone. specific block(s). Another external force that has

The factor of safety of this slope can be been considered in the worked example in Section

found by increasing the friction angles until the 9.4.6 are rock anchors that are secured in stable

base blocks are just stable. It is found that the ground beneath the toppling mass and then

required friction angle for limit equilibrium con- tensioned against the face.

◦ A feature of limit equilibrium analysis is that

ditions is 39 , so the factor of safety as given any number of forces can be added to the

by equation (9.30) is 0.97 (tan 38.15/ tan 39). analysis, provided that their magnitude, direction

The analysis also shows that the required and point of application are known. Figure 9.11

tension in an anchor installed horizontally in shows a portion of a toppling slope in which there

block 1 to just stabilize the toe blocks is 500 kN is a sloping water table. The forces act-ing on

per meter length of slope. This compares with block n include the force Q inclined at an angle

the maximum value of P (in block 5) equal to ψQ below the horizontal, and three water forces

4837 kN/m.

V1, V2 and V3, as well as the forces Pn and

If tan φ is reduced to 0.650, it will be found

that blocks 1 to 4 in the toe region will slide Pn−1 produced by the blocks above and below.

while blocks 5 to 13 will topple. The tension in By resolving all these forces normal and parallel

an anchor installed horizontally through block to the base of the blocks, it is possible to modify

1, required to restore equilibrium, is found to be equations (9.23) and (9.25) as follows. Consider-

2013 kN/m of slope crest. This is not a large ing rotational equilibrium, the force Pn−1,t that is

number, demonstrating that support of the just sufficient to prevent toppling of block n has

“keystone” is remarkably effective in increas- the value:

ing the degree of stability. Conversely, remov-

Pn−1,t = {Pn(Mn − x tan φd)

ing or weakening the keystone of a toppling

slope that is near failure can have serious

+ Wn/2(yn sin ψp − x cos ψp)

consequences.

2

When the distribution of P forces has been + V1yw /3 + γw x /6

defined in the toppling region, the forces Rn and

× cos ψp(zw + 2yw ) − V3zw /3

Sn on the base of the blocks can be calcu-lated

using equations (9.21) and (9.22). Assuming + Q[− sin(ψQ − ψp) x/2

[Qn−1 = Pn−1 tan φs], the forces Rn and Sn −1

can also be calculated for the sliding region. + cos(ψQ − ψp)yn]}Ln (9.31)

Figure 9.10(c) shows the distribution of these Assuming that the blocks are in a state of limiting

forces throughout the slope. The conditions equilibrium, the force just sufficient to prevent

214 Toppling failure

Pn

y

n

n d

+

1

M n

n

P

n

P

n–1

y

w d

n –

Wn 1

1

–

V n

1 P

R

b n

S

n Ln

V3

Zw

V

2

x

sliding of block n has the value The limit equilibrium stability analysis then

proceeds as before using the modified versions

Pn−1,s = Pn + {−W (cos ψp tan φp − sin ψp) of the equations for Pn−1,t and Pn−1,s.

+ V1 − V2 tan φp − V3 9.5 Stability analysis of flexural toppling

+ Q[− sin(ψQ − ψp) tan φp

Figure 9.3(b) shows a typical flexural toppling

+ cos(ψQ − ψp)]} failure in which the slabs of rock flex and

maintain fact-to-face contact. The mechanism of

−1

× (1 − tan φp tan φd) (9.32) flexural toppling is different from the block top-

where pling mechanism described in Section 9.4. There-

fore, it is not appropriate to use limit equilibrium

stability analysis for design of toppling slopes.

1 2

V1 = 2 γw cos ψp · yw ; Techniques that have been used to study the sta-

V 1 bility of flexural toppling include base friction

2 = 2 γw cos ψp(yw + zw ) x

models (Goodman, 1976), centrifuges (Adhikary

1 2 et al., 1997) and numerical modeling (Pritchard

V3 = 2 γw cos ψpzw (9.33)

Toppling failure 215

(a) (b)

50 m

50 m

(c)

50 m

Figure 9.12 UDEC model of toppling pit slope: (a) pure flexural toppling deformation with grid point velocity

vectors; (b) contours of horizontal displacement for toppling slopes; (c) area of failed nodes due to flexure

(Pritchard and Savigny, 1990).

and Savigny, 1990, 1991). All these models information, for example, on the allowable face

show the common features of this failure angle for specific geological conditions and

mechanism including interlayer shearing, slope height.

obsequent scarps at the crest, opening of

Alternatively, computer simulations of block

tension cracks that decrease in width with

movement provide a means of studying a wide

depth, and a limiting dip angle ψb for the base range of geometric and material properties. One

of toppling. As discussed in Section 9.4.1, dip of the most suitable computer programs for these

angle ψb is steeper than the plane ψp (normal simulations is UDEC (Universal Distinct Element

◦ Code), which has been developed by Itasca

to the dip of the slabs) by about 10–30 .

The centrifuge modeling by Adhikary et al. has Consulting Group in Minnesota (Itasca, 2000).

been used to develop a series of design charts that Figure 9.12 shows the results of an analysis

relate stability to the slope face angle, the dip of the carried out on an open pit mine slope (Pritchard

blocks into the face and the ratio of the slope height and Savigny, 1990). The main features of UDEC

to the width of the slabs. Another input parameter is analysis for studying toppling slopes are that it

the tensile strength of the rock, because bending of

the slabs induces tensile crack-ing in their upper • incorporates a number of materials each

face. The design charts provide with differing strength properties;

216 Toppling failure

• recognizes the existence of contacts or failure is described by Davies and Smith (1993).

interfaces between discrete bodies, such as The toppling occurred in siltstones in which the

slabs of rocks formed by discontinuities dip- beds were very closely spaced and dipped at

ping steeply into the slope face; ◦ ◦

between 90 and 70 into the face. Excavation for

• calculates the motion along contacts by a bridge abutment resulted in a series of ten-sion

assigning a finite normal stiffness along the cracks along the crest, and stabilization of the

discontinuities that separate the columns of slope required the installation of tensioned rock

rock. The normal stiffness of a discontinuity bolts and excavation to reduce the slope angle.

is defined as the normal closure that occurs

on the application of a normal stress and 9.6 Example Problem 9.1:

can be measured from a direct shear test toppling failure analysis

(see Figure 4.17);

• assumes deformable blocks that undergo Statement

bending and tensile failure;

Consider a 6 m high slope with an overhanging

• allows finite displacements and rotations of ◦

the toppling blocks, including complete face at an angle of 75 There is a fault, dipping

◦

detachment, and recognizes new contacts at an angle of 15 out of the face, at the toe of

automatically as the calculation progresses; the slope that is weathering and undercutting

• uses an explicit “time”-marching scheme to the face. A tension crack, which is wider at the

solve the equation of motion directly. This top than at the bottom, has developed 1.8 m

allows modeling of progressive failure, or behind the crest of the slope indicating that the

face is marginally stable (Figure 9.13). The

the amount of creep exhibited by a series ◦

of top-pling blocks for a chosen slope friction angle φ of the fault is 20 and the

condition, such as excavation at the toe of cohesion c is 25 kPa. The slope is dry.

the slope. Note that the time step in the

analysis is not actual time but a simulation Required

of progressive movement; and (a) Calculate the factor of safety of the block

• allows the user to investigate different against sliding if the density of the rock is

stabilization measures, such as installing 3

23.5 kN/m .

rock bolts or installing drain holes, to

(b) Is the block stable against toppling as

determine which scenario has the most

defined by the relation:

effect on block movements.

x/y > tan ψp—stable?

Because of the large number of input paramet-

ers that are used in UDEC and the power of the

analysis, the most reliable results are obtained if

the model can be calibrated against an existing

toppling failure in similar geological conditions to

those in the design slope. The ideal situation is in

mining operations where the development of the Y=6m

topple can be simulated by UDEC as the pit is

deepened and movement is monitored. This =15

allows the model to be progressively updated

with new data. Chapter 10 discusses numerical

x = 1.8 m

modeling of slopes in more detail.

The application of kinematic stability tests Figure 9.13 Toppling block illustrating Example

and reinforcement design for a flexural toppling Problem 9.1.

Toppling failure 217

(c) How much more undercutting of the fault (b) From the dimensions given on Figure 9.13,

must occur before toppling failure takes the following values are obtained to test

place? stability conditions:

(d) What stabilization measures would be

appropriate for this slope?

x/y = 0.3

Solution tan 15 = 0.27

(a) The factor of safety against sliding is

determined by the methods described in

The block is stable against toppling because

Chapter 6; the equation for a dry slope is

0.3 > 0.27.

=

cA + W cos ψp · tan φp (c) If weathering results in the width of the

FS base of the block being reduced by a

W sin ψp

further 0.2 m, toppling is likely to occur

= ((25 × 1.8) + (254 × cos 15 because: x/y = (1.8 − 0.2)/6 ≈ 0.27.

× tan 20))/(254 × sin 15) (d) Stabilization measures which could be

used on this slope include the following:

≈ 2.0

limit build-up of water pressure both in

A = base area of block the tension crack and on the fault at the

2 toe.

= 1.8 m /m • Application of reinforced shotcrete to the

W = weight of block/m fault to prevent further weathering.

• Trim blasting to reduce the slope angle

= 23.5 × 1.8 × 6

and the dimension y, if this can be

= 254 kN/m achieved without destabilizing the block.

Chapter 10

Numerical analysis

1

Dr Loren Lorig and Pedro Varona

material. Elastic–plastic models use strength

The previous four chapters discussed limit equi- parameters to limit the shear stress that a zone

librium methods of slope stability analysis for rock 2

bounded by specified slide planes. In con-trast, may sustain.

this chapter discusses numerical analysis The zones may be connected together,

methods to calculate the factor of safety without termed a continuum model, or separated by

pre-defining slide planes. These methods are discontinuities, termed a discontinuum model.

Discontinuum models allow slip and separation

more recent developments than limit equilibrium

at explicitly located surfaces within the model.

meth-ods and, at present (2003), are used

Numerical models tend to be general purpose in

predomin-ately in open pit mining and landslides

nature—that is, they are capable of solving a wide

studies, where interest often focuses on slope

variety of problems. While it is often desirable to

displace-ments rather than on the relative

have a general-purpose tool available, it requires

magnitude of resisting and displacing forces.

that each problem be constructed individually. The

Numerical models are computer programs that

zones must be arranged by the user to fit the limits

attempt to represent the mechanical response of

of the geomechanical units and/or the slope

a rock mass subjected to a set of initial conditions

geometry. Hence, numerical models often require

such as in situ stresses and water levels, bound-

more time to set up and run than special-purpose

ary conditions and induced changes such as

tools such as limit equilibrium methods.

slope excavation. The result of a numerical

There are several reasons why numerical

model sim-ulation typically is either equilibrium or

models are used for slope stability studies.

collapse. If an equilibrium result is obtained, the

resultant stresses and displacements at any point

in the rock mass can be compared with • Numerical models can be extrapolated

measured values. If a collapse result is obtained, confid-ently outside their databases in

the predicted mode of failure is demonstrated. comparison to empirical methods in which

Numerical models divide the rock mass into the failure mode is explicitly defined.

zones. Each zone is assigned a material model • Numerical analysis can incorporate key geo-

and properties. The material models are idealized logic features such as faults and ground water

stress/strain relations that describe how the providing more realistic approximations of

material behaves. The simplest model is a linear behavior of real slopes than analytic models.

elastic model, which uses the elastic properties

2 In numerical analysis the terms “elements” and “zones”

are used interchangeably. However, the term element is

1 Itasca Consulting Group, Inc., Minneapolis, Minnesota used more commonly in finite element analysis, and the

55401 USA. term zone in finite difference analysis.

Numerical analysis 219

In comparison, non-numerical analysis meth- strength to the reduced shear strength at failure.

ods such as analytic, physical or limit equi- This shear-strength reduction technique was

librium may be unsuitable for some sites or used first with finite elements by Zienkiewicz et al.

tend to oversimplify the conditions, possibly (1975) to compute the safety factor of a slope

leading to overly conservative solutions. composed of multiple materials.

• Numerical analysis can help to explain To perform slope stability analysis with the

observed physical behavior. shear strength reduction technique, simulations

• Numerical analysis can evaluate multiple are run for a series of increasing trial factors of

pos-sibilities of geological models, failure safety (f ). Actual shear strength properties,

modes and design options. cohesion (c) and friction angle (φ), are reduced

for each trial according to the equations

Many limit equilibrium programs exist to

determine factors of safety for slopes. These

execute very rapidly, and in the case of the method 1

of slices for circular failure, use an approximate ctrial = f c (10.1)

scheme in which a number of assumptions are 1

made, including the location and angle of inter-slice

forces (see Section 8.2). Several assumed slide φtrial = arctan f tan φ (10.2)

surfaces are tested, and the one giving the lowest

factor of safety is chosen. Equilibrium is satis-fied If multiple materials and/or joints are present, the

only on an idealized set of surfaces. With numerical reduction is made simultaneously for all materi-

models, a “full” solution of the coupled als. The trial factor of safety is increased

stress/displacement, equilibrium and constitutive gradually until the slope fails. At failure, the factor

equations is made. Given a set of properties, the of safety equals the trial factor of safety (i.e. f =

system is found either to be stable or unstable. By FS). Dawson et al. (1999) show that the shear

performing a series of simulations with vari-ous strength reduction factors of safety are generally

properties, the factor of safety can be found within a few percent of limit analysis solutions

corresponding to the point of stability. when an associated flow rule, in which the friction

The numerical analysis is much slower, but angle and dilation angle are equal, is used.

much more general. Only since the late 1990s, The shear strength reduction technique has two

with the advent of faster computers, has it main advantages over limit equilibrium slope sta-

become a practical alternative to the limit bility analyses. First, the critical slide surface is

equilibrium method. Even so, while limit found automatically, and it is not necessary to

equilibrium solu-tions may require just a few specify the shape of the slide surface (e.g. circular,

seconds, numerical solutions to large complex log spiral, piecewise linear) in advance. In general,

problems can take an hour or more, particularly the failure surface geometry for slopes is more

when discontinuum behavior is involved. The complex than simple circles or segmented sur-

third section of this chapter presents typical faces. Second, numerical methods automatically

safety factor analyses for the most common satisfy translational and rotational equilibrium,

discontinuum failure modes in rock slopes. whereas not all limit equilibrium methods do sat-isfy

For slopes, the factor of safety often is defined as equilibrium. Consequently, the shear strength

the ratio of the actual shear strength to the min- reduction technique usually will determine a safety

imum shear strength required to prevent failure. A factor equal to or slightly less than limit equilibrium

logical way to compute the factor of safety with a methods. Itasca Consulting Group (2002) gives a

finite element or finite difference program is to detailed comparison of four limit equilibrium

reduce the shear strength until collapse occurs. The methods and one numerical method for six different

factor of safety is the ratio of the rock’s actual slope stability cases.

220 Numerical analysis

such as slices

Stresses Computed everywhere using field Computed approximately on certain

equations surfaces

Deformation Part of the solution Not considered

Failure Yield condition satisfied everywhere; Failure allowed only on certain

slide surfaces develop “automatically” pre-defined surfaces; no check on

as conditions dictate yield condition elsewhere

Kinematics The “mechanisms” that develop satisfy A single kinematic condition is

kinematic constraints specified according to the particular

geologic conditions

A summary of the differences between a assumed within deformable blocks. The most

numerical solution and the limit equilibrium widely used discrete element codes for slope

method is shown in the Table 10.1. stability studies are UDEC (Universal Distinct

Element Code; Itasca Consulting Group, 2000)

and 3DEC (3-Dimensional Distinct Element

Code; Itasca Consulting Group, 2003). Of fun-

10.2 Numerical models

damental importance in discontinuum codes is

All rock slopes involve discontinuities. Represent- the representation of joint or discontinuity

ation of these discontinuities in numerical models beha-vior. Commonly used relations for

differs depending on the type of model. There are represent-ing joint behavior are discussed later

two basic types of models: discontinuum models in this section.

and continuum models. Discontinuities in discon- Continuum codes assume material is con-

tinuum models are represented explicitly—that is, tinuous throughout the body. Discontinuities

the discontinuities have a specific orientation and are treated as special cases by introducing

location. Discontinuities in continuum mod-els are interfaces between continuum bodies. Finite

represented implicitly, with the intention that the 2

element codes such as PHASE (Rocscience,

behavior of the continuum model is sub-stantially 2002c) and its predecessor PHASES (P lastic

equivalent to the real jointed rock mass being H ybrid Analysis of Stress for Estimation of

represented. Support) and finite difference codes such as FLAC

Discontinuum codes start with a method designed (F ast Lagrangian Analysis of Continua; Itasca

specifically to model discontinua and treat Consulting Group, 2001) cannot handle gen-eral

continuum behavior as a special case. Dis- interaction geometry (e.g. many intersecting joints).

continuum codes generally are referred to as Their efficiency may degenerate drastic-ally when

Discrete Element codes. A Discrete Element code connections are broken repeatedly. Typical

will typically embody an efficient algorithm for continuum-based models may have less than ten

detecting and classifying contacts, and maintain a non-intersecting discontinuities. Of fun-damental

data structure and memory allocation scheme that importance to continuum codes and deformable

can handle many hundreds or thousands of blocks in discrete element codes is representation

discontinuities. The discontinuities divide the of the rock mass behavior. Continuum relations

problem domain into blocks that may be either rigid used to represent rock mass behavior are discussed

or deformable; continuum behavior is later in this section.

Numerical analysis 221

Finite element programs are probably more 10.2.1 Joint material models

familiar, but the finite difference method is

The material model used most commonly to

perhaps the oldest numerical technique used to

represent joints is a linear-elastic–perfectly-plastic

solve sets of differential equations. Both finite

model. The limiting shear strength is defined by

element and finite difference methods produce a

the usual Mohr–Coulomb parameters of friction

set of algebraic equations to solve. While the

angle and cohesion (see Section 4.2). A peak

methods used to derive the equations are differ-

and residual shear strength relation can also be

ent, the resulting equations are the same. Finite

spe-cified for the joints. The residual strength is

difference programs generally use an “explicit”

used after the joint has failed in shear at the peak

time-marching scheme to solve the equations,

strength. The elastic behavior of the joints is spe-

whereas finite element methods usually solve

cified by joint normal and shear stiffnesses, which

systems of equations in matrix form.

may be linear or piece-wise linear.

Although a static solution to a problem is usu-ally

of interest, the dynamic equations of motion are

10.2.2 Rock mass material models

typically included in the formulation of finite

difference programs. One reason for doing this is to It is impossible to model all discontinuities in a

ensure that the numerical scheme is stable when large slope, although it may be possible to model

the physical system being modeled is unstable. With the discontinuities for a limited number of

non-linear materials, there is always the possibility benches. Therefore, in large slopes much of the

of physical instability—for example, the failure of a rock mass must be represented by an equivalent

slope. In real life, some of the strain energy is continuum in which the effect of the discontinu-

converted to kinetic energy. Expli-cit finite-difference ities is to reduce the intact rock elastic properties

programs model this process directly, because and strength to those of the rock mass. This is

inertial terms are included. In contrast, programs true whether or not a discontinuum model is

that do not include inertial terms must use some used. As mentioned in the introduction to this

numerical procedure to treat physical instabilities. chapter, numerical models divide the rock mass

Even if the procedure is successful at preventing into zones. Each zone is assigned a mater-ial

numerical instability, the path taken may not be model and material properties. The material

realistic. The con-sequence of including the full law models are stress/strain relations that describe

of motion in finite difference programs is that the how the material behaves. The simplest model is

user must have some physical feel for what is a linear elastic model that uses only the elastic

happening. Explicit finite-difference programs are properties (Young’s modulus and Poisson’s ratio)

not black boxes that will “give the solution.” The of the material. Linear elastic–perfectly plastic

behavior of the numerical system must be stress–strain relations are the most commonly

interpreted. used rock mass material models. These models

FLAC and UDEC are two-dimensional finite- typically use Mohr–Coulomb strength paramet-

difference programs developed specifically for ers to limit the shear stress that a zone may

geomechanical analysis. These codes can simu- sustain. The tensile strength is limited by the spe-

late varying loading and water conditions, and cified tensile strength, which in many analyses is

have several pre-defined material models for rep- taken to be 10% of the rock mass cohe-sion.

resenting rock mass continuum behavior. Both Using this model, the rock mass behaves in an

codes are unique in their ability to handle highly isotropic manner. Strength anisotropy can be

non-linear and unstable problems. The three- introduced through a ubiquitous joint model,

dimensional equivalents of these codes are which limits the shear strength according to a

FLAC3D (F ast Lagrangian Analysis of Continua Mohr–Coulomb criterion in a specified direction.

in 3 Dimensions; Itasca Consulting Group, 2002) The direction often corresponds to a predominant

and 3DEC (Itasca Consulting Group, 2003). jointing orientation.

222 Numerical analysis

A more complete equivalent-continuum model to the failure criterion and that the flow rule is

that includes the effects of joint orientation and isotropic, whereas the Hoek–Brown criterion is not.

spacing is a micropolar (Cosserat) plasticity Recently, Cundall et al. (2003) has proposed a

model. The Cosserat theory incorporates a local scheme that does not use a fixed form of the flow

rotation of material points as an independent rule, but rather one that depends on the stress level,

parameter, in addition to the translation assumed and possibly some measure of damage.

in the classical continuum, and couple stresses Real rock masses often appear to exhibit pro-

(moments per unit area) in addition to the clas- gressive failure—that is, the failure appears to

sical stresses (forces per unit area). This model, progress over time. Progressive failure is a com-

as implemented in FLAC, is described in the plex process that is understood poorly and diffi-

context of slope stability by Dawson and Cundall cult to model. It may involve one or more of the

(1996). The approach has the advantage of using following component mechanisms:

a con-tinuum model while still preserving the

ability to consider realistic joint spacing explicitly. • Gradual accumulation of strain on principal

The model has not yet (as of 2003) been structures and/or within the rock mass;

incorporated into any publicly available code. • Increases in pore pressure with time; and

The most common failure criterion for rock • Creep, which is time-dependent deformation

masses is the Hoek–Brown failure criterion (see of material under constant load.

Section 4.5). The Hoek–Brown failure criterion is an

empirical relation that characterizes the stress Each of these components is discussed briefly

conditions that lead to failure in intact rock and rock later in the context of slope behavior.

masses. It has been used successfully in design Gradual accumulation of strain on principal

approaches that use limit equilibrium solutions. It structures within the rock mass usually results from

also has been used indirectly in numerical models excavation, and “time” is related to the excavation

by finding equivalent Mohr– Coulomb shear strength sequence. In order to study the pro-gressive failure

parameters that provide a failure surface tangent to effects due to excavation, one must either introduce

the Hoek–Brown fail-ure criterion for specific characteristics of the post-peak or post-failure

confining stresses, or ranges of confining stresses. behavior of the rock mass into a strain-softening

The tangent Mohr– Coulomb parameters are then model or introduce similar characteristics into the

used in traditional Mohr–Coulomb type constitutive explicit discontinuities. In practice, there are at least

relations and the parameters may or may not be two difficulties asso-ciated with strain-softening rock

updated during analyses. The procedure is awkward mass models. The first is estimating the post-peak

and time-consuming, and consequently there has strength and the strain over which the strength

been little direct use of the Hoek–Brown failure reduces. There appear to be no empirical guidelines

criterion in numerical solution schemes that require for estimat-ing the required parameters. This means

full constitutive models. Such models solve for dis- that the properties must be estimated through

placements, as well as stresses, and can continue calibration. The second difficulty is that, for a

the solution after failure has occurred in some loc- simulation in which the response depends on shear

ations. In particular, it is necessary to develop a localization and in which material softening is used,

“flow rule,” which supplies a relation between the the res-ults will depend on the zone sizes. However,

components of strain rate at failure. There have it is quite straightforward to compensate for this

been several attempts to develop a full con-stitutive form of mesh-dependence. In order to do this,

model from the Hoek–Brown criterion: for example, consider a displacement applied to the boundary of

Pan and Hudson (1988), Carter et al. (1993) and a body. If the strain localizes inside the body, the

Shah (1992). These formulations assume that the applied displacement appears as a jump across the

flow rule has some fixed relation localized band. The thickness of the band

Numerical analysis 223

contracts until it is equal to the minimum strength reduction process, these zones should

allowed by the grid, that is a fixed number of be considered as a new material with lower

zone widths. Thus, the strain in the band is strength, but no further softening should be

allowed due to the plastic strains associated to

the gradual reduction of strength.

ε = u/n z (10.3) Increases in pore pressure with time are not

where n is a fixed number, u is the common in rock slopes for mines. More com-

displacement jump, and z is the zone width. monly, the pore pressures reduce due to deepen-

If the softening slope is linear, the change in ing of the pit and/or drainage. However, there are

a property value p is proportional to strain, the cases in which the pore pressures do increase

change in property value with displacement is: with time. In such cases, the slope may appear to

fail progressively.

p s Creep, which is time-dependent deformation of

u=n z (10.4) material under constant load, is not commonly

· considered in the context of slope stability. It is

where s is the softening slope. much more common in underground excav-ations.

In order to obtain mesh-independent results, a Several material models are available to study creep

scaled softening slope s can be input, such that behavior in rock slopes. These include classical

viscoelastic models, power law models, and the

Burger-creep viscoplastic model. Applic-ation of a

s=s z (10.5) creep model to the study of slope beha-vior at

where s is constant. Chuquicamata mine in Chile is discussed later in

In this case, ( p/ u) is independent of z. If the this chapter (see Section 10.5.2).

softening slope is defined by the critical strain,

s

ε , then

crit 10.3 Modeling issues

1 Modeling requires that the real problem be ideal-

s

εcrit ∝ z (10.6) ized, or simplified, in order to fit the constraints

imposed by factors such as available material

For example, if the zone size is doubled, the models and computer capacity. Analysis of rock

critical strain must be halved for comparable mass response involves different scales. It is

results. impossible—and undesirable—to include all fea-

Strain-softening models for discontinuities are tures, and details of rock mass response

much more common than similar relations for mechan-isms, into one model. In addition, many

rock masses. Strain-softening relations for dis- of the details of rock mass behavior are unknown

continuities are built into UDEC and 3DEC, and and unknowable; therefore, the approach to

can be incorporated into interfaces in FLAC and model-ing is not as straightforward as it is, say, in

FLAC3D via a built-in programming language other branches of mechanics. This section

such as FISH functions. Strain-softening models discusses the basic issues that must be resolved

require special attention when computing safety when setting up a numerical model.

factors. If a strain-softening constitutive model is

used, the softening logic should be turned off

10.3.1 Two-dimensional analysis versus

during the shear strength reduction process or

three-dimensional analysis

the factor of safety will be underestimated. When

the slope is excavated, some zones will have The first step in creating a model is to

exceeded their peak strength, and some amount decide whether to perform two-dimensional

of softening will have taken place. During the or three-dimensional analyses. Prior to 2003,

224 Numerical analysis

three-dimensional analyses were uncommon, of 1. The average slope height was 100 m. Piteau

but advances in personal computers have and Jennings (1970) found that the average slope

permitted three-dimensional analyses to be angle for slopes with radius of curvature of 60 m

◦ ◦

performed routinely. Strictly speaking, three- was 39.5 as compared to 27.3 for slopes with a

dimensional analyses are radius of curvature of 300 m.

recommended/required in the following: Hoek and Bray (1981) summarize their experi-

ence with the stabilizing effects of slope

1 The direction of principal geologic structures curvature as follows. When the radius of

◦ curvature of a concave slope is less than the

does not strike within 20–30 of the strike of

the slope. ◦

height of the slope, the slope angle can be 10

2 The axis of material anisotropy does not steeper than the angle suggested by

◦

strike within 20–30 of the slope. conventional stability ana-lysis. As the radius of

3 The directions of principal stresses are not curvature increases to a value greater than the

neither parallel nor perpendicular to the slope height, the correc-tion should be

slope. decreased. For radii of curvature in excess of

4 The distribution of geomechanical units twice the slope height, the slope angle given by a

varies along the strike of the slope. conventional stability analysis should be used.

5 The slope geometry in plan cannot be rep- To better quantify the effects of slope curvature

resented by two-dimensional analysis, which on stability, a series of generic analyses were

assumes axisymmetric or plain strain. per-formed. All analyses assumed a 500 m high

◦

dry slope with a 45 face angle excavated in an

Despite the forgoing, most design analysis for iso-tropic homogeneous material with a density of

slopes assumes a two-dimensional geometry 3

2600 kg/m . Initial in situ stresses are assumed

com-prising a unit slice through an infinitely

to be lithostatic, and the excavation was made in

long slope, under plane strain conditions. In

40 m decrements beginning from the ground sur-

other words, the radii of both the toe and crest

face. For these conditions, pairs of friction angle

are assumed to be infinite. This is not the

and cohesion values were selected to produce a

condition encountered in practice—particularly factor of safety of 1.3 using circular failure chart

in open pit mining where the radii of curvature number 1 (see Section 8.3). A factor of safety of

can have an important effect on safe slope 1.3 is a value that is frequently used in the design

angles. Concave slopes are believed to be of slopes for open pit mines. The actual values

more stable than plain strain slopes due to the used are shown in Table 10.2.

lateral restraint provided by material on either A series of analyses was performed using FLAC

side of a potential failure in a concave slope. for different radius of curvature for both concave

Despite its potential importance in slope sta- and convex slopes. For concave slopes, the radius

bility, very little has been done to quantify this of curvature is defined as the distance between

effect. Jenike and Yen (1961) presented the

results of limit theory analysis of axisymmetric

slopes in a rigid, perfectly plastic material. Table 10.2 Janbu’s Lambda coefficient for various

However, Hoek and Brown (1981) concluded combinations of friction angle and cohesion

that the analysis assumptions were not

Friction angle Cohesion λ = γ H tan φ/c

applicable to rock slope design. (MPa)

Piteau and Jennings (1970) studied the influ-

ence of curvature in plan on the stability of slopes 45 0.22 59

in four diamond mines in South Africa. As a result 35 0.66 14

25 1.18 5

of caving from below the surface, slopes were all

15 1.8 2

at incipient failure with a safety factor

Numerical analysis 225

= 15 =

25

= 35

FS plane strain

curved

FS slopes

= 45

–10 –8 –6 –4 –2 0 2 4 6 8 10

Slope height/radius of curvature

Figure 10.1 Results of FLAC axisymmetric analyses showing effect on factor of safety of slope curvature.

the axis of revolution and the toe of the slope. beneficial effects of slope curvature should not

For the convex slopes, the radius of curvature be ignored—particularly in open pit mines,

is defined as the distance between the axis of where the economic benefits of steepening

revolu-tion and the crest of the slope—not the slopes can be significant. The same is true for

toe. Under both definitions, cones have a convex slopes, which are also more stable

radius of curvature of zero. than plane strain slopes. This goes against

Figure 10.1 shows the results with observed experience in rock slopes. If the slide

FS/FSplane strain versus height/radius of curvature surface is defined in terms of active (top) and

(H /Rc), which is positive for concave and negat- passive (bottom) wedges, the ratio of the

ive for convex slopes. The figure shows that the surface (and weight) of the passive wedge to

factor of safety always increases as the radius of the active wedge in a convex slope is greater

curvature decreases, but the increase is faster for than the plane strain condition. However, this

concave slopes. One unexpected result is that as only applies to a homo-geneous Mohr–

the friction angle increases, the effect of Coulomb material that might be found, for

curvature decreases. One possible explanation is example, in waste dumps. The reason why

that as Janbu’s lambda coefficient (λ = γ H tan “noses” in rock slopes are usu-ally less stable

φ/c) increases, the slide surface is shallower with may be related to the fact that they are more

only a skin for purely frictional material. This, exposed to structurally controlled failures.

makes the slope less sensitive to the confining

effect in concave slopes, and to the ratio of

10.3.2 Continuum versus

active/passive wedges for the convex ones.

discontinuum models

One reason that designers are reluctant to

take advantage of the beneficial effects of con- The next step is to decide whether to use a

cave slope curvature is that the presence of continuum code or a discontinuum code. This

discontinuities can often negate the effects. decision is seldom straightforward. There appear

However, for massive rock slopes, or slopes to be no ready-made rules for determining which

with relatively short joint trace lengths, the type of analysis to perform. All slope stability

226 Numerical analysis

another. However, useful analyses, particularly of

Initial conditions are those conditions that

global stability, have been made by assum-ing

existed prior to mining. The initial conditions of

that the rock mass can be represented as an

importance at mine sites are the in situ stress

equivalent continuum. Therefore, many analyses

field and the ground water conditions. The role

begin with continuum models. If the slope under

of stresses has been traditionally ignored in

consideration is unstable without structure, there

slope analyses. There are several possible

is no point in going to discontinuum models. If, on

reasons for this:

the other hand, a continuum model appears to be

reasonably stable, explicit incorporation of

principal structures should give a more accurate • Limit equilibrium analyses, which are widely

estimation of slope behavior. used for stability analyses, cannot include the

Selection of joint geometry for input to a model effect of stresses in their analyses. Neverthe-

is a crucial step in discontinuum analyses. Typic- less, limit equilibrium analyses are thought to

ally, only a very small percentage of joints can provide reasonable estimates of stability in

actually be included in a model in order to create many cases, particularly where structure is

models of reasonable size for practical analysis. absent, such as soil slopes.

Thus, the joint geometry data must be filtered to • Most stability analyses have traditionally

select only those joints that are most crit-ical to been performed for soils, where the range

the mechanical response. This is done by of pos-sible in situ stresses is more limited

identifying those that are most susceptible to slip than for rocks. Furthermore, many soil

and/or separation for the prescribed loading con- analyses have been performed for

dition. This may involve determining whether suf- constructed embankments such as dams,

ficient kinematic freedom is provided, especially where in situ stresses do not exist.

in the case of toppling, and calibrating the ana- • Most slope failures are gravity driven, and

lysis by comparing observed behavior to model the effects of in situ stress are thought to be

response. minimal.

• In situ stresses in rock masses are not

routinely measured for slopes, and their

10.3.3 Selecting appropriate zone size effects are largely unknown.

The next step in the process is to select an

appropriate zone size. The finite difference zones One particular advantage of stress analysis pro-

assume that the stresses and strains within each grams such as numerical models is their ability to

zone do not differ with position within the zone— include pre-mining initial stress states in stability

in other words, the zones are the lowest-order analyses and to evaluate their importance.

elements possible. In order to capture stress and In order to evaluate the effects of in situ

strain gradients within the slope adequately, it is stress state on stability, five cases were run

necessary to use relatively fine discretizations. using a simple model similar to the model

shown in Figure 10.2. For each of the five

By experience, the authors have found that at ◦

least 20 (and preferably 30) zones are required cases, the slope angle was 60 , the slope

over the slope height of interest. As discussed height was 400 m, the material density was

later, if flexural toppling is involved, a minimum of 3 ◦

2450 kg/m , the friction angle was 32 , and the

four zones across the rock column are required. cohesion was 0.92 MPa. The results of FLAC

Finite element programs using higher-order analyses are shown in Table 10.3.

elements likely would require less zones than the In general, it is impossible to say what effect

constant strain/constant stress elements common the initial stress state will have on any partic-

in finite difference codes. ular problem, as behavior depends on factors

Numerical analysis 227

Legend

1000

Step

–138.9 < x < 2639 45

–1264 < y < 1514 500

Grid plot

0 500

Water table 0

axis (m)

–500

tical

Ver

Friction angle = 35

Cohesion = 660 kPa

–1000

Figure 10.2 Problem geometry used to determine the effect of in situ stresses on slope stability.

Table 10.3 Effect of in situ stress on slope stability are not particularly important in slope

[x—horizontal in-plane direction; y—vertical in-plane studies.

direction; z—out-of-plane direction] • Initial horizontal stresses in the plane of ana-

In-plane Out-of-plane Factor of lysis that are less than the vertical stresses

horizontal stress horizontal stress safety tend to slightly decrease stability and reduce

σ σ σ σ the depth of significant shearing with respect

xx = yy zz = yy 1.30

to a hydrostatic stress state. This observa-tion

σxx = 2.0 σyy σzz = 2.0 σyy 1.30

σxx = 0.5 σyy σzz = 0.5 σyy 1.28 may seem counter-intuitive; smaller hori-

σxx = 2.0 σyy σzz = 0.5 σyy 1.30 zontal stresses would be expected to increase

σxx = 0.5 σyy σzz = 2.0 σyy 1.28 stability. The explanation lies in the fact that

the lower horizontal stresses actually provide

slightly decreased normal stress on potential

such as orientation of major structures, rock shearing surfaces and/or joints within the

mass strength and water conditions. However, slope. This observation was confirmed in a

some observations on the effects of in situ UDEC analysis of a slope in Peru where in

stress on stability can be made: situ horizontal stresses lower than the ver-tical

stress led to deeper levels of joint shear-ing in

• The larger the initial horizontal stresses, the toppling structures compared to cases

larger the horizontal elastic displacements. This involving horizontal stresses that were equal

is not much help, as elastic displacements to or greater than the vertical stress.

228 Numerical analysis

• It is important to note that the regional can be made regarding the displacement bound-

topography may limit the possible stress aries near the toe of any slope. One assumption

states, particularly at elevations above is that the displacements near the toe are

regional valley floors. Three-dimensional inhibited only in the horizontal direction. This is

models have been very useful in the past in the mech-anically correct condition for a problem

addressing some regional stress issues. that is perfectly symmetric with respect to the

plane or axis representing the toe boundary.

Strictly speak-ing, this condition only occurs in

10.3.5 Boundary conditions slopes of infinite length, which are modeled in

Boundaries are either real or artificial. Real two-dimensions assuming plane strain, or in

boundaries in slope stability problems corres-pond slopes that are axi-ally symmetric in which the pit

to the natural or excavated ground surface that is is a perfect cone. In reality, these conditions are

usually stress free. Artificial boundaries do not exist rarely satisfied. Therefore, some models are

in reality. All problems in geo-mechanics, including extended laterally to avoid the need to specify

slope stability problems, require that the infinite any boundary condi-tion at the toe of the slope. It

extent of a real prob-lem domain be artificially is important to note that difficulties with the

truncated to include only the immediate area of boundary condition near the slope toe are usually

interest. Figure 10.3 shows typical a result of the two-dimensional assumption. In

recommendations for locations of the artificial far- three-dimensional models, this difficulty generally

field boundaries in slope sta-bility problems. Artificial does not exist.

boundaries can be of two types: prescribed The far-field boundary location and condi-tion

displacement, or prescribed stress. Prescribed must also be specified in any numerical model for

displacement boundaries inhibit displacement in slope stability analyses. The general notion is to

either the vertical direction or horizontal direction, or select the far-field location so that it does not

both. Prescribed displace-ment boundaries are used significantly influence the results. If this criterion is

to represent the condi-tion at the base of the model met, whether the boundary is prescribed-

and toe of the slope. displacement or prescribed-stress is not important.

Displacement at the base of the model is always In most slope stability studies, a prescribed-

displacement boundary is used. The authors have

fixed in both the vertical and horizontal directions to

used a prescribed-stress bound-ary in a few cases

inhibit rotation of the model. Two assumptions

and found no significant differences with respect to

the results from a prescribed-displacement

boundary. The mag-nitude of the horizontal stress

W >W for the prescribed-stress boundary must match the

assumptions regarding the initial stress state in

order for the model to be in equilibrium. However,

following any change in the model, such as an

H excavation increment, the prescribed-stress

boundary causes the far-field boundary to displace

toward the excavation while maintaining its original

stress value. For this reason, a prescribed-stress

>H /2 bound-ary is also referred to as a “following” stress,

or constant stress boundary, because the stress

does not change and follows the displacement of

Figure 10.3 Typical recommendations for the boundary. However, following stresses are most

locations of artificial far-field boundaries in slope likely where slopes are cut into areas where the

stability analyses.

Numerical analysis 229

topography rises behind the slope. Even where concentrations near the toe of a slope, and slightly

slopes are excavated into an inclined over-predicts the pore pressure behind the toe by

topography, the stresses would flow around the ignoring the inclination of equipotential lines.

excavation to some extent, depending on the Seepage forces must also be considered in the

effective width of the excavation perpendicular analysis. The hydraulic gradient is the difference

to the downhill topographic direction. in water pressure that exists between two points

A summary of the effects of boundary condi- at the same elevation, and results from seepage

tions on analysis results is as follows: forces (or drag) as water moves through a porous

medium. Flow analysis automatically accounts for

• A fixed boundary causes both stresses and seepage forces.

displacements to be underestimated, whereas To evaluate the error resulting from specifying a

a stress boundary does the opposite. water table without doing a flow analysis, two

• The two types of boundary condition identical problems were run. In one case, a flow

“bracket” the true solution, so that it is pos- analysis was performed to determine the pore

sible to conduct tests with smaller models pressures. In the second case, the pressures were

to obtain a reasonable estimate of the true determined using only a piezometric surface that

solution by averaging the two results. was assumed to be the phreatic surface taken from

the flow analysis. The material properties and

A final point to be kept in mind is that all open pit geometry for both cases are shown in Figure 10.2.

slope stability problems are three-dimensional in The right-hand boundary was extended to allow the

reality. This means that the stresses acting in and far-field phreatic surface to coincide with the ground

around the pit are free to flow both beneath and surface at a horizontal distance of 2 km behind the

around the sides of the pit. Therefore, it is likely toe. Hydraulic conductivity within the model was

that, unless there are very low strength faults assumed to be homogeneous and iso-tropic. The

parallel to the analysis plane, a constant stress or error caused by specifying the water table can be

following stress boundary will over-predict the seen in Figure 10.4. The largest errors, under

stresses acting horizontally. prediction of up to 45%, are found just below the

toe, while over prediction errors in pore pressure

10.3.6 Incorporating water pressure

values behind the slope are gen-erally less than 5%.

The effect of water pressure in reducing effective The errors near the phreatic surface are

stresses and, hence, slope stability is well under- insignificant, as they result from the relatively small

stood. However, the effect of various assump- pore pressures just below the phreatic surface

tions regarding specification of pore pressure where small errors in small values result in large

distributions in slopes is not as well understood. relative errors.

Two methods are commonly used to specify pore For a phreatic surface at the ground surface at a

pressure distributions within slopes. The most distance of 2 km, a factor of safety of 1.1 is

rigorous method is to perform a complete flow predicted using circular failure chart number 3 (refer

analysis, and use the resultant pore pres-sures in to Section 8.3). The factor of safety determined by

the stability analyses. A less rigorous, but more FLAC was approximately 1.15 for both cases. The

common method is to specify a water table, and FLAC analyses give similar safety factors because

the resulting pore pressures are given by the the distribution of pore pressures in the area behind

product of the vertical depth below the water the slope where failure occurs is very similar for the

table, the water density and gravity. In this sense, two cases. The conclusion drawn here is that there

the water table approach is equivalent to is no significant differ-ence in predicted stability

specifying a piezometric surface. Both methods between a complete flow analysis and simply

use similar phreatic surfaces. However, the water specifying a piezometric surface. However, it is not

table method under-predicts actual pore pressure clear if this conclusion

230 Numerical analysis

Legend 800

Pore pressures

underpredicted by

Step 20000 water table in this

Cons. time 1.3380E + 11 area

400

Pore pressure error

–4.5

–3.5

–2.5

–1.5 0

–5.0

i m

x (

s )

a

Contour interval = 5.00E-02

(zero contour omitted)

V

e

a

c

r

t

i

l

–400

Pore pressures

overpredicted by

water table in this

area

–800

Figure 10.4 Error in pore pressure distribution caused by specifying water table compared to performing a

flow analysis.

for example, anisotropic flow. number of excavation stages has evolved over

the years. Using this approach, only one, two

10.3.7 Excavation sequence or three excavation stages are modeled. For

each stage, two calculation steps are taken. In

Simulating excavations in numerical models the first step, the model is run elastically to

poses no conceptual difficulties. However, the remove any iner-tial effects caused by sudden

amount of effort required to construct a model removal of a large amount of material. Second,

depends directly on the number of excavation the model is run allowing plastic behavior to

stages simulated. Therefore, most practical ana- develop. Following this approach, reasonable

lyses seek to reduce the number of excavation solutions to a large number of slope stability

stages. The most accurate solution is obtained problems have been obtained.

using the largest number of excavation steps,

because the real load path for any zone in the

10.3.8 Interpretation of results

slope will be followed closely. In theory, it is

impossible to prove that the final solution is inde- As noted in the introduction, finite difference

pendent of the load path followed. However, for programs are not black boxes that “give the

many slopes, stability seems to depend mostly solution.” The behavior of the numerical system

on slope conditions, such as geometry and pore and results from finite difference models must be

pressure distribution at the time of analysis, and interpreted in much the same way as slope

very little on the load path taken to get there. movement data are interpreted. Finite difference

Numerical analysis 231

programs record displacements and velocities at • Toppling failure—block and flexural; and

nominated points within the rock mass. During • Flexural buckling failure.

the analysis, the recorded values can be

examined to see if they are increasing, remaining The two-dimensional distinct element code,

steady, or decreasing. Increasing displacements UDEC, was used for most of the analyses, and

and velocit-ies indicate an unstable situation; the characteristics of the slopes that were

steady displace-ments and decreasing velocities modeled are as follows:

indicate a stable situation. In addition, velocity

and displacement vectors for every point in the Slope height 260 m

◦

model can be plot-ted. Fields of constant velocity Slope angle 55

and displacement indicate failure. Water pressure none/dry

3

The authors have found that velocities below Density 2660 kg/m

1e−6 indicate stability in FLAC and FLAC3D; 3

(26.1 kN/m )

conversely, velocities above 1e−5 indicate ◦

Rock mass friction angle 43

instability. Note that no units are given for velocit- Rock mass cohesion 675 kPa

ies. This is because the velocities are not real, due Rock mass tension 0

to the damping and mass scaling used to achieve Rock mass bulk modulus 6.3 GPa

static solutions. While the displacements are real, Rock mass shear modulus 3.6 GPa

the velocities are not, and there is no information on ◦

Joint friction angle 40

the “time” the displacement occurs.

It is also possible to examine the failure (plas- ◦

Dilation angle 0

ticity) state of points within the model, where A maximum zone size of 15 m was used

failure is defined as failure in tension or shear. except where noted. In all cases, the factor of

Care must be used in examining the failure state safety was estimated by simultaneously reducing

indicators. For example, local overstressing at the the rock mass properties and the joint friction

base of toppling columns can appear to form a until failure occurs using the procedure shown in

deep-seated slip surface when, in reality, it is just equa-tions (10.1) and (10.2). The safety factor

compressive failure of the columns. There-fore, was assumed to be the reciprocal of the

the failure (plasticity) indicators must be reviewed reduction required to produce failure. For

in the context of overall behavior before any example, if the strengths must be reduced by

definitive conclusions can be drawn. 25% (i.e. 75% of their “real” strength) in order to

achieve failure, the safety factor is 1.33.

10.4 Typical stability analysis

In this section, typical stability analyses for a vari- 10.4.1 Rock mass failure

ety of failure modes are discussed. The objective Numerical analysis of slopes involving purely

of this section is to show how numerical models rock mass failure is studied most efficiently using

can be used to simulate slope behavior, and com- 2

continuum codes such as PHASE , FLAC or

pute safety factors for typical problems. Modeling

FLAC3D. As mentioned in the previous section,

issues important in each of the following failure

discontinuities are not considered explicitly in

modes are discussed:

continuum models; rather, they are assumed to

be smeared throughout the rock mass. Assuming

• Rock mass failure; that rock mass shear strength properties can be

• Plane failure—daylighting estimated reasonably, the analysis is straightfor-

and non-daylighting; ward. The process for initially estimating the rock

• Wedge failure—daylighting mass properties is often based on empirical rela-

and non-daylighting; tions as described, for example, by Hoek and

232 Numerical analysis

500

Legend

m

(

)

Boundary plot 400

User defined grid value

a

x

s

i

Displacement contour

V

e

a

c

r

t

i

l

interval = 0.5 300

0.0

0.5

200

1.0

1.5

2.0

100

2.5

–100

0 100 200 300 400 500 600

Figure 10.5 Rock mass failure mode for slope determined with UDEC.

modified, as necessary, through the calibration Disturbance factor, D = 0.7

process.

Failure modes involve mainly shearing The tensile strength is estimated to be 0.012

through the rock mass. For homogeneous MPa. For the Bishop’s analysis method, the

slopes where the slide surface is often Mohr– Coulomb strength is estimated by fitting

approximately circular, intersecting the toe of a straight line to the curved Hoek–Brown

the slope and becom-ing nearly vertical near failure envelope at the normal stress level

the ground surface. The failure mode for the estimated from the slope geometry. Using this

parameters listed earlier is shown in Figure procedure, friction angle and cohesion were

10.5. The calculated safety factor is 1.64.

The following is a comparison of a slope ◦

stability analysis carried out using limit equilib- φ = 43

rium circular failure analysis (Bishop method) and c = 0.145 MPa

numerical stability analysis. In Chapter 8, the

stability of a benched slope in strong, but closely The mass density of the rock mass and water

fractured, sandstone including a water table and 3 3

were 2550 kg/m (25.0 kN/m ) and 1000

tension crack is described (see Figure 8.19). The 3 3

kg/m (9.81 kN/m ) respectively. The phreatic

rock mass is classified as a Hoek–Brown material

surface is located as shown in Figure 8.19.

with strength parameters:

Based upon these parameters, the Bishop

method produces a location for the circular

mi = 0.13 slide surface and tension crack, as shown in

Figure 8.19, and a factor of safety of 1.39.

GSI = 20

Numerical analysis 233

In using FLAC to analyze the stability of the FLAC solution includes the effect of stress

slope in Figure 8.19, the slide surface can evolve redis-tribution and progressive failure after

during the calculation in a way that is represent- movement has been initiated. In this problem,

ative of the natural evolution of the physical slide tensile failure continues up the slope as a

surface in the slope. It is not necessary to make result of the tensile softening. The resulting

an estimate for the location of the circular slide factor of safety allows for this weakening effect.

surface when beginning an analysis, as it is with

limit equilibrium methods. FLAC will find the slide

10.4.2 Plane failure—daylighting

surface and failure mechanism by simulat-ing the

and non-daylighting

material behavior directly. A reasonably fine grid

should be selected to ensure that the slide Failure modes that involve rigid blocks sliding

surface will be well defined as it develops. It is on planar joints that daylight in the slope face

best to use the finest grid possible when studying are most efficiently solved using analytical

prob-lems involving localized failure. Here, a zone methods. For comparison purposes, a UDEC

size of 2 m was used. ◦

analysis is per-formed for blocks dipping at 35

The FLAC analysis showed a factor of safety of out of the slope. The joints are assumed to

1.26, with the slide surface closely resembling have a cohesion of 100 kPa and a friction

that produced from the Bishop solution (Figure ◦

angle of 40 . The resulting safety factor is 1.32,

10.6). However, the tensile failure extends farther which agrees with the ana-lytic value given by

up the slope in the FLAC solution. It is important equation (6.4) in Chapter 6, assuming that no

to recognize that the limit equilibrium solution only tension crack forms. The plane failure mode in

identifies the onset of failure, whereas the the UDEC analysis is shown in Figure 10.7.

Legend 80

60

Boundary plot

Vertical axis (m)

40

0 20

Max. shear strain increment

0 20

0.004

0.008

0.012 0

0.016

0.020

0.024 –20

Contour interval = 0.002

– 40

– 40 –20 0 20 40 60 80 100

Figure 10.6 Failure mode and tension crack location determined with FLAC for slope in closely

fractured sandstone slope (refer to Figure 8.19).

234 Numerical analysis

500

Legend

Cycle 487840 400

Block plot

axis (m)

Velocity vectors

Maximum = 0.043 300

Vertic

0 0.2 200

al

100

–100

Figure 10.7 Plane failure mode with rigid blocks determined with UDEC.

If a tension crack does form, then the factor of analyses must be performed in three dimensions.

safety is slightly reduced. Deformable blocks with As with plane failure, sliding analysis of day-

elastic–plastic behavior are required to form lighting rigid blocks is best solved using analytic

tension cracks within the UDEC analysis. When methods, as described in Chapter 7. Analyses

deformable zones are used, the resultant safety involving formation of tension cracks and/or non-

factor is 1.27, similar to the value of 1.3 given by daylighting wedges require numerical analysis.

the analytic solution. The difference may be that Candidate codes include FLAC3D and 3DEC.

the analytic solution assumes a vertical tension The plasticity formulation in FLAC3D uses a

crack, whereas the UDEC analysis indicates that mixed discretization technique and presently

the tension crack curves where it meets the provides a better solution than 3DEC in cases

sliding plane (see Figure 10.8). where rock mass failure dominates. On the other

Similar analyses can be performed for non- hand, setting up problems involving more than

daylighting failure planes. In this case, failure one sliding plane in FLAC3D is more difficult and

involves sliding on discontinuities and shearing time consuming than similar problems in 3DEC.

through the rock mass at the toe of the slope, as

shown in Figure 10.9. Here, the cohesionless

◦

slid-ing planes dip at 70 and are spaced 20 m

apart. The resultant safety factor is about 1.5.

10.4.4 Toppling failure—block and flexural

Toppling failure modes involve rotation and thus

usually are difficult to solve using limit equilib-

10.4.3 Wedge failure—daylighting

rium methods. As the name implies, block top-

and non-daylighting

pling involves free rotation of individual blocks

Analyses involving wedge failures are similar to (Figure 9.3(a)), whereas flexural toppling involves

those involving plane failures, except that the bending of rock columns or plates (Figure 9.3(b)).

Numerical analysis 235

500

Legend

X displacement contours

Contour interval = 0.2

300

(zero contour line omitted)

0.2

0.4 200

0.6

0.8

1.0 100

1.2

1.4

1.6

1.8 0

Block plot

–100

Figure 10.8 Plane failure mode with deformable blocks determined with UDEC.

500

Legend

400

Cycle 1246860

axis (m)

Time 1.286E + 03 sec

300

Y displacement contours

Vertic

al

–0.6

100

–0.4

–0.2

0.2

0

Block plot

–100