SCHOOL OF

CIVIL ENGINEERING

INDIANA
DEPARTMENT OF HIGHWAYS

:••••••••••••••••••••••••••••••••••••••••••••••••:•
: :
-

:• :•
. !
JOINT HIGHWAY RESEARCH PROJECT ¡•
JHRP-85-17

STABL5...THE SPENCER METHOD 1

OF SLICES: FINAL REPORT •1
1 J. R. Carpenter 1
l........................................" I

UNIVERSITY
 Digitized by the Internet Archive
in 2011 with funding from
LYRfSIS members and Sloan Foundation; Indiana Department of Transportation

http://www.archive.org/details/stabl5spencermet00carp
JOINT HIGHWAY RESEARCH PROJECT
JHRP-85-17
STABLS.,.THE SPENCER METHOD
OF SLICES: FINAL REPORT

J. R. Carpenter
 FINAL REPORT

STABLS•••THE SPENCER METHOD

OF SLICES

by

J. R. Carpenter
Graduate Instructor in
Research

Joint Highway Research Project

Project No.: C-36-361

File No.: 6-14-12

Prepared as Part of an Investigation

Conducted by

Joint Highway Research Project

Engineering Experiment Station
Purdue University

in cooperation with the

Indiana Department of Highways

Purdue University
West Lafaye
tte, Indiana
August 28, 1985
 FINAL REPORT

STABLS.••THE SPENCER METHOD OF SLICES

TO: H. L. Michael, Director August 28 , 1 985

Joint Highway Research Project
File: 6-14-12
FROM: C,W. Lovell, Research Engineer
Joint Highway Research Project Project: C-36-361

The attached report is the final one for the JHRP project entitled
"Incorporating Spencer's Method of Slices in Program STABL." The work
was perf ormed by J. R. Carpenter under the direction of Professor C. W.
Love11.

The Spencer version of STABL will be designated as either STABLS

(ma inframe) or PC STABLS (mi cro-computer). The Spencer version of STABL
is more rigorous than that previously available in STABL4, and accord-
ingly requires more running time.

The IDOH is encouraged to test the new program against STABL4,

before adopting it for routine use.

Respectfully submitted,

C< t<J l &c..e¿7/ /}(',?717

C, w. Lovell
Research Engineer

CWL:kr

ce: A.G. Altschaeffl M.A. Hunter C.Fº Scholer

J.M. Bell J.P. Isenbarger K.C. Sinha
W,F. Chen J.R. McLaughlin J.R. Skinner
w.1. Dolch R.D. Miles C.A. Venable
R.L. Eskew P.L. Owens L.E. Woods
J.D. Fricker B.K. Partridge
G,T. Satterly
 iv

TABLE OF CONTENTS

Page

HIGHLIGHT SUMMARY.. V

STABL5 AND PCST A BL5

Stability Equa ti ons 2

STABL Method of Solution - Linear App roximat ion
Me thod . . . 11
Line of Thrust .. 17
Spencer Options in STABL 20

SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 23

Summary and Con c lusion s 23

Recommendations 24

LIST OF REFERENCES . 25

References Cited 25
 1

STABLS ANO PCSTABLS

In order to increase the versa ti 1i ty of STABL,

S pe nc e r ' s method of slices has been i mp l e me n t e d in the

program. Spencer's method was chosen since it satisfies

complete equilibrium of the sliding mass; i.e. equilibriurn

with respect to moment and force equilibrium. In con t r a s t ,

the Simplified Janbu method satisfies only vertical and

horizontal force equilibrium and not mornent equilibrium,

while the Simplified Bishop method satisfies only ve r t i c a l

force and overall moment eq u i l i br i u m but not ho r i zo n t a l

force equi li bri um. These me t hod s a r e eas i ly s o l ve d a nd

typically give conservative va l ue s f o r t he FOS wh e n c om p a r e d

to the more accurate methods of s l i ces sa t i s f y i ng com p l e t e

equilibrium such as the S pe nc e r or Mor ge ns t e r n a nd Pr i c e

methods (Wright et. al., 1973; S ha r ma a nd Lo ve l l , 1 98 3 ) .

Spencer's method of s l i ces is es pec i a l l y we l l suited

for handling hor i zon t a l or i nc l i ne d loads suc h as tieback loads

s i nc e the me t hod satisfies complete equ i l i b r i u m . Sp e nc e r

' s method d i s t r i b u t e s t he force from a load such as a t i eba c k

be t we e n s l i c e s t h r oug h the interaction of the i n t e r s l i c

e side forces. Therefore, there is no need to use a t ech n i q

ue such as t he Load Distribution Method to account

 2

far the presence of horizontal or inclined loads . The LDM

was developed far use in conjunction with the Simplified

Bishop or Simplified Janbu methods which do no consider the

interaction of the interslice side forces.

Spencer's method is not used as frequently as the

simpl i fied methods since it requires more computation time

and also since convergence of the solution is also often a

problem. However, for problems with horizontal or inclined

loads, Spencer's method is more appropriate than the

simplified methods. Convergence problems have been avoided

using the Linear Approximation Method of solution which will

be described later.

The addition of Spencer's method complements the

Simplified Janbu and Sirnplified Bishop methods existing in

STABL. Implementation of this method allows the STABL user

to search for critical potential failure surfaces using

either the Simplified Janbu or Simplified Bishop methods and

reanalyze any critical potential failure surface with

Spencer's method to obtain a more accurate value of the FOS.

Stability Equations

Spencer (1967) developed a limiting equilibrium method

of slices which satisfies complete equilibrium for circular

failure surfaces assuming a constant ratio of the interslice

normal and shear forces. This assumption leads to the

formation of parallel interslice side forces inclined at a

constant angle, e, on each slice. Spencer (1973) found that

 3

a reasonably reliable value far the FOS can be obtained by

assurning parallel interslice forces. The method was later.
extended to potential failure surfaces of a general or
irregular shape (Wright, 1969; Spencer, 1973).
The slice forces considered in the derivation of
Spencer's rnethod of slices are shown in Figure 1. As with
other lirniting equilibriurn rnethods, the factor of safety on
each slice is assurned to be the sarne such that all slices of
the the sliding rnass will fail sirnultaneously. For all
slices of a sliding rnass to fail sirnultaneously, the load
frorn one slice must be transmitted to the next slice through
the interslice side forces. The interslice forces z1 and zr
are inclined from the horizontal at an angle 0. The
interslice forces acting on both sides of each slice can be
replaced with a single statically equivalent resultant
interslice force, QF, acting through the rnidpoint of the
base of the slice and inclined atan angle 0 (Figure 2).
Sununing the forces normal and tangential to the base of
each slice provides two equations of force equilibriurn:

6N' + 6Ua + QFsin(a-0) + 6W(khsina - (1-kv)cosa)

- t.U cos(a-S) - t.Qcos(a-o) - 6Tsin(a-i) = 0 ••• (1a)
6

6Sr - QFcos(a-0) - 6W((l-kv)sina - khcosa)

+ t u sin (a-S) + 6Qsin(a-o) + t.Tcos(a-i) = 0 ( 1 b)
6
 4

o.x

1
6N

\6Ua
·- [ca'
1
FS
+6N' TANcpj

FIGURE 1 . Slice Forces Considered for Spencer's Method of

Slices
 5

OiX

N'
QF

\ ua
=-
1
FS
[ca' + N' TAN<f>j

FIGURE 2. Slice Forces Considered in Derivation of

Stability Eguations for Spencer's Method
of Slices
 6

The expression far the effectiv e normal force on the

base of each slice may be obtained from equation 1a.

AN' • AW((l-kv)cosa - khsina) - AUa + AU cos(a-S)

8
+ il.Qcos(a-.S) - QFsin (a-0) + il.Tsin(a-i) •.. (2)

The expression for the mobilized resisting shear force

at the base of a slice is given by:

[Ca ' + il.N' tan4>' l

( 3)
FS

where:

FS = Factor of safety: assumed equivalent on all

slices

e'a = Cohesion force= c'•(dx)/cosa

il.N' = Effective normal force acting on the base of
a slice
4>' = Effective angle of shearing resistance

Substi tuting equation 2 far the effecti ve normal force

into the expression far the resisting shear force at the
base of each slice (Eqn. 3), and substituting the resulting
expression into equation 1 b yields the expression for the
resultant of the interslice side forces on each slice:

QF =
cos(a-e) [ 1 + s3/ FS l • • • • • • • • • • • • • ( 4)
 7

where:

ea ' + tan '[AW((l-k v )cosa -khsina) - AU a

+ AU cos(a-B) + AQcos(a-o) + ATsin(a-i))
6
s = dU sin(a-B) - AW((l-kv)sina + khcosa)
2 6
+ dQsin(a-o) + ATcos(a-i)

s3 = tan •tan(a-0)

FS = Factor of safety: assumed equivalent on all

slices

e'a = Cohesion force= c"•(dx)/cosa

If the overall moment produced about an arbitrary point

by all externa! forces is zero, then the overall moment of

the interna! forces must also be zero, thus:

n
QF[Rcos(a-6)) =
I: 0 ••••••••••••••• (5)
i=l

where R is the distance from the center of rotation about

which moments are summed to the center of each slice. For

circular potential failure surfaces, the value of R is

constant and may be taken out of the summation:

n
I: [QFcos(a-6)) = 0 ( 6)
i=l

For surfaces of a general shape where no common axis

exists, moments may be taken about a different axis far each

 8

slice in turn. It is often convenient to take moments about

the center of the base of each slice far irregular surfaces

rather than about an arbitrary center of rotation. The

approach adopted in obtaining t he eq u i l i b r i u m equations does

not affect the final solution to a given problem (Spencer,

1970).

If overall force equilibrium is satisfied, the n the

summation of the interna! forces in two mutually exclusive

directions must be zero. He nc e , for f o r ce equilibrium in

the horizontal and vertical directions:

n
¡; [QFcose] = 0 •• , •••••••••••••• (7a)
i=l

n
¡; [QFsine] = 0 .••••••••.••••••• (7bl
i=l

The inclination of the resultant side f o r c e s c a n be

expressed as:

ei=0f(x) • • • • • • • • • • • • • • • • • • • ( 8)

where e is a scaling angle of inclination and f (x) is an

arbitrary function which defines ho w ei va r i e s wi t h the x

position of a s l i c e . Pa r a l l e l s i d e f o r c e s oc c ur whe n f(x) =

1 for all va l ue s of x. T he assumption of parallel resultant

side forces is equ i va l e n t to the Morgenstern and Price

Me t h o d (1 96 5 ) whe n f ( x ) = 1 , thus making Spencer's tane

equ i va l e n t to Mor ge n s t e r n and Pr i c e ' s A (Spencer, 19 73 ) .

 9

If the slope of the resultant interslice side forces is

assumed to be parallel; i.e., ei = constant, equations 24a

and 24b become identical and can be expressed as:

n
E [QF] 0 (9)
i=l

Two factors of safety are obtained when equations 5

and 9 are solved assuming a value of e. Equation 5 yields

a FOS satisfying moment equilibrium (Fm), while equation 9

yields a FOS satisfying force equilibrium (Ff). There is a

unique value of the FOS and corresponding value of e which

satisfies both force and rnoment equilibrium (Figure 3b ) .

Equations 5 and 9 are solved using values of e until F
m
and Ff are equal corresponding to equilibrium of forces and

moments for the sliding mass.

It should be noted that ate= 0, theFOS with respect

to rnornent equilibrium (Fm) corresponds to the Simplified

Bishop FOS, while the FOS with respect to force equilibrium

( Ff) corresponds to the Simplified Janbu FOS. It can be

seen from Figure 3b that the FOS wi th respect to mornent

equilibrium (Fm) is much less sensitive to the side force

assumption (value of 0), than the FOS with respect to force

equilibrium (Ff). From this figure it can also be seen that

the Simplified Bishop FOS yields rather accurate values of

FOS when compared to complete equilibrium methods. This is

 10

ooo
d

o
e:,
d
".,
4,.
o
ci
l/J "'
X e:,
< od
10

>- e' • 300 psf

o ,. • 40°
d 't' • 125 pcf
"' ru • O. 5

o
o 30 . 00 60.00 90.00 120 . 00 ISO.DO !B0.0C 210. 00 240 . 00

X - AX I S ( f t )

(a)

1 , •O
D F■
1, 3 5 .o. Ff'

(Jl 1, 30
□la..
l . 25

1, 20
CII
4,.
e 1.15
Ul
...o 1.10
10S • ,l ll S
1

l. 1
o 1. 0 5
+J
u
1
1...
..o.. 1.00 1:
0. 95 I=
1
0. 90
o 10 15 20 25 3D 35 •o
e C da9 >

(b)

FIGURE 3. Var iation of Fm and Ff with e

 11

due to the in sensitiv ity of the Fm curve to the assumpt ion

of the slope of the interslice forces.
Reasonableness of the solution can be judged by
examining the positi on of the li ne of th rust and the
magnitude of the int ersl ice shear stresses. Both are
obtai ned from the moment egui li bri urn eguat ions far the
indiv idual slices. This topic will be add ressed lat er in
this chapter.

STABL Metho d of Solu tion - Li near Ap proximatio n Method

Numerous it erat iv e schemes have been used to solve for
the FOS satisfy ing complete eguilib riu m. tt is import ant
to use an iter ati ve scheme that read i ly converges and also
mini mizes the number of iterations reguired to produce a
solut ion. Techniques ha ve i ncluded: 1) mathemat icall y
sophisticated analyses such as the Newton-Ra phson numerical
technique (Wrigh t, 1969; Boutrup, 1977); 2) arbitrar i ly

assumi ng several values of e, calculating the corr espondi ng

values of F
m and Ff, and using a regression analysis to find
the int erse ction of the Fm and Ff cur ves (Fredlund, 1974);
3) assumi ng a value of 0, calculat ing F
m and Ff' and
choosing a new value of e based on the relative magnitude of
F
m and Ff for a given e (Fredlund, 1981); and 4) assuming a
value of e and FOS, calculating Ff, setting Fm equal to Ff,
solv in g for the new value of 0, and substituting that value
into the F
m eguation (Spencer, 1973; Maksimovic, 1979).
 12

None of the iterative schemes outlined above

complemented the routines already present in STABL.
Therefore, a new iterative method has been developed which
rapidly and accurately determines the FOS satisfying

complete equilibrium while avoiding problems of non-

convergence. The new method is called the Linear
Approximation Method (LAM), and utilizes the INTSCT routines
in STABL which calculate the intersection of two straight
lines. The method uses values of Fm' Ff and 0 to
approximate the Fm and Ff curves with straight lines and
calculates their intersection. An accurate value of the FOS
satisfying complete equilibtium is obtained by successive
approximations of the Fm and Ff curves with straight lines
for several values of e. Due to the shape of the Fm and Ff
curves, convergence is rapid and often occurs within three
iterations. Unlike sorne of the iterative techniques
outlined previously, the LAM is easily comprehended and
minimizes the number of iterations required for solution.
Equations 4 and 8 are first solved with initial
estimates of e and FOS. The initial value of e is taken as
one half the approximate slope angle, which is input by the
user. Spencer (1967) found that the angle of the resultant
interslice side forces satisfying complete equilibrium was
less than the slope angle. Therefore, STABL utilizes a user
input estimate of the slope angle to begin iteration for the
FOS and corresponding angle of the resultant interslice side
 13

forces. The solution is not sensitive to the value input

for the slope angle; however, a reasonable estímate will

minimize i terat io n time. The initial estímate of the FOS is

obtained by first calculating the FOS by either the

Simplified Bishop or Simplified Janbu method depending on

the type of analysis being performed.

Using the initial estimates of e and FOS, equations 8

and 4 are solved for the sum of the resultant interslice

forces and their corresponding moments. Based on the

relative magnitude of the sum of the forces and the sum of

the moments, a second value of FOS is chosen and the sum of

the forces and moments are recalcula ted. The factor of

safety satisfying force equilibrium for the given value of 0

is found by calculating the intersection of the line through

the sum of the forces previously calculated with the EQF = 0

axis as shown in Figure 4a. The value of Ff satisfying

force equilibrium for the given e is checked by using that

value of FOS to calculate the sum of the forces. The value

of Ff is recalculated using a straight line intersection of

the two previous values of the sum of the forces with the

EQF = 0 axis. Normally three trials are all that are

required to find the value of Ff within a tolerance of

0. 001. The value of Fm for a gi ven e is found in the same

manner.

The calculation of the sum of the resultant interslice

forces and moments for a given value of e requires little

 14

(+)
., 9- constant
<11
...o
<)

""
....
<<11
)

.....,.
.....q¡

..e..:
...e :
.. . o FOS
-
":,"'
"...'.
<11

o
e
:,
CI)

(-) (a)

(+)

...e : 9 - consta nt

. ..
-. ,,..,.
:,
a.,
<11
<)

"'
.... i:...
o
o
., .
...... a.,
<

e: ......
) ."..' o
o a.,.., FOS
X
....o ..e..:
e
:,
lll

(b)
(-)

FIGUR E 4. Determination of Ff and F for a Gi ven Value of

0 by the Linear Approxi marion Method
 15

calculation time. In addition, calculation of the

intersection of two straight lines is simple and also
requires very little computation time. The combination of
these two facts leads to a very efficient procedure for
accurately and rapidly determining the values of Ff and Fm.

A second value of e is taken as three-fourths the input

slope angle. The force and moment equations are again
solved for new values of Ff and Fm corresponding to the new
value of e. A second value of 0 equal to three-fourths the
input slope angle was found to lead to rapid solution of the
FOS by the LAM, and was thus chosen.
After two iterations, the Fm and Ff curves are
approximated by straight lines and the intersection of the
these lines is calculated ((0int' Fint); Figure 5a). It
can be seen from Figure 5a that the intersection of the
approximation of the Ff and Fm curves by straight lines
leads to a very accurate estimate of the value of e
satisfying complete equilibrium and a rather good estímate
of the FOS. The difference between e. t and the nearest
in
value of e used (in this case e2) is calculated and another
value of e is selected such that e3 = eint + (eint - 02)•
The value of 03 is used along with Fint to calculate Fm and
Ff corresponding to e3• Using the new and previous values
of Fm' Ff ande, the intersection of the two curves is again
approximated by the intersection of the straight lines
representing the Fm and Ff curves (Figure 5b). This
 16

). 'º
J . 35

U) 1. 30
o
u.
1. 25

...
QI 1.20

o J. 15
U)

...o 1.10

l..
o J. 05
+ >
ou J. 00
u.
0. 95

0.90
0 5 10 15
8 <dr.9)
20 25 30 35
'º
(a)

1. •o

1. 35

Vl 1. 30
o
u.
1. 25

..o.
QI
1. 20

V lJ. 15
...
o J. 10
l..
o J. 05
+ >
u :;.,-
o J. 00
l.&.

0.95

0.90
o 5 10 15 20 25 30 35 •o
8 <dr.g>
(b)

FIGURE 5. Determination of the FOS and e Satisfying

Complete Equilibrium Using the Linear
Approximation Method
 17

process is repeated unt i l the difference between the current

FOS and the previous FOS is less than 0.001 and the

difference between the last two values of e is less than

0.001 radians, or the difference between Ff and Fm is less

than 0.001 for a gi ven e. This avoids unnecessary

calc ulat io ns when the value of e being used happens to

correspond to the value of e satisfying complete

equ ili br iu m. The program is structured such that a new

value of 0 wi ll be determined if 0int lies to the left of 0

1
or between e1
and 0 .
2
Due to the shape of the curves, convergence is rapid

and often occ urs wi thin three i terations. No problems have

been indicated with respect to non-convergence of a solution

using the LAM. It is believed that STABL is the only known

slope stability program to contain the Linear Approximation

Method.

Line of Thrust

As mentioned previously, attention should be paid to

the position of the line of thrust (location of the line of

action, or points of application, of the interslice side

forces on the slices) to check the reasonableness of the

solution. A satisfactory solution is one in which the line

of thrust passes through the middle thi rd of the slices.

Tensile forces are indicated within the slope if the line of

thrust lies outside the middle third of the slice. The

 18

location of a satisfactory line of t h r us t is shown in Figure

6 for the example problem shown in Figure 3a .

Once the slope of the intersl ice forces e and the FOS

satisfying complete egu i l i b r i u m have been determined, the

-
line of thrust may be calculated. The values of the

resultant interslice f o r c e s , ( Z , Zr), for each slice are

1
calculated by substituting the values of FOS and e
s a t i s f y i ng complete equ i li b r i um into equation 4 . Working

from the first slice to the last, the points of action of

the interslice forces are found by taking moments about the

center of the base of each s l i ce in turn.

Spencer (1973) indicated that suitable lines of thrust

can be obtained assuming that a tension crack filled with

water e xi s t s at the upper end of the slip surface. The

depth of the tension crack may initially be taken as the

depth of zero active effective stress:

2c' 1 + sin •
z = 1- m • • • • • • • • • • • • (10)
o yFS(l-ru) sin 'm

S pe nc e r demonstrated that rel iable factors of safety

can be obtained assumlng the slope of the interslice forces

are parallel. However, he recommended that the slope of the

i nt e r s l i c e forces should be reduced at the upper end of the

slip su r f a c e in arder to obtain r asonable positions of the

line of thrust. Spencer' s method as programmed in STABL

follows these recommendations.

 19

o
o

1:
1 º
"::l'
"E-<'
.""
e
..... J:
o

..
111
i::

..,¡
J: ,....
+J
+l

UI
::J
1-1
10 4-- .e
'-' E-<

1 U)
. . ...
o
(LI
e:
X ··•

:1
< .J

ci
X '°
i:,J
Ol
e,:;
::,
t!)
o 1-1
o r...
c:i
\O

o
o
c:i
(T)

o
00 ºD!; I 00·02r □□·□6 oo·og oo·oE o
n.:D SIXV - Á
 20

Spencer Options in STABL

The Spencer option may be invoked by spec ify ing the

comma nd "SPENCR " and an estima te of the slope angle. The

SPENCR command precedes specification of the surface type

and method of solution; i,e., SURFAC, SURBIS, CIRCLE,

CIRC L2, RANDOM, BLOCK or BLOCK2.

Spencer's method has been implemented in the STABL

program for the primary purpose of obtaining a more accurate

value of the FOS and line of thrust for specific surfaces of

in teres t. For cr i tic a! surfaces, the Spencer method of

sl ices is preferred over the Simplified Janbu or Simplified

Bish op methods since it sat isfies complete equi li br ium of

the slid i ng mass and yields a slightly more accurate FOS.

Sin ce determination of the FOS by Spencer's method requires

appro xi mately six times more calculation time, it is

i ntended that only specific surfaces of interest will be

analyzed utilizing Spencer's method. However, Spencer's

method may be used for analysis of either user input

specific surfaces, or randomly generated surfaces.

The most efficient use of STABL's capabilities will be

realized if the user investigates a number of potential

fai lure surfaces using one of STABL's random surface

generation techniques and determines the FOS by either the

Simpli fied Janbu or Simplified Bishop method of slices.

Once critical potential failure surfaces have been

identified, they may be analyzed using the SPENCR option in

 21

conjunction with either the SURFAC or SURBIS option to

obtain a more accurate value of the FOS and to gain insight

into the reasonableness of the solution through examination

of the line of thrust.

When a user input potential failure surface is

analyzed, the program will output the values of Ff, Fm and 0

calculated during iteration along with the value of FOS and

0 satisfying complete equilibrium. The user may use this

information to construct a graph similar to that of Figure

50b. When analyzing a user input potential failure surface,

the coordinates of the line of thrust, the ratio of the

height of the line of thrust above the sliding surface to

the slice height for each slice, and the values of the

interslice forces are all output. This information allows

the user to quickly determine whether or not the line of

thrust, and hence the solution, is satisfactory.

The Spencer option may also be used wi th the STABL

options that generate surfaces randomly. However, when the

Spencer option is used in conjunction wi th randomly

generated surfaces, only the FOS and angle of the interslice

forces satisfying complete equilibrium are output for the

ten most critical surfaces. Information regarding the line

of thrust, interslice forces or values of Ff, Fm and 0

calculated during iteration is not output for randomly

generated surfaces; hence the reasonableness of the solution

obtained for a randomly generated surface will not be

 22

apparent. When the reasonableness of the solution of a

randomly generated surface is desired, the surface must be

analyzed using the SPENCR option in conjunction with either

the SURBIS or SURFAC options.

The STABLS/PCSTABLS User Manual (Carpenter, 1985b),

further describes the Spencer options, input format,

restrictions, and error codes. This document also describes

sorne minor program enhancements and provides an example

problem using the Spencer option.

 23

SUMMARY, CO NCLUSIONS AND R EC OMMENDA TJ O NS

Summary and Conclus ions

S pen ce r's method of slices wa s imp l em en ted in lhe progra m s

S T A BL 5 and PCST A BL5 to provide addit i onal v ersati lit y to the

ST A B L pr ogra ms. S pencer 's method of slices satisfies c o mp l ete

equ i l i br iu m and is capable of transferring the J oad from one

sl ice to another through the interactio n of t he i n ter sl ice

shear and normal side forces. Therefore, S pencer 's method of

sl ices is p ar ti c ularly well suited for analysis of slo pes and

re t a i n i ng walls subj ec t ed to ti e back ]oads si n ce it dis tr ibut es

t he l oad fro m a tieback between sl ices. The de v elo pment of

S pen c er's method was revi ew e d and its imp le men ta t i on in ST A BL

presented. A new ite ra t ive tech n iqu e was developed by the author

for d e t er m i nin g the factor of safety and angle of the interslice

forces satisfying co mple t e equ ili brium of a s lidin g mass co rn puted

by S pencer's method. The new iterative technique is called the

L i near Approximation Method.

The Spencer rnethod of slices is preferred for ana lysi s of

tiedback slopes and walls over the sirnplified methods since it

satisfies co mp l ete eq ui li brium and accounts for th e interaction

of the i n te rslic e side forces between slices. The FOS obtained

by S pence r's me t hod is typically sl ightly higher than the FOS

obt a i ned by the less rígorous Simplified Bishop or Simplified

 24

Janbu methods. Since S pen c er's method satisfies co m plete

eq uil i briu m of the sJ i d i ng m ass , it is especially well suited

for analysis or t,iedllack slopes and walls. S pen c er's solutlon

is more rigorous than t h e simplified m et ho ds and requires more

comput at t on time.

The Linear Approximation M e th od provides a reltahle m et hod

fo r determining the FOS satisfying complete equilibrium. The

method not only co n v erges readily, it also sea rches for the FOS

satisfying co mpl ete equilibrium, thus m inim iz i ng the nu m ber of

iterations required to obtain a solution. It is believed that

the LAM is unique to the STABL programs.

R eco mm en d at ions

Since S pencer's method of sli ces sa tis fies com pl ete equ111-

brium, this m etho d s houl d be use d to ana l yze the stabllity or

tiedback slopes an d ret a ini ng str uctu res when e v er poss i ble.

Since S p encer' s m et hod of sl ices requires more co mpu ter

time to arri v e at a sol ut i on , it is rec ommended that S pen cer' s

method be used to analy ze onl y those c ritica l po te nt i al fa i l ur e

surfaces found b y analy s is of random l y generated surfac es us i ng

the Simplified Bis ho p or S imp lifie d Janbu meth od of slic es.

Followi ng t h is reco mm enda ti on will lead to the mos t eff ici en t

uti li zatio n of STABL's capabil iti es and the eng in eer's t im e.

It is st rong ly recommen d ed t hat t he new S pe ncer routines be

thoroug h ly te st ed pr ior to pub lic release of the STAB L5 and

PCST A B L5 prog rams.

LIST OF REFERENCES
 25

LIST 0 F R EFE R E N CES

R efere n ce s C it ed

Bo ut r u p , E. ( 1977 1 , "Co m put erize d Slope Sta bi l i ty Ana l ys is

fer I nd i ana H ig hw ays " , MSCE Thes is , P urdu e U n iv ers ity ,
West La faye tte , Ind iana , 1977 .

Ca r penter , J. R . ( 19 8 5b l , "STAB L5/ PCST AB L5 User M anu al" ,

Joint H i g h way R es ear c h Proj ect No. 85- , S c hool of
Civil Eng in eer in g , Purdu e U n i v ersi ty , We s t Lafay ette ,
I nd ia na , J une , 198 5.

Fre d l und , D.G. (1974 ), "S lo p e Stability An alys is User's

M an ual " , Co m pu t er Doc umentati on CD -4 , Tr ans p or tati on
and G eo tec hni c al Gro u p , D e p ar t m en t of C ivi l
Eng ineering , U n i v ers i ty of Sask a tc h ewan, S as k a toon ,
Ganad a , D ec em ber , 1974.

Fred l u nd , D.G. (198 1 1 , "S L0 P E-II Co m p ute r Pr og ram" , Us er' s

Ma nua l S- 10 , G eo-S lo p e Progra mm i ng Lt d. , Ca lga ry ,
G a nada , 1981.

M aks imo vi c , M . ( 1979 1 , " Li m it Eq u ilib ri um fo r Non l i n ear

Failu re Envelope and A r bitr ary S l i p S urfa ce " , T h ird
Inte r n a tio nal Co n feren c e on Nume ric al Methods in
G eo m ec ha nics , A ac hen, April, 1979 , pp. 769-777 .

Morg ens te rn , N. R. an d Pr ice , V . E. ( 1965 1 , "The An a lys is of

t he S ta b il ity of G ener al S li p S urf aces" , G eo te c hnique ,
Vol. 15, No. 1, M arc h, 1965 , pp. 79-93.

Sh arma , S . S. and Lo vell , C. W. ( 198 3 1 , "Strengths and

We akness es of S l o p e S t a b il it y Ana l ys is " , Pro c ee dings ,
34th A nn ual H ig h way G eolo gy Symposium, Atlanta, G
eorg ia , 1983 , pp. 215-232.

S p en c er, E. ( 1967 ), "A Method of Analysis of the Stability

of Embank men ts Ass um i ng Pa ral l e l I nter-S l i c e Fo rc es" ,
G eo tec hn iq ue , V ol. 17 , No. 1 , Ma rc h, 1967, pp. 11-26.

S penc er , E. ( 1970 ), "The A n alys is of the S t ab il i ty of Emb

ank men ts by the Method of Sl ices" , Ph.D. Thesis,
University of Manches t e r , 1970.
 26

S p e n c e r , E. (1973), "T h r u s t Line Criterion in Emb a n kme n t

Stability A n a 1 y s i s " , Ge o t e c h n i q u e , Vol. 23, No. 1,
March, 1 9 7 3 , pp. 8 5 - 1 00 .

Wright, S. G. ( 1 9 6 9 ) , "A S t u d y of S1ope Stability and t h e

U n d r a i n e d Shear St r e ng t h of Cl a y S h a l e s " , Ph.D. Thesis,
U n i v e r s i t y of Ca l i f o r n i a , Berkeley, 1969.

Wr i g h t , S. G., Kulhawy, F. H . , and D u n c a n , J. M. ( 1 9 73 ) ,

" A c c ur a c y o f Eq u i l i b r i u m Slope St ab i l i t y A n a l y s i s " ,
J o u r n a l o f thc Soi1 M e c h a n i c s and Fo u n d a t i o n Di v i ion,
ASCE, Vo l. 9 9 , N o . S Ml 0 , Oc t o b e r , 1973, pp. 78 3 - 7 9 2 .