Sie sind auf Seite 1von 18

Session l/6

GENERAL THEORY OF STABILITY


OF SLOPES
bY
0. Ii. FR~~HLICH, Dr Techn, Dipl. Ing.

INTRODUCTION
During the last four decades, numerous efforts have been made to deal with the problem
of the stability of slopes, with the aim of computing the factor of safety with respect to the
sliding of slopes of cuts and earth dams. As the properties of the materials in which sliding
phenomena can occur deviate very much from those of elastic solids, governed by the well-
developed theories of elasticity and strength of materials, all investigators based their efforts
on radically idealized soils, with the result that their procedures must be considered as working
hypotheses whose results represent rough estimates only. This statement holds good even
for homogeneous soils.
In consequence, the title of this Paper, chosen by the Organizing Committee of the
European Regional Conference in Stockholm, must not be considered too precisely ; a more
accurate title would have been : “ The existing theories of the stability problem of slopes.”

I’RINCIPLES OF ESTABLISHING 24 THEORY OF THE STABILITY PROBLEM OF SLOPES


The stability of an earth mass is guaranteed if, at no point of it, the shearing strength of
the material is exceeded. In order to investigate whether this is the case or not, one has to
analyse the state of stresses at least within the potential zone of failure, which may be approxi-
mately known by observation of slides that have taken place. Furthermore, the so-called
condition of jaw failure or condition of Plasticity of the involved soil must be known. If, at a
single spot, in a plane section of the mass, the principal stresses fulfil the condition of flow
failure, then the limiting elastic equilibrium at this point, and, at the same time, the critical
ratio of the principal stresses, are attained. If the greater principal stress increases further,
movement of the surrounding soil particles starts in a direction perpendicular to the greater
principal stress.
This is the beginning of plastic flow. If the point in question is located at some distance
from the free boundary of the mass or underneath the toe of the slope, the plastic flow mobilizes
stresses due to earth resistance, which counteract the plastic flow and maintain the critical
ratio of the principal stresses to a certain point. A further increase of the principal stresses
at this point is then impossible and the beginning of Progressive failure is reached.
Ultimate failure occurs if the plastic flow condition is realized over a continuous line or
zone within the considered section of the mass of soil, by which a sliding part of the soil is
separated from a remaining one, which is not affected by the failure.
The conditions in all planes parallel to the considered section of the mass of soil may be
identical, so that one has to deal with a two-dimensional stress problem, taking no account
of the mean principal stress acting normal to the considered section. By this assumption, the
analysis of the stability problem is simplified to a certain extent, without prejudicing the
applications of the theory to most of the practical problems involved in the investigation of
cuts and dams.

EXISTING THEORIES OF STABILITY OF SLOPES


1. The method of approaching the present problem suggested above (known as the
“ point-to-point ” stress analysis of dams) has been employed by two investigators, J. H. A.
Brahtz 1 and P. T. Bennett.* Brahtz overcomes the difficulty of establishing the real state
1 The references are given on p. 47.
37

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
38 0. K. FR6HLICH

of stresses in every point of the section of a dam, its slope, and its foundation by stress func-
tions similar to those being used in plane stress problems of the theory of elasticity. To
account for the initial ratio of horizontal and vertical stress in the mass of earth, Brahtz
50
introduces a so-called comparation factor K = G’ where G is the weight per cubic foot of

saturated soil. The condition of plasticity employed by Brahtz is :


r,=c+utan+ . . . . . . . . . (‘1
where 7, represents available shearing strength,
C >> cohesion,
u 3, normal stress at the shearing surface,
and + ,, angle of shearing resistance.
The safety factor is defined by the ratio of the available shearing strength to the shearing
stress at every point of the potential zone of failure. This point-to-point stress analysis aims
at finding out zones of failure, where the elastic limit of the soil is reached or exceeded,
characterized by lines of equal factors of safety against flow of failure.
The factors of safety in this analysis are point functions and cannot be compared with
the mean value of safety against total rupture along a continuous sliding surface.
The derivation of the distribution of stresses is somewhat arbitrary and is not based on
the stress-strain relationship of the actual soil ; it is one out of the infinite number of possible
distributions which are compatible with the conditions of static equilibrium of this highly
statically indeterminate problem. If the critical ratio of the principal stresses at a single
point of the potential zone of failure is reached (and according to the location of this point
passive earth stresses cannot be mobilized), the surplus load must be taken over by the points
in the neighbourhood until the critical ratio at these points is also attained. This, however,
is only possible by an alteration of the original stress distribution-on which the whole
analysis is based. This final stage of the point-to-point analysis is not taken into considera-
tion by existing procedures of this kind and seems to bc the most difficult and time-consuming
part of further development.
P. T. Bennett advocates point-to-point stress analysis, without using stress functions but
using the elastic stress distribution laws up to the limit of elastic equilibrium. He admits
that the quantitative determination of stresses within the soil of a slope or dam is impossible,
but recommends the study of the probable magnitude and orientation of principal stresses by
which potential sliding surfaces may be detected. Finally, he recommends that the point-
to-point stress analysis, as well as the sliding surface procedure, may be used in the analysis
of any important embankment on a weak foundation and that the results be given equal
weight in a final judgement of the safety factor.
2. Several authors 324 see the solution of the stability problem of slopes in the determina-
tion of the shape and location of the sliding surface. It must be admitted that the problem
of finding the shape and location of the sliding surface is a very attractive one. Frontard
based his investigations on the Rankine plane state of stresses in cohesive masses, but neg-
lected the fact that the stresses parallel to the surface of an infinitely high slope acting on
vertical sections of a sliding mass must be constant for a given depth. It should be sufficient
to refer to the critical remarks on these theoretical efforts given by Terzaghi,s but the
Authors wishes to characterize the difference between the critical heights in a very simple case,
as given by Frontard’s method, and those given by the procedure based on a plane sliding
surface. Frontard explains 3 that failure of a vertical slope must occur if the height amounts
to

. . . . . . . . . (2)

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
GENERAL THEORY OF STABILITY OF SLOPES 39

whereas the critical height according to Coulomb is

!z=%.tan i+g . . . . . . . . . .
Y ( 1
It is readily seen that at the foot A of the vertical slope in Fig. 17 (a) the critical ratio of the
principal stresses is reached when the height AB equals h,. But at all other points of the
plane sliding surface AC the shearing stress is smaller than the shearing strength (s = c + Gtan #)
and, at C, the shearing stress along AC is even zero.

r*, /
B c

-L-F A4
$4

Fig. 17

If one builds up a slope as shown in Fig. 17(b) th e critical ratio of the principal stresses is
attained in all points of the lower part of the sliding plane AC, whereas, if along CD a joint
is inserted into the cohesive mass, the conditions along CE are identical to those in Fig. 17(a).
If one closes this joint by the same material which is present within the triangle ABC, a
further load, represented by the triangle BDF in Fig. 17(c), is necessary to produce failure.
It occurs along the plane AE, everywhere along which the critical ratio of the principal
stresses is attained.
The vertical shearing force along DC in Fig. 17(c) equals exactly the weight of the triangle

BDF, i.e., i . PO tan Therefore, failure occurs only if the critical height

(= 1
4C
Y 4 2
h = 2ho = - . tan i- 4
is attained.
It is interesting to note that, in practical examples, the critical height of slopes on the
base of curved surfaces of rupture, calculated on the same basis as equation (3), exceeds
Frontard’s values by more than 100%. For the reasons set out above, the existing
computations of the shape and location of the sliding surface of slopes cannot be considered
satisfactory.
3. The third possibility of attacking the problem of the stability of slopes consists of
choosing any continuous surface of rupture along which failure may occur and of placing it
into the earth mass in such a way that there is a maximum danger of sliding. Observations
of slides that have taken place show that, in many cases, the shape of the sliding surface is
approximately circular. In other slides, a logarithmic spiral gives better agreement with
actual findings.
Petterson, Hultin, and W. Fellenius 6 based their efforts to compute the danger of sliding
on the assumption that the actual sliding surface may be replaced by a circular cylinder.

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
40 0. K. FRiiHLICH

11’. Fellenius, approximately thirty years ago, developed the procedure for investigating the
stability of slopes now known as the “ Swedish Method ” or “ Slip Circle Method.” A great
number of authors have tried to improve and simplify this method, so that in every country-
a variant of the “ Swedish Method ” may be encountered. The principle of this method is
as follows :
Equilibrium of the sliding mass is attained if at every point of the sliding circle the shearing
stresses fulfil the condition of flow failure :
s=c+otan+ . . . . . , . . . (4)
where c and 4 are constant values, characterizing the rectilinearity of the shearing strength
as a function of the normal stress along the sliding surface ; these values are neither the
true cohesion nor the true angle of internal friction. The problem is a two-dimensional one
and is statically indeterminate. The radius and the location of the centre of the slip circle
are unknown and have to be determined in such a way that the safety factor is a minimum.
The most plausible definition of the factor of safety (F) against sliding is :
available shearing strength
I;= . . . . . .
actual shearing stress
for every point of the sliding surface. (Up to some years ago, the way generally used to find
the factor of safety against sliding was an indirect one.)
For any given value of c, the required value, &i, which was necessary to establish
equilibrium for an arbitrarily chosen slip circle was found. By trial and error the radius of
this circle and the location of its centre was varied until &,q, was a minimum. The result
of this computation was equilibrium of the sliding mass for c, &, valid for the critical circle ;
the factor of safety, according to this procedure, was unity, thus F = 1. In this way,
numerous cases of plane slopes, using toe circles, have been treated, and the results are given
in diagrams 6, 7, s which refer exclusively to cases where F = 1. In order to find the factor
of safety on the basis of the above-mentioned diagrams, Fellenius 6 has given a general rule
which represents a very good approximation to reality. The “ Fellenius rule,” which was
also used by Krey and Taylor without giving its derivation, states the following fact, expressed
schematicallv :
Characteristics of the Factor of safety
shearing strength of a slope (F)

cl,c$l . . . . . . 1
c,+. . . . . .7)
Condition :

This rule may be employed if a great number of computed cases of sliding slopes is tabulated
or represented by a diagram as, for instance, shown in Fig. 45 of p. 159, of Terzaghi’s book
on Theoretical Soil Mechanics 5.
A proof, that Fellenius’s rule is a good approximation will be given later.
There is another method of establishing approximately the factor of safety for a case of
sliding, if the values of the characteristics of the shearing strength c,,~., + = ,-,(“ pure cohesion “)
and L,., c = o (“ pure friction “) are known. This rule was given by Ohde 9 and reads :
tan 4
F=C + . (6)
CW, # = 0 tan &eq.,c=O ’ . . . .
whose derivation will be given below.
These two rules for the establishment of the safety factor are the necessary tools for the
application of the indirect method of computing it. There is also a direct method of finding

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
GENERAL THEORY OF STABILITY OF SLOPES 41

the safety factor, which, at the same time, satisfies the three conditions of equilibrium of the
two-dimensional system of forces, and which takes into account the fact that the problem is
statically indeterminate. This simple procedure will now be explained graphically by an
example.
In Fig. 18 (a) an irregular slope AB and a potential slip circle AB with its centre 0 is shown.
The self weight of the sliding mass, combined with uplift, seepage pressure, and hydrostatic
excess pressures, etc., is represented by the force R. The distance of R from 0 being a.
The given shear strength characteristics of the mass are c and 4. The factor of safety is to
be computed. The result of the cohesional stresses along AB can easily be given :
R,=cAB

Its location parallel to the chord AB is fixed by the distance pC. It is known that pC = P$~

(2~ = cAOB, Y = radius of the slip circle).


The resultant of all frictional stresses along AB is R,, the resultant of all normal stresses
is N. The forces R, RC, R,, and N must be in equilibrium. The first two conditions are
satisfied when the polygon of these four forces is closed, see Fig. 18 (b). From this figure the
magnitude and direction of R, and N can be taken. The normal stress resultant N must
pass through the centre 0 of the slip circle (see Fig. 18 (a)), the frictional stress resultant R4 is

.____
_.____ _.._...d_..___._..
_.___._ _ ...d
NCM+(p'O NVW.C*

.dB
v-0 c-o

a*
‘png c.o=
tsf?-qc-0
-0334

_..~,v_;..R
_ .._.
mm ._......_. --
-2JomT-l
%ng.c*
R-?tXSlnl?l

Fig. 18

normal to N at a distance of p+ 2 Y from 0. This distance is not known and it depends on the
distribution of the normal stresses along the am AB. In the abstract case p+ = Y, N is
concentrated at a point of the circle and the general procedure, set out here, degenerates into
the “friction circle method,” very often used. The real distance p+ can be computed for any
chosen distribution of the normal stresses along the arc AB.lo~ii~ 1s
From the intermediate resultant 1-3 of R and R,, on the one hand, and of N and R,, on
the other, which are equal, parallel, and at a distance e, it is to be concluded that the third
condition of equilibrium of the two-dimensional system of forces is satisfied only if the force R
is displaced parallel towards the right over a distance e, such that R passes the point of
5”

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
42 0. K. FRijHLICH

intersection of N and K,. The factor of safety according to the above given definition
equation (5) gives :

. . . . . . . .

The numerator represents the moment of the available shearing forces about the centre 0,
the denumerator is the moment of the actual shearing stresses about the same point. From
the third condition of equilibrium it follows that

R, . pe + R, . pd = R(a + e) . . . . . . . . . (8)
therefore,

F=Nil’=l+f 61
. . . . . . . . (9)
From Figs 18 (a) and 18 (b) one gets :
‘/ e = 13.8 mm, a = 30.5 mm,
therefore,

F-1+$1.45 . . . . . . . . . (94

Dem’vation of Fellenius’s rule


The shearing strength characteristics of a slope in limiting equilibrium (F = 1) may be
ci and &. If these values are replaced by proportionally increased values c and 4 satisfying
the conditions :

and
(10)
where 71signifies a constant, the factor of safety according to equation (7) gives :
Rc pc + R, . pd
F=- (1’)
R,, . pc + R+, . p$l ’ ’ ’ ’ ’ ’ ’
This equation may be written as follows :

AB.c.p,+N.tan4.p,
F= . . . . , (114
tan 4
A2 f.pe+% . __ . P41
rl rl
If the difference between q5 and $i is not very great, one can put N1 = N and pdl = ps ;
then equation (lla) reads :
F=v . . . . . . . . . . . . (llb)
which is the result of .Fellenius’s rule. This rule does not represent a special definition of
the factor of safety, but it is approximately in agreement with the definition given in
equation (5).

Derivation of Ohde’s rule (equation 6)


According to the direct method of establishing the factor of safety (Figs 18 (a) and (b)),
it is very simple to compute the required “ cohesion ” c,,,, +,= ,,, when the material of the
slope is assumed to be purely cohesive. The forces R, N, req.,$ = ,, and R, req.,.+= ,, must be in
equilibrium which is accomplished by the polygon of forces, Fig. 18 (c). To the original scale
of this figure, R, req.,.$= o is 23 mm. Fig. 18 (a) shows the force polygon for R AT4req,, C_ 0 and

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
GENERAL THEORY OF STABILITY OF SLOPES 43
R, = o, 6 req.9when the material of the slope is perfectly cohesionless. The angle of shearing
resistance required is obtained at tan &q,, c = o = 0384 that is $_, c = o = 21”.
If the actual strength characteristics are c and 4 we can, according to equation (7), write :
Rc . pc -I- R, . p,+
I;= R . . . . . . . . . (12)
Creq.,+ = 0 . PC
R
and ;* PC -I- R, . ~4
F;= . (13)
,$req.,c=o. P$ * . * . . * . *
From equations (12) and (13) we deduce the equations :

Rc . pc %
F= (134
R+req.,c=o . PS + ROreq.,c=o ’ ’ ’ ’ . *
The second term of equation (12a) on the right-hand side and the first right-hand term of
equation (13a) together give F, because the numerators of both terms are equal to Ra, There-
fore, the addition of equations (12a) and 13a) gives :
RC %
FER
creq.,4=0 +R4req,,c=0
’ * - . . . * . (14)
Sow we have :
RC c
z---- . . . . . . . . (15)
R Creq.,$=O creq.,4 = 0
and

R, =- N . tan 4
R 9req.,c=O . (16)
Nd,req.,c=O.tan~,,q.,,=o ’ .
If the difference between 4 and $req.,c= o is not very great, we may write :
N = N4req.,c=o . . . . . . . . . . (17)
and, after substitution of equation (17) into equation (16),
R, tan+
R9req.,c=. = tan &,q., C= o ’ ’ ’ . a . . . . (“)
Substituting equations (15) and (18) into equation (14) we get :
tan 4
FE-f---j-
creq.,$ = 0 tan4req.,c=0 ’ ’ ’ ’ . * . (19)
which proves that Ohde’s rule, equation (6) approximately holds good. This rule may be
used to determine graphically the factor of safety for any given couple of the values c and 4.
This is shown in Fig. 19. If one inserts the values c,,,., + = o and &req.,c=. into the shearing
strength diagram (a, s) resulting from laboratory tests, one gets a vertical scale for F through
the point 1. It is then very simple to determine different couples of c, 4 for any desired
factor of safety on the basis of the values c,,~.,4 = o and &.q., C= o which follow from the simple
triangles of forces shown in Figs 18 (c) and (d). The numerical example of Fig. 18 treated
by equation (19) gives :
RC tan I$ 15.5 0.325
+ - = 1.52
F=R--
creq.,+=o tan I$~~~.,
C= o = %O + 0384
This value is 4.8% higher than the factor of safety according to the direct method using
equation (9).

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
44 0. I<. FRSHLICH

The two rules equations (llb) and (19) are both based on the definition of the factor of
safety according to equation (5) and are perfectly consistent. Therefore, it is possible to
derive, for instance, Fellenius’s rule from equation (19), which reads :
tan q5
l&L+
LX&.4 = 0 tan ~&a.;C= 0

3 scale

-6
Fig. 19

If the shearing characteristics c and tan 4 are proportionally decreased by an arbitrary divisor
7 the factor of safety also will be decreased, giving :
C tan +

F +-+ ’
new
req., 4 = 0 tanheq., c = 0
or, with equation (19),
1
F new= - F
rl
If we desire to find the value 77,which makes F,, equal to unity, we obtain
rl=F . . . . . . . . . . . . . (lib)
as Fellenius’s rule requires.

Analytical Solution of the Direct Method


The direct method of computing the factor of safety leads very simply to an analytical
expression for the factor of safety.10

Notation :
2a angle AOB (Fig. 18 (a)).
Y radius of slip circle
R resultant of external forces
6 inclination of R to axis OE
a distance of R from 0
c, (bcharacteristics of shearing strength
E auxiliary angle

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
GENERAL THEORY OF STABILITY OF SLOPES 45
According to Friihlich 10 it is :
2r . c . sin a
tan E = tan 6 - . (2Oj
R.cosS . . ’ . . . *
2ar2c cos 6 . sin 4 B
and Fyj-$
COSE ‘a ....*I (21)
where & = Ycosc;~~; E, . . . . . (22)
The parameter v is the characteristic of the distribution of normal stresses along the sliding
surface.
With the notations given in Fig. 18 (a) this parameter is :
w
v=F@ . . . . . . . . . . . . cw
Using the substitution
RC
u=- . . . . . . . . . . . .
R . (24)
and the relation
R, = 2rc . sin a . . . . . . . . . (25)
we get :
2rc . sin a
R= zG . . . . . . . . . . . (26)

Introducing equation (26) into equation (21) we obtain with regard to equation (22) :
cos 6 . sin fj cos v($ - E)
. . . . . W-4
COSE * cos vu >
and equation (20) changes into

tan Z = tan S - J- . (20a)


COSS ** ‘**” l

The application of these equations to the numerical example Figs 18 (a) and (b) (direct method)
is shown below :
Y 80
- = C5 = 2.62 6 =33”
a
a = 55” tan 6 = 0.649, cos 6 = 0.839
a 0460 0.220
= - zz.z1.17 tan E = 0.49 - - = 0387
sin 0.819 0.831

E =21”

+ = 18” cos E = 0.933


sin + = O-309 4- E = 18” - 21” 10’ = - 3” 10’
From Fig. 18 (a) it follows that w = 22” 30’ (a + j3) = 58” ;
therefore,
w 22.5
v= - = __ = 0.388
a+B 58

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
46 0. K. FR6HLICH

IJ(# - l) = - 1” 15’
cos v (4 - e) = 1
VCI= 0.38855” = 21 o 20’
cos va = 0.931
Equation (21a) furnishes the factor of safety :
0.839 . 0.309 1
F = 2.62 1.17. 0.22 + 0.933 ’ 0931
= 2+62{0.258 + 0.298) = 145 . . . . . . . . . (9b)
This is exactly the value of the graphical procedure of the direct method, see equation (9a).
The factor v which is characterized by the relation
05v51 . . . . . . . . . . . . (27)
governs in a very simple manner the whole range of the statical indeterminateness of the slip
circle method. The special value v = 0 leads to the results of the “ friction circle method”;
therefore the parameter v may be called the “ friction circle correction,” to use an expression,
introduced by Taylor.8 For practical cases the mean value V, = 0.5 gives plausible and
reliable results.10
The graphical or analytical direct method of finding the factor of safety must be repeated
several times with different choices of the slip circle in order to find the minimum value of
the factor of safety belonging to the critical sliding circle. Since much work has been done
by Fellenius, Krey, Taylor, Terzaghi, and others with respect to the location of the centre
of the critical circle, the practical foundation engineer is in a position to choose the first sliding
circle properly and to save much time in designing slopes and dams by using the direct method
of computing the factor of safety. Equation (21) or (21a) can be easily adapted to the cast
of rapid draw-down of the reservoir level behind a dam.12
Recently a most interesting procedure based on the “ Swedish method ” has been sug-
gested by A. Casagrande, called “ Resistance Envelope Procedure,” worked out and
extended by N. Janbu.13 The derivation of the method is based only on the third condition
of equilibrium of the plane force system (EM = 0) ; the two others (6X = 0, EY = 0) are
not taken into account. The statical indeterminateness of the problem is wiped out by the
use of average normal stresses along the slip circle. Nevertheless, the numerical results
obtained quickly by this procedure agree very well with those computed, for instance, by the
direct method set out above. By application of this procedure to the example represented
in Fig. 18 one obtains a factor of safety approximately 7% higher than that given by
equations (9a) and (9b).

The Logarithmic-spiral Method


Replacing the real sliding surface by a logarithmic spiral, it is possible to establish a direct
method of computing the factor of safety graphically and analytically as weILl

CHOICE OF SHE_4RING-STRENGTH CHARACTERISTICS FOR THE APPLICATION OF


THEORIES OF THE STABILITY OF SLOPES
This is the most difficult point of all attempts aimed at giving approximate estimate of
the danger of sliding of slopes. The different procedures dealt with above assume the shearing
strength to be a rectilinear function of the normal stress. In reality, this is only approxi-
mately true. Therefore, if the test results show a curved line for the shearing strength plotted
against normal stress, one is compelled to replace it by a straight line within the region of the
average normal stress along the sliding surface, or by two straight lines representing two

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
GENERAL THEORY OF STABILITY OF SLOPES 47
limiting couples of c and 4. If the factor of safety has been found for .one couple, the nomo-
gram Fig. 19 gives immediately the factor of safety for my other couple with satisfactory
approximation. The shearing tests (quick or slow) have to be carried out in such a way that
the relation between shear strength and normal stress for several points of the potential
sliding surface is known for the time of construction and later on, as the designer has to
compute not only a single value of the safety factor but a continuous line of the safety-factor
value showing its variation with time and changing loads.
It is impossible in the limited space permitted to deal with all recent contributions to this
subject and the reader is referred to L. Bjerrum’s general report on “ The Stability and
Deformation of Slopes, Earth Dams and Groundwater Problems ” (Proc. 3rd Int. Conf. Soil
Mech. 1953, 2 : 362).

REFERENCES
1. M.~Y, D. R., a& BRAHTZ, J. H. A., 1936. Proposed Methods of Calculating the Stability by Earth
Dams. 2nd Congr. Large Dams, 4 : 539.
2. BENNETT, P. T., 1951. Notes on Embankment Design. Tvans. 4th Cong. Large Dams, 1 : 223.
3. FRONTARD, J., 1922. Cycloides de Glissement des Terres. Conapt. Rend. Hebdom. Acad. Sci., 174: 520.
4. JAKY, J., 1936. Stability of Earth Slopes. Proc. 1st I&. Conf. Soil Me&., 2 : 125.
5. TERZAGHI, K., 1946. Theoretical Soil Mechanics, New York, p. 151.
6. FELLENIUS, W., 1927. Erdstatische Berechnungen, Berlin.
7. KREY, H., 1936. Erddruck, Erdwiderstand und TragfLhigkeit des Baugrundes, Berlin.
S. TAYLOR, D. W., 1937. Stability of Earth Slopes. J. Bostola Sot. Civ. Engrs, 24 : 197.
9. OHDE, J.. 1943. Einfache erdstatische Berechnungen der Standsicherheit von BGschungen, Berlin
and Dresden. Arch. Wasserw., No. 67.
10. FRBHLICH, 0. K., 1950. Sicherheit gegen Rutschung einer Erdmasse auf kreiszylindrischer Gleitfl&he
mit Beriicksichtigung der Spannungsverteilung in dieser FlZiche. Federhofer and Girkman. Fest-
schrift, Wien, p. 181.
11. FRBHLICH, 0. K., 1954. Sobre a avalia@o do Coefficiente de Seguranqa relacionado corn o Escorrega-
mento dum maciqo de Terra coerente ao longo duma superficie circular. Bol. Ord. Eng., Lisboa, 3 : 6.
12. FRBHLICH, 0. K., 1951. On the Danger of Sliding of the Upstream Embankment of an Earth Dam,
Caused bv Comolete or Partial Discharge of the Reservoir. Trans. 4th Cow. Laree Dams, 1 : 329.
13. JANBU, N., 3954. AStability Analysis of Slopes with Dimensionless Parameters. H&avd Sdil Meek.
Series No. 46.
14. FRBHLICH, 0. K., 1953. The Factor of Safety with respect to Sliding of a Mass of Soil along the Arc
of a Logarithmic Spiral. Proc. 3rd Int. Conf. Soil Me&, 2 : 230.

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
48 DISCUSSION ON STABILITY OF EARTH SLOPES

DISCUSSION
Opening the discussion, Professor 0. K. Friihlich stated that there were upper and lower
limits to the range of the factor of safety F associated with the infinity of statically permissible
distributions of the normal stress (Taround the slip surface, and this enabled a decision to be
made whether any value of F computed for a particular problem was possible or not. He had
examined the problem considered by Dr Bishop on p. 14 of his Paper, in the manner shown
in Fig. 19, which indicated the relevant data. The minimum value of the factor of safety,
statically possible for this slope was :

F e= Rcpc + TN tan4 = l .387

aR
while the maximum value was :

F _ =
max
RCPC
+ r(NA + NB) tan 4 = i.537
aR
These two limits were abstract values because the normal pressures N, Na, and Nb could not
possibly be concentrated at single points. Nevertheless, any statically possible value of F
must lie within the limits of 1387 and 1537. The value of Fminwas given by the “ friction
circle ” method when the forces satisfied the three conditions of equilibrium. The real value
of the safety factor Frealmight lie in the inner third of the range of Fminto Fmax, thus :
1.437 < F,.eal < 1.487

Dr Bishop had found F = 138 by the simplified analysis and F = 1.59 by his more
accurate method using the same critical slip circle given in Fig. 3 (a), p. 14. The factor of
safety by the improved method, therefore, lay beyond the range Fd,,-+ Fmax, and there
must be a fundamental error in Dr Bishop’s procedure which he would try to indicate.
Equation (1) of Dr Bishop’s paper was based on his Fig. 1 (b), p. 8, and used Fellenius’s
rule. That rule was correct for a plane sliding surface, and for a curved surface if 4 = 0.
If 4 > 0 the rule would be exact only if the sum of all normal stresses D along the slip circle
for F = 1 were equal to the sum of the analogous stresses for F > 1, which was not so.
Replying to Professor Frohlich, Dr. Bishop said, the purpose of his Paper had been to
examine the errors introduced into the “ slices ” method of analysis by the simplifying
assumptions usually made, and to propose a reasonably accurate alternative method which
could be a#died even when the parameters c’ and q5’ varied along the slip surfaces. Professor
Frohlich had recalculated the factor of safety for the case of the homogeneous dam quoted as
his first example but, having obtained a different value, had concluded that there was a mistake
in procedure.
There was, however, a fundamental difference between the definition of factor of safety
adopted by the Professor and himself. While his definition* was in agreement with that
adopted by Professor Fellenius, Professor Taylor and Mr A. Caquot, Professor Friihlich had
chosen to consider the additional disturbing moment which could be applied to the sliding mass
before equilibrium was reached, and to base his definition on this. Dr Bishop himself felt
that this definition had only a restricted field of application, and might be misleading in some
cases, since in frictional soils the additional disturbing moment would alter the magnitude and
distribution of the normal stresses on the slip surface.
In making his calculations, Professor Frohlich had taken a different circle from that
* F is the factor by which the shear strength parameters c’ and tan @ would have to be reduced to bring
about limiting equilibrium.

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
DISCUSSIOS ON STABILITY OF EARTH SLOPES 49

Fig. 19

referred to in his Paper due to a misunderstanding of the text ; Dr Bishop had now calculated
the factors of safety on this circle (i.e., Fig. 19), and found that the conventional method gave
F = 1.41, and the more exact method la%.* Moreover, Professor Frijhlich had taken
the value of the force due to the pore-water pressure to be 40% of the total weight of the
sliding mass, instead of summing vectorially the elemental forces acting across the slip
surface. The important difference, however, was that of definitions, and this should not be
allowed to obscure the main issue which was the limited accuracy of the conventional method
under certain conditions commonly met with in practice.
Observing that it was current practice to estimate the factor of safety of a slope in the
manner defined by Professor Fellenius, Mr A. Lazard declared that such a factor of safety was the
ratio of the strength of the soil to the value to which the soil strength was mobilized to provide
limiting equilibrium of the particular slope. However, in a given construction the soil was
neither homogeneous nor were the forces completely known. Thus it was unrealistic to give
a unique value to the strength of the soil and to the strength required to be mobilized ; rather
a range should be given for each. If these ranges overlapped, a possibility of failure would
exist and an actual failure might result if circumstances were unfavourable. The factor of
safety to be chosen should thus bear some relation to the risk to be accepted.
* Since in this example $’ is constant along the slip surface the modified friction circle method (Taylor,
1948) can be used, and also gives F = l-55.
6*

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
50 DISCUSSION ON STABILITY OF EARTH SLOPES

Mr J. Frontard mentioned that he was one of the first to consider stability problems.
He had treated slips in banks as shear movements in soils, and had concluded that the
slip surface was cycloidal in shape. The assumption that a slip surface had the form of a
circular arc was simpler, but should be used only for minor cases, and then with large factors of
safety. Regarding Professor Frohlich’s criticism of his calculation of the critical height of a
vertical bank, he thought everything depended on the nature of the material forming that
bank. If the material were capable of resisting tension, then he realized his figure could be
exceeded, but for soils he considered his own theory the more correct.
Professor FrBhlich observed that the inequality of the two frictional forces
Y + X tan 4 >_-x
2
tan # + cot 55
given by Mr A. Caquot (p. 31) could not be taken as giving a clear picture of the danger of
sliding. The ratio of the two sides of this equation was not the factor of safety as generally
accepted. Mr Caquot had recommended the Fellenius rule for finding the factor of safety,
but this was only an approximation ; in fact, it was unnecessary to apply any rule for deter-
mining the factor of safety, and this he would show by considering the examples given in
Mr Caquot’s Paper.

Case I. (No port pressure)


F = 668 700 + (7 111 - 189) = l.16
1 736 400
Case II.
F = 668700 + (7 111 - 1240)189 = 1 024
1 736 400
Case III
F = 668 700 + (7 111 - 3 030)189 = o.829
1 736 400
The last figure was close to Mr Caquot’s factor of safety of 0.825 because the Fellenius rule
gave a reliable result when $ was fairly small. However, he could not agree with Mr Caquot’s
finding that in Case I the safety was more than sufficient. With regard to the figure on p. 29
(1) the forces P and S must pass through the centre 0, and
(2) the friction force F could not intersect the slip circle, but must lie outside it.
Dr I. T. Rosenqvist then observed that at the Norwegian Geotechnical Institute they had
found the fine clay fractions to be isotropic, but there was a certain parallel arrangement of
the coarse particles in marine clays. The isotropy seemed dependent upon the salt content of
the water from which the clay sedimented. He would be interested to know if only post-
glacial clays in Sweden were homogeneous and the older clays varved.
Mr B. Jakobson considered that Professor FrShlich had made an error in his criticism of Dr
Bishop’s paper. The forces in Professor Friihlich’s figure would satisfy equilibrium only
when friction was fully mobilized, otherwise the force N would have a different position, and
he therefore proposed applying the following modification to Professor Frohlich’s method.
X chosen factor of safety F, for the cohesion force R, is introduced, and a new resultant P (see
Fig. 20) obtained. The lower limit of the safety factor is then obtained by drawing a circle of
centre 0 to touch the force P ; the required angle of friction being calculated from the radius
of this circle. The safety factor for the friction forces is
F = tan I# existing
tan 4 required

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
DISCUSSION ON STABILITY OF EARTH SLOPES 51

This procedure is repeated a number of times with different values of Fe until F = Fe is


reached, and this gives the value of Fd. To find Fmax, the circle of centre 0, which is tangent
to the forces P, and Pa, is found by trial and error, otherwise the procedure is the same as for
finding Fmin. Using this method for Dr Bishop’s problem, he had found Fmh = 1*58 and
F nlax= l-81, and it would appear that Bishop’s revised method gave too low a value. He
also questioned the correctness of the term - $y. w . d” in Dr Bishop’s equation 23b, and
suggested it should be zero.

V
G
FORCE Dl*GRAn

P)‘R Fig. 20

In answer to Mr Jakobson, Dr Rosenqvist said that in referring to a homogeneous clay he


had meant a post-glacial clay, and when alluding to a glacial clay, a varved. He then gave
details of the homogeneity of clays which had settled out in various lakes.
Dr L. Bjerrum was of the opinion that an important step forward had been made by Dr
Bishop, and that the stage had been reached when comparison of theory with slides in the field
was needed rather than further theoretical refinements. Dr Bishop’s method was now widely
used in Norway, and as an illustration of its application he would mention a calculation of the
factor of safety at the end of construction of an earth dam. Neglecting forces between slices,
a factor of safety of 1.2 to 1.3 was found, while Dr Bishop’s method gave 1.6 to 1.7. This
example illustrated the magnitude of the error in the simplified method and emphasized
the economic importance of Dr Bishop’s method in earth dam design.
Professor Suklje then referred to the definition given by Professor FrGhlich, in his equation
5 (p. 40) which he felt was generally acknowledged but which later he had applied in the form :
Resisting moment
Factor of safety =
Disturbing moment
With this, however, he could not agree. It was inapplicable where cc was small or the slip
surface plane, and inaccurate where the surface was a logarithmic spiral; he considered that the
graphical method of Taylor* was a useful approximation.
* Fundamentals of Soil Mechanics, D. W. Taylor, Kern York, 1948. p. 447.

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
52 DISCUSSION ON STABILITY OF EARTH SLOPES

Mr W. Kjellman said that he could not agree with Mr Frontard’s statement that the
Swedish method of stability analysis was unreliable for soils which lose part of their strength
at rupture. A large number of cases had been analysed by this method for clays having
this property, and although many gave a safety factor only slightly greater than 1.0, no
failures had occurred.
Dr J. Brinch Hansen said that while in some instances, notably in the direct shear test,
failure took place not along a slip surface but along a surface of zero elongation, he could not
agree that it was always true; even if it were, the problem was not solved by using an
apparent friction angle and proceeding otherwise as usual. No simple method existed yet for
the determination of the critical surface of zero elongation. Mr Kjellman in reply recommen-
ded the procedure of using the apparent friction angle in the customary methods of stability
analysis for cases in which the severely stressed region was narrow, and where the rupture
layer was compelled to follow the imagined slip surface-as in a direct shear test. He believed
that the procedure was correct in such instances. Mr S. Odenstad mentioned the reference
made in Mr Kjellmann’s Paper (p. 22) to Professor Prager’s work on plasticity. Prager’s
continuum had only a positive dilatancy and, therefore, it would appear to have only a
restricted application to soils though he thought that this restriction could be disregarded
in view of the small deformations envisaged in Prager’s theory. Dr Samsioe then emphasized
the importance of model tests on the stability of slopes to provide a basis for a satisfactory
theory of internal stresses in earth masses.
In answer to Mr Kjellman and Dr Samsioe, Mr Lazard showed how the exact calculations
of Mr Frontard and the approximate circular method of Professor Fellenius gave nearly
the same results. Differences were not greater than 1 y0 or 2% with normal slopes.
Professor Frohlich stated that he believed that Professor Fellenius considered his own rule
to be only an approximation. This he felt could be concluded from the original text :*

“ Im dritten Falle (Reibung und Kohasion) bestimmt man passungsweise die


Reibungs- und Kohasionswerte, die, in dem gleichen Verhaltnisse zu den entsprechenden
wirklichen Werten vermindert, fiir gleichgewicht eben nur erforderlich sind. Da es
sich in diesem Falle immer urn Kleine Werte der Reibungswinkel handelt, wird der Sicher-
heitsgrad gleich dem invertierten Wert des so gefundenen Verminderungs-verhaltnisses.”

Therefore in his opinion the rule was proposed in order to calculate a value of F which would
closely approximate to the real factor of safety so long as the available angle of shearing resistance
4 was not much greater than the required angle &. Dr Samsioe then raised the question of
the meaning of a factor of safety. In the case of a concrete dam having a factor of safety of
2, or a bank with a factor of 1.5 it was not possible to imagine the system of loads needed to
mobilize the full resistance. He therefore suggested that the factor should indicate the
reduction in the resistance of the structure to be made because of deficiencies in knowledge.
In reply, Professor Frijhlich thought that by considering the additional disturbing force
(for a plane sliding surface) or additional moment (circular slip surface) required to bring
the slope to failure, a measure of the factor of safety was obtained. Indeed, in the case of a
plane sliding surface, the usual definition of the safety factor was obtained in this way. Mr
Kjellman added that the factor of safety must always be related to the uncertaintiesinfluencing
the stability. If, for instance, the cohesion were the only uncertain quantity, it would be
logical to define the factor of safety as the ratio of the probably available cohesion to that
necessary for equilibrium. If some external load were the only uncertain quantity, the factor
of safety could be defined as the ratio of the load which would produce rupture to the probable
load. If two or more quantities were uncertain, it would be impossible to define a factor
* Erdstatische Berechnungen mit Reibung und Kohasion (Adhasion) und unter Annahmene
Kreiszylindrischer Gleitkhen. 1927, W. Ernst & Son, Berlin.

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
DISCUSSION OX STABILITY OF EARTH SLOPES 53

of safety logically. Probabilities must then be considered, and he would emphasize that
two probabilities could be combined by calculation, but not two factors of safety. Neither
Fellenius’s rule, Ohde’s rule, nor Frbhlich’s definition of factor of safety was satisfactory if,
for instance, the pore pressure was the main uncertain quantity in a problem.
Mr Lazard was glad to hear Mr Kjellman’s comments, introducing the idea of probability.
He showed how it was possible to consider the variations in the applied forces as well as the
variations in the resistance of the material in order to arrive at an idea of the probability of
failure. It was the duty of engineers to make the probability of failure sufficiently small
for the risk to be acceptable having regard to time, money, and possible accident. Dr Samsioe
thought that it was a mistake to consider only the possibility of circular arc slip surfaces.
The information presented to the Conference concerning actual slides did not support the view
that they were circular, and he believed that the resistance to sliding would be smaller on
surfaces of other shapes. Discussing the conception of a factor of safety, Mr C. van der Veen
pointed out that it contained at least three elements : it was partly a question of allowable
deformations, partly of chance, and partly an allowance for uncertainties or mistakes. He
thought a reasonable philosophical approach could be made on these lines towards a definition.
Commenting on Mr Kjellman’s paper, Dr Bishop* said that probably most people would
assert that slip surfaces actually existed because they have observed them in the field. He
had found in triaxial tests on loose sands that while only a bulging type of deformation occurred
when the failure resulted from an increase in the axial stress, this was not always so if the failure
was caused by reduction of the lateral stress. Moreover, the latter mode of failure was more
usual in slope stability problems. Tests on model walls retaining loose sand (Jenkin, 1931)
suggested that visible rupture surfaces might be formed. He also believed that the illustra-
tion of the relationship given by Kjellman between the angle of internal friction 4 of sand and
the coefficient of friction p between the actual grains was an over-simplification. The follow-
ing solution to the problem had been given by Mr Caquot in 1934 :

With p = tan 25”, the above equation gives + = 36” which is in better agreement with
experimental results than the value obtained by Mr Kjellman. Recently he had obtained a
solution, based upon energy considerations, in which a failure surface was not postulated, the
method distinguishing between values of $ which would be measured in triaxial and plane
strain tests. In these two cases it led to 38” and 44” respectively for constant volume tests
(Bishop, 1954). Mr Kjellman’s substitution of thin rupture layers for slip surfaces led, he felt,
to a number of difficulties since in an undrained triaxial test it would lead to a prediction of
rupture layers at 45”, but laboratory tests did not support this view.
Dr Brinch Hansen emphasized that in any stability analysis the rupture figure must be kine-
matically possible, i.e., the movement of the different zoncsmust be compatible with each other.
The circular slip surface, for instance, was kinematically possible, but it was not generally so
when the rupture surface was composed of arbitrary circular arcs and straight lines. He
suggested the use of logarithmic spirals instead of circular arcs for the slip surface in soils
with internal friction, since the unknown reactions no longer entered into the problem and,
therefore, no division of the sliding mass into slices was necessary.
Mr Zehnle thought that the application of the formulae of Fellenius and FrGhlich was
hardly possible where the slip surface was not circular. And the direct comparison of the
disturbing and resisting forces was complicated. Therefore, in his opinion it would be better
to develop the ideas put forward by Dr Samsioe, and to replace the customary factor of safety
by a function which enabled the calculation of how much the values of certain of the pro-
perties, such as cohesion and friction, should be diminished before limiting equilibrium was
* Correspondence, GCotechnique, 4 : 1 : 43.

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.
54 DISCUSSION ON STABILITY OF EARTH SLOPES

reached. The method was essentially that of finding a series of limiting equilibria, of
calculating the limiting values for the various properties in question, and then deducing the
function without considering any ratio of moments or forces.
Dr Bishop then said that Mr Jakobson had questioned the correctness of equation 23b in
his Paper on the use of the slip circle. This equation is only invoked in obtaining a complete
balance of the internal forces and does not, therefore, affect the accuracy of the routine method
which had been put forward. He thought, however, that, if it is recalled that the horizontal forces
E, and -&+i are total forces including the water pressure, then the result of the summation
must be that given in his Paper.
In answer to a question put by Professor Suklje, Professor FrGhlich considered that his
method of computing the factor of safety for a plane sliding surface gave results agreeing
with values obtained by all investigators. He mentioned that he had dealt with logarithmic
slip surfaces in his Paper in the Proceedings of the Third Conference on Soil Mechanics and
Foundation Engineering, 1953, 2 : 230.

Downloaded by [ UC San Diego Libraries] on [12/09/16]. Copyright © ICE Publishing, all rights reserved.

Das könnte Ihnen auch gefallen