Source: GEOTECHNICAL ENGINEERING

21 Slope Stability and Landslides

21.1 CUT SLOPES VS. BUILT-UP SLOPES

21.1.1 The Slippery Slope

Flat ground may be widespread, but sloping ground can be wider and deeper,
where the land has been incised by streams and rivers. If the slope is too high or
too steep, or if the soil is too weak, it will find ways to correct the deficiency.
A common corrective measure is the landslide.

Landslides go by many different names from rockfalls to mudlows, but the point
is that they all slide and they all involve shearing resistance. In terms of lives and
property lost, landslides are among the most tragic and devastating events that
involve soils. Landslides occur naturally as streams cut valleys deeper and wider,
and as lakes and oceans erode their shores. Landslides are triggered by shaking
from an earthquake, or by willful indiscretions of man.

The most common naturally occurring slopes are stream valley walls and eroding
shorelines, but slopes also occur as exposed sides of faults and uplifted mountain
ranges. Depositional slopes include sand dunes, deltas, and volcanic cinder cones.
The most common depositional slopes are man-made—road embankments, earth
dams, piles of sand, grain, or industrial or mining waste. Other man-made slopes
include cuts made by trenching and by excavations for foundations, basements, or
roadcuts. All slopes have one thing in common: if conditions are right, they have
it in them to host a landslide.

Geotechnical engineers frequently are called on to analyze the mechanics of

landslides and propose fixes. Engineering geologists also are closely involved with
landslides, but their emphasis is more on recognition and defining landslide
hazard areas. Landslide hazard maps of many critical areas now are available
from geological surveys and on the Internet.

590 Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 591

This chapter will discuss the soil mechanics of naturally occurring slopes and
landslides, followed by design considerations for cut slopes and embankments to
prevent slope failures.

21.1.2 When Push Exceeds Drag

Landslides occur when the downslope component of soil or rock weights exceeds
the maximum resistance to sliding along a particular surface. Landslides can be
shallow or deep. They differ from erosion by wind or running water because the
main driving force is gravity acting on the weight of the soil.

The best time to stop a landslide is before it starts, because shearing and
remolding of soil in the slip zone causes it to lose a substantial part of its shearing
strength. Therefore, after sliding starts it may speed up, depending on geometric
factors and how much strength has been lost. Sliding then continues until the
shearing resistance improves or the ground reaches a configuration that com-
pensates for the loss of soil strength. Sensitive soils move farther and much more
rapidly that those that are less sensitive, sometimes becoming devastating mud
flows that move so fast that they are virtually inescapable.

Normally for a landslide to stop it must move to a flatter surface. This often takes
place gradually, as the sliding soil moves along a concave surface that is nearly
vertical at the top and nearly horizontal at the bottom. Passive resistance of soil
building up at the bottom stops the slide. If that soil is removed, either naturally
or as a result of man’s activities, the slide continues.

One of the worst things that can be done to encourage a landslide to continue also
seems to be the most intuitive, which is to put the soil back where it belongs.
Curiously it sometimes takes two or even three attempts before owners are willing
to admit defeat. Putting the soil back not only removes passive resistance, it puts
weight where it will be most effective at renewing sliding.

21.1.3 Reaching for More

A landslide releases lateral pressure on soil at the top of the slope, so soil
that has not yet moved becomes less stable and more susceptible to sliding.
Thus, as shown in Fig. 21.1, a landslide area will enlarge headward by biting off
more ground until the geometry changes or the landslide runs out of soil.

The uphill sequence sometimes stops at a basement wall because soil

removed to make a basement usually weighs more than the structure. However,
the weight of the structure is carried by the foundation that is exposed by the
landslide, which can lead to a bearing capacity failure. Putting soil back will
reactivate the slide, but if necessary the structure often can be saved by
underpinning.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

592 Geotechnical Engineering

Figure 21.1
Causes of this landslide were excavation at the toe of the slope and doing too much laundry at
the top (waste water went into a septic drainfield). The slide enlarged uphill in steps as sliding
removed lateral support from soil exposed in the slide scarp. This landslide was stabilized with
drilled lime.

Man-made slopes obviously must be designed to be safe from landslides.

An active landslide automatically selects the most critical slip surface, but in a
stable slope that surface is not known. Because of the infinite number of possible
slip geometries the problem is statically indeterminate, so a slip surface either is
assumed or is arrived at by computational trial and error.

Active landslides are more readily analyzed because the slip surface can be
located by inspection and by drilling. The analysis can determine the causes, or
if more than one causal factor is involved, prorate responsibility to the several
contributors. Analysis also is required to compare the effectiveness of different
repair methods. Homeowner insurance policies almost inevitably have a waiver
in fine print that excludes damages from ground movements, placing them in
the category with nuclear wars and volcanic eruptions.

21.1.4 Infinite Slopes

The simplest example of an unstable slope is sand dumped off the end of a
conveyor belt, or any other grains dribbling down a surface that is appropriately
named the slip face. As discussed in Chapter 18, the material is deposited at the
‘‘angle of repose.’’

The forces involved in an angle of repose are shown in Fig. 21.2, where W is the
weight of an element. As shown in the figure, the weight may be resolved into N,

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 593

Figure 21.2
A simple case of
borderline slope
stability: sand at
the angle of
repose.

a force normal to the slip surface, and Sa, a force acting parallel with the surface
and tending to cause sliding. The acting force is

Sa ¼ W sin i ð21:1Þ

and the force normal to the shear surface is

N ¼ W cos i ð21:2Þ

From the Coulomb equation, if cohesion is zero, the resisting force is only
a function of N:

Sr ¼ N tan

ð21:3Þ
¼ W cos i tan

At the instant of sliding Sa ¼ Sr, or

W sin i ¼ W cos i tan

tan i ¼ tan

i¼
ð21:4Þ

The angle of repose of a cohesionless granular material, i, therefore

equals the angle of internal friction,
. Because
tends to increase
with an increasing thickness of overburden, shearing is concentrated near the
surface.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

594 Geotechnical Engineering

21.1.5 Factor of Safety for a Slope in Cohesionless Material

The factor of safety against sliding may be defined as the resisting force divided
by the acting force, or
Sr N tan
W cos i tan

FS ¼ ¼ ¼
Sa W sin i W sin i

tan

FS ¼ ð21:5Þ
tan i
If the factor of safety is 1.0, then i ¼
(eq. 21.4).

Example 21.1
A highway slope is to be cut through a stable sand dune with a factor of safety of 1.2
against slipping. The sand is predominately quartz having an angle of internal friction of
258. What is the design slope angle assuming no influence from water?
Answer: tan i ¼ tan 258/1.2 ¼ 0.389; i ¼ 218.

21.1.6 Seepage Force

As viscosity holds back the flow of water through soil pores, there is an equal and
opposite push exerted by the water on the soil. This was discussed in relation
to the formation of quicksand in Section 14.10, where the following formula is
derived:
S ¼ i w w ð14:44Þ
where S is the seepage force which, as in the case of a unit weight, is expressed in
force per unit volume of soil, iw is the hydraulic gradient, or head loss per unit
length that the water travels, and  w is the unit weight of water.

Equation (21.1) can be written for a unit volume of soil oriented in the direction
of seepage (Fig. 21.3), so the soil unit weight can be substituted for W. Because
the soil is submerged, the submerged unit weight is used to calculate the
acting force. Then
Soil: Sa ¼ ð
w Þsin i ð21:6Þ

Figure 21.3
Seepage force S
acting within a
unit volume of soil
having weight W.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 595

where Sa is parallel with the slope. With seepage parallel to the slope the hydraulic
gradient is
iW ¼
hW ¼ sin i ð21:7Þ
Seepage force S ¼  w sin i. Adding this to the soil acting force gives
Sa þ S ¼  sin i ð21:8Þ
The submerged unit weight also is used for the calculation of normal force:
N ¼ ð
w Þcos i ð21:9Þ
The factor of safety is
Sr N tan
ð
w Þcos i tan

FS ¼ ¼ ¼
Sa  sin i  sin i

ð
w Þ tan

FS ¼ ð21:10Þ
 tan i
If the factor of safety is 1.0,
ð
w Þ
tan i ¼ tan
ð21:11Þ

It will be noted that the acting force, in the denominator, depends on the total unit
weight of the soil, whereas the resisting force in the numerator depends on
(
 w), the submerged unit weight.

As an approximation  w ¼ 0.5. Making this substitution,

tan

FS  0:5 ð21:11aÞ
tan i
Seepage forces are real forces. They also can be analyzed from a consideration of
effective stresses, as discussed later in this chapter.

Example 21.2
Assume a worst moisture condition for the highway cut in the previous example and
determine the corresponding slope angle. Is this value appropriate for design?

Answer: 1.2 ¼ 0.5 tan 258/tan i; solving for i gives 118. This would be conservative because
fully saturated conditions are not likely to occur in a sand dune.

21.1.7 Creep Angle

Equation (21.11a) also may be used to estimate a creep angle where cohesion is a
viscosity function that approaches zero with a very slow shear rate. At incipient
creep with FS ¼ 1,
tan i ¼ 0:5 tan
ð21:12Þ

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

596 Geotechnical Engineering

Example 21.3
Estimate the creep angle for a saturated cohesive soil with
¼ 208.

Answer: tan i ¼ 0.5 tan 208; i ¼ 108, or approximately 1 vertical to 5.5 horizontal.

21.2 SLOPES IN COHESIVE SOILS

21.2.1 Complications from Cohesion

The infinite slope analysis does not apply when a cohesion variable is added
to the resisting force in eq. (21.3). Shearing no longer is along a shallow sur-
face parallel with the slope, but takes a path that minimizes the combined
influences of cohesion, which acts on the length of a slip surface, and internal
friction, which depends on the normal force on that surface. In most landslides
the failure path is curved and the problem is statically indeterminate, so solu-
tions are obtained through a series of approximations and trial and error. This
following discussion begins with a simple case of a planar slip that follows
along the path of a weak stratum or fault, and then proceeds to more complicated
geometries.

21.2.2 Slip Surface Fixed by an Inclined Weak Layer

A simple case that illustrates the method of analysis by separating soil weight
into acting and resisting force components is shown in Fig. 21.4, where the slip
surface has been fixed by a weak layer of soil inclined at an angle  with
horizontal. The acting force tending to cause slipping is the component of
soil weight that is parallel with the slip surface:
Sa ¼ W sin  ð21:13Þ

Figure 21.4
Sliding of unsaturated soil on an inclined weaker layer, where W is the weight of the sliding soil.
This was the case for a tragic failure at Vaiont Dam in northern Italy, where the slip surface was an
old fault plane.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 597

Thus, the steeper the  angle, the higher the acting force. Friction depends on the
weight component acting normal to the slip surface:

N ¼ W cos  ð21:14Þ

In this case the steeper the  angle, the lower the normal force and the lower the
frictional resistance to sliding. The Coulomb equation for maximum resisting
force, which is the same as eq. (18.6) but on a total force basis, is

Sr ¼ cL þ N tan
ð21:15Þ
where c is the soil cohesion, L is the length of the slip surface, and
is the friction
angle. Hence

Sr ¼ cL þ W cos  tan
ð21:16Þ

If Sr exceeds Ss and therefore is not fully mobilized, the factor of safety is

Sr cL þ W cos  tan

FS ¼ ¼ ð21:17Þ
Sr W sin 

If Sr is fully mobilized and equals Sa, the factor of safety is 1.0 and the soil mass
will slide. In this form eq. (21.17) does not include pore pressure, which normally
plays a critical role in landslides, so the equation is applicable only for
unsaturated conditions.

Example 21.4
The size of the triangular mass of soil in Fig. 21.4 is scaled from the drawing: L ¼ 35 ft
(10.7 m) and  ¼ 248. The height of the triangle normal to L is h ¼ 7 ft (2.13 m). Strength
parameters for the inclined weak layer are determined to be approximately c ¼ 50 lb/ft2
(2.4 kPa) and
¼ 208. The unit weight is  ¼ 100 lb/ft3 (15.7 kN/m3). What is the factor of
safety against sliding?

Answer: The area of the triangle is

A ¼ ½ð35Þð7Þ ¼ 122:5 ft2 , or ½ð10:7Þð2:13Þ ¼ 11:4 m2 :

The soil weight is

W ¼ ð122:5 ft2 Þð100 lb=ft3 Þ ¼ 12,250 lb=ft of length,

or ð11:4 m2 Þð15:7 kN=m3 Þ ¼ 179 kN=m:

Substituting the appropriate values into eq. (21.17) gives

50ð35Þ þ 12,250 cos 24 tan 20 1750 þ 4073

FS ¼ ¼ ¼ 1:2, or
12,250 sin 24 4862
2:4ð10:7Þ þ 179 cos 24 tan 20 25:7 þ 59:5
¼ ¼ ¼ 1:2
179 sin 24 72:8

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

598 Geotechnical Engineering

Question: Which is the more important component of the resisting force, friction
or cohesion?

21.2.3 Planar Slip in Homogeneous Soil

Steep slopes in homogeneous soils may fail along a slip surface that is approx-
imately planar. The analytical solution usually is attributed to Culmann, who
published it in 1866, but it appears to have been first developed by Française
in 1820 (see Geotechnique 1(1), 66, 1948). In shallower slopes the restricted
geometry of the Culmann method causes it to overestimate the factor of safety
and is on the unsafe side for design.

The notation in Fig. 21.5 is the same as in Fig. 21.4, and eqs. (21.13) to (21.17)
still apply, except that  is not known. It therefore is a simple matter to apply
eq. (21.17) to a series of trial values of , plot the factor of safety versus the
trial value, and pick off the most critical value. An analytical solution is more
precise:

From trigonometric considerations the soil weight is

W ¼ ½HL csc i sin ði
Þ ð21:18Þ

Equation (21.18) may be rewritten in terms of a developed cohesion, cd,

and a developed friction angle, d, for which the factor of safety would be 1.0.
Then
W sin  ¼ cd L þ W cos  tan
d ð21:19Þ

Substituting the value of W from eq. (21.18) into eq. (21.19) and solving for cd
gives

cd ¼ ½H csc i sin ði
Þðsin 
cos  tan
d Þ

An additional substitution is

tan
d ¼ sin
d = cos
d

Figure 21.5
Culmann’s
method to define
the most critical
slip angle for
planar failure in
homogeneous
soil.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 599

By use of a trigonometric identity, sin ð

d Þ ¼ sin  cos
d
cos  sin
d

cd ¼ ½½H csc i sin ði
Þ sin ð

d Þ= cos
d ð21:20Þ

Setting the derivative of cd relative to  equal to zero establishes the most critical
value of , which is
c ¼ ½ ð i þ
d Þ ð21:21Þ
That is, the most critical failure angle for a planar failure is the average of the slope
angle and the angle of internal friction. The steeper the slope, the higher the failure
angle. This applies only to a planar failure surface.

Example 21.5
According to eq. (21.21), what is the most critical failure angle for a vertical slope? How
does this compare with the angle for Rankine active failure?

Answer: c ¼ ð90 þ
d Þ ¼ 45 þ
d =2, which is identical to the Rankine active case that
assumes parallel planar failures.

21.2.4 Critical Height of a Vertical Cut

Vertical cuts (Fig. 21.6), have a considerable practical significance as they are
common during construction, for example for utility trenches and basements.
Unsupported cuts still contribute to construction fatalities from trench cave-ins
despite many precautions and regulations. It is critically important that contractors
and construction workers be aware of the dangers and how to avoid them.

Figure 21.6
Three versions of
critical height for a
vertical cut. When
the cliff lets go,
better toss the
umbrella and get
out of there.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

600 Geotechnical Engineering

One method for predicting the maximum height of a vertical cut can be obtained
by substituting eq. (21.21) into eq. (21.18) and setting the factor of safety equal to
1.0, but it must be emphasized that this is not an appropriate model, as discussed
later in this section. This substitution gives
4c sin i cos

Hcul ¼
 ½1
cos ði

Þ
where Hcul is the Culmann critical height. Setting i ¼ 908 gives
4c cos

Hcul ¼
 1
sin


 ð21:22

mathematically correct but unsafe!Þ
4c

Hcul ¼ tan 45

 2

Equation (21.22) is unsafe for a vertical cut because it does not address the most
likely failure mechanism.

Terzaghi’s Correction
As shown in Fig. 21.6, Terzaghi (1943) indicated that the Culmann Hcul is unsafe
because vertical tension cracks running parallel to the faces of an open vertical cut
reduce the length and the amount of restraint along the slip plane. He therefore
reduced the critical height by one-third. While that is a substantial improvement,
it is empirical and inadequate. An analytical solution should be more accurate
if the model is successfully drawn.

Modification for Unconfined Compressive Strength

An earlier edition of this book pointed out that, because of the lack of lateral
confinement, soil at the bottom of a vertical cut should yield if the overburden
pressure exceeds the soil unconfined compressive strength, qu. As shown at the
bottom of Fig. 21.6, the tendency for a block of soil to settle and rotate will create
a mechanism whereby tension cracks will develop, as observed by Terzaghi.

Let us therefore redefine the critical height, Hc, in terms of the unconfined
compressive strength, qu. At depth Hc in the face of a vertical cut,

qu ¼ Hc 
Hc ¼ qu = ð21:23Þ

The equation for unconfined compressive strength is

qu ¼ 2c tan ð45

=2Þ ð18:13Þ
which gives
2c
Hc ¼ tan ð45

=2Þ ð21:24

Safer but still requires a factor of safetyÞ


Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 601

It will be noted that this is the critical height at failure and does not include a
factor of safety, yet is only one-half of that obtained with eq. (21.22), emphasizing
the importance of using an appropriate model for an analytical treatment.

The analysis based on unconfined compressive strength incorporates the concept

of progressive failure, since shear failure is not assumed to be simultaneous all
along a failure surface as in the case of the Culmann method, but starts with
compression failure at the bottom that creates a bending moment that must be
resisted by tension, which is low or nonexistent in soil. Unconfined compressive
strength tests are routinely included in geotechnical reports.

Equation (21.24) still can be on the unsafe side if a trench cuts below a groundwater
table because seepage forces will be directed toward the the open cut. This situation
also is unsatisfactory from a construction standpoint. Restrictive regulations now
require that workers in a trench deeper than a specific height must be protected by a
steel ‘‘trench box,’’ or the sides of the trench must be cut back to a stable angle.

Example 21.6
A trench is to be excavated in soil that is reported to have an unconfined compressive
strength of 500 lb/ft2 (23.9 kN/m2) and a unit weight of 120 lb/ft3 (18.8 kN/m3) What is Hc
with a factor of safety of 1.2?

Answer: According to the qu criterion Hc ¼ 500/120 ¼ 4.2 ft, or 23.9/18.8 ¼ 1.27 m. The
allowable height is Ha ¼ 4.2  1.2 ¼ 3.5 ft (1.1 m).

21.3 LANDSLIDES

21.3.1 A Slip Is Showing

A landslide does a credible job of selecting the most critical slip surface and
establishing that, at the instant of sliding, the factor of safety was very close to 1.0.
The factor of safety also is 1.0 when the landslide stops. It is a serious mistake to
ignore an existing slip surface and impose some idealized shape such as a circle for
the analysis, which can lead to some extremely misleading results. In one case
a self-styled expert reported that the factor of safety was 0.2, which would provide
a surplus of energy sufficient to carry the slide up and over the next hill.

21.3.2 Field Investigation

The first step in analysis of a landslide is a field investigation that involves (1) an
elevation survey, usually along a traverse down the central axis of the landslide,
and (2) exploration borings at selected points along that traverse. In natural
landslides more than one soil type usually is involved, and it is important to
identify weak strata encountered in borings that may reveal the location of the slip
zone, and impermeable strata that can contribute to a perched groundwater table.
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

602 Geotechnical Engineering

Particularly important is (3) to measure the position of the groundwater table in

each boring, and (4) to extend borings deep enough that they encounter and
penetrate through the slip zone.

An active landslide should be checked for sounds of running water and for any
other evidence of broken water or sewer pipes indicated by welling-up of the
groundwater table and local wet spots. Nearby residents should be asked if they
have had any problems with sewer backups. Leakage is stopped by shutting the
water off.

Landslides often are spoon-shaped with the long dimension of the spoon oriented
across the slope. A three-dimensional analysis is possible but adds complexity to
an already complicated problem. If the landslide is wide, two or more traverses
may be investigated.

Borings, cone probings, and/or in-situ shear tests, discussed in Chapter 26, are
made at selected locations along a traverse. Difficult terrain requires the use of
a track vehicle. Shallow landslides may most conveniently be investigated by
hand augering. Undisturbed soil samples also may be taken for laboratory tests,
but it may be difficult or impossible to obtain duplicate samples, in which case
tests can be performed in situ.

Ground elevations and soil boring information are plotted to scale to develop
a cross-section, as shown in Fig. 21.7. There should be no vertical exaggeration
because slip angles may be measured from the graph. The cross-section should
show soil strata, the slip surface, and the position of the groundwater table.
Average unit weight of the sliding mass of soil is needed, and cohesions and
friction angles of the slip zone soil are required for an analysis.

Landslides generally are analyzed on an effective stress basis. Pore water pressures
are determined from groundwater levels in the borings with the aid of flow nets.
Pore pressures also can be monitored in the field with piezometers, although
active sliding can destroy a piezometer. Anomalies in groundwater elevations
can be critically important clues to broken sewers or water pipes, or to aquifers
and artesian effects, all of which can be causal factors.

The factor of safety is 1.0 when a landslide starts, and immediately decreases
as soil in the slip zone is remolded and loses strength. By the time a field inves-
tigation is performed the landslide probably will temporarily have stopped so the
factor of safety still is 1.0, or it may be slightly higher if the pause has allowed
water to drain out.

The analysis will provide valuable benchmark values that will enable predictions
of the effectiveness of different repair methods. The design factor of safety will
be influenced by drainage conditions. The worst-case scenario will involve full
saturation of the sliding mass.
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 603

Figure 21.7
Landslide cross-section divided into arbitrary vertical slices for analysis. The factors of safety for
individual slices shown at the bottom define active and passive parts of the landslide—the part doing
the pushing and the part being pushed. This is an important consideration for stabilization.
The analysis did not include vertical shearing stress, which adds to the factor of safety where the
slip surface curves.

21.3.3 From Head to Toe

The landslide in Fig. 21.7 is typical in that the soil mass has moved down onto
a more nearly level surface so soil in the toe area can buttress against further
sliding. An unusual feature that should be noted in a field investigation is the river
aggressively removing soil from the toe and thereby acting to promote further
sliding.

The exposed part of a landslide slip surface at the top of a landslide is the
landslide ‘‘scarp.’’ The field investigation should include a walk along and above
the scarp to examine for tension cracks running parallel to the scarp, as these
normally will delineate the next part of the hill to go. The situation is aggravated
by surface water running down into the cracks, which therefore should be sealed
by cutting in a soil wedge with a spade.

21.3.4 Progressive Failure

Obviously the most favorable time to stop a landslide is before it starts.
Most landslides involve progressive failure, like tearing a sheet of paper out of
a notebook, so there are some worrisome clues such as development of a tension
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

604 Geotechnical Engineering

crack prior to failure. Another evidence for progressive failure is that back-
calculation of the resisting force favors the residual, remolded shear strength
instead of the peak shear strength (Duncan and Wright, 2005). Thus, as one part
of the slide moves sufficiently to reduce the soil strength to a remolded state,
additional stress is transferred to the remaining areas until the slip zone becomes
continuous and the entire mass slides. The strength reduction also is confirmed
by Borehole Shear tests that inevitably show lower strength in the shear
zone compared with the same soil outside of the zone. Conventional analysis
methods nevertheless assume simultaneous shear all along a slip surface, and
progressive failure is taken into account by using the residual shear strength in
calculations. If residual strength data are not available, cohesion may be assumed
to be zero.

21.3.5 Development of a Shear Zone

As sliding continues, the slip zone gradually incorporates more soil until it
constitutes a shear zone. One theory to explain this is that the first slipping
involves the weakest set of grain contacts, so when movement occurs the contacts
attain more nearly average strength so that slipping stops there and shifts to the
next weakest set of contacts. This has been successfully simulated in laboratory
models (Logani, 1973).

The development of a remolded zone at the base of a landslide has a parallel

in grinding of rocks in a fault zone, where it has been found that the thickness
of ground rock, or gouge, is between 0.1 and 10 percent of the distance
moved, averaging about 1 percent (Scholtz, 1967). In a landslide, substantial
additional thickening occurs in the toe area if the slip zone soil is close to its liquid
limit, so higher overburden pressures upslope and confinement by overlying soil
cause it to squeeze out at the bottom.

Example 21.7
What average thickness of slip zone can be expected in a landslide that has moved
30 ft (10 m)?

Answer: 0.3 ft ¼ 4 in. (0.1 m ¼ 100 mm), not enough for replicate samples. Adequate for a
Borehole Shear test.

21.3.6 After Sliding Stops

Nature’s remedy for a landslide is to build up enough passive resistance from soil
at the toe that the landslide stops. This of course can be overruled by nature or by
man, by a river or by a bulldozer. If the landslide is left alone, in time it will
partly repair itself through drainage and redevelopment of cohesive strength
of soil in the shear zone. However, drainage can only return the slip zone soil
to a normally consolidated state that is permanently weaker than when it was

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 605

overconsolidated prior to sliding. An old landslide therefore remains a silent foe

hiding under the trees and ready to harvest new victims. Often all that is necessary
to reinvigorate an old landslide is to put a structure such as a house or apartment
building on top. This was the case in Fig. 1.3 where affected structures were a total
loss.

21.3.7 Earthquakes and Liquefaction

An earthquake shakes the ground back and forth like noodles in a frying pan.
The movement is a vivid demonstration of Newton’s Second Law, f ¼ ma, where
f is force, m is mass, and a is acceleration provided by the earthquake. The inertia
of the soil mass contributes to horizontal body forces that are in addition to
horizontal components of gravitational and seepage forces, and therefore can
trigger many landslides, particularly in marginally stable ground.

Soil liquefaction is a highly destructive phenomenon when shaking of a satu-

rated, low-density granular soil breaks down its structure so that it collapses in on
the soil water. Intergranular contact stress is transferred to pore water pressure
so the soil becomes a dense liquid the same as quicksand. After a granular soil
has liquefied and shaking stops, soil grains suspended in water settle out so
excess pore water rises to the ground surface, carrying along sand to make small
‘‘sand volcanoes.’’ The existence of ejected sand in layered sediments allows
dating of earlier earthquakes in order to determine their periodicity. The influence
of earthquakes and liquefaction is discussed in more detail in Chapter 27;
see also Seed (1968).

21.3.8 Emergency Measures

Landslides can require immediate action to save properties from complete
destruction:

1. If a structure is in imminent danger, local authorities must be informed

and endangered structures evacuated.
2. Utility companies must be informed so that they can disconnect services
or install flexible loops in the connections.
3. Lawn watering must be stopped on and above the slide area.
4. Transverse tension cracks in the soil are plugged with soil to prevent entry
of surface runoff water.
5. Surface runoff water is diverted away from the slide area with trenches,
and downspouts are extended away from the slide area.
6. Backward tilting of the individual segments ponds water that infiltrates
into the soil, so channels should be cut to drain the water.
7. As a temporary emergency measure a slope may be given a plastic raincoat
that must be securely staked down to prevent blowing.
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

606 Geotechnical Engineering

8. Consider moving structures to a different site before it is too late and they
become involved in the landslide.

More permanent repair measures then can be applied, such as removing load
from the top, adding load at the toe, installing drains or structural measures,
or chemical stabilization in responsive soils.

21.4 SLOPES WHERE THE MOST CRITICAL SLIP SURFACE

IS NOT KNOWN

21.4.1 Slipping into the Modern Era

In 1918 a commission was established in Sweden to investigate a series of soil
failures that had resulted in a number of deaths. The upper limit of each landslide
was defined by an exposed near-vertical surface, the slide scarp, and downslope
the slip surface was found to approximate a circle. The ‘‘Swedish method’’ or
‘‘Ordinary Method of Slices’’ was devised by Fellenius to find the most critical
circle and predict the stability of existing slopes.

Because the soil mass is assumed to rotate about a common center, solution
is based on moment equilibrium—that is, the sum of the acting forces along the
failure arc times the radius to the center equals the sum of the resisting forces
along the same arc times the same radius. Since the radius is constant, in polar
coordinates this is the equivalent of a force equilibrium of tangential forces.
Moment equilibrium still remains the basis for modern computer solutions,
including those for which the slip surface is not a circle and the radius is not
constant. However, there also are limitations to a moment equilibrium. For
example, it cannot be used when part of the slip surface is flat because the radius
is infinite. Then a force equilibrium is used, as in the preceding discussions.

21.4.2 Adaptation of the Ordinary Method of Slices to

Noncircular Surfaces
Although the ordinary method was devised for a circular slip, the method of
analysis also can be used with noncircular surfaces using a force equilibrium on
orthogonal xy axes instead of a moment equilibrium. In this text this will be
referred to as the ‘‘simple method of slices.’’

The first step is to draw a cross-section of a slope (or landslide) to scale, and
include the actual slip surface or, if the slip surface is not known, a trial surface.
The mass of soil above the slip surface is then divided into vertical slices as in
Fig. 21.7, and forces on each slice are calculated using the same principles
and formulas as previously presented for a triangular mass of soil slipping along

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 607

a planar surface. Analysis involves summing resisting and acting forces on all
slices and dividing to obtain a factor of safety.

The relationships in eqs. (21.13) to (21.17) may be used to define acting and
resisting forces along the base of each slice. A factor of safety then may be
calculated that does not include side forces and can define active and passive
zones, as in Fig. 21.7. Horizontal side forces are required for equilibrium, and it is
these forces and resulting friction between slices that are ignored in the simplified
method of slices. The result is to underpredict the overall factor of safety,
particularly for deep slip surfaces and long slices.

21.4.3 Pore Water Pressure

Saturation and the resulting pore water pressure often become critical factors
affecting slope stability. Pore water pressures can be measured in the field with
piezometers, and can be estimated from a flow net. A flow net then can be
modified to predict the influence of drainage on stability.

Whereas in static equilibrium the pore water pressure is simply the depth below a
groundwater table times the unit weight of water, in a dynamic situation involving
seepage, the resistance to seepage causes a reduction in pore pressure. In Fig. 21.8
this is indicated by sloping equipotential lines. Therefore the pore water pressure
at point 8 on the slip surface does not equal the depth below the groundwater
table, but a depth Hw below the intersection of the equipotential line with the
groundwater table.

21.4.4 Calculation Procedure for Each Slice

The following steps are required to evaluate the stability of each slice and
of all slices, and may be used as a model for a spreadsheet. Slices are vertical
(Figs. 21.7 and 21.9). The same procedure is used whether the location of the

Figure 21.8
Flow net in a
landslide. Upward
seepage force in
the toe area further
destabilizes that
area, which may
be too weak to
support the weight
of a person.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

608 Geotechnical Engineering

Figure 21.9
Dimensions and
forces in the
simplified method
of slices.

slip surface is known or is represented by a trial surface. Variables are identified in

Fig. 21.9. Distances and slip angles are measured from a plotted cross-section,
and soil properties are determined from density and shear strength tests.

Soil Data:
1. Slice number.
2.  ¼ soil unit weight.
3.
0 ¼ soil friction angle measured on an effective stress basis.
4. c0 ¼ soil cohesion measured on an effective stress basis.

Slice Data (Measured on the Cross-Section):

5. x ¼ slice width.
6. z ¼ slice depth.
7.  ¼ slip angle from horizontal.
8. hw ¼ pore pressure in units of water head.

Calculations:
9. Weight W ¼ xz.
10. Acting force Sa ¼ W sin .

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 609

11. Normal force N ¼ W cos .

12. Length L ¼ x cos .
13. Pore water pressure u ¼ hw w, where  w ¼ unit weight of water.
14. Effective normal force N0 ¼ N – uL.
15. Resisting force Sr ¼ c0 L þ N0 tan
0 .
16. Slice factor of safety FSi ¼ Sr/Sa for the individual slice.

Summations: P
17. Sum of acting forces is Sa.
P
18. Sum of resisting forces is Sr.
P P
19. Global factor of safety ¼ FSt ¼ Sr/ Sa for the slope.

Notes:
Each of the steps can be set up as a column in a spreadsheet, and appro-
priate numbers and formulas copied vertically to enable a solution for each
slice.

In Step 2 the soil unit weight is the total unit weight including water.

In Steps 3 and 4 the soil cohesion and internal friction are on an effective
stress basis, which means that they either are measured with the soil in a fully
drained condition, or stresses have been corrected by subtracting pore water
pressures measured in a test specimen at the moment of failure.

In Step 8 the pore water pressure is measured from the equipotential line as
shown in Fig. 21.9, which is not the same as the depth below the phreatic
line because of seepage restraints of the water.

In Step 14 pore water pressure acts across the bottom of each slice, so it
must be multiplied by length L to give the force acting in opposition to N.

Steps 1–17 are calculated for each slice, and Steps 18–20 are summations that
are performed only once per analysis.

After a spreadsheet has been set up it is a simple matter to change selected

variables and instantly determine the effects on the factor of safety. For example,
hw values can be decreased by installing drains, or shearing strengths
can be increased by incorporating aggregate columns or by chemical stabili-
zation. Slices can be changed by adding soil to increase passive resistance
at the bottom of the landslide, or tension members can be introduced to
transfer part of the acting forces by anchoring into stable soil outside of the
shear surface.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

610 Geotechnical Engineering

Example 21.8
Sum forces and calculate a factor of safety for the following slice:
1. Slice number ¼ 5.
2.  ¼ soil unit weight ¼ 125 lb/ft3 or 19.6 kN/m3.
3.
0 ¼ 208.
4. c0 ¼ 500 lb/ft2 or 23.9 kN/m2.
5. X ¼ 10 ft or 3.05 m.
6. Z ¼ 16 ft or 4.88 m.
7.  ¼ 328.
8. hw ¼ 8 ft or 2.44 m.

Calculations:
9. W ¼ XZ ¼ 10(16)(125) ¼ 20,000 lb/ft length, or 3.05(4.88)(19.6) ¼ 292 kN/m.
10. Sa ¼ W sin  ¼ 20,000 sin 328 ¼ 10,600 lb/ft, or 292 sin 328 ¼ 155 kN/m.
11. N ¼ W cos  ¼ 20,000 cos 328 ¼ 16,960 lb/ft, or 292 cos 328 ¼ 248 kN/m.
12. L ¼ X sec  ¼ 10 sec 328 ¼ 11.8 ft, or 3.05 sec 328 ¼ 3.60 m.
13. N0 ¼ N
hw w L ¼ 16,960
8ð62:4Þð11:8Þ ¼ 16,960
5891 ¼ 11,070 lb=ft, or
¼ 248
2:44ð9:81Þð3:60Þ ¼ 162 kN=m:
14. Sr ¼ c0 LþN0 tan
0 ¼ ð500Þð11:8Þ þ 11, 070 tan 20 ¼ 5900 þ 4029
¼ 9930 lb=ft, or ¼ ð23:9Þð3:60Þ þ 162 tan 20 ¼ 86 þ 59 ¼ 145 kN=m:
15. FSi ¼ 9930/10,600 ¼ 0.94, or 145/155 ¼ 0.94.

Question: Is this slice in the active or the passive zone?

21.4.5 Additional Horizontal Forces

Horizontal acting and resisting forces can be added to each slide to account for
reinforcing members such as soil nails, or inertial forces developed from
earthquakes. The latter are calculated from Newton’s Law, f ¼ ma, where f is
force, m is mass, and a is maximum acceleration from the earthquake, which is
discussed in more detail in Chapter 27. The mass of each slice is obtained
from the same formula, m ¼ W/g, where W is the weight and g is the acceleration
of gravity.

21.4.6 Seepage Force and Effective Stress

Seepage forces are automatically incorporated into an analysis made on an
effective stress basis. This can be demonstrated from a consideration of forces on
a single slice as shown in Fig. 21.10. Assume a unit volume of fully saturated
cohesionless soil with weight W ¼ . Then

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 611

Figure 21.10
Diagram
to illustrate
equivalence of
seepage force and
effective stress
analysis.

Sa ¼  sin 
N ¼  cos 
The normal force on an effective stress basis is

N0 ¼ N – U, where U ¼ uL, where L ¼ 1/cos 

From the equipotential line u ¼  w hw, where hw ¼ ab cos  and ab ¼ cos . Then
 
U ¼ w cos2 = cos  ¼ w cos 
N0 ¼  cos 
w cos  ¼ ð
w Þ cos 
Sr ¼ N0 tan
0 ¼ ð
w Þ cos  tan
0

ð
w Þ cos  tan
0 ð
w Þ tan
0

FS ¼ ¼ ð21:10Þ
 sin   tan 

which is the equation that was developed on the basis of seepage forces, with
the notation on
0 indicating an effective stress basis. This illustration does not
include soil cohesion as cohesion is not affected by seepage forces or buoyancy.

Generally the effective stress method is preferred for spreadsheet or computer

programmed solutions as being easier to manipulate. Where seepage is upward, as
at the toe of the slope in Fig. 21.8, the hydraulic gradient is negative and the slope
angle is negative, so seepage still adds to the acting force. Seepage force remains
an important consideration that should be separately evaluated and controlled in
critical areas such as at the toe of a levee or earth dam, where vertical seepage can
cause quicksand (cf. Chapter 14) or excessive seepage can cut a path through the
dam and lead to eventual failure through piping.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

612 Geotechnical Engineering

21.4.7 Solution for a Worst-Case Scenario

A worst-case saturation condition with full mobilization of seepage forces is not
unusual in an active landslide. A simple way to estimate this condition is to use the
submerged soil unit weight to calculate resisting force and the total unit weight for
acting force, with soil shear strength parameters measured on an effective stress
basis. This errs on the safe side because it assumes a static condition for the pore
water, such that the pore pressure is a function of depth below the ground surface
instead of being determined from a flow net. For best accuracy the saturated
and buoyant unit weights can be calculated using a block diagram, as discussed
in Section 9.2.

Example 21.9
Use the buoyant unit weight to estimate a fully saturated factor of safety for the slice in the
preceding example, using 125 lb/ft3 (19.6 kN/m3) as the saturated unit weight.

Answer:
Sa ¼ W sin ¼ 10,600 lb=ft or 155 kN=m ðunchangedÞ
W 0 ¼ XZ 0 ¼ ð10Þð16Þð125
62:4Þ ¼ 10,020 lb=ft ðbuoyant weightÞ or
¼ ð3:05Þð4:89Þð19:6
9:81Þ ¼ 146 kN=ft
N0 ¼ 10,020 cos 32 ¼ 8490 lb=ft or 146 cos 32 ¼ 124 kN=m ðbased on buoyant
weightÞ
cL ¼ 5900 lb=ft or 86 kN=m ðunchangedÞ
Sr ¼ 5900 þ 8490 tan 20 ¼ 8990 lb=ft or 89 þ 124 tan 20 ¼ 134 kN=m
FSi ¼ 8990=10,600 ¼ 0:85 or 134=155 ¼ 0:86

compared with the previous answer of 0.94 for a partially saturated condition.

Example 21.10
Calculate a fully saturated factor of safety for the slice in the preceding example using
effective stresses.

Answer:
Sa ¼ 10,600 lb=ft or 155 kN=m ðunchangedÞ
hw ¼ 16cos2 32 ¼ 11:5 ft or 4:89 cos 232 ¼ 3:52 m
u¼ 11:5ð62:4Þ ¼ 718 lb=ft2 or 3:52ð9:81Þ ¼ 34:5 kN=m2
N0 ¼N
uL¼ 16,960
ð718Þð11:8Þ ¼ 8187 lb=ft or
¼ 248
ð34:5Þð3:60Þ ¼ 124 kN=m
Sr ¼ c0 LþN0 tan
0 ¼ ð500Þð11:8Þ þ 8187 tan 20 ¼ 5900 þ 3089 ¼ 8989 lb=ft or
¼ ð23:9Þð3:60Þ þ 124 tan 20 ¼ 86:0 þ 45:1 ¼ 131 kN=m
FSi ¼ 8989=10,600 ¼ 0:85 or 131=155 ¼ 0:85

In this case the solution involving submerged unit weight is simpler because of possible
ambiguity in evaluating hw.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 613

21.5 SOLUTIONS FOR CIRCULAR AND

LOG SPIRAL SHEAR FAILURES

21.5.1 Stability Numbers and Friction Circles

Dimensionless numbers such as the Reynolds number are used to reduce the
number of experimental variables. A dimensionless number called the stability
number can be defined for slopes:
c
s¼ ð21:25Þ
H
where s is Taylor’s stability number, c is the soil cohesion,  the soil unit weight,
and H is the height of the slope from the base to the crest. Solving for H gives
c
H¼ ð21:26Þ
s
Which shows that the higher the stability number, the lower the height at failure.
It is instructive to compare results from various methods for a homogeneous
soil without any influence from a groundwater table. In Table 21.1, s is the Taylor
stability number. Equation (21.26) may be compared with eq. (21.24) for
progressive failure of a vertical cut, in which case it may be shown that for this
special case s ¼ 0.5(tan 458

/2), which is included in the table. Also included in
the table are results from a graphical solution called a friction circle, which
has largely been replaced by computer programs employing methods of slices.
From these data a planar slip is not the most critical surface and should not be

Table 21.1
Taylor stability numbers for various failure surfaces for soil having a friction angle
¼ 258. The higher
the number, the lower the indicated stability. Answers that are in close agreement are bracketed—but
they still only apply to simultaneous failure along the entire slip surface
Slope i, degrees Surface Method s , degrees

90 (vertical) Plane Culmann 0.159 57.5

" Circle Ord. slices 0.165 # Variable
Circle
circle 0.166 Variable
Log spiral Taylor 0.165 Variable
Progressive Unconfined 0.319 Variable

30 Plane Culmann 0.002 27.5

Circle Ord. slices 0.012 Variable
h i
Circle
circle 0.009 Variable
Log spiral Taylor 0.008 Variable

Note: Partly after Taylor (1948).

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

614 Geotechnical Engineering

used unless it is directed by a weak sloping stratum. There is little to distinguish

between a circular and a logarithmic spiral failure surface, but progressive failure
must be simulated by using residual shear strength. With the lower slope angle
with a deep circular failure, the Ordinary Method of Slices is overly conservative
as it ignores frictional restraint from side forces on the slices, but this effect will
diminish as the sliding mass becomes shallower.

21.5.2 Partially Mobilized Resisting Force

If the resisting force Sr is fully mobilized, the resultant makes an angle

(which equals the obliquity angle) with a normal to the slip surface. However,
Bishop (1955) pointed out that in a stable slope Sr is only partially mobilized,
and the acting and normal force components of a slice weight depend on the
obliquity angle, not on the friction angle, which is larger. A smaller angle increases
the portion of weight W that is converted to N, increasing friction, while a smaller
portion is allocated to the acting force. This affects the calculated factor of safety
for a stable slope but not for one that is actively sliding.

Because the obliquity and the factor of safety affect both the resisting force, the
solution is by trial and error. Bishop proposed a simplified method that rapidly
converges. The method was developed specifically for circular slip surfaces.

Bishop’s concept is expressed mathematically as follows: in Fig. 21.11, vertical

side forces are assumed to be equal and opposite so they do not affect the
summation. Then
W ¼ N cos  þ Srd sin  ð21:27Þ
where Srd is the developed resisting force, which is the maximum resisting force
divided by the slice factor of safety:
Sr sin 
W ¼ N cos  þ , or ð21:28Þ
FSi

Figure 21.11
The simplified
Bishop method
assumes a
circular slip
surface and
adjusts the factor
of safety
for a partially
developed
resisting force Sr.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 615

Sr sin 
N cos  ¼ W
ð21:29Þ
FSi
Thus as the factor of safety increases, N decreases. By assuming that the factor
of safety is constant along the entire slip plane, Bishop obtained the following
expression:
X c0
l cos  þ ðW
u
l cos Þtan
0  X
FS ¼  W sin  ð21:30Þ
cos  þ ðsin  tan
0 Þ=FS

This may be solved with a computer program, or by iteration on a spreadsheet

by substituting different values of FS.

If
¼ 0, eq. (21.30) reduces to

P 0
c
1
FS ¼ P
W sin 

This is the same as the Ordinary Method of Slices. Bishop’s method does not
specify the radius or position of the center of the most critical circle, which are
determined by repeating the calculations for different trial surfaces and selecting
the one having the lowest factor of safety. This is quickly done with a computer
that can be programmed to find the 10 most critical circles. An example solution is
shown in Fig. 21.12, and it will be seen that a diversity of circular shear paths gives
almost the same factor of safety. The actual path therefore can be expected to be
sensitive to small variations in soil strength but can deviate from the most critical
circle without greatly influencing the factor of safety.

The Bishop simplified method is an improvement over the ordinary method

but generally has been replaced by other more rigorous and/or more general
solutions. Nevertheless many newer methods still assume a circular slip surface.

Figure 21.12
Trial circular
surfaces from a
STABL computer
run using the
simplified Bishop
method of slices.
All factors of safety
are in a range
1.5 percent, the
most critical being
shown with a
heavy line.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

616 Geotechnical Engineering

Some methods incorporate a force as well as a moment equilibrium, and solutions

are obtained using charts or computers.

21.5.3 Spencer’s Method

Spencer (1967) approached the problem of too many unknowns by assuming that
all forces acting on the sides of slices are parallel and inclined at a constant angle 
with horizontal. Two equations result, one representing a force equilibrium and
the other a moment equilibrium, and there are two variables, the  angle and the
factor of safety. Solution is by trial and error, so Spencer presented a graphical
summary shown in Fig. 21.13. The analysis now is more accurately accomplished
with a computer.

Figure 21.13
Spencer’s (1967)
charts for
estimating the
factor of safety of
a slope in
homogeneous soil
with selected pore
pressure ratios.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 617

Incorporation of pore water pressure into the analysis was simplified by defining a
pore pressure ratio, defined as the ratio of pore water pressure at the base of each
slice to the weight of the slice:
u
ru ¼ ð21:31Þ
h
where u is the pore water pressure at the slip surface, and h represents the
overburden pressure.

As a convenient approximation the pore pressure ratio is assumed to be constant

throughout the potentially sliding mass. Spencer’s charts in Fig. 21.13 show
stability factors for pore pressure ratios of 0, 0.25, and 0.5. The first is for
unsaturated conditions, and the last approximates a full saturation condition
since the unit weight of water is roughly one-half that of soil.

Spencer’s method was derived for a circular slip surface but also can be extended
to noncircular surfaces. Duncan and Wright (2005) indicate that this is the
simplest complete solution for estimating a factor of safety.

Example 21.11
Estimate the factor of safety for a 35 ft (10.7 m) high slope at 1 vertical to 3 horizontal, in
soil having a friction angle of 258, cohesion of 300 lb/ft2 (14.4 kN/m3), and unit weight of
120 lb/ft3 (18.8 kN/m3). Assume a worst saturation condition.

Answer: The slope angle is i ¼ tan
1 0.333 ¼ 18.48. For full saturation, the lower graph in
Fig. 21.11 gives
c 300 14:4
¼ ¼ 0:028, or ¼
FH Fð120Þð35Þ Fð18:8Þð10:7Þ
where F is the factor of safety. Solving for F gives 2.5.

Question: What is the factor of safety with c ¼ 0?

Answer: If c ¼ 0 the failure is parallel to the soil surface and Spencer’s charts
cannot be used. The factor of safety for a continuous slope in cohesionless soil is
discussed at the beginning of this chapter. From eq. (21.5), FS ¼ tan
/tan i ¼ 1.4.

21.6 SOLUTIONS FOR NONCIRCULAR SURFACES

21.6.1 Internal Restraint from Slice Side Forces

Some vertical repositioning of slices is necessary if they move downslope on
a surface having a changing curvature. An analogy is a stack of books held
vertically and allowed to slide along a curved surface: if the surface is circular

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

618 Geotechnical Engineering

and the books are allowed to rotate, there is no book-to-book friction, but if
the curvature changes, individual books must slide up or down relative to one
another. This vertical resistance to slipping adds to the factor of safety.

21.6.2 Janbu’s Procedure

Janbu’s (1973) simplified model, eq. 21.11, assumes that the line of thrust of side
forces runs at an angle through lower parts of sides of the slices, analogous to
pressures on a retaining wall. In addition the side forces are allowed to vary
between slices, which is important because it recognizes active and passive zones,
but it also increases the number of variables to be evaluated by trial and error.

Janbu’s solution became increasingly popular after being incorporated into a

computer program STABL developed by Siegle at Purdue University. The original
program is in the public domain, and recently has been modified commercially to
enable use with newer computer operating systems. Janbu’s method includes both
force and moment equilibria and can be applied to any shape slip surface.

21.6.3 Simplified Janbu

Janbu’s simplified procedure assumes that interslice forces are horizontal,
i.e., with zero friction. An empirical correction then is applied based on com-
parisons of results from both the complete and simplified procedures. As might
be expected, the correction is largest for deep slip surfaces where slices are
longer and have larger side forces. Correction factors are shown in relation
to the thickness of the slide compared with its length, d/L, in Fig. 21.14.
The correction is empirical and also can be applied to solutions by the ordinary
method of slices.

Figure 21.14
Janbu’s empirical
correction to
factors of safety to
account for
mobilization of
side forces.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 619

Example 21.12
An analysis of the landslide in Fig. 21.7 by the ordinary method of slices gave a factor of
safety of 1.0 but the slide temporarily was stable. What is the Janbu correction?

Answer: d/L is approximately 0.1, so the factor of safety should be multiplied by 1.05.

21.6.4 Other Methods

A comprehensive survey of various analytical procedures is beyond the scope
of this book and is presented by Huang (1983) and by Duncan and Wright (2005).
Many are included in commercially available computer programs. Assumptions
made in some of the most commonly used are as follows: Lowe and Karafiath
performed a force equilibrium assuming an inclination of side forces that is the
average of the slope and slip angle for each slice, which appears to be a fairly
realistic assumption. Morgenstern and Price and Chen and Morgenstern present
rigorous solutions with both force and moment equilibria. Horizontal seismic
forces and horizontal restraints from tensile reinforcement, nailing, etc., can be
applied to slices in the various methods. Sarma’s solution readily incorporates
seismic influences.

21.6.5 Importance of Having Reliable and Accurate Soil Data

The variability of factors of safety from the various methods and procedures fades
in comparison with the variability of the soil strength properties and accuracy in
measurements. Soil unit weight is less variable, but its variability is magnified by
the location of the groundwater table.

Slope stability analyses require averaged cohesions and angles of internal friction
that can be obtained only with large numbers of tests that are specific for these
values. The problem is both simplified and made more difficult in the case of an
active landslide—simplified because remolding in the shear zone tends to move
shearing strengths toward an average so that fewer tests are required, and more
difficult because the shear zone usually is too thin to obtain multiple identical
samples for laboratory testing. In that case a stage test, either a triaxial stage test
or the in-situ Borehole Shear test, may be the only practical solution.

Where adequate soil data are lacking, as often is the case, analyses sometimes are
performed using assumed values for cohesion and friction angle, or values obtained
on the basis of empirical correlations such as shown in Fig. 19.2. The degree of
sophistication of a computer program then has little to do with the reliability of the
results, as any analysis based on a guess remains a guess. Chapter 27 presents
somewhat more reliable estimates of friction angle based on penetration test data.

It is helpful to note that in an active landslide, soil cohesion is viscous or rate-

dependent and therefore can be assumed to be zero. Cohesion is regained through

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

620 Geotechnical Engineering

thixotropic recovery if the landslide stops. Still, it is on the safe side to assume that
it is zero.

21.6.6 Appropriate Factors of Safety

Engineers appreciate that a factor of safety is partly a ‘‘factor of ignorance’’
because it must take into account both the imprecision of the data and assump-
tions that may be made for analysis. An active landslide narrows the target area
because the factor of safety is 1.00. Therefore a factor of safety that is only
1.05 based on a worst-case scenario with fully saturated conditions can give
a reasonable assurance of stability, so long as nothing is done to aggravate the
situation. On the other hand, if the factor of safety is known to only two
significant figures, which often is the case, that variability must be incorporated
into a design factor of safety.

In a stable slope, additional unknowns include location and shape of the most
critical slip surface, location and anticipated changes in the groundwater table,
and indeterminacy of the analysis. Because of these additional unknowns the
design factor of safety usually is of the order of 1.2 or more.

Because of the potential for damages, an earth dam reasonably should be

designed with a higher factor of safety than a roadcut. However, flattening the
slopes of a large embankment increases costs geometrically so greater care is used in
construction and control testing, and a moderate safety factor is tolerated. Design
often is based on strengths measured on laboratory specimens shortly after
compaction and therefore does not include long-term thixotropic strength gain,
which can easily double the shear strength. Nevertheless dams do fail. Because of
the additional care used in design and construction of large earth dams, the
frequency of failures diminishes with increasing size.

21.7 RUMMAGING AROUND FOR THE BEST SLIP SURFACE

21.7.1 Computer-Drawn Trial Surfaces

Because of the ease and speed of electronic computers, it is a simple matter to let
the computer step off trial surfaces and determine those having the lowest factors of
safety. This procedure can be used with different computational methods, with
nonhomogeneous soils, and with complicated slope geometries. A range in starting
and ending positions is specified so that the search will not become infinite, and
a number of steps are selected. With modern computers the number of trials readily
is increased to over a hundred, compared with three or four using hand
calculations.

As the number of computer runs increases, the computer can produce some
unrealistic quirks in the slip surface. These probably relate to the assumptions made

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 621

in the analysis, so even a ‘‘perfect’’ solution to an indeterminate problem is not

perfect. It is doubtful that any model, whether analytical, numerical or physical,
can duplicate the random perfection of reality. Nevertheless some idealized shapes
will be considered, in part because of their application to foundation bearing
capacity, discussed in the next chapter.

21.7.2 Log Spirals

A logarithmic spiral is shown in Fig. 21.15, and is most commonly used to predict
foundation bearing capacity. Spirals also are applied to soil deformations on
active and passive sides of retaining walls, as shown in Fig. 19.15.

A unique property of the log spiral is that the angle between the radius and the
spiral is constant. If that angle is assigned to the angle of internal friction,
any radius for the spiral defines the direction of the resultant force at failure.
The corresponding equation is

r ¼ r0 e0
! tan
ð21:32Þ

where r ¼ any radius;
r0 ¼ a reference radius;
! ¼ the angle between r and r0;
e ¼ the base of natural logarithms; and

¼ the angle of internal friction.

The log spiral is most appropriate where pressure is externally applied to a

soil mass, as in the case of foundations and passive pressure on retaining walls.
The spiral is less suitable where driving forces change with depth because of

Figure 21.15
Ideal slip surfaces
and (a–c) some
developments
involving cycloids.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

622 Geotechnical Engineering

weight of a soil mass, such as in slopes or the active side of retaining walls.
From eq. (21.32) and Fig. 21.15 will be seen that a circle is a special case of
a log spiral with
¼ 0.

21.7.3 Cycloids
A cycloid is generated by a point on the rim of a wheel that is moving
along a horizontal surface (Fig. 21.15(a)). This does not appear particularly
relevant, but the cycloid has some other features that have interesting parallels
for slope stability.

Classic studies of landslides by a French engineer, Alexandre Collin (1846), led to

a conclusion that the natural shape of the slip surface is most closely represented
by a cycloid. This work was overlooked for many years until rediscovered and
championed by Frontard (1948). Terzaghi objected that as the cycloid is unrelated
to the friction angle, it can present unrealistic stress distributions. Ellis (1973)
applied the cycloid to vertical cuts in trench excavations.

As pointed out by Fox (1981), the cycloid was shown by Bernoulli to be the
solution to the ‘‘brachistochrone’’ problem that had intrigued Galileo: what is the
shape surface that will provide the fastest and most efficient path for a ball rolling
down a slope? The problem is illustrated in Fig. 21.15(b). Bernoulli used an
analogy to solve the problem by noting a similarity to the path of a beam of light
penetrating at an angle through a medium with a gradually increasing index of
refraction (Fig. 21.15(c)). The solution was obtained by integrating Snell’s Law of
refraction. There also is a striking analogy in soil mechanics to a shear path that
encounters increasing resistance with depth because of the increasing overburden
pressure and friction. Although analogy can lead to new innovations and insights
(Koestler, 1964), it is important to remember that, regardless of the depth of
philosophical conjecture, it does not constitute a proof. However, the analogy
that can be reinforced by reasoning can come closer.

Conceivably the energy-efficient cycloidal path for a rolling ball might also
describe the most energy-efficient path for the slip surface in a landslide. This
would be consistent with a concept of progressive failure, as each new segment of
a slip surface searches for the most energy-efficient path. Fox (1981), who pointed
out many of the features in this discussion, also pointed out that catenary curves,
which according to arching theory are trajectories of major principal stress,
for practical purposes are orthogonal to cycloids.

The x and y parameters of a cycloid with a starting point at the lower end
and verticality at the upper end are
x ¼ að þ cos Þ and y ¼ að1
cos Þ ð21:33Þ
where  is in radians.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 623

21.8 STOPPING LANDSLIDES

21.8.1 How to Be a Slide Stopper

The most obvious way to stop a landslide is to mitigate one or more of the
contributing factors by (a) removing load from the top, which seldom is feasible,
(b) adding load at the toe, (c) draining, or (d) increasing the average shearing
strength in the slip zone. The effectiveness of different options is predicted
using a method of slices. Some methods used to stop landslides are depicted in
Fig. 21.16.

21.8.2 Drainage
Drainage is one of the oldest methods for stopping a landslide, and is often
accomplished with corrugated plastic pipe that has been perforated with narrow
slots to let water in. However, the slots also can let water out, and a tile line that
crosses an active slip zone will be stretched, bent, compressed, or sheared off,

Figure 21.16
Some repair
methods for
landslides. The
one at the bottom
is too frequently
used and can only
make matters
worse.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

624 Geotechnical Engineering

so it may drain directly into the slip zone. Trenching to install the tile is a risky
adventure because it can reactivate the slide. Trenching can be done only during
dry periods when the water table is low, and should be done so that only short
sections are open at any one time.

Directional drilling such as shown in Fig. 21.17 is expensive, but can extend much
deeper than is possible with trenching, and avoids the dangers associated with
open trenches. Directional drilling is accomplished in soils by pushing a beveled
mandrel using water jetting. The mandrel carries electronic direction sensors so
that directions can be changed by changing the orientation of the bevel. A small-
diameter perforated plastic pipe is attached to the mandrel at the exit point, and
then pulled back and left in the boring.

Gravity drainage requires an exit point, and the groundwater table also can
be lowered by the use of well points, vacuum well points, or electro-osmosis.
The latter processes are relatively expensive but may be the only effective ways
to drain low-permeable clay soils. Vertical wells can be effective but require
intermittent pumping.

A key trench filled with granular material at the toe ideally will cross the slip zone
and act as a drain. Another advantage is that part of the soil at the toe is replaced
with soil having a higher friction angle, a factor that is readily included in stability
calculations. However, a key trench is not necessarily a perfect solution because
the landslide still can go over the top.

Drainage of a toe area can be improved by installation of gabions, which are

interconnected woven wire baskets filled with rocks. They have the advantage
of flexibility should the ground continue to move.

Figure 21.17
Directional drilling
can install
perforated plastic
drain lines without
trenching.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 625

21.8.3 Soil Reinforcement

Construction operations on a landslide preferably are conducted during dry
weather while the landslide has stopped. The area may be impassable for all but
track-mounted equipment, and toe areas may be a quagmire. However, this lower,
passive part of a landslide will not move without being pushed. Therefore the most
effective means for stabilization will address the upper, active part of a landslide,
which also can aid in stopping headward expansion. Headward expansion also can
be stopped by soil nailing into the exposed scarp. Stabilizing the passive part of
a landslide does not prevent it from being overrun.

One method for stabilization is to install vertical ‘‘stone columns’’ on a grid

pattern and extending through the landslide slip zone. Design is based on an area
replacement basis, substituting crushed stone with a friction angle of about 358 for
soil that may have a friction angle of only one-half or one-third of that amount.
‘‘Rammed Aggregate Piers’’ are similar but smaller in diameter and have a higher
friction angle, about 48–508. These are discussed in Chapter 24 on intermediate
foundations. Piles are more expensive and less effective unless used in large
numbers because the sliding soil readily can go around. A formula for area
replacement assuming that cohesion is zero in the slip zone is
m tan R þ ð1
mÞ tan S þ ¼ F tan S ð21:34Þ
where m is the fractional replacement of soil S by R, and F is the desired factor
of safety.

Example 21.13
The factor of safety of an active landslide is assumed to 1.0 and with full saturation the
friction angle in the slip zone is back-calculated to be 128. What percent replacement will be
needed to increase the factor of safety to 1.2, (a) using stone columns having a friction
angle of 358, (b) using Rammed Aggregate Piers having a friction angle of 488?

Answer:
ðaÞ Stone columns: m tan 35 þ ð1
mÞ tan 12 þ ¼ 1:2 tan 12
0:70 m þ 0:21
0:21 m ¼ 0:26
m ¼ 0:10 or 10%

ðbÞ Rammed Aggregate Piers: m tan 48 þð1
mÞ tan 12 ¼ 0:26
1:1m þ 0:21
0:21m ¼ 0:26
m ¼ 0:06 or 6%

21.8.4 Drilled Lime

Seeding a landslide slip zone with quicklime was first tried in the early 1960s
and has since been widely used where soils contain active clay minerals, smectites
or montmorillonites. Pebble quicklime is introduced into borings made on a grid
pattern in the active part of a landslide. Skill and experience are required in
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

626 Geotechnical Engineering

handling the chemical to prevent personal injuries. If the soil is too soft to hold
an open boring, reverse augering is used to push lime down into the soil, and
immediately creates a sheath of stabilized soil. In extreme cases lime can be
introduced in holes created with a concrete vibrator, which has been used to save
houses that were audibly creaking from pressures from landslides.

The drilled lime process works by immediately drawing water from the soil to
hydrate the lime, and sliding immediately stops. However, this reduction in water
content of the soil is only temporary. The lime column approximately doubles in
diameter as the lime hydrates, drying and creating radial tension cracks in the
surrounding soil. The cracks then are filled by a paste of hydrated lime, greatly
enhancing its distribution into the soil (Handy and Williams, 1967). For eco-
nomical reasons borings may only cover the upper part of the active zone so long
as no other changes are made that would reduce the factor of safety. The boring
diameters and spacing are designed to introduce from 1 to 3 percent lime,
depending on the clay content of the soil. Excess lime is used in a pozzolanic
reaction with soil clay minerals that creates hydrated silicates and aluminates
that are the same products as in hydrating Portland cement.

A simple field test to determine applicability of the drilled lime method for a
particular soil is to place small quantities of soft, saturated soil from the slip zone
into two sealable plastic bags. A few percent hydrated lime is added to soil in one
of the bags, and both are sealed and the samples remolded by hand. Within a few
minutes a reactive soil will become crumbly and appear to dry out even though no
moisture has been lost, as the plastic limit increases and exceeds the soil natural
moisture content. If that does not occur, either the moisture content is too high or
the soil does not contain the necessary active clay minerals.

21.9 SPECIAL PROBLEMS WITH EARTH DAMS

21.9.1 Relevance
Earth dams vary from some of the smallest to some of the largest man-made
structures in the world. Design of large dams takes on additional importance
because of safety issues. Many topics such as compaction, settlement, slope failure,
dispersive soils, piping, etc., are covered elsewhere in this text. Because of the
severity of failures and the continuing construction of new earth dams, particularly
in developing countries, the topic merits some additional consideration.

21.9.2 Causes and Frequency of Dam Failures

A study by ICOLD (1995) indicates that of 5268 dams built between 1951 and
1986, 2.2 percent failed. Most failures involved small dams, as the probability of
dam failure decreases with increasing dam height according to an approxi-
mate logarithmic relationship. Small dams are more likely to fail because

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 627

Figure 21.18
Causes of failures of earth dams. Slope failures include those occurring both in
the dam and around the reservoir; internal failures include piping and improper
compaction. (From data of Milligan, 1999.)

the consequences are not nearly so severe, and there typically is a lower level of
care in design and construction. Small dams also are much easier to repair and
restore to operating condition.

Most failures occurred before the dams were 10 years old, perhaps in part because
soil aging results in a gradual gain in shear strength. A statistical study indicates
that the odds of failure of a dam are about 1 in 7000 or 0.014 percent for any
particular year, but the odds are influenced by seismicity as well as the size of the
dam. In the typical 100-year useful life of a dam the probability is 1.4 percent,
but the odds are improved after the dam has survived its first decade.

The pie chart in Fig. 21.18 indicates that slope failures are not the major cause
of distress in earth dams, being eclipsed by foundation failures and overtopping.
Nevertheless the basic tenets are followed regardless of size: using properly
selected soil, having an adequate site preparation, and compaction of the
embankment at a moisture content that is at or above the optimum. Some
problems specific to earth dams and levees are discussed below.

21.9.3 Overtopping and Wave Action

Overtopping of an earth dam is a short ticket to disaster because of the large
hydraulic gradient from the reservoir level to the base of the dam. The effect of the
hydraulic gradient on erosion is effectively illustrated in Fig. 8.8 where a large
section of the dam washed away. Overtopping usually is the result of an insuf-
ficient spillway capacity, and most dams have an emergency spillway during
periods of unusually high rainfall. However, wave action also can be a factor.

Upstream faces of earth dams are covered with coarse stone, broken concrete,
stabilized soil, or grout-filled plastic mattresses to prevent wave erosion. The dam
obviously must be high enough above the reservoir level to prevent overtopping
by waves, whose height depends on wind velocity and the distance the wind blows
across water, or ‘‘fetch.’’ The influence of fetch is illustrated by a guide to
freeboard for small dams (Fig. 21.19).
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

628 Geotechnical Engineering

Figure 21.19
Suggested
freeboards for
earth dams.
(Adapted from
U.S. Bureau of
Reclamation,
1960.)

Figure 21.20
Rate of movement
of the ground
surface along a
traverse running
downslope on the
Vaiont landslide.
(Modified after
Leonards, 1987,
p. 608.)

21.9.4 Hydraulic Gradient

Hydrological factors influencing the internal stability of earth dams were
previously discussed in relation to dispersive soils and a critical hydraulic gradient
that can cause quick conditions. It should be emphasized that the design hydraulic
gradient may not be that which exists in the field, because if water finds a less
restrictive path, pressure will be transferred that will increase the hydraulic gradient
for the remainder of the path. For example, an uninterrupted gravel layer that can
be expected to occur under an earth dam will conduct water with very little decrease
in head, and the gradient may become critical for the remainder of the flow path.
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 629

Figure 21.21
Some
representative
earth dam designs
to direct and
control seepage.
Riprap protects the
faces from erosion;
berm adds to
stability of the
downstream slope.
(Adapted from
Creager et al.,
1948.)

The resulting erosion and piping can have much more serious consequences than
simply a leak.

21.9.5 Hydraulic Fracturing

High groundwater pressure near a filled reservoir can induce hydraulic fracturing,
particularly in abutment areas that are not confined under the weight of the dam.
Hydraulic fracturing is intentionally used in the petroleum industry to increase
production from oil wells, and occurs in soils during pressure grouting, indicated
by a sudden decrease in pumping pressure and increase in pumping rate. Hydrau-
lic fracturing can create a bypass channel that in turn affects leakage and the
hydraulic gradient.
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

630 Geotechnical Engineering

The probability of fracturing is reduced by performing detailed ‘‘dental work’’ on

foundation rock before a dam is built. This involves digging out and filling open
seams and weathered zones that might yield and erode out under pressure.

The tip of a hydraulic fracture becomes a focal point for tension that can be
calculated from a stress concentration factor from elastic theory and mechanics of
materials. The smaller the radius at the tip of the fracture, the higher the tensile
stress. Hydraulic fracturing is prevented by equalizing seepage pressure with
pore water pressure, which means slow first-filling of a reservoir. Piezometers
monitor pore water pressures within the dam and also preferably within the
abutments to determine if they are as predicted from the flow net—a low pressure
being conducive to hydraulic fracturing, and a high pressure indicating that
fracturing already may have occurred. Piezometers are monitored throughout the
life of a large dam to show any deviations from expected behavior.

21.9.6 Rapid Drawdown

An important factor contributing to slope instability of an earth dam is the
possibility for a rapid drawdown, where the reservoir level is lowered sufficiently
rapidly that soil in the dam and in slopes around the reservoir slope remains
fully saturated, creating destructive downslope seepage forces. Reservoir levels
therefore are lowered at a rate that will allow drainage of water from soil in the
slopes. As pointed out earlier in this chapter, seepage forces can reduce a factor of
safety against sliding by as much as 50 percent.

21.9.7 Leaky Reservoirs

While reservoir leakage does not constitute a failure on the same scale as the other
factors discussed in this section, a reservoir leakage can cause the dam to be
ineffective for its intended purpose, which insofar as the investors are concerned
constitutes a failure. The likelihood of developing leaks is magnified because the
reservoir area in contact with impounded water is orders of magnitude larger than
the contact area with the dam itself. Reservoir leakage is most critical in karst
areas where limestone is invaded by having underground caverns and sinks, and
can be troublesome where a reservoir is in contact with areas of fractured rocks or
gravel deposits.

This potential for leakage through soils and rocks underneath a dam is evaluated
with pressure tests performed in borings, and these are an important part of site
exploration. Underdam leakage is most effectively prevented by use of a cutoff wall
that can be concrete, clay, or driven sheet pile. Grouting is frequently used but is
difficult to direct to the desired area, and in severe cases of underdam leakage the
grout is washed away as fast as it is injected, particularly after a reservoir is filled
with water. In some cases the volume of grout pumped has exceeded the volume of
the dam itself, resulting in a huge cost overrun. Leaky reservoirs also impact an

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 631

entire neighborhood by raising the groundwater table and flooding nearby fields
and basements.

In recognition that a river often flows on a bed of sand that grades downward into
gravel, an upstream clay blanket may be used to seal off the river channel prior to
reservoir filling. If the hydraulic conductivity of the potential seepage layer is 10
times that of the average soil underneath the dam, the clay blanket can extend a
distance upstream equal to 10 times the width of the dam to achieve the same
resistance to flow. Some common methods for controlling through-dam and
under-dam seepage are illustrated in Fig. 21.21.

21.9.8 Case History of a Slope Failure

Vaiont Dam is a concrete arch dam that was subjected to a catastrophic landslide
in a slope above the dam. The dam itself survived but the reservoir did not, as the
tremendous splash overtopped the dam and swept away over 2000 lives in the
valley below.

Vaiont Dam occupies a deep valley in the Dolomite Alps in northern Italy, and is
one of the highest concrete arch dams in the world, standing tall at 860 ft (262 m).
The landslide area was about 2 km wide by 1.5 km long (1.2  0.9 miles).
The speed of sliding was back-calculated to be about 110 km/hr (70 mph), and
a combination of wind and water from the slide blew out windows in a town
100 m above the reservoir on the opposite hillside. The failure in 1963 remains
one of the worst dam tragedies to date.

The landslide at least in part developed along a prehistoric slip zone, and as
the lake was being filled a smaller landslide developed in the same area. During
the first filling of the reservoir livestock on the hillside were observed to be very
nervous, probably as a result of rock noises that can be a prelude to shear
failure but are inaudible to the human ear. Ground surveys were conducted
and showed that the hillside was moving at a very slow rate, as shown at the left
in Fig. 21.20. Attempts were made to reduce the rate of movement by reducing
the lake level, but the lake level ominously kept rising as the hill squeezed in on the
reservoir.

The engineers then were faced with a serious dilemma: should they increase the
rate of drainage, which would create a rapid drawdown condition, or should they
try and maintain the reservoir at a constant elevation, and/or should they require
an evacuation? There were no easy answers, particularly as attempts to lower or
maintain the reservoir level were frustrated by movement of the landslide.

The data in Fig. 21.20 present a classic example of different stages of soil creep,
although this may not have been fully appreciated at the time. During primary
creep, which is the mechanism of secondary consolidation, the rate is decreasing
and may stop. During secondary creep the rate is constant, and during tertiary
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

632 Geotechnical Engineering

creep the rate is increasing and is preliminary to rapid shear failure. From the
graph it appears that tertiary creep initiated about 2 months prior to failure and
was assured by a month prior to failure, which is when warnings would have been
appropriate.

The sudden and rapid rate of sliding was not expected, and it later was conjec-
tured that heat generated from friction might have contributed by decreasing the
viscosity. Steam pressure also was suggested but is unlikely under the huge
confining pressure. A more likely explanation for the speed of the sliding comes
from an examination of measurements at different points on the sliding mass.
These indicated that the upper part of the slide had moved 4.03 m while the lower
point moved only 2.87 m. This indicates that the rock mass between those two
points had compressed 1.16 m, confirming the thesis that a landslide has an active,
pushing part and a passive part that is being pushed. The separation distance
between the points was about 700 m, so the compressive strain was about
0.17 percent as stored strain energy. Being elastic, that energy would be subject
to instant release and conversion to acceleration.

Several approaches might have helped avert this tragedy, including a better
appreciation of the geological factors contributing to the slide. Modern
instrumentation, for example the use of inclinometers and acoustic micro-
phones, could have provided a more stringent warning. The cows had the
right idea.

Problems
21.1. Describe two situations when the factor of safety of a landslide is exactly 1.0,
and explain how this observation can be used to estimate soil sensitivity.
21.2. Explain the mobilization of shearing stresses within a sliding mass of soil,
and two situations where they theoretically will not be mobilized.
21.3. Analysis of an active landslide yields a factor of safety of 2.8. Is that pos-
sible? In your opinion is this more likely to be due to erroneous data or to the
method of analysis? What procedures should be used to check the results?
21.4. The friction angle of a soil is 258. Calculate the angle of repose on the
Moon where gravity is about one-fifth that on Earth.
21.5. A deep railroad cut intersects the plane of contact between a loess soil and
an impermeable clay paleosol. The plane of contact makes an angle of 128
with horizontal. Soil information is given in Fig. 21.22. Assume that
cohesion is zero and calculate the factor of safety with no contribution from
seepage.
21.6. Repeat Problem 21.5 assuming a worst-case scenario for seepage.
21.7. In the preceding problem even though there is a plane along which sliding
may occur, there is no assurance that a more critical curved slip surface

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 633

Figure 21.22
Conditions for
Problems 21.5 and
21.6.

Shear component Normal component Table 21.2

Data for Problem
Slice No. kN (lb) kN (lb)
21.8
1 14.31 (980) 8.17 (560)
2 14.90 (1020) 15.91 (1090)
3 12.57 (860) 29.76 (2040)
4 7.74 (530) 45.54 (3120)
5 5.97 (410) 49.06 (3360)
6 4.66 (320) 47.44 (3250)
7 3.94 (270) 40.00 (2740)
8 2.92 (200) 23.36 (1600)
9 2.20 (150) 18.37 (1260)
10 0.59 (40) 12.11 (830)

may not exist in the soil above that plane. Use Spencer’s method with three
different degrees of saturation to estimate factors of safety. Which do you
feel is most appropriate for design of a roadcut along a county road? Of a
cut bank behind a house?
21.8. The stability of a slope is being analyzed by the method of slices. On a
particular curved surface through the soil mass, the shearing component of
the weight of each slice and the normal component of the reaction at the base
of each slice are presented in Table 21.2. The length of the curved surface is
29.9 m (98 ft). If the friction angle of the soil is 128 and the cohesion is
2.39 kPa (50 lb/ft2), what is the factor of safety along this particular curved
surface?
21.9. A 21 m (70 ft) roadcut has side slopes of 2 to 1. The base of the cut is soil
in limestone. The soil has the following properties:  ¼ 19.6 kN/m3
(125 lb/ft2); c 0 ¼ 26.8 kPa (560 lb/ft2);
0 ¼ 208. Draw a trial circular failure
with the center at the crest of the cut and intersecting the cut at the toe. Use
the simplified method of slices to determine the minimum factor of safety
if seepage is not a factor, and apply the Janbu correction.
21.10. Recalculate the factor of safety in Problem 21.8 using Spencer’s method.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

634 Geotechnical Engineering

Table 21.3 Slice H X  W  Sai

Spreadsheet No. (m) (m) (kN/m3) (MN/m) (deg.) (MN/m)
calculation of acting 1 4.9 11 17.3 0.93
40
0.60
forces in Fig. 21.23 2a 7.3 20 17.3 2.53
b 4.9 20 19.8 1.94
Total 4.47
21
1.60
3a 3.7 15 17.3 0.96
b 15.2 15 19.8 4.51
Total 5.47
13
1.23
4 24.4 15 19.8 7.25
3
0.38
5a 6.1 30 18.4 3.37
b 26.2 30 19.8 15.56
Total 18.93 9 2.96
6a 3.0 30 18.4 1.66
b 23.8 30 19.8 14.14
Total 15.80 28 7.41
7a 4.3 30 18.4 2.37
b 18.3 30 19.8 10.87
Total 13.24 33 7.21
8 surcharge 0.04
a 8.5 30 18.4 4.69
b 12.2 30 19.8 7.25
Totals 11.95 40 7.68
P
Sa ¼ 21.45

Figure 21.23
Landslide problem
for the method of
slices data in
Tables 21.3 and
21.4.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 635

Table 21.4
Spreadsheet calculations of resisting forces in Fig. 21.23
Slice
L c c
L N u u
L N0
N 0 tan ’ Sri FSi
No. (m) (kPa) (MN/m) (MN/m) (kPa) (MN/m) (MN/m) (deg.) (MN/m) (MN/m) (Sr/Sa)

1 14.4 45.5 0.655 0.712 48 0.69 0.02 0 0 0.655
1.09

2 21.4 38.3 0.819 4.17 120 2.56 1.61 25 0.751 1.57
0.98
3 15.4 38.3 0.590 5.33 185 2.85 2.48 25 1.156 1.75
1.42
4 15.0 38.3 0.574 7.24 239 3.59 3.65 25 1.702 2.28
5.99
5 30.4 38.3 1.164 18.70 257 7.82 10.88 25 5.073 6.24 2.11
6 34.0 38.3 1.302 13.95 233 7.93 6.02 16 1.726 3.03 0.41
7 35.8 38.3 1.371 11.47 179 6.41 5.06 16 1.451 2.82 0.39
8 39.2 38.3 1.501 9.15 120 4.69 4.46 16 1.29 2.78 0.36
Tot. 205.6 7.976 13.178 21.13
FS¼ 21.13  21.45 ¼ 0.99

Note : Negative FSi for individual slices results when Sa and Sr act in the same direction.

21.11. Repeat the calculations of Problem 21.8 using Bishop’s method.

21.12. Set up a spreadsheet, check and if necessary correct the calculations in
Tables 21.3 and 21.4 for the slope shown in Fig. 21.23. Now make changes
to account for drains that will draw the groundwater table down to a level
DD0 and determine the effect on factor of safety.
21.13. Drilled lime stabilization of soil in Fig. 21.23 Slice 8 increases cohesion
50% and the friction angle to 258. Use the spreadsheet to recalculate the
factor of safety. (This may be done without a computer by substituting
values.)
21.14. Calculate the effect on factor of safety if 25% of the soil in Fig. 21.23 Slice
8 is replaced by Rammed Aggregate Piers having zero cohesion and a
friction angle of 508.
21.15. Set up a table with recommended factors of safety for landslides, roadcuts,
road embankments, small earth dams, and large earth dams.
21.16. In Fig. 21.23 what is the most likely explanation for the relatively flat
ground surface on Slice 6? (b) How might your explanation be confirmed?
(c) Might this subtle observation have any bearing on location of a drain
line? Explain.
21.17. Discuss relative merits of circular, log spiral, and cycloid failure surfaces in
homogeneous soils.
21.18. Carefully examine soil at and beyond the top of a steep slope for tension
cracks. Measure the height and slope angle and prepare a cross-section,
and sketch in a possible failure surface.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

636 Geotechnical Engineering

21.19. The slip surface of an active landslide is planar at the top and thickens to
more than a meter downslope. Explain.
21.20. Look to sources to review geological factors that may have influenced the
failure at Vaiont Dam.
21.21. Prepare a table with columns for quicksand, dispersive clay, and hydraulic
fracturing and entries describing (1) flow rate required, (2) flow direction,
(3) factors governing hydraulic pressure, (4) prevention, (5) high, medium,
or low potential danger to a dam, and (6) clues that the phenomenon is
occurring.
21.22. What should be looked for when inspecting an earth dam? When should
inspection be most rigorous, during high or low level of the reservoir? How
often should an earth dam be inspected (a) the first 10 years, (b) after 10
years?

References and Further Reading

Bishop, A. W. (1955). ‘‘The Use of the Slip Circle in Stability Analysis of Earth Slopes.’’
Ge´otechnique 5(1), 7–17.
Collin, A. (1846). Landslides in Clay. Transl from French by W. R. Schriever. University of
Toronto Press, Toronto, 1956.
Creager, W. F., Justin, J. D., and Hinds, J. (1948). Engineering for Dams, Vol. III: Earth,
Rock-Fill, Steel and Timber Dams. John Wiley & Sons, New York.
Duncan, J. M., and Wright, S. G. (2005). Soil Strength and Slope Stability. John Wiley &
Sons, Hoboken, N.J.
Ellis, H. B. (1973). ‘‘Use of Cycloidal Arcs for Estimating Ditch Safety.’’ ASCE J. Soil
Mech. and Foundation Eng. 99(SM2), 165–180.
Engineering Foundation Conference (1975). Safety of Small Dams. ASCE, New York.
Fox, D. E. (1981). ‘‘Behavioral Predictions of Compacted Embankments on Rigid
Foundations.’’ Unpublished Ph.D. dissertation, Iowa State University, Ames, Iowa.
Frontard, J. (1948). ‘‘Sliding Surfaces and Stability Calculations for Soil Masses of
Curvilinear Profile.’’ Proc. 2d Intn. Conf. on Soil Mechanics and Foundation
Engineering, Rotterdam, II,5.
Handy, R. L. (1986). ‘‘Borehole Shear Test and Slope Stability.’’ ASCE In Situ 86,
161–175.
Handy, R. L. (1995). The Day the House Fell. ASCE Press, New York.
Handy, R. L., and Williams, W. W. (1967). ‘‘Chemical Stabilization of an Active
Landslide.’’ ASCE Civil Engineering, 37(8), 62–5.
Huang, Y. H. (1983). Stability Analysis of Earth Slopes. Van Nostrand Reinhold,
New York.
ICOLD (International Commission on Large Dams) (1995). Ruptures de barrages —
Analyse statistique – (Dam Failures — Statistical Analysis). ICOLD Bulletin 99.
Janbu, N. (1973). ‘‘Slope Stability Computations.’’ In Embankment-Dam Engineering:
Casagrande Volume. John Wiley & Sons, New York.
Jumikis, A. R. (1962). Soil Mechanics. Van Nostrand, Princeton, N.J.
Koestler, A. (1964). The Act of Creation. Macmillan, New York.
Leonards, G. A. (1987). Dam Failures. Elsevier, Amsterdam. Reprinted from Engineering
Geology 26(1–4).

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
 Slope Stability and Landslides

Slope Stability and Landslides 637

Logani, K. L. (1973). ‘‘Dilatancy Model for the Failure of Rocks.’’ Unpublished. Ph.D.
dissertation, Iowa State University, Ames, Iowa.
Milligan, V. (1999). ‘‘An Historical Perspective of the Development of the Embankment
Dam.’’ Canadian Dam Association. Conference, Sudbury, Ontario.
Morgan, A. E. (1971). dams (sic) and other disasters. Porter Sargent Publisher, Boston.
Scholtz, C. H. (1967). ‘‘Wear and Gouge Formation in Brittle Faulting.’’ Geology 15,
493–495.
Seed, H. B. (1968). ‘‘Landslides During Earthquakes Due to Soil Liquefaction.’’
Fourth Terzaghi Lecture. ASCE J.Soil Mech. and Foundation Eng. Div. 94(SM5),
1053–1122.
Skempton, A. W. (1964). ‘‘Long Term Stability of Clay Slopes.’’ Geotechnique 14(2),
75–102.
Spencer, E. (1967). ‘‘A Method of Analysis of the Stability of Embankments Assuming
Parallel Interslice Forces.’’ Ge´otechnique 17(1), 11–26.
Taylor, D. W. (1948). Fundamentals of Soil Mechanics. John Wiley & Sons, New York.
Terzaghi, K. (1943). Theoretical Soil Mechanics. John Wiley & Sons, New York.
U.S. Department of Interior Bureau of Reclamation (1960). Small Dams. U.S. Govt.
Printing Office, Washington, D.C.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)

Copyright © 2007 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.