See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281587174

Pore water pressure manipulation in computerised slope stability analysis

Conference Paper · January 1985

CITATIONS READS

5 180

1 author:

Edward Nicholas Bromhead

150 PUBLICATIONS   1,253 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Edward Nicholas Bromhead on 30 January 2016.

The user has requested enhancement of the downloaded file.

Papers by E. N. Bromhead

in the

Midland Geotechnical Society

Symposium on

Computer Applications in
Geotechn ical Eng ineering

April 16 & 17, 19Bo

E. N. Bromhead, (1986). Pore water pressure

manipulation in computerised slope stability
analysis.
Proc. Midland Geotechnical Society Conference on
Computer Applications in Geotechnical
Engineering. 175-184.

E. N. Bromhead, (1986). A suite of programs for

groundwater modelling on microcomputers. Proc.
Midland Geotechnical Society Conference on
Computer Applications in Geotechnical
Engineering.
I 85-194.

Reproduced from original

camera-ready copy
Edward N. Bromhead PhD MSc DIC BSc I1ICE CEng FGS

Reader in Geotechni-cal Engineering, Kingston polytechnic

Pore water pressure manipulation in computerised slope stability analysis

SYNOPSIS

Si-mpIe concepts to describe pore water pressure (p.w.p.) information in

slope stability analysis such as ,the r,, valuer, or ,the ground water
table, can lead to the use of incorrect distributions of pore water
pressures. Pore pressure ratio values, in particular, involve averaging,
which smooths out both high and low pore pressure zones alike. In turn,
this leads to the selection of inappropriate failure mechanisms.

Examples are given, showing that the proper treatrnent of the pore water
pressure values is of much more fundamental importance to the rtcorrectness,l
of a computer analysis than are olher factors whi-ch coriventionally receive
greater attention from the geotechnical engineer. changes of p.w.p. in a
slope during its "lifetimerr are reviewed, and are applied to the solution
of the general problem of specifying data for slope stability analysis.

INTRODUCTION

In his paper of 1955, Bishop laid the foundations of modern slope stability
analysis by computer. Here was a method of sufficient accuracy, based
firmly on an effecii're stress approach. While simple enough for hand calcu-
lations, or using such aids as then existed, it was nevertheless tedious,
and seemed an ioeal subject for the then novel principle of electronic
digital- computation (Lit.tle & Price, 1958). Bishoprs interest in dams, and
the generation of p.w.p.s by the construction process, (the nett resulting
Pore water pressure could be related to the overburden stress pv via the
pore pressure parameter E) iea him to couch his equations at least in part
in terrns of E, and this was subsequently taken up by Little & price in
their comPuter Program as a simple and economical vehi-cle for handling the
pore water pressure data. Nowadays, we would use the term pore pressure
ratio, given the symbol.,r, in the general case, where pore water pressures
 Conventional flow net Lines of constant vertical total stress

water Ievel

toe drai
\ \-' ,/ \\ \
flov line
__ _/ / ./ \---
t equipotential
,/ ./ .a'- ---
I I
-t ,/ \\-
impermeable bedrock

Lines of equal pory ater pressure Lines of equal pore ressure ratio r

water 1eve1

1n
I rapid
I
vari.ation cf
r in this
\- \
U
area
0.5 -o.t

Figure 1. Iso-lines of quantities related to the r rat io.

l!

(n, is the rpiezometric headr for piezometer J)

piezometric line

ground level A
" slip surface
piezometer
/<'.'
piezometric line does not appiy for
this piezometer, or for slip surfaces at depth

Figure 2. The piezometric line, and how it is relative t.o a particular

s I ip surface.
are relaled to the overburden stress, but may or rnay not result from it;
leaving E to denote the generation element in the poreviater pressure. We
will see later that the use of this ratj-o as the sole or prime means of
communicating p.w.p. data to a slope stability program can give rise to
some serious difficulties.

THE USE OF r AND THE PIEZOMETRIC LINE.

The main advantages of using the parameter ru are in simplicity and

economy: simple for the analyst, (only one parameter is required. to define
the p.w.p"s in large sections of the slope); economv, because data storage
is minimised (only one extra data item per soil layer is required), and
computing the p.w.p. for a given slice uses code required to evaluate the
slice weight and to select relevanl shear strength parameters for it.
Regret.ably, these advantages are offset by the resulting inaccuracies.
Evaluation of an r,, factor involves averaging which eliminates both high
and low areas in the p.w.p. regime. This might be expected not to affect
the calculated factors of safety significantly, since the calculation of a
factor of safety by a slices method is itself a form of averaging. Indeed,
if the r,,u averaging is carried out along the slip surface, little damage to
the computed factor of safety occurs, but it is exceptional for that to be
the situaLion. Usi-ng r,, in most cases indicates the analyst.'s wish to
force the pore Pressure distribution to fit the zonation of the soil slope
section into I'typesr!, selected on the basis of shear strength parameters,
unit weight, particle size, colour or some other fac""or of secondary impor-
tance. (Note that the pore pressures assumed in desi-gn and analysis have a
much greater impact on computed shear strengths than do the shear strength
parameters tnemselves - even an extremely poor laboratory test can produce
c', Ot values well within the confidence range for p.w.p.s). Quite often,
the p.w.p. distributi-on is not in any way relat_ed to the subdivision of the
soil section into zones, and when further the distrlbution of total verti-
cal stress in the slope is j-ncluded, any initial relationship is mad.e less
applicable. Eigure I demonstrates this.

The combinati-on of a distortion in the p.w.p. distribution both by the

averaging procedure, and by a forced fit to an a priori- zone subdivision of
the slope, has the resuit that the actual p.or.p. distribution computed for
a slip surface may bear little relationship to the analystrs source data, &
is systematically biased away from the excreme p.w.p. values. It is not. so
much that factors of safety are computed in error (although that can
easily
occur when the subdivisi-on of the p.w.p. data is coarse) but it is a more
subtle and potentially more misleading one: the modes of failure indicated
for the critical slip surface are distorted. Thus, Ehe significance of
(for example) a particular horizon in a soil zone where high p.w.p.s
are
generated may well not be appreciated at the design stage. using ru is
often excused by stating that it is merely a means of extending lj-mited
data' This can be satisfactory, provided that the analyst underst.ands that
the use of r. does imply a great deal about the distribution of p.!r.p.s in
the slope as well as their magnitude.

oddly enough, the r.,, technique is reast likely to give problems to the
earth dam engineer, who will probably have reconciled himself at the outset
to the use of a fine subdivision of the slope section, and will in all
probability have a surfeit of pi-ezometer readings to deal with, no two of
which will yield the same r ratio to be applied in analysis as a blanket
f igure.

Finally' a note of caution applied to partly submerged slopes. A technique

widely used for dealing with these (from Bishoprs (1955) paper) is ofren
mi-sunderstood and misapplied. Recent attempts have been made to clarify
this (Bromhead, 1984, I986). Even when correctly performed, the technique
is difficult to appry where the p.w.p.s are derived from an r, value: in
the submerged part of the slope, for instance, should one calculate the
p.w.p.s using the totalr or the partly bouyant weight of the slice? There
is no concensus on this, and the unwary analyst uses this combination at
his peril.

An alternative method of dealing with p.w.p.s also goes back to the early
days of stability analysis, but reflects the work on natural slopes.
Figure 2 shows a slide, with piezometers located on the slip surface. The
piezometric line joining all the individual pressure heads forms the sim-
plest means of interpolating between the d.ata point.s. Like r,, values , this
too is a simple and economic technique for representing p.w.p. information.
Sometimes only a few pairs of coordinates are needed. The p.w.p. head for
any slice may be computed from the level difference between it and the
slip surface. A normal refinement is to forbid the compuLation of a nega-
tive p.w.p. where the piezometric line goes beneath the slip surface.
However, even the simplesE fLow pattern must yierd d.ifferent p ie zome tr ic
lines for different slip surfaces. A single piezomeLric rine will only
apply to a crosely related family of slip surfaces. The method is often
adequate, but proves to be almost useless where pore pressure regimes of
any complexity exist.

In the stability analysis of slopes under seismic conditions, lateral loads

from seismic acceleration are simulated by variety of techniques.
"
Simplest of these, allowing the use of standard software, is the rslope
rotation' technique where the slope section is rotated. to bring the resul-
tant of the seismic (horizontal) and gravitational (vertical) accelerations
into the vertical. This simulates the effect of the seismic acceleration
on total forces on1y, and p.w.p.s must be modified separately. An inadver-
tant effect of the slope rotation on the p.w.p. distribution can be either
to change the total stresses, thus altering p.w.p.s computed via rrri or to
tilt the piezometric line itself, (rigure 3), so that p.w.p. heads no
longer relate to the vertical distance from slip surface to piezometric
line in the new (tittea) frame of reference.

AN ALTERNATIVE.

There are a number of alternative methods remaining. The first of these,

is to treat the specification of pore water pressures completely indepen-
dently from the specification of the slope section. Such a separate speci-
fication could be a regular or irregular grid of points on the section at
whj-ch p.w.p.s are defined. Interpolation facilities within the program
could then evaluate pore pressures for any part of the slip surface. Such
a method is uneconomic, because both a duplication of coordinate input
information is required, and the interpolation of p.w.p.s is on a different
basis than that for slope topology.

It seems that the ideal method. is to link p.w.p. information with the
coordinated points used in slope section definition. Then, the same inter-
polat.ion procedures used withi-n the program to extend. the slope geometry
are available to treat the p.w.p.s. Economy of both data storage, and
program code resuLts, without the loss of any fac,ility to input arbitrary
p.w.p. distributions, if at the cost of some apparent repet.ition of the
slope section details (Figures 4 and 5).
Rather than use p.w.p. heads, Fu values could be used at each of the
coordinated points, and interpolated in much the same h/ay. Howeve!, there
seems little merit in reducing p.w.p. head information to this format,
merely to reconvert at some later stage. Caution needs to be observed when
dealing with phreatic surfaces, although with this method, the subdivision
of soil zones in terms of saturated and unsaturated unit weights is more
natural than when using a piezometric surface for the di_viding rine.

APPLICATION TO MICROCOMPUTERS.

Examine the application of these concepts to micro-computer based analysis.

Two major strategies exist for processing the slope section: in the first
of these, a detailed analysis of the slope secti-on geometry is made before
attemPting to locate the individual slip surface(s) on this slope. It is
an a priori subdivision of the section into tsli-cesr. This demands a
substantial amount of memory in the computer, and means that 16 bit (or
better) architectures are to be preferred, since these a11ow a larger RAM
to be accessed r'rithout resort to data compression techniques (demand.ing a
major, cost-ineffective, programming input) or backing store, e.g. floppy
disks (i-ncrease in execution times). It is not normally possible to use
this technique in 8-bit computers while still retaining the capability to
attempt realistic problems.

The alternative is to retain only the coordinated input (and this, using
the normal cartesian coordinate syst.em, is the least error-prone method of
topological input), which is usually compact, and t.o subdivide into slices
for each slip surface. Thj-s can be attractive, and certainly leads to a
more natural slj-ce subdivision. Where only a limited range of slip surfaces
are to be investigated, the resulting increase in program execution time
through the repetition of largely the same sequence may not be too disadv-
antageous. No particularly high denands on RAM capacity are made, so it is
equally suited to micro-computers of any architecture. Users of inter-
preted BASIC, in Particular, will be compelled to adopt the latter strate-
gy, since even if the computer they use has enough RAII, often only a 64k
byte data space is accessible via the interpreter. The former is often
adopted in 'slip circler programs, and the latter in 'non-circular slip'
progralns. Adoption of r., values in a program capable of handling say zo
strat.a, needs some B0 bytes of RAll to hold the values. It is obviously
compact. Similarly, the RAII to hold a piezometric line i-s miniscule, typi-
cally 2t-+0 bytes f or a 30 point limit. on the orher hand, inputting a
p.w.p. head at each point in the slope section increases the memory
requirement by some 507., and may be an important control on the selection
of which strategy to adopt in handling the section data.

EXA}IPLES.

Some examples may demonstrate the arguments. Take first the case of an
embankment built on a soft. foundation. Generation and dissipation of
p-w-p-s Eherein are the controlling factor for overall stability. If the
fill is a free draining granular material, embankment side slope stability
may be seen by inspection alone to be irrelevant in comparison with the
foundation failure mode, and therefore not be the subject of specific
stability calcurations. rn the foundation, the p.w.p. dissipation may
proceed in a variety of ways. Where the subsoil layer is thin, drainage is
one dimensional, and at some stage before completion of consolidation, the
conventional parabolic-shaped excess p.w.p. isochrone, superimposed on the
initial ground wat.er pressure profile rnay yield the composite shown in
Figure {+, inser (a). Alternatively, the installation of vertical sand
drains may lead to the dissipation process resulting in p.w.p. / depth
profiles as shown in inset (b). In both cases a piezometric line can only
represent the p.w.p. conditions for a restrict.ed range of slip surfaces. In
the former, a thin layer approach to the subdivision of the foundation soil
may permit the use of the r, method. Nor does the discontinuous p.w.p.
distribution in the latt.er in principle forbid the use of r,r, but it wilr
be found that the actual distribution of p.w.p.s makes it imperative to
utilise a fine subdivi-sion of soil zones in the foundation to represent
what is, after all, only a simple p.w.p. distribution.

rn the above, the p.w-p- dissipation involved onLy changes of magnitude,

and changes in distribution were negligible. Where the subsoil is deep, or
is anisotroPic r lateral migration of the porewater occurs, the necessary
subdivision of the foundation into zones to enable t.he use of .r_, will
change as consolidation progresses, and thus introduce additional complexi-
ti-es. Proper use of ru may well require such additional subdivision of the
slope section that the economv of t.he method is completely 1ost.
The p'w'P' distribution in a clay fill may be an initial state of suction
(see, for instance, Crabb & West, I9S5) which dissipates
to a positive pore
pressure state as the foundation equilibrates to a lower piezometric
condi-
tion' Most software always forbids the utilisation of negative p.w.p.
conditions via the piezometric line method, and often with r. values
too.
Neither method is a satisfactory method of investigating the effect
on
shear strength and stability of negative pore pressure states.

A second example concerns the general cut slope case. Even a simpre anaLy_
sis demonstrates the existence of suctions within the body of a cut slope
to a depth (at the toe) of the order of the cutEing depth, and of extensive
depresseci p.w.p-s throughout the remainder of the section (".g. Bishop
&
Bjerrum, 1960). Recent detailed work (Vaughan & Walbancke, Lg75, Bromhead
& Dixon, 198/+) shows the existence of these depressed p.w.p. regimes
to be
fact, not merely speculation. Therefore to investigate the effects of their
dissipation, one needs to handle comptex distributions of p.w.p., iocruding
suctions' The impossibility of doing this with trivial methods is obvious.

The final exampLe concerns dams

in general, although for an exampre a dam
on a rock foundation and with relatively free ciraining shoulders has been
chosen- This has an unorthodox shape of core. p.w.p.s in the core respond
to the undrained loading, starting from suctions incorporated as a result
of compaction (close to the working surface level), so that there is
some
initial correlation with overburden stress. where construct.ion extends
over
several seasons, consolidation of the core occurs, and the relationship
of
P'w'p' to total stress changes. In order to represent this with r* values,
different soil zcne subdivisions within the core are needed for
the
different times, but if p.w.p. values are used, no such difficulty arises.

r have used this technique frequently in earth dam stability computations.

The input to the programs used treats the zone to zone contact line
as a
rchainr of coordinated points, and permits
the association of a pore water
pressure head with each coordinated point. Artificial rzone contacts, can
be introduced where p.w-p-s wi-thin a larger soil mass (".g. the core
in the
present context) must be represented. It is often convenient to rchaj-nl
from piezometer position to piezometer position (Figure 5), although a
little care is important. if the chaining process is not to read to an
incongruous subdi-vi_sion of the (core) zone, that is, to a subdivision
within which a simple interporation of the input p.w.p. head values
is
 y c0ordinate axis in tilted frame of reference
ground
level

correct soii thickness to

use in caiculations with r tnc0rr ect
assumed piezo
soil thick NESS
he
to use with
r
correct piezo-
metri
hea;[//
surface
-/
rtiltr angle:
a function of seismici ty

x c0ordinate axis in tilted frame of reference

Figure 3. P.w.p. errors i_nduced in the tilted slope method

vertical section dovn which piezometric pressures are sholtn

embankment fili
parabolic excess pore water
pressure isochrone after consolidaiiq3,rith vertical drainage

raf ter consoli-

dation I excess
pressure head in the initial excess piezometric heads
sand drain case ( ootteo)

Figure 4 Pore pressures in an embankment foundation with consolidation.

LtGtND

piezometer position o
coordinated point a
(not piezometer)

upstream downstream
shoulder shoulder

z0Nr (1) (1)

core (8)
(s)
(8)
)n
foundation (10) piezometer

Figure 5. Coordinate defini tion in a dam using piezometer

locat ions as coordinated nn i n f c
 inadmissable. Further changes in p.w.p., including the wetting of the up-
stream shoulder, impoundment, and rapid drawdown: frdy all be accomodated
equally as easily within a slightly amended version of the basic dataset.

Features described in this paper as being desirable for the treatment on

P.!/-p-s in slope stability analysis, as well as the simplified treatment of
p-w-p-s Lhrough the use of r. values and piezometric lines are available in
computer programs BISHOP and MORGEN approved for use not only in UK highway
schemes through the HECB of the DoE, but also for use in Hong Kong through
approvals granted by the Geotechnical Control Office.

REFERENCES.

BISHOP, A.W. (I955). The use of the slip circle in the stabitity analysis
of earth slopes . Geotechnique, 5 1-17 .

BISHOP, A.W. & BJERRUM, L. (1960). the relevance of the triaxial test to
the solution of stability problems. Proceedings of the Res. Conference on
on the shear strength of cohesive soils , Denver, Colorado, (Special PubL.
1 /^

American Society of Civil Engineers), 439-501.

BROMHEAD, E.N.(1984). Slopes & embankments. Chap. 3 in ',Ground Movements &

their effect on structuresr'.Eds. R.K. Taylor & p;B. Attewell, Blackies.

BROI{{EAD, E-N. & DIXON, N. (l-984). Pore water pressure observations in the
coastal clay cliffs at the Isle of Sheppey, England. Proceedings 4th Inter-
national Symposium on landslides, Toronto. 1, 385-390.

BROMHEAD, E.N. (1986). Srabiliry of slopes. Surrey Univ. press, (Blackies)

CRABB,G.I. & WEST, G. (1985). Monitoring pore water pressures i-n an embank-
ment srope. Tect^r"Note 4, Proc.rnt.symp. Failures in Earthworks , ICE,London.

LITTLE, A.L- & PRICE, V.E. (f958). The use of an electronic comDuter for
slope stability analysis. Geotechni_que, 8, I13-120.

VAUGIIAN,P.R.& WALBANCKE,H.J.(1973). Pore pressure changes anci delayed fail-

ure of cutting slopes in overconsolidated clay. Geotechnique, 23, 531-539.

View publication stats