Stability of Slopes and Embankments

under Static Load

Indrajit Chowdhury
Petrofac International Limited;
Sharjah, UAE.
[email protected]

Shambhu P. Dasgupta
Department of Civil Engineering,
Indian Institute of Technology Kharagpur, India
[email protected]

ABSTRACT
Conventional slope stability analysis usually consists of method originally proposed by
Fellenius (1936) wherein failure profile of the slope is assumed to be an arc of a circle.
Considering moment of driving force and the resistive force about center of the arc, it is
ensured that factor of safety (FOS) is greater than unity.
The present paper puts forward a rational theory on the characteristics of the failure curve
based on classical mechanics and proves that circular curve that was chosen as a convenience
is not strictly correct. The trial and error method prevalent with slip circles was avoided and
the solution comes up with a logical basis of iteration, one that converges very fast once the
weakest plane has been identified.
It takes into cognizance deficiencies in both Fellenius and Bishop’s formula in terms of pore
pressure and also circumvents the issue of instability that occurs with Bishop’s formula in
certain cases. No- homogenous layered soil can also be considered with equal ease. That the
results match in certain cases is coincidental, and may be due to the fact that circular curve is
also generically a 2nd order curve.
The solution is direct. If one wishes, can well workout the problem in a spreadsheet and no
special software is needed.
KEYWORDS: Slope Stability, non-circular slip surface, pore pressure

INTRODUCTION
Despite significant development in our understanding of its behavior, the stability of natural
or man-made slopes still remains a topic of significant research among the civil engineers.
Conventional design of slope stability analysis usually consists of methods originally
proposed by Fellenius (1936) wherein failure profile of the slope is assumed to be an arc of a
circle. Considering moment of driving force and the resistive force about center of the arc, it is
ensured that factor of safety (FOS) is greater than unity. This theory was pioneered by Fellenius
- 5351 -
 Vol. 18 [2013], Bund. V 5352

based on observations of a number of failures along slopes of Alps Mountain in Switzerland. It

was observed that failure of a soil slope usually takes place along a curved surface. Fellenius’
theory was based on homogeneous clayey soil (i.e. φ=0).
Considering the soil to be of c-φ type, and has layers with variable properties especially for a
natural deposit, the method was extended to the method of slices by Bishop (1955) where variable
soil properties including pore water pressure were taken into consideration by numerical
summation of all forces over the slices considered along an assumed circular arc of failure.
Bishop actually proposed two methods; 1) a rigorous method, and 2) a simplified method. In the
rigorous method, normal and friction forces are considered on the face of each slice. In simplified
case, Bishop ignored this normal and friction forces from the analysis. Based on these two
analysis, Bishop demonstrated that other than ease of calculation due to ignoring of normal and
shear force along the slice face, FOS derived was conservative and lower bound. Thus for static
loads, Bishop’s method of simplified analysis considering a circular arc of failure has remained
most popular in design office practice.
Taylor (1937) proposed another variation of Fellenius’ method called friction circle technique
deriving a stability number based on which one could predict the free standing height of clayey
soil when excavated in slope.
Morgenstern and Price (1965) developed a generalized method for circular as well as non
circular failure profile where they solved for the factor of safety using summation of forces along
normal and tangential direction to the base of slice and taking summation of moments about
center of the base of each slice. The equations are written for a slice of infinitesimal thickness.
Force and moment equations are combined and a modified Newton-Raphson numerical technique
was used to solve for factor of safety satisfying the force and moment equilibrium. This method
requires an extensive use of computer to solve the problem.
In recent past, finite element method (FEM) as a tool of analysis for stability of slopes, has
gained increasing popularity. Originally used by Chopra and Clough (1966) to predict the
stability of earth dams under earthquake, the procedure is sound in terms of mathematical basis
and it predicts failure along the path of principal stress, though cannot always give a feel like the
traditional method of slices, as to what is the FOS inherent in the system? Considering FEM’s
capability to cater the layered soil having variable soil property, it certainly makes a very
attractive choice to an engineer in terms of analysis, but is not without its own problem. First, it
requires a very careful and exhaustive investigation of in situ soil properties which are not always
possible to asses in detail (and also could be very expensive). Moreover, considering the problem
is in the domain of limit equilibrium, it would require a rational constitutive model that can be
fitted into the analysis. For many soils, especially those which are naturally deposited, developing
a constitutive model in the laboratory is always not possible and in many cases undisturbed
samples are difficult to collect. Finally, even if developed, they cannot always be fitted into a
commercially available FEM package unless special purpose software is written for this and
plugged into the program. Also, considering the intense computational effort necessary in terms
of model generation, input data and interpretation of output, FEM has certainly not been an
automatic choice unless the problem is too complex to be handled by conventional method.
Other than this, number of researchers namely, Janbu (1954, 1968), Lowe and Karafiath
(1960), Spencer (1967), Whitman and Bailey (1967) and Sarma (1973) proposed different
methods for slope stability analysis. Abramson et al. (2002) have critically reviewed all these
techniques and came up with the conclusion that despite many of them being mathematically
rigorous, do not provide a significant improvement in results compared to Bishop’s (1955)
 Vol. 18 [2013], Bund. V 5353

technique and are far more laborious to apply, irrespective of being done electronically or
otherwise.
Thus, notwithstanding its limitation, Bishop’s simplified slip circle method continues to
remain the most popular method in practice.

PROBLEM WITH THE TRADITIONAL SLIP CIRCLE

ANALYSIS
Irrespective of its continued popularity and sound mathematical basis many slopes and
embankments designed by method of slices or Swedish Slip Circle Method, have misbehaved
under static as well as earthquake loads. Major lacunae in the method possibly calumniates from
the assumption of considering failure along a circular path which makes the analysis basically a
trial and error of finding the curve having least resistance and its center of curvature. Even with
the best judgment of finding out the curve of least resistance for a soil system, uncertainty still
gnaws the designer that there could indeed be another curve having lower resistance whose center
has eluded him. As the method opted by her/him to locate the center has been trial and error
(though presently, a number of commercially available software has significantly reduced this
risk but yet not full proof). Moreover that a circular curve itself is the most optimum curve along
which failure should take place—there is no comprehensive mathematical proof of the same
furnished in literature. Nature of the curve even under homogenous condition could well be
different.
Chen et al. (1971), based on limit equilibrium in plasticity analysis, have modeled the failure
curve as a logarithmic spiral and came up with very satisfactory result against field observed data.
However due to complicated nature of the curve, it is rarely put into practice and requires trial
and error method to locate the primary radius r0, as radius of the curve varies as an exponential
function ( r = r0eθ tan φ ) of the angle θ subtended at center of the curve and angle of soil friction φ.

The question thus arises is, what was the reason behind choice of a circular curve in the first
place?
Plausible reasons that could be attributed to this query are as follows:
1. Observation, that failure takes place mostly along a curvilinear path;
2. As a matter of convenience, circular curves are easy to handle having established
geometric properties;
3. Era in which the technique was pioneered by Petterson (1916), Fellenius (1937) and
Taylor (1937) application of Graphic Static was at its peak. With no computers available,
most of the problems dealing with complex force system were solved graphically when a
curvilinear failure profile can most conveniently be represented by a circular arc.
Analytically speaking, a circular curve suffers from one serious drawback. Considering its
center lies at the co-ordinate (-g,-f), equation of a circle can be represented in a Cartesian
coordinate system by the equation

x 2 + y 2 + 2 gx + 2 fy + c = 0 (1)

Thus, to represent a circle or its arc it requires three boundary conditions that should be
known.
 Vol. 18 [2013], Bund. V 5354

• It should pass through three known points that lie on the circle.
• It may pass through two known points and radius r be known
where r = g2 + f 2 − c .
• It may pass through one known point that shall lie on circle and coordinate
of its center (-g,-f) is known.
Unfortunately for a sloped embankment the circular arc passes through two known points
(actually only one to start with) that intersect the embankment/slope with center of curvature and
the radius remaining unknown. Thus geometrically, there can be an infinite number of such
circular arcs that can be drawn through these two points having different center and radius, to the
extent that it could also have different radius at two ends of the curve. Because of this limitation
of insufficient boundary conditions, an analyst has no other choice but to resort to trial and error
to locate the center and radius of the circular arc that again would hopefully provide a FOS that
would be a minimum.
Present paper attempts to propose a method that can circumvent many of the problems that
are cited above and come up with a more rational basis of analysis.

PROPOSED METHOD
We start with a simple case shown in Figure 1 before digressing to more complex and
practical cases.
Shown in Figure 1 is a sloped ground resting on a firm base/foundation. Experience shows
that when resting on firm base, failure usually occurs through the toe at point A and becomes our
first basis of analysis (toe failure is the most common occurrence with embankments).
Now as per Figure 1 through point A, if we draw a number of lines like AF, AE, AD, AC
etc., subtending an angle θ with horizontal (0 < θ < α ) , there would invariably be one plane
among them say AE through which probability of failure is maximum. In other words, factor of
safety of its stability against sliding down the path AE (say) is a minimum.

B C D E F

α
A θ Firm Base

Figure 1: A sloped ground resting on firm soil.

Assuming the failure plane a straight line is valid as the factor of safety (FOS) is represented
by
FOS = FR FD (2)
 Vol. 18 [2013], Bund. V 5355

where FR = resistive force along plane AE; FD = driving force along plane AE.
Thus, as per Equation (2) for FOS to be a minimum, FR has to be a minimum. Again for FR to
be a minimum between point A and E, path of least resistance has to be a straight line.

B E(xe,H)
θ
180-α

H
α
A θ Firm Base
(0,0)
Figure 2: Slope with weakest plane AE making an angle θ with horizontal.

As shown in Figure 2, let AE be the weakest plane making an angle θ with horizontal. For an
unstable situation, soil mass made up of ∆ ABE will have a tendency to slide down the path AE.
Phenomenon explained above can then be modeled as shown in Figure 3.

E
FD= W sinθ
W cos θ

FR W
A θ
B

Figure 3: A body of weight W sliding down a wedge of angle θ.

From Figure 3 we see that soil above plane AE of weight W tries to slide down the triangular
wedge having an inclination θ.
Based on basic principle of mechanics the driving force FD can be expressed as

FD = W sin θ (3)

For general c-φ soil the resistive force can then be expressed as

FR = µW cos θ + cL (4)

Here L = length of the line AE and can be expressed as L = H.cosecθ with µ = tanφ, where φ is
friction angle of the soil; c = cohesion property of the soil, H = height, EB.
 Vol. 18 [2013], Bund. V 5356

The factor of safety (FOS) can then be expressed as

µW cos θ + cL
FOS = (5)
W sin θ

and this should be greater than 1.0 for the soil mass to be stable.
Referring to Figure 2, weight of soil constituting the triangle ( ∆ ABE) can be expressed as

1 sin(α − θ )
W = γ sH 2 (6)
2 sin α sin θ

where γs = unit weight of soil, and H = height of embankment.

Substituting Equation (6) in Equation (5), the FOS can finally be expressed as

 H c  sin α 
µ cos θ +   
 H  sin(α − θ ) 
FOS = (7)
sin θ

In Equation (7), Hc= 2c/γs , the critical free standing height as defined by Taylor.
Having established Equation(7) one can now take a value of θ say starting from 2o and
progressively increase it to the angle α (moving counter clock wise in Figure 2) to obtain FOS for
different values of θ and find out the critical value of θcr for which FOS is minimum. FOS for
various values of alpha is shown in Figure 4.

Figure 4: FOS curve for a slope various values of alpha

A typical FOS curve for different values of θ based on Equation (7) for a soil profile of slope,
α = 35o, height H =15m having φ = 26o, c = 10 kN/m2 and unit weight γs = 18 kN/m3 is shown in
Figure 5. The graph shows that critical plane AE is at θ = 26o with horizontal, where FOS is
minimum (FOS=1.62).
 Vol. 18 [2013], Bund. V 5357

FOS linear along linear plane

18.000
16.000
14.000
12.000
FOS(linear)
10.000
FOS

8.000
6.000
4.000
2.000
0.000
00

48

95

43

8
.9

.3

.8

.3

.8

.2

.7

.2

.7

.1
2.

4.

6.

9.
11

14

16

19

21

24

26

29

31

34
Theta in degrees

Figure 5: Typical FOS curve for a slope of 35o

Now based on Equation (7) and Figure 5, if FOS would have been less than 1.0, the slope
should have failed along straight line AE as it provides the path of least resistance. Yet,
observation paradoxically shows that in most cases failure takes place along a curvilinear path.
The query that arises then is, why does the failure follow a curvilinear path when logically it
should have followed a straight line?
This puzzle is not difficult to explain. When the body is in static equilibrium, stability of the
body is indeed maintained through resistance incurred along plane AE. However when it becomes
unstable and mobilizes motion, it seeks out a path within the body ABE along which, it can come
down in shortest possible time – in other words, it tries to follow the “Brachistochrone curve”.
The curve of steepest descent otherwise known as the Brachistochrone Problem remains one
of the most challenging problems in realms of mechanics and advent of variational calculus. Thus
before proceeding further, it would perhaps not be too much out of place to outline (for clarity)
the problem itself, and also narrate briefly the fascinating history behind it when some of the
unquestionable geniuses of mathematics and physics waged their intellectual battle against each
other to unravel this problem.

THE BRACHISTOCHRONE PROBLEM

As we describe today, the problem is exactly 316 years old. The word Brachistochrone is
actually a derivative of two Latin words Brachisto (Shortest) + Chrono (Time).
On 24th December 1696, trying to establish the supremacy of Swiss and German
mathematicians, Johann Bernoulli challenged all the mathematicians in Europe (including Sir
Isaac Newton from England) by posing a problem he asked them to solve (Johann has previously
solved this problem himself based on Fermat’s principle), stating that “it would only be the best
and cleverest of the mathematicians who would be able to solve it”. He gave one year time i.e. up
to Christmas, however later extended the challenge till the Easter next year.
The problem was that “if a frictionless body moves from a point E to A (vide Figure 6) in a
gravitational field, what is the path it will take to come down at shortest possible time?”
 Vol. 18 [2013], Bund. V 5358

E
Direction of motion

Which one is the path?

Figure 6: Brachistochrone Problem as posed by Johann Bernoulli (1697)

The problem though apparently looks quite innocent is not so easy to solve (Chandrashekar,
2012). There were only four mathematicians then in the whole of Europe who could solve it
correctly. They were,
1) Jacob Bernoulli (Johann’s brother), 2) Gottfried Leibniz (from Germany),
3) Isaac Newton (from England) and 4) L’Hopital(from France) (what a crowd!!).
All their individual solutions where published in Acta Eruditorum, the most prestigious
scientific German journal of that time (Bardi, 2006). It is said that Newton received the challenge
letter from Bernoulli as late as 16th January 1698, and it took him only one night to come up with
the correct answer – that the path would be a cycloid. Newton sent the solution to Bernoulli albeit
in a terse manner anonymously (On seeing the solution it is said that Bernoulli stated “It has to be
Mr. Newton, I can recognize the lion by its paw”).
Though Newton et al. solved the problem for mass moving along a friction less surface the
problem becomes immensely complex when friction is involved ( Leonhard Euler tried to solve
this problem but could not come up with an exact answer).
Parnovsky (1997) showed that considering friction along surface AE, the curve does not
remain a cycloid but become something different.
Having described the Brachistochrone problem briefly above, we hope one can foresee some
congruence with problem we are discussing herein, as this problem in classical mechanics
becomes a backbone to our solution to the problem in hand.

THE SOLUTION
Shown in Figure 7 is a sloped surface where based on Equation (7), say we have identified
the weakest plane AE making an angle θ with horizontal. Now considering point A as origin i.e.
A (0,0), coordinate of point E can be expressed as (xe, H) where xe = H cotθ. Here obviously H
and θ are known a priori.
 Vol. 18 [2013], Bund. V 5359

B E1 E(xe,H)

Y
y = mx y =kxn
H
δs
α X
θ ydx

A(0,0) F(xe,0)

Figure 7: Critical path of failure of the slope along a curvilinear profile.

Now having some foreknowledge of the works mentioned, we can infer that failure takes
place along a curve AE rather than a straight line AE, though what is the equation of this curve
we cannot say for sure (at the moment) as both friction and cohesion is involved in it. However
we know that the curve would pass through points A (0,0) and E(xe,H).
Considering the straight line AE, its general Equation can be expressed as,
y = mx + p (8)
here m is the gradient of the line.
Considering it passing through A (0,0), we have p = 0 thus,
y = mx where m = H xe or simply tanθ.
Proceeding in analogous manner let us assume equation of the curve along AE as
y = kx n + p where n ≥ 1 and can be a real number or an integer. Considering it passes
through A(0,0) → p = 0 when we have,

y = kx n (9)

Considering it passes through E(xe,H), it satisfies Equation (9) when,

k = H xe (10).
n

We now re-look at Equation (5) and simplify it to,

cL
µ cosθ +
FOS = W (11)
sin θ

Further, to start with we ignore the cohesion effect for the time being (this will be plugged in
at a later stage) when Equation (11) can be expressed as,
 Vol. 18 [2013], Bund. V 5360

µ cosθ
FOS = (12)
sin θ

Equation (12) is valid when the plane is linear and does not hold good for a curvilinear path
as because the gradient becomes variable from point to point. However if we take an
infinitesimally small segment δs on curve AE as shown in Figure 6 supporting a weight say δW
we can represent this segment as shown in Figure 8.

δs
δy
δθ X
δx

Figure 8: Representation of the segment δs in Cartesian coordinate.

Based on Figure 8 as the segment δs is infinitesimally small it may be considered linear and
one can say that,

=
lim δx → 0 ds dx 2 + dy 2 (13)

2
 dy 
→ ds = 1 +   dx (14)
 dx 

Equation (12) can then be represented as

µ cos dθ
d (FOS ) = (15)
sin dθ

Equation (15) is valid on segment ds as it is considered linear.

Now based on Equation (15) and Figure 8, we have,

dx
µ
→ d (FOS ) = ds (16)
 dy  dx 
  
 dx  ds 

Now considering Equation (9) we have,

 Vol. 18 [2013], Bund. V 5361

dy
= knx n −1 (17)
dx

From Equation (14) we can say that,

ds
= 1 + k 2 n 2 x 2 ( n −1) (18)
dx

Thus Equation (16) which is valid over the segment ds only can be expressed as,

1
µ
→ d (FOS ) = 1 + k n x 2 ( n −1)
2 2
(19)
knx n −1
1 + k 2 n 2 x 2 ( n −1)

To determine FOS over the full curve AE we have to integrate Equation (19) between xe to 0
(Figure 7) which gives,

xe
dx
µ∫
0 1 + k 2 n 2 x 2 ( n −1)
FOS = (20)
knx n −1
xe

∫
0 1 + k 2 n 2 x 2 ( n −1)
dx

To Equation (20) we now plug in the cohesive effect when we have

xe 
dx cL
µ∫ + 
0 1 + k 2 n 2 x 2 ( n −1) W
FOS = (21)
knx n −1
xe

∫
0 1+ k n x 2 2 2 ( n −1)
dx

For linear path AE, L can be expressed as L= xe/cosθ. Thus for the curvilinear path AE,
 xe
L = xe ∫ 1 + k 2 n 2 x 2 ( n −1) dx , which finally gives,
0
 Vol. 18 [2013], Bund. V 5362

xe xe
dx cx
µ∫ + e ∫ 1 + k 2 n 2 x 2 ( n −1) dx
2 2 2 ( n −1)
0 1+ k n x W 0
FOS = xe n −1 (22)
knx
∫
0 1 + k 2 n 2 x 2 ( n −1)
dx

Here, referring to Figure 7,


W = weight of soil mass constituting ∆ AEB plus weight of soil mass that lie within the line
AE and curve AE. This gives,

2 sin(α − θ )
 1 1
xe
W = γ sH + γ s xe H − γ s ∫ ydx (23)
2 sin α sin θ 2 0

Equation (23) on simplification gives

 1 sin(α − θ ) 1  n − 1
W = γ sH 2 + γ s xe H  (24)
2 sin α sin θ 2  n + 1

As an engineer, one can well stop the solution at this point stating that the integrals in
Equation (22) can be solved numerically and considering n =1,1.1,….1.5,2 etc., one can find out
the value for which FOS will be a minimum. A typical trial and error method deployed and would
possibly meet the approval of many geotechnical engineers.
However putting the hat of a physicist we state that:
For FOS to be a minimum, value of n has to be unconditionally 2. This shall be valid
irrespective of if the soil is homogenous, layered or otherwise.
Having made somewhat an audacious statement above, it surely needs some clarification.
However, before we delve into this explanation it would be worth evaluating what happens to
Equation (22) when n = 2?
For n = 2, Equation (22) can be expressed as

xe xe
dx cx
µ∫ + e ∫ 1 + 4k 2 x 2 dx
0 1 + 4k 2 x 2 W 0
FOS = (25)
2kx
xe

∫
0 1 + 4k 2 x 2
dx

Surprisingly the integrals in Equation (25) are all standard integrals that can be solved
explicitly when we have
 Vol. 18 [2013], Bund. V 5363

  2H + x 2 + 4H 2    2H + x 2 + 4H 2 
xe 2   cxe  1 xe 2  
µ log e  e
+ 
 W  2 e x 2
+ 4 H 2
+ log e
e

 2H  xe 4H  xe
FOS =      
(26)
xe xe 2 + 4 H 2 − xe 2
2H

Equation (26) will give minimum factor of safety among all the curves possible between
points A and E.
However, computation shows that Equation (26) may not always be the absolute
Brachistochrone plane, especially when the soil is cohesive having high value of c. For
predominantly cohesive soil, depending on its value, the soil body seeks a slightly different plane
to come down in shortest possible time. This makes it slightly move to left of the plane AE to
AE1 [vide Figure 7], thus in the process increasing the angle θ, that increases the driving force
(Mg.sinθ) and also minimizes the resistance by shortening the curve length L, through which cL
gives resistance__ though character of curve still remains y = kx2.
For pure cohesion less soil (i.e. c = 0) the weakest plane AE for linear and curvilinear path
remains unaltered. It is only for general c-φ soil as explained above it can well be a little different.
However as a first trial seeking θAE is critical, as because it becomes an excellent starting point
for search of the true Brachistochorne plane for c-φ soil, and would converge to it rapidly by
varying xe towards xB as it usually lies within about 3 to 4 degree to the left of θAE.
The above phenomenon is explained further by Figure 9 below (the scale is exaggerated for
clarity).

B E1 E

FOS=4.067

FOS=2.658
FOS= 2.552 (Brachistochrone Plane)
A X

Figure 9: Variation of FOS for predominantly cohesive soil having

φ = 4o and c = 20 kN/m2; embankment slope 1:2.
Figure 9 shows variation of FOS for a typical embankment of slope 1:2 having φ = 4o and c =
20 kN/m2. It is observed that the curvilinear path has made a shift from AE to AE1. By Equation
(7) minimum FOS in this case is 4.067 at θΑΕ =13.67o. The corresponding FOS by Equation (26)
at θΑΕ =13.67o is 2.658.
 Vol. 18 [2013], Bund. V 5364

The true Brachistochrone plane however actually lies at θΑΕ1=16.74o (i.e.3.07o to the left of
AE), whereas per Equation (26) the FOS is 2.552 which is the minimum FOS for the problem.
For cohesion less soil the Brachistochrone plane does not move from AE to AE1.

Why n has a unique value of 2 for FOS to be a minimum

along any plane AE?

To this end let us consider Figure 10.It shows a body of mass say M that is moving down a
curved path y = kxn in a gravitational field. The path is assumed frictionless.
Now at point A from where the body starts its potential energy is PE = MgH and kinetic
energy KE = 0. When the body reaches point O (0,0) which is origin of the Cartesian system,
potential energy of the body PE = 0 and kinetic energy KE= ½ Mv2, where v = 2 gH .

B Y A
M

y=kxn
H

C
X
O(0,0)

Figure 10: A body moving down the curve y= xn in a gravitational field

On reaching point O it must travel along the path OC till it reaches point B before coming to
rest when height of OA and OB are equal. This has to be followed to conform to the law of
conservation of energy. Now if there is friction and some cohesion along the path, the body
beyond point O will move up to somewhat lesser distance than B but energy has to be still
conserved in terms of KE at C = Potential energy at A - work done by the friction and cohesive
force along path AOC.
Based on above one can conclude that for the body on reaching O to move up the path OB
represented by the Equation y = kxn, n must have such a value, that for any value of x<0, y must
have a value >0 i.e. positive.
 Vol. 18 [2013], Bund. V 5365

One can immediately conclude that n cannot be an odd integer. For any odd value of n when
x<0, y is also < 0 and this flouts the law of conservation of energy, which is not possible.
For n a rational number like 1.1,1.2,1.5 etc there are no real solution for x<0.Either y
becomes a complex number or something undefined.
Thus the condition of conservation of energy is only satisfied when n is an even integer. Of
all the even numbers 2,4,6.. etc n = 2 is the least and would give the shortest path and steepest
gradient. Thus a soil body when under motion in a gravitational field from a point E to A (vide
Figure 7) will unconditionally seek a path expressed by y = kx2 to satisfy the law of conservation
of energy.
One can also conclude from above that the condition is not dependent on the character of
mass like whether it is homogeneous or other otherwise. To satisfy the law of conservation of
energy the curve is indeed unique.
The concept explained above is produced graphically in Figure 11.

Odd and even curves of y=kx^n

40
30
20
y=kx^2
10
y=kx^3
0
y

y=kx^4
-10
0

0
10

20

30

40

50

60

y=kx
-6

-5

-4

-3

-2

-1

-20
-30
-40
x

Figure 11: Odd and even curves of y = kxn

Figure 11 clearly shows that beyond origin the curves y = kx and y = kx3 is in violation to
principle of conservation energy. And among the even curves for n = 2, the path is shortest
having steepest gradient.

Effect of pore pressure and hydrostatic force

In many cases like banks of unlined canals, natural slopes along rivers edges, earth dams etc,
hydrostatic effect of water and pore pressure plays a significant role on the stability factor. This
issue though remains a bit uncertain in terms of mathematical modeling. First, measurement of
pore pressure in situ is not an easy task and in many a cases are co-related with other soil
parameters that may or may not be realistic to the actual field condition.
Finally, two most popular techniques that are in vogue for stability analysis (Fellenius and
Bishop), overestimate the driving force and underestimates the resistive force respectively. Under
high pore pressure both gives a lower bound value of FOS (especially Fellenius).
 Vol. 18 [2013], Bund. V 5366

So before we delve into the topic it would be worth having a look into the Equations of
Fellenius and Bishop to see where the snag is.
As per Fellenius the factor of safety is expressed as,

cL + tan φ ( N − U )
FOS= (27)
W sin θ

where c = cohesion of soil; tanφ = friction angle of soil; N = normal reaction on plane θ; U =
force due to pore pressure in soil; W = weight of soil within the slip circle.
In Fellenius’s formulation Wsinθ is represented as Σ T where Σ T.R is the actuating moment.
In Equation (27), while Fellenius takes reduction in resistive force due to pore pressure, does not
take into account the reduction in driving force that takes place especially under steady seepage
condition to it due to buoyancy. This gives a lower value of FOS (as the numerator reduces while
the denominator do not) under certain conditions asking for a far more conservative vis-à-vis
expensive design.
While Bishop takes into cognizance the reduction in driving force due to buoyancy, it
however fails to consider the additional hydrostatic pressure that acts in tandem to the pore
pressure that goes on to reduce the resistive force for those slices below water line (Terzaghi and
Peck, 1982). Moreover it has been observed that for low factor of safety on highly cohesive soil
having large pore pressure, the normal thrust on slice base sometimes becomes negative or zero
as per Bishop’s formula and fails to come up with a solution.
In terms of the proposed method as shown in Figure 12, let Z-Z be the water level along an
unlined canal embankment say. It is obvious that due to seepage the water level within the slope
and that in the canal must be same under steady state (as water seeks it own level). Let the water
line cut the embankment slope and the failure curve at point C and D respectively.

B (Hcotα,H) E(xe,H)
Water level C (Hwcotα,H) D
Z Z
(Hw -y)
H Hw
Y
y
α θ

A(0,0) X
Figure 12: Slope with water under steady seepage.

Under this condition the shear stress τ along the curved surface AE is expressed as,

τ = c + (σ N − γ w ( H w − y ) − u )µ (28)
 Vol. 18 [2013], Bund. V 5367

W + W  
→ τ = c +   d  b  cosθ − γ w ( H w − kx 2 ) − u  µ (29)
  LAE  

where Wd = dry weight of soil above water line and Wb= submerged weight of soil below water
line.

Now multiplying each of the term by its influencing length over arc length L we have,

    Wd + Wb     
τ × LAE = cLAE +     LAE cosθ − γ w ( H w − kx 2 ) LAD − uLAD  µ
 (30)
  LAE  

Thus resistive force FR is expressed as,


(   
FR = cLAE + (Wd + Wb )cosθ − γ w H w LAD + kx 2 LAD − uLAD µ ) (31)

The driving force is expressed as,

FD = (Wd + Wb )sin θ (32)

Hence,

FOS=

(   
cLAE + (Wd + Wb )cosθ − γ w H w LAD + γ w kx 2 LAD − uLAD µ ) (33)
(Wd + Wb )sin θ

Considering We = Wd+Wb the effective weight, factor of safety can be expressed as,

 γ H  γ k  u   c 
 cosθ − w w LAD + w x 2 LAD − LAD  µ + LAE
FOS =  We We We  We
(34)
sin θ
In terms of proposed method, Equation (34) can be expressed as,

xe xd xd xe xd
µdx  µγ H x   µγ kx  2 µuxd
∫ ∫ ∫ ∫ ∫ f ( x)dx
cx
−  w w d  f ( x)dx +  w d  x f ( x)dx + e f ( x)dx −
f ( x)  We 0  We 0 W e We
FOS = 0
xe
0 0
(35)
2kx
∫
0
f ( x)
dx

Here, f ( x) = 1 + 4k x , xd = xe ψ , ψ = H w / H
2 2

Equation (35) can be expressed as,

 Vol. 18 [2013], Bund. V 5368

γ H   µγ k  c µx xd
µI1 − µ  w w  I 2 +  w  I 3 + I4 − d ∑ u ( x) f ( x)
FOS =  We   We  We We 0
. (36)
I5

It will be observed that in Equation (36) pore pressure coefficient has been represented as a
numerical summation rather than an integral. This is because in most cases these values are
obtained and collated based on field observed data. Thus for practical application it is preferable
that this be separately integrated numerically and finally plugged into Equation (36). Terms I1 to
I5 are integral factors that can be derived explicitly in terms of xe and are as follows:

 2H + x 2 + 4H 2 
xe 2
I1 = log e  e  (37)
2H  xe 
 

  2 H ψ + x 2 + 4 H 2ψ 
xe 3  2 Hψ 2 2  
I2 = xe + 4 H ψ + ψ log e   (38)
e
4 H  xe 2  xe
  
  2 H ψ + x 2 + 4 H 2ψ 
( )
3
xe 7 ψ  4 H ψ 2H ψ  
I3 = xe 2 + 4 H 2ψ 2 − xe 2 + 4 H 2ψ − log e  e
 (39)
64 H 3  xe 4 xe 2  xe
  
  2H + x 2 + 4H 2 
xe3  2 H  
I4 = + log e
e
 (40)
4 H  xe 2  xe
  

x e x e 2 + 4 H 2 − xe 2
I5 = . (41)
2H
Having established the integral factors, only unknown that need to be identified is We.
Here area within the curve ABEA vide Figure 12 is expressed as,
1 2 sin(α − θ ) 1
HA1 = + xe H (42)
2 sin α sin θ 6
1  sin(α − θ ) 1 
→ A1 = H 2  + cot θ  (43)
2  sin α sin θ 3 
By proportion area within curve ACDA below water line is expressed as,
1 2  sin(α − θ ) 1 
A2 = Hw  + cot θ  (44)
2  sin α sin θ 3 
Thus area of curve above ACDA and below line AE can now be expressed as,

Anet =
1 2
2
(
H − Hw
2
) sin(α −θ) 1 
+ cot θ  (45)
sin α sin θ 3
 
Thus dry weight portion of the soil can now be expressed as,

Wd =
γs
(H 2
− Hw )
2  sin(α − θ ) 1 
+ cot θ  (46)
2  sin α sin θ 3 
Submerged or buoyant weight of soil is expressed as,
 Vol. 18 [2013], Bund. V 5369

(γ sat − γ w ) H 2  sin(α − θ ) 1 
Wb =
2
w  sin α sin θ + 3 cot θ  . (47)
Thus,
γ s H 2  sin(α − θ ) 1    γ sat − γ s − γ w  2 
We = + cot θ  1 +  ψ  . (48)
2  sin α sin θ 3   γw  
Now considering xe = H cot θ , and varying θ as 0 ≤ θ ≤ α one can easily find out the
Brachistochrone plane which gives the minimum FOS from Equation (36).
A point may be noted at this juncture. In Equation (36), considering the FOS is only in one side
of the Equation unlike Bishop’s Equation (which has FOS on both sides of the Equation) no trial
and error is required. And for high value of cohesion as well as pore pressure the solution does
not become unstable. Thus other than being computationally more efficient it well circumvents
the problem as faced with Bishop’s formula.

Case of deep seated failure

This is the other common occurrence of slope failure when an embankment is underlain by
weaker soil.

B(xb,H) C

1
H

A(0,0)

h 2 Weak layer of soil

D Firm ground from here

Figure 12: Typical deep seated failures in embankments.

Figure 12 shows typical deep seated failure in embankments when underlain by weaker soil.
Observation in field has shown that bottom plane of weaker soil usually becomes a tangent to the
failure curve at point D which is exactly at midpoint of slope AB (Fellenius 1937, Smith 2013).
To start with we only know the coordinates of point A and B as shown in Figure 12.
Considering bottom line of weaker section is the datum line i.e. here potential energy is zero, as
such we shift our reference system from A to point D, when based on shift of axis the new co-
ordinates are shown in Figure 13.
 Vol. 18 [2013], Bund. V 5370

C(xc,H+h)

W2 W1 H
1 DH
A α
G(0,h) F
2
Y h
θ Weak layer of soil
D(0,0) X Firm ground from here

Figure 13: Failure profile deep seated failure of embankment.

Based on Figure 13 it is apparent that point D is the origin of failure profile.

In this case the driving force FD is again Wsinθ along line CD. The resistive force FR may be
expressed as,

FR = µW cos θ + c.DC (49)

The factor of safety can thus be expressed as,

µW cos θ + cL
FOS = (51)
W sin θ − Pp

Here Pp= Resistive Force of soil beyond point D.

Equation (51) based on coordinates can be expressed as,

FOS =
( )
µ1W1 cos θ1 + µ 2W2 cos θ 2 + c1 xc − h k sec θ + c2 h k sec θ
(52)
W1 sin θ1 + W2 sin θ 2 − Pp

The coordinates in terms of known variables are tabulated in Table 1 for ease of computation.
 Vol. 18 [2013], Bund. V 5371

Table 1: Coordinates of Different points in the embankment.

Point x y Remarks
D 0 0
C xc HT xc is variable
B Hcotα/2 HT HT=H+h
E 0 ½(HT+h)
F xc h H T h
G 0 h
H xc(h/HT) h

It will be observed now that in Equation (52) the only unknown variable is xc. Now starting
with an arbitrary value of xc well away from point B, if we now keep on decreasing it in steps of
say half a meter then based on Equation (52) we will get one unique value xc, for which FOS will
be minimum and will define the weakest plane CD, the limiting boundary condition being that
while moving xc towards B, point I shall not move beyond B, i.e. h k ≥ (H / 2 ) cot α .

Factor of safety in terms of the curvilinear path CD can then be expressed as,

xf xf
 h 
xc xc

∫ ∫ ∫ ∫
dx dx h
µ1W1 + µ 2W2 + c1  xc − 1 + 4k 2 x 2 dx + c2 1 + 4k 2 x 2 dx
 
xf 1 + 4k 2 x 2 xd
2 2
1 + 4k x  k  xf k
xd
FOS = xf
xc
 2kx  2kx
W1 ∫
xf 1 + 4k x2 2
dx + W2 ∫
xd 1 + 4k 2 x 2
dx − Pp

(53)
Putting the exact coordinates we have,
h h
 xc   h
k xc k
dx dx h
µ1W1 ∫ + µ 2W2 ∫ + c1  xc −  ∫ 1 + 4k 2 x 2 dx + c2
 ∫ 1 + 4k 2 x 2 dx
2 2 2 2
h 1 + 4k x 0 1 + 4k x  k h k 0

FOS = k k

2kx
hk
  2kx
xc
W1 ∫ 1 + 4k x 2 2
dx + W2 ∫
0 1 + 4k 2 x 2
dx − Pp
hk

(54)
Equation (54) can be expressed as,
   h h
µ1 I1W1 + µ 2 I 2W2 + c1  xc −  I 3 + c2 I4

k k
FOS =  (55)
(W )I
1 5 + (W2 ) I 6 − Pp

Here the integrals I1,I2 ….I6 can be explicitly expressed as,

 2kx + 1 + 4k 2 x 2 
1
I1 = log e  c c  (56)
2k  2 kh + 1 + 4kh 
 
 Vol. 18 [2013], Bund. V 5372

I2 =
1
2k
(
log e 2 kh + 1 + 4kh )
(57)

  2kx + 1 + 4k 2 x 2 
kx 1 + 4k 2 xc 2 − kh(1 + 4kh ) 1  
I3 =  c + log e  c c
 (58)
 4k 4k  2 kh + 1 + 4kh
  

I4 =
1
4k
[ (
2 hk (1 + 4kh ) + log e 2 kh + 1 + 4kh )]
(59)

1 
I5 =  1 + 4k 2 xc 2 − 1 + 4kh 
2k   (60)

I6 =
1
2k
[ 1 + 4kh − 1] (61)

Where k
= ( H + h ) / x c2 (62)

 
In Equation (55) W1 and W2 are as elaborated hereafter,

 2 h 
W1 = γ s1  xc (H + h ) − xc H 
 3 H + h 
(63)

 h γ H2 2 h
W2 = γ s1 xc H − s1 cot α + γ s 2 xc h (64)
H +h 8 3 H +h

Equation (55) thus gives the minimum FOS in this case with lim xi → xb
Here,

2 2
h γ s1K p 2 H γ s2 K p2h
Pp = µ 2W p I 2 + W p I 6 + c2 xc I4 + cot α + + 2c2 h K p 2
H +h 8 2 (65)

where,

2 h γ H2
Wp = γ s 2 xc h + s1 cot α and K p 2 = (1 + sin φ2 ) (1 − sin φ2 ) (65a)
3 H +h 8

Two practical cases often faced by engineers

Having solved two base cases constituting slope stability we now try to provide solution of
two practical cases often faced by engineers especially designing industrial plants that puts them
into significant problem.
 Vol. 18 [2013], Bund. V 5373

The problems are pictorially shown in Figure 14. In many cases due to the local topography
or for the reason of flood protection, plants are build on embankments (i.e. higher elevation than
the surrounding) to ensure their smooth operation. One major concern then is slope stability of the
embankment on which the plant is being built that generates a surcharge load over it.

Backfilled soil

Original contour

Figure 14: Embankment with surcharge load on surface from plant and point load due
to foundation.

The situation is as shown below in Figure 15.

q E(xe,H)
B

Y y = mx
H
y =kx2
α θ X
A(0,0) F(xe,0)

Figure 15: Embankment with surcharge load.

In this case the only change is, Equation (24) that gets modified to,

 1  sin(α − θ ) 1  n − 1
W =  γ s H 2 + qH  + γ s xe H  (66)
2  sin α sin θ 2  n + 1

where q = surcharge load on embankment @ kN/m2 per meter width.

Rest of the steps remains same as explained before.
Let us now consider the second case as shown in Figure 16. The case cited here, one often gets
confused with positional vector of the load and would often try to seek a solution with FEM
 Vol. 18 [2013], Bund. V 5374

(often with half-baked data at his disposal) and finally get totally confused with output
interpretation. Aesthetically pleasing stress contour are nice to look at, but does not always
answer the bottom line “What is the FOS?”.

B K P1 J(x ,H) E(xe,H)

j
P2

I(xi,yi) y = mx
Y
H
y = kx2
α θ X
A(0,0) F(xe,0)

Figure 16: Embankment with Point load on surface and within the body.

It should be remembered that our basic assumption in this case is that the triangular block
ABE fails as a rigid body where positional vectors of P1 and P2 has no bearing other than it only
goes on to add additional force to the whole soil body whose sine and cosine component enhances
the driving and resistive force respectively.
Yes, it is possible that if P1 or P2 are of high value a local failure could occur through line AJ
or AK (vide Figure 16) and can well be checked in terms of straight line basis as a preliminary
check but for most of the cases where the load is moderate (compared to the weight of mobilized
soil mass) the failure line AE will constitute of a plane where both P1 and P2 will also contribute.
In this case, resistive force is expressed as,

FR = µW cosθ + µP1 cosθ + µP2 cosθ + cL (67)

Driving force is expressed as,

FD = W sin θ + P1 sin θ + P2 sin θ (68)

Thus factor of safety is expressed as,

µW cosθ + µP1 cosθ + µP2 cosθ + cL

FOS = (69)
W sin θ + P1 sin θ + P2 sin θ

 P1 P2  cL
µ 1 + +  cosθ +
W W
FOS = 
W
(70)
 P P 
1 + 1 + 2  sin θ
 W W
 Vol. 18 [2013], Bund. V 5375

P1 P2
Considering λ = 1 + + a dimension less term we see that both friction and acceleration
W W
component gets increased by a factor λ which finally gives,

cL
µλ cosθ +
FOS = W (71)
λ sin θ
Equation (70) is valid for the plane AE. For the curvilinear path Equation (70) can be expressed
as,

 cL
µλ cosθ + 
FOS =  W (72)
λ sin θ

Here W is as expressed by Equation (24),

 P P
and λ = 1 + 1 + 2 (73)
W W

Substituting Equation (73) in Equation (26) we finally have,

  2 2    2 2 
 xe 2  2 H + xe + 4 H  cxe 2
 1 x 2 + 4 H 2 + xe log  2 H + xe + 4 H 
µλ  loge   +  2 e e 
2H  xe  W 4H  xe 
FOS =      
  . (74)
 xe xe 2 + 4 H 2 − xe 2  
λ
 2H 
 

Procedural Steps for computation of minimum FOS by the

proposed method

Procedural steps to be adapted by the proposed method can be summarized as hereafter.

1. Select the Height (H) and slope (α) of the embankment.
2. Read out the soil properties c and φ from soil report.
3. Compute the critical plane θAE from Equation (7) where 0 ≤ θ AE < α for which FOS is a
minimum.
4. Compute xae = H.cotθAE as the critical distance.
5. Refine the FOS now by adapting Equation (26) starting with xae as obtained in step 4 and
decreasing x AE → x AE − ε ≤ xB to obtain the minimum FOS for the problem. Here ε is
 Vol. 18 [2013], Bund. V 5376

decreasing step that may be chosen by the analyst. The convergence usually takes place
within 3 to 4 degrees of shift to the left from θAE if the soil is c-φ or c type. For sandy soil
θAE is the Brachistrchrone plane and no further refinement is required.

For base failure the steps to be followed are furnished hereunder.

1. Select the Height (H) and slope (α) of the embankment and also the height h of
weaker soil lying beneath
2. Determine xB = ½ H cotα and other relevant co-ordinates from Table 1 except co-
ordinate C where xc remains a variable.
3. Obtain the weakest plain CD from Equation (30) by varying xc for a minimum FOS
ensuring xi ≤ xb which is the limit.
4. With xc obtained in step 3 arrive at the minimum FOS vis-à-vis the true
Brachistochrone plane of the problem.
For other two problems the steps are same as mentioned above except that appropriate FOS terms
are to be used.

RESULTS AND DISCUSSIONS

To validate the results, a real life problem has been chosen.
An embankment of height 5 meters having soil unit weight of 18.0 kN/m3 and has a slope of
1:2(V:H). The FOS has been compared for different values of c / γ s H and φ with data provided
by Bishop and Moregenstern (1960). The embankment actually also had a surcharge of 20 kN/m2
due to built up of an oil and gas plant. The problem was analyzed further in the general purpose
software SLOPEW and compared with proposed method in Figures 17, 18 and 19.
In Figures 20 and 21, base failure with D/H ratio greater than 1.0(vide Figure 12) is
considered and compared with data of Bishop and Morgenstern (1960).
The results are as depicted hereafter.

Variation of FOS for c/gH=0 Variation of FOS for c/gH=0.025

2.5
2
1.8
1.6 2
1.4
1.2 1.5 Bis hop &
Bishop &
FOS
FOS

1 Morgenstern Morgens tern

Proposed method 1 Propos ed m ethod
0.8
0.6
0.4 0.5
0.2
0 0
20 25 30 35 40 20 25 30 35 40
Phi Phi

Figure 17: Comparison of FOS for toe failure c/γH=0.0 and 0.025.
 Vol. 18 [2013], Bund. V 5377

Variation for FOS for c/gH=0.05

Variation of FOS for c/gH=0.075
3
3.5
2.5 3
2 2.5
Bis hop &
FOS

Morgens tern 2 Bis hop &

1.5

FOS
Morgens tern
Propos ed m ethodi 1.5 Propos ed m ethod
1
1
0.5 By SlopeW
0.5
0 0
20 25 30 35 40 20 25 30 35 40
Phi Phi

Figure 18: Comparison of FOS for toe failure c/γH=0.05 and 0.075.

Variation of FOS for c/gH=0.1 Variation of FOS for c/gH=0.1, with reduced phi

3.5 3.5
3 3
2.5 2.5 Proposed
2 Bishop &
FOS

2 method(reduced phi)
FOS

1.5 Morgenstern
Proposed method 1.5 Bishop & Morgenstern
1 1
0.5 0.5
0 0
20 25 30 35 40 20 25 30 35 40
Phi Phi

Figure 19: Comparison of FOS for toe failure c/γH=0.1 peak φ vis-à-vis reduced φ.

Com parison of FOS for D/H=1.25 and c=2.25 kN/m 2 Comparison of FOS for D/H=1.25 and c=4.5 kN/m2

3.5
3.5
3 3
2.5 Bishop & 2.5 Bishop &
Morgenstern Morgenstern
FOS

2
FOS

2
1.5 Proposed Method 1.5 Proposed Method
1 1
0.5 0.5
0 0
10 15 20 25 30 35 40 10 15 20 25 30 35 40

Phi(degrees) Phi(Degrees)

Figure 20: Comparison of FOS for base failure for D/H=1.25 and c/γh= 0.025 and 0.5.
 Vol. 18 [2013], Bund. V 5378

Com parison of FOS base failure D/H=1.5 and c= 4.5 kN/m 2

3 Bishop &
FOS Morgenstern
2
Proposed Method
1

0
10 15 17.5 20 25 30 35
Phi(Degree)

Figure 21: Comparison of FOS for base failure for D/H=1.50 and c/γh= 0.5.

From Figures 17 to 21 it may be observed that for c/γh varying from 0.025 and 0.075 the
proposed results are in commendable agreement with Bishop and Morgenstern’s values. For
c/γh=0, 0.05 and 0.1 the proposed values are slightly more than that of Bishop. But this cannot be
construed as poor results as because it is an established fact that Bishop’s result are lower bound
by as much as 15% (Smith, 2013).
Average variance in each of the cases for c/γh=0, 0.05 and 0.1 are 6%, 7% and 12% upper
bound respectively. Thus it may be stated that results of proposed method are possibly a little bit
more realistic than Bishop.
However a point to be noted here which we believe is important. Soil having c/γh = 0.1 and
φ ≥ 20o is a rare occurrence in nature. For value of c around 9 to10 kN/m2 or more, the value of φ
is usually 3 to 4o. It is possibly in case of reclaimed land with some exceptional cases one would
encounter such type of soil. When soil has high cohesive value and also a high value of φ, when
incipient motion develops, strain induced in soil would be far higher than when a normal soil is
mobilized. In such cases it would be prudent to consider the residual friction value of the soil
rather than peak value which is normally used to calculate the factor of safety. Clause 2.2.4 of BS
8002(1994) code has furnished values residual φ against SPT value (N) and peak φ value of the
soil. Adapting this value the FOS was also calculated for c/γh = 0.1 and shown in Figure 19.
Average variance in this case for FOS over full range of φ is 0.37% only.
We finally show results of the same slope with surcharge of 20 kN/m2 having c = 4.5 kN/m2
that was run in the software SLOPEW (Using Moregenstern & Price(1965) technique) vis-à-vis
proposed method in Figure 22.
 Vol. 18 [2013], Bund. V 5379

2
Variation of FOS for sucharge Load 20 kN/m and gc/H=0.05

2.5

2
SlopeW
FOS

1.5
Proposed Method

0.5

0
20 25 30 35 40
Phi

Figure 22: Comparison of FOS for c= 4.5 kN/m2 and surcharge of 20 kN/m2.

Again results are found to be in commendable agreement with average variance over full
range of φ is about 6% lower bound with SLOPEW results.
For other cases, it is observed that proposed method gives results that are slightly upper
bound to Bishop, but in this case FOS computed by the proposed method is lower than the FOS
computed by traditional method. This apparent anomaly can well be explained by the fact that in
both Morgenstern and Bishop’s method numerical error is generated in the way it considers the
surcharge load.
In the conventional method of slice, the surcharge load is broken up into concentrated force
acting as a vertical load at the center of each slice. The number of points in which this uniformly
distributed load (UDL) is broken into is equal to the number of slices considered for the analysis.
Considering the failure surface is broken into finite number of slice (say n), it underestimates the
total resistive as well as driving force. The matter can be quantitatively explained further as
follows.
If we have a simply supported beam say of span Lb, with an UDL w, the maximum deflection
3
at center is given by δ = WLb / 76.8 EI (Here W = w.Lb).Now if in the same beam, the UDL is
broken up into 5 internal nodes (i.e. n = 5) having concentrated loads, whose sum is equal to W,
3
then maximum displacement in this case is δ n=5 = WLb / 79.36 EI . When n = 10, i.e. divided
3
into 10 nodes, δ n=10 = WLb / 77.5 EI . Thus we see that δ n=5 < δ n =10 < δ n = ∞ .This implies that for
finite number of nodes considering equivalent concentrated load in lieu of UDL, underestimates
the overall loading. Lesser the value of n, more the result is lower bound. It is only when n → ∞ ,
the values match.
In the same token, for finite number of slices, having surcharge UDL q over it, as the
numbers of slices considered are finite, it underestimates the load q.xe. It is only when numbers of
 Vol. 18 [2013], Bund. V 5380

slices tend to infinity the value will converge to q.xe. It is for this both Bishop and Moregensten
will give an upper bound solution in this case.
In the proposed method as it is not necessary to break up the UDL into equivalent
concentrated load the solution is exact and lower bound to the traditional methods- but possibly
more realistic.
One may reflect at this point that, despite its indubitable popularity, neither Bishop’s nor
Moregenstens’ method is sacrosanct. As such, one cannot definitively comment on superiority of
the proposed or the traditional method.
It would suffice to say that both the methods give comparable results. However with the
proposed method, a tangible advantage being, one need not do trial and error at every step and
convergence to the Brachistochrone plane is very fast when the analysis is carried out in a
computer.
It also explains logically based on the basics as to why the failure curve should be a second
order polynomial and not necessarily a circular arc that was conceived more as a convenience
rather than by design.

CONCLUSION
Having presented the proposed solution, we have possibly reinstated Abramson’s (2002)
prophecy that “Of all the available procedures, irrespective of the method one chooses, results do
not significantly improve over Bishop’s technique”.
So what makes the proposed method amelioration to the problem in hand?
The value addition can be summarized as below:
1. It puts forward a rational theory on characteristics of the failure curve based on classical
mechanics and proves that circular curve that was chosen as a convenience is not strictly
correct. That results match in certain cases is coincidental, and possibly due to the fact
that circular curve is also generically a curve of 2nd order.
2. It circumvents the trial and error techniques prevalent with slip circles, and comes up
with a logical basis of iteration that converges very fast once the weakest plane has been
identified.
3. The results are possibly slightly more realistic than Bishop.
4. It takes into cognizance deficiencies in both Fellenius and Bishop’s formula in terms of
pore pressure and also circumvents the issue of instability that occurs with Bishop’s
formula in certain cases.
5. There is no problem in catering to non homogenous layered soil when in lieu of
analytical integration the FOS values can be numerically integrated (though mostly
during practical application considering a weighted average of the soil data is usually in
vogue).
6. Considering the solution is direct if one wishes, can well workout the problem in a
spreadsheet, and no special software really needs to be developed.
Vol. 18 [2013], Bund. V 5381

REFERENCES
1. Abramson L.W, Lee T.S, Sarma S & Boyce G.M.(2002) Slope stability and Stabilization
Methods, John Wiley & Sons Inc. N.Y.,USA.
2. Bardi J (2006) Calculus Wars; High Stake Publication, U.K.
3. Bishop A.W. (1955); Use of slip circle in the stability of slopes, Geotechnique (5), 7-17.
4. Bishop A.W, Moregensten N.R.(1960) Stability coefficient of earth slopes,
Geotechnique, 10(4), 164-169.
5. Chen W.F, Giger M.W. Limit analysis of stability of slopes Journal of Soil Mechanics
and Foundation Engineering ASCE 97(SM1) 19-26
6. Chopra A.K, Clough R.W (1966) Earthquake stress analysis in Earth Dams, J.Engg
Mechanics ASCE Vol. 92, EM2 197-212.
7. Chandrashekhar S (2012) Newton’s Prinicipia for Common Man, Oxford University
Press U.K.
8. Fellenius W(1936) Calculations of the stability of Earth Dams; Transactions of the 2nd
International Conference on Large Dams 4, Washington USA.
9. Janbu N (1954) Stability analysis of slopes with dimensionless parameters Harvard Soil
Mechanics Series 46, 811.
10. Janbu N(1968) Slope Stability computations. Soil Mechanics and Foundation
Engineering; Technical University of Norway.
11. Krahn J(2004) Stability Modeling with SLOPEW and Engineering Methodology, Geo-
Slope/W International Limited.
12. Lowe J, Karafiath L (1960) Stability of Earth dams upon drawdown. Proceedings 1st Pan
American conference on Soil Mechanics and Foundation engineering Mexico 537-552.
13. Moregensten N.R, Price V.E.(1965) The analysis of the stability of generalized slip
surface, Geotechnique 15 79-93.
14. Sarma S (1973) Stability analysis of embankment and slopes, Geotechnique 23(3) 423-
433.
15. Petterson K.J.(1916) Kajraseti Goteborg des 5te Mars Tekniske Tidskift Vol 46.
16. Parnovsky A.S.(1998) Some Generalization of the Brachistochrone problem Acta
Physica Polinica A-93; S55 1-9.
17. Smith I (2006) Elements of Soil Mechanics, Blackwell Science Publications U.K.
18. Spencer E (1967) A method for analysis of the stability of embankments assuming
parallel inter slice forces, Geotechnique 17(1) 11-26.
19. Taylor D.W.(1937) Stability of Earth slopes, J. Boston Society of Civil Engineering 24,
197-246.
20. Terzaghi K, Peck R.B.(1984) Soil Mechanics in Engineering Practice, John Wiley
Publication N.Y. USA.
Vol. 18 [2013], Bund. V 5382

21. Whitman R.V, Bailey W.A.(1967) Use of computers for slope stability analysis; ASCE
Journal of Soil Mechanics and Foundation Engineering Vol 93(SWM4).

© 2013, EJGE