Professional Documents
Culture Documents
u
All limit equilibrium methods of slope stability analysis have four characteristics in common
(Duncan and Wright, 1980):
1.? All use the following definition of the factor of safety ():
(1)
2.? 0lacing a factor on shear strength is appropriate because evaluation of the shear strength
typically involves the greatest uncertainty in practical applications of slope stability
analyses. Note, however, that by definition the factor of safety is the same at all points
along the potential slip surface. This is reasonable only at failure; that is, when the factor
of safety equals unity. Because the factor of safety is taken to be the same at all points
along the potential slip surface even when the factor of safety is greater than unity, limit
equilibrium methods of analysis cannot model the mechanism of progressive failure.
3.? All assume that the strength parameters are independent of stress-strain behaviour.
4.? All use some or all of the equations of equilibrium to calculate the average values of
and on each slice, where is the normal stress on the base of the slice. is
required to determine the shear strength using the following equation:
(2)
in which 2 and are Mohr-Coulomb strength parameters. Since the forces involved in
equilibrium methods are statically indeterminate, all methods employ assumptions to
make up the balance between the number of equilibrium equations and the number of
unknowns in the problem.
The most commonly used slope stability analysis methods divide the mass above an assumed slip
surface into vertical slices. This is to accommodate conditions where the soil properties and pore
pressures vary with location throughout the slope. The forces acting on a typical slice are shown
in Figure 1.
î] weight of slice
î] seismic force applied at center of slice
] mobilized shear forces at base of slice
] effective normal forces on base
] water pressure force on base
] resultant top boundary forces
]vertical side force
] horizontal side force
(3)
Moment equilibrium around the center of the circular slip surface is the only condition of
equilibrium satisfied by this method. The value of given in Equation 3 is a low approximation,
which leads to a conservative value of . In cases of flat slopes with high pore pressures, the
error in the value of may be as much as 50%. In total stress analyses the error is typically not
more than 10%.
In
Ê
Ê the effect of the inter-slice force is eliminated by assuming that
the vertical component of the inter-slice forces is zero. The forces on a typical slice are shown in
Figure 3 and consist of:
a.? the slice weight, î
b.? the pseudo-static seismic force, î, in which is the seismic coefficient
c.? the pore pressure force, À = R ¦
x
d.? the effective normal force on base,
(4)
(5)
Once the normal force at the base of each slice is found, overall moment equilibrium yields and
implicit expression for the factor of safety:
(6
)
Whitman and Bailey (1967) point out a rare numerical difficulty with this method. When is
negative the possibility exists that the denominator in Equation 5 could be negative, or worse,
zero. When this occurs, a warning is printed and the resisting moment is set to zero.
2
Ê assumes that the inter-slice forces are parallel (Spencer, 1967). A typical slice
and the corresponding force polygon are shown in Figure 4. The forces on the slice are:
a.? the slice weight, î
b.? the pseudo-static seismic force, î, in which is the seismic coefficient
c.? the pore pressure force, À = R ¦
x
d.? the effective normal force on base,
(8)
(9)
Solving Equation 8 for and substituting the expression into Equation 9 and then solving
Equation 9 for ÿ yields:
(10)
If the external forces on the slope are in equilibrium, the vectorial sum of the inter-slice forces
must be zero to assure overall force equilibrium. Since the inter-slice forces are all parallel, this
requirement reduces to:
(11)
Furthermore, the normal force and the weight of each slice are assumed to be coincidental at a
point on the slip surface with the same X-coordinate as the slice's center of gravity. For each
slice to be in moment equilibrium the resultant ÿ of the inter-slice forces must be concurrent
with the remaining forces acting on the slice. In other words, ÿ must act through the point on the
base of each slice where the normal and weight forces act, with sufficient modifications made to
account for top-of-slice boundary forces or pseudo-static seismic forces.
If the sum of the moments of the external forces about an arbitrary point, say the origin, is zero,
then the sum of the moments of the inter-slice forces about this point must also be zero:
(12)
where Ú and are the coordinates of the point on the base of the slice where the forces are acting.
Satisfaction of Equations 11 and 12 assures that equilibrium is fully satisfied for each slice. Once
a solution is found to these two equations, the line of thrust can be calculated for each slice.
A solution to Equations 11 and 12 is obtained by simultaneously varying and until the two
equations are satisfied. For the initial assumed values of and , the equations may be in error
by the amounts
and
respectively, that is:
(13)
(14)
(15)
The angle is determined in the calculation process. To extend Spencer's Method to allow non-
parallel side forces we let:
(17)
(18)
The forces on a typical slice and the force polygon are shown in Figure 5. The side forces are
calculated using:
(19)
where ÿ is calculated using Equation 10 in which is replaced by . The horizontal force and
moment imbalance are calculated and and are varied until the imbalances are within
acceptable limits.
dissipation of the excess pore pressures set up, the stability can be determined by the
R analysis, using the undrained strength obtained in the laboratory or from in-situ
tests. This is essentially an end of construction method, and in the majority of foundation
problems, where the factor of safety increases with time, it provides a sufficient check on
stability. For cuts, on the other hand, where the factor of safety generally decreases with
time, the long term stability must be calculated by the effective stress method."
3.? "For saturated soils the values of 2 and are obtained from drained [triaxial] tests or
consolidated undrained tests with pore pressure measurements, carried out on undisturbed
samples. The range in stresses at failure should be chosen to correspond to those in the
field. Values measured in the laboratory appear to be in satisfactory agreement with field
records with two exceptions. In stiff fissured clays the field value of 2 is lower than the
value given by standard laboratory tests; in some very sensitive clays the field value of
is lower than the laboratory value."
These 1960 guidelines are still generally valid but increases in our understanding, particularly of
undrained strengths, since that time now allow us to do more accurate analyses albeit at the
expense of some complications!
0roblems in slope stability can be broadly grouped in two classes: short-term problems and long-
term problems. When a saturated or partially saturated soil with a low permeability undergoes a
change in stress there will generally be a corresponding change in pore pressure. The stage at
which the excess pore pressures (positive or negative) resulting from the change in stress are
fully developed is referred to as the short-term condition. With the passage of time these out-of-
balance pore pressures are redistributed until eventually they are everywhere in equilibrium with
the steady state pore pressures appropriate for the new stress conditions. This final stage is
referred to as the long-term condition and the continuing stability of the slope under gravity or
applied loads is a problem with drained loading conditions.
èÊ Ê
Ê
are usually simpler than
Ê Ê R
Ê
since they always involve drained or effective stress strength parameters and
for a given soil these do not vary very much with the type of test that is used to determine them.
However, it should be noted that even the effective stress Mohr-Coulomb envelope is curved,
rather than straight, for most soils and that the values of are thus lower at higher confining
pressures. The curvature of the Mohr-Coulomb envelope should normally be taken into account
for slopes higher than about 100 feet since use of values of determined at the usual confining
pressures of about 1 t.s.f. will then lead to errors on the unsafe side.
In general, short-term stability problems involve undrained loading and they can be addressed
using total stresses and undrained strengths or effective stresses, drained strengths and pore water
pressures. It is commonly believed that both approaches should give the same answer but this is
not necessarily so. As noted by Bishop and Bjerrum (1960):
"for factors of safety other than 1 the two methods will not in general give numerically equal
values of . In the effective stress method the pore pressure is predicted for the stresses in the
soil, under the actual loading conditions, and the value of expresses the proportion of 2 and
tan then necessary for equilibrium. The total stress method on the other hand implicitly uses a
value of pore pressure related to the pore pressure at failure in the undrained test."
Since the limit equilibrium method is most applicable at failure, in effective stress analyses one
should in fact use the pore pressures "at failure," rather than the "actual" pore pressures for the
short-term loading condition, and then both approaches will give similar answers. In this
connection one should note that the pore pressures specified in effective stress analyses affect
only the resisting forces that are computed and not the driving forces. This occurs because the
total normal force at the base of each slice is an unknown and an increase in the specified pore
pressure decreases the effective normal force but has no effect on the total normal force.
Similarly, changes in the pore pressures created by shearing under undrained loading conditions
are not included as driving forces in total stress analyses.
The traditional argument for using effective stress analyses for short-term, undrained problems is
that it is the effective stresses which really count in determining deformations and therefore
effective stress analyses provide greater insight into the problem at hand. However, effective
stress analyses require determination of either the "actual" pore pressures or the pore pressures
"at failure" and this is no easy task. Indeed, it is about as easy as it is to determine the undrained
strength that should be used in a total stress analysis since the reason that undrained strengths
vary with sample orientation, the type of test and the details of the loading conditions is largely
that the excess pore pressures are sensitive to these factors. In other words, it is about equally as
difficult to predict excess pore pressures for use in effective stress analyses as it is to determine
the appropriate undrained strengths for use in total stress analyses.
In practice various methods may be used to either predict excess pore pressures or to determine
undrained shear strengths for use in short-term stability analyses. Excess pore pressures in fully
saturated soils are most commonly predicted using the pore pressure coefficients A and B
(Skempton, 1954). 0rediction of excess pore pressures in partly saturated soils is extremely
difficult. Fredlund and Morgenstern (1977) and Fredlund (1979) discuss various approaches.
Special methods have been developed for particular problems such as the end-of-construction
condition for embankments, as discussed subsequently.
The preferred method for determining undrained strengths has changed over the years. In 1960
Bishop and Bjerrum recommended the use of UU triaxial tests and cautioned against the use of
CU triaxial tests. For a time, use of the vane shear test was popular but it is now recognized that
correction factors should normally be applied to the measured strengths (Bjerrum, 1972; Duncan
and Buchignani, 1973; Larsson, 1980). Deficiencies in the UU test were also subsequently
recognized (e.g. Bjerrum, 1973) and it is now generally agreed that use of UU triaxial tests
should normally be restricted to those cases where local experience has shown that use of UU
strengths leads to safe and economical construction. More generally, R should be obtained from
tests on reconsolidated samples using the test equipment and rate of loading which best
reproduces the in-situ stress and deformation conditions, using anisotropic consolidation if
necessary to represent the initial shear stresses on potential failure planes in the field and using
procedures such as the SHANSE0 method (Ladd and Foott, 1974) to minimize the effects of
sample disturbance.
Some of the issues involved in selecting appropriate pore pressures and/or strengths for slope
stability analyses are discussed further in the following sections in terms of the common classes
of slope stability problems.
u
1.? The simplest slope stability problem is a dry embankment as shown in Figure 6(a). The
pore pressures are equal to zero and the effective stress strength parameters, 2 and ,
should be used. Consolidated-drained (CD) tests should be performed to determine 2 and
.
2.? A partially submerged slope is shown in Figure 6(b). In this case, the water table is static
and the pore pressures are easily determined by taking the depth below phreatic surface
and multiplying by the unit weight of water. Effective stress strength parameters should
be used as determined by CD or consolidated-undrained (CU) tests with pore pressure
measurements. This problem may be solved two ways:
a.? Use total unit weights throughout, apply the boundary water pressure and specify
the pore pressures in the slope.
b.? Use buoyant unit weight below the water table and neglect the boundary water
pressure and pore pressures. This type of analysis is demonstrated in Example 3i
for TSLO0E.
Note that if a pseudo-static seismic loading is subsequently applied, method (a) must be
used because the correct inertia forces are obtained only by using total unit weights.
3.? The classic long-term stability problem is the steady state seepage condition shown in
Figure 6(c). This represents, for instance, the most critical condition for the downstream
slope of a dam with a full pool and with steady seepage through the dam. Again, use 2
shear strengths, the procedure termed the R analysis by Bishop and Bjerrum. The
undrained strength R used in such problems is commonly expressed in terms of the in situ
vertical effective stress and the overconsolidation ratio (OCR) (e.g. Ladd and Foott, 1974),
but in more sophisticated analyses the position of the element on the potential sliding surface
should also be taken into account (e.g. Ladd et al., 1977). Cuts in saturated clay, Figure 7 (b), can
also be analyzed for short-term stability using the R method; however, a long-term
effective stress analysis should also be performed as this is usually the more critical case.
The end-of-construction condition for a constructed embankment, Figure 7(c), is also a problem
in short-term stability. This problem may be analyzed using total or effective stress methods. The
total stress method normally involves determination of the undrained strengths using UU triaxial
tests and the effective stress method commonly relies on the use of the procedure for computing
pore pressures developed by Hilf (1961). The total stress method is shown in Example 1 for
TSLO0E. Johnson (1974) provided a detailed discussion of the relative merits of various
approaches to end-of-construction stability problems.