Landslides and Engineered Slopes Chen et al.

(eds)
2008 Taylor & Francis Group, London, ISBN 978-0-415-41196-7

Landslide stabilizing piles: A design based on the results of slope failure

back analysis

M.E. Popescu
Illinois Institute of Technology, Chicago, Illinois, USA

V.R. Schaefer
Iowa State University, Ames, Iowa, USA

ABSTRACT: It is generally accepted that shear strength parameters obtained by back analysis of slope failures
ensure more reliability than those obtained by laboratory or in-situ testing when used to design remedial measures.
In many cases, back analysis is an effective tool, and sometimes the only tool, for investigating the strength
features of a soil deposit. Procedures to determine the magnitude of both shear strength parameters (c and  )
or the relationship between them by considering the position of the actual slip surface within the failed slope are
discussed. Using the concept of limit equilibrium the effect of any remedial measure (drainage, modification of
slope geometry, restraining structures) can easily be evaluated by considering the intercepts of the c tan 
lines for the failed slope (c0 , tan 0 ) and for the same slope after installing some remedial works (cnec , tan nec ),
respectively. The above outlined procedure is illustrated to design piles to stabilize landslides taking into account
both driving and resisting force acting on each pile in a row as a function of the non-dimensional pile interval
ratio B/D. The accurate estimation of the lateral force on pile is an important parameter for the stability analysis
because its effects on both the pile-and slope stability are conflicting. That is, safe assumptions for the stability
of slope are unsafe assumptions for the pile stability, and vice-versa.

1 INTRODUCTION modification of slope geometry, as well as some novel

Correction of an existing landslide or the prevention
of a pending landslide is a function of a reduction
in the driving forces or an increase in the available
resisting forces. Any remedial measure used must
involve one or both of the above parameters. The
IUGS Working Group on Landslides (Popescu, 2001)
has prepared a short checklist of landslide remedial
measures arranged in four practical groups, namely:
modification of slope geometry, drainage, retaining
structures and internal slope reinforcement, as shown
in Table 1. As many of the geological features, such as
sheared discontinuities, are not well known in advance,
it is better to put remedial measures in hand on a
design as you go basis. That is the design has to
be flexible enough for changes during or subsequent
construction of remedial works.
Although slope instability processes are gener-
ally seen to be engineering problems requiring
engineering solutions involving correction by the
use of structural techniques, non-structural solutions
including classical methods such as drainage and
methods such as lime/cement stabilization, grouting
or soil nailing, are increasingly being used (Popescu,
1996). The cost of non-structural remedial measures
is considerably lower when compared with the cost of
structural solutions.
Terzaghi (1950) stated that, if a slope has started
to move, the means for stopping movement must be
adapted to the processes which started the slide. For
example, if erosion is a causal process of the slide, an
efficient remediation technique would involve armor-
ing the slope against erosion, or removing the source
of erosion. An erosive spring can be made non-erosive
by either blanketing with filter materials or drying up
the spring with horizontal drains, etc.
Morgenstern (1992) followed this theme when he
noted that post-failure analyses can be used to pro-
vide a consistent explanation for landslide causal
events. The back-analyses can then be used as a basis
for design of the stabilizing measures if engineering
works are required. This approach has the added
appeal that the remedial design is normalized in terms
of the post-failure analytical model.

1787
Table 1. A brief list of landslide remedial measures. Most landslides must usually be dealt with sooner
or later. How they are handled depends on the pro-
1. MODIFICATION OF SLOPE GEOMETRY cesses that prepared and precipitated the movement,
the landslide type, the kinds of materials involved,
1.1. Removing material from the area driving the
the size and location of the landslide, the place or
landslide (with possible substitution by
components affected by or the situation created as a
lightweight fill)
result of the landslide, available resources, etc. The
1.2. Adding material to the area maintaining stability
technical solution must be in harmony with the natu-
(counterweight berm or fill)
ral system, otherwise the remedial work will be either
1.3. Reducing general slope angle
short lived or excessively expensive. In fact, landslides
2. DRAINAGE are so varied in type and size, and in most instances,
so dependent upon special local circumstances, that
2.1. Surface drains to divert water from flowing for a given landslide problem there is more than one
onto the slide area (collecting ditches and pipes) method of prevention or correction that can be success-
2.2. Shallow or deep trench drains filled with free- fully applied. The success of each measure depends, to
draining geomaterials (coarse granular fills a large extent, on the degree to which the specific soil
and geosynthetics) and groundwater conditions are prudently recognized
2.3. Buttress counterforts of coarse-grained in an investigation and incorporated in design.
materials (hydrological effect) In this paper a methodology involving back analy-
2.4. Vertical (small diameter) boreholes with sis of the slope and the use of piles to remediate the
pumping or self draining landslide are presented.
2.5. Vertical (large diameter) wells with
gravity draining
2.6. Subhorizontal or subvertical boreholes
2.7. Drainage tunnels, galleries or adits 2 BACK ANALYSIS OF FAILED SLOPES
2.8. Vacuum dewatering TO DESIGN REMEDIAL MEASURES
2.9. Drainage by siphoning
2.10. Electroosmotic dewatering 2.1 Failure envelope parameters
2.11. Vegetation planting (hydrological effect)
A slope failure can reasonably be considered as a
3. RETAINING STRUCTURES full scale shear test capable to give a measure of the
strength mobilized at failure along the slip surface.
3.1. Gravity retaining walls The back calculated shear strength parameters, which
3.2. Crib-block walls are intended to be closely matched with the observed
3.3. Gabion walls
real-life performance of the slope, can then be used
3.4. Passive piles, piers and caissons
3.5. Cast-in situ reinforced concrete walls in further limit equilibrium analyses to design reme-
3.6. Reinforced earth retaining structures with dial works. The limit equilibrium methods forming
strip/sheet polymer/metallic reinforcement the framework of slope stability/instability analysis
elements generally accept the Mohr-Coulomb failure criterion:
3.7. Buttress counterforts of coarse-grained
material (mechanical effect) f = c +  tan  (1)

3.8. Retention nets for rock slope faces where f and are the shear stress and effective nor-
3.9. Rockfall attenuation or stopping systems mal stress respectively on the failure surface and c and
(rocktrap ditches, benches, fences and walls)  are parameters assumed approximately constant for
3.10. Protective rock/concrete blocks
against erosion
a particular soil.
A significant limitation in the use of this criterion
4. INTERNAL SLOPE REINFORCEMENT is that the constant of proportionality is not really a
constant when wide range of stress is under considera-
4.1. Rock bolts tion. There is now considerable experimental evidence
4.2. Micropiles to show that the Mohr failure envelope exhibits signif-
4.3. Soil nailing icant curvature for many different types of soil and
4.4. Anchors (prestressed or not)
compacted rockfill. Therefore, if the assumption of a
4.5. Grouting
4.6. Stone or lime/cement columns linear failure envelope is adopted, it is important to
4.7. Heat treatment know what range of stress is appropriate to a particu-
4.8. Freezing lar slope instability problem. To avoid this difficulty
4.9. Electroosmotic anchors a curved failure envelope can be approximated by the
4.10. Vegetation planting (root strength following power law equation:
mechanical effect)
f = A( )b (2)

1788
which was initially suggested by De Mello (1977) for
compacted rockfills and subsequently found appropri-
ate for soils (Atkinson and Farrar, 1985).

2.2 Procedures for back analysis of slope failures

Shear strength parameters obtained by back analysis
ensure more reliability than those obtained by labora-
tory or in-situ testing when used to design remedial
measures. In many cases, back analysis is an effective
tool, and sometimes the only tool, for investigating
the strength features of a soil deposit (Duncan, 1999).
However one has to be aware of the many pitfalls
of the back analysis approach that involves a num-
ber of basic assumptions regarding soil homogeneity,
slope and slip surface geometry and pore pressure
conditions along the failure surface (e.g. Leroueil &
Tavenas 1981). A position of total confidence in all
these assumptions is rarely if ever achieved.
While the topographical profile can generally be
determined with enough accuracy, the slip surface is
almost always known in only few points and inter-
polations with a considerable degree of subjectivity
are necessary. Errors in the position of the slip sur-
face result in errors in back calculated shear strength
parameters. If the slip surface used in back analysis is
deeper than the actual one, c is overestimated and 
is underestimated and vice-versa.
The data concerning the pore pressure on the slip
surface are generally few and imprecise. More exactly,
the pore pressure at failure is almost always unknown.
If the assumed pore pressures are higher than the
actual ones, the shear strength is overestimated. As
a consequence, a conservative assessment of the shear
strength is obtainable only by underestimating the pore
pressures.
Procedures to determine the magnitude of both
shear strength parameters or the relationship between
them by considering the position of the actual slip
surface within a slope are discussed by Popescu and
Yamagami (1994). The two unknowns i.e. the shear
strength parameters c and  can be simultaneously
determined from the following two requirements:
a. F = 1 for the given failure surface. That means the
back calculated strength parameters have to satisfy
the c tan  limit equilibrium relationship;
b. F = minimum for the given failure surface and the
slope under consideration. That means the factors Figure 1. Shear strength back analysis methods.
of safety for slip surfaces slightly inside and slightly
outside the actual slip surface should be greater than
one (Fig.1a).
different trial sliding surfaces, is drawn and the unique
Based on the above mentioned requirements, Saito values c and tan  are found as the coordinates of
(1980) developed a semi-graphical procedure using the contact point held in common by the envelope
trial and error to determine unique values of c and and the limit equilibrium line corresponding to the
tan  by back analysis (Fig.1b). An envelope of the actual failure surface. A more systematic procedure to
limit equilibrium lines c tan  , corresponding to find the very narrow range of back calculated shear

1789
strength parameters based on the same requirements
is illustrated in Fig.1c.
The procedures discussed above to back calculate
the linear strength envelope parameters, c and  in
equation (1) can be equally applied to back calculate
the nonlinear strength envelope parameters, A and b in
equation (2) (Popescu et al., 1995). The fundamental
problem involved is always one of data quality and con-
sequently the back analysis approach must be applied
with care and the results interpreted with caution. Back
analysis is of use only if the soil conditions at failure are
unaffected by the failure. For example back calculated
parameters for a first-time slide in a stiff overconsol-
idated clay could not be used to predict subsequent
stability of the sliding mass, since the shear strength
parameters will have been reduced to their residual
values by the failure. In such cases an assumption of
c = 0 and the use of a residual friction angle, r is
warranted (Bromhead 1992). If the three-dimensional
geometrical effects are important for the failed slope
under consideration and a two-dimensional back anal-
ysis is performed, the back calculated shear strength
will be too high and thus unsafe.

2.3 Design of remedial measures based on back

analysis results
In order to avoid the questionable problem of the
representativeness of the back calculated unique set
of shear strength parameters a method for designing
remedial works based on the limit equilibrium relation-
ship c  rather than a unique set of shear strength
parameters can be used (Popescu, 1991).
The method principle is shown in Fig. 2. It is con-
sidered that a slope failure provides a single piece of
information which results in a linear limit equilibrium
relationship between shear strength parameters. That
piece of information is that the factor of safety is equal
to unity (F = 1) or the horizontal force at the slope toe
is equal to zero (E = 0) for the conditions prevailing
at failure. Each of the two conditions (F = 1 or E = 0)
results in the same relationship c tan  which for
any practical purpose might be considered linear.
The linear relationship c tan  can be obtained
using standard computer software for slope stability
limit equilibrium analysis by manipulations of trial
values of c and tan  and corresponding factor of Figure 2. Limit equilibrium relationship and design of slope
safety value. It is simple to show that in an analysis remedial measures.
using arbitrary  alone (c = 0) to yield a non-unity
factor of safety, F , the intercept of the c tan  line
(corresponding to F = 1) on the tan  axis results as: assuming  = 0 and an arbitrary c value which yield
to a non-unity factor of safety, Fc :
tan 0  = tan  /F (3)
c0  = c /Fc (4)
Similarly the intercept of the c tan  line (cor- Using the concept of limit equilibrium linear rela-
responding to F = 1) on the c axis can be found tionship c tan  , the effect of any remedial measure

1790

Figure 3. Driving vs. resisting force for stabilizing piles.
(drainage, modification of slope geometry, restraining
structures) can easily be evaluated by considering the
intercepts of the c tan  lines for the failed slope
(c0  , tan 0  ) and for the same slope after installing
some remedial works (c nec , tan  nec ), respectively
(Figure 2). The safety factor of the stabilized slope is:
 
c0  tan 0 
F = min Fc =  , F = (5)
c nec tan  nec
Errors included in back calculation of a given slope
failure will be offset by applying the same results, in
the form of c tan  relationship, to the design of
remedial measures.
The above outlined procedure was used to design
piles to stabilize landslides (Popescu, 1991) taking
into account both driving and resisting force. The
principle of the proposed approach is illustrated in
Figure 3 which gives the driving and resisting force
acting on each pile in a row as a function of the non-
dimensional pile interval ratio B/D. The driving force,
FD , is the total horizontal force exerted by the slid-
ing mass corresponding to a prescribed increase in
the safety factor along the given failure surface. The
resisting force, FR , is the lateral force correspond-
ing to soil yield, adjacent to piles, in the hatched area
1791
shown in Figure 3. FD increases with the pile interval
while FR decreases with the same interval. The inter-
section point of the two curves which represent the
two forces gives the pile interval ratio satisfying the
equality between driving and resisting force.
The accurate estimation of the lateral force on pile
is an important parameter for the stability analysis
because its effects on both the pile-and slope stability
are conflicting. That is, safe assumptions for the stabil-
ity of slope are unsafe assumptions for the pile stability,
and vice-versa. Consequently in order to obtain an eco-
nomic and safe design it is necessary to avoid excessive
safety factors.
The problem is clearly three-dimensional and some
simplification must be accepted in order to develop a
two-dimensional analysis method based on the prin-
ciples outlined above. However the only simplicity
to be accepted and trusted is the simplicity that lies
beyond the problem complexity and makes all details
and difficulties simple by a sound and profound
understanding.
3 APPLICATION
3.1 Site conditions
The described methodology is applied to a landslide
in Ohio in the United States. The site is located along
the Ohio River in south-central Ohio. A replacement
bridge was proposed at the site and site preparations
reactivated an ancient slide. The cross section is shown
in Figure 4. The slope consists of shale bedrock over-
lain by shale weathered to a residual clay. Overlying the
residual clay is alluvial silts and clays. Construction
activities at the site led to the reactivation of an ancient
slide. The slip plane discerned from surface scarps
and inclinometer data is shown in Figure 4. It can be
seen that the failure surface is planar in nature and
occurs just above the shale bedrock in the weathered
residual clay.
620
600
New Fill Existing Fill
580
560
Ohio River
Alluvial Silts
Height (ft)
540 & Clays
Shale
520
Failure Surface
500
Residual Clay
480
460
0 100 200 300 400 500 600
Length (ft)
Figure 4. Cross section of slope.

15
12
Friction Angle, deg
9 Driving Force
No Pile
6
B/D = 0.75
3
B/D = 0.50
0
0.5
0 1
Cohesion, ksf
Figure 5. Back analyzed relationship between friction angle
and cohesion.
To accommodate the new bridge a fill was pro-
posed on the existing slope, which was now moving
and would have exacerbated the instability. Hence the
use of piles to stabilize the slope was proposed.
3.2 Back analysis
Back analyses were conducted of the slope failure
using limit equilibrium techniques as described previ-
ously. The back analyzed friction angle and cohesion
for the residual clay and the failure surface are shown
in Figure 5. The resulting strength parameters vary
depending upon the water level in the Ohio River. The
relationship between the friction angle and the cohe-
sion are shown in Figure 5 for cases of no pile and pile
B/D ratios of 0.75 and 0.5. The back analyzed fric-
tion angle of about 13 for the no pile case compares
favorably with residual shear test results.
The results for the B/D ratios in Figure 5 were
obtained using the methodology proposed by Liang
(2002) and coded into an Excel spreadsheet.
3.3 Driving and resisting forces
The driving forces were determined using limit equi-
librium analyses utilizing the program XSTABL
(Interactive Software Designs, Inc. 1994) and spread-
sheet analyses. The driving forces are shown in
Figure 6 for various B/D ratios. The resisting forces
were determined using the Ito and Matsui (199x)
method as outlined by Popescu (1995). The resisting
forces are shown in Figure 6 for various B/D ratios.
From the results in Figure 6 it can be seen that
the resisting force and driving force cross at a B/D
1792
6000
5000
Force, Driving & Resisting, kips
4000
Resisting Force
3000
2000
1000
0
0.30 0.40 0.50 0.60 0.70 0.80
B/D Ratio
Figure 6. Driving and resisting forces as a function of B/D
ratio.
ratio slightly larger than 0.5 with a required resisting
force of about 1800 kips. A shear force of this magni-
tude could be obtained using six-foot diameter shafts;
however, eight-foot diameter shafts were selected to
provide a margin of safety for the drilled shafts.
4 CONCLUSIONS
This paper has outlined an approach to back analyzing
the strength parameters in a slope failure and determin-
ing the force required to stabilize a slope using piles
considering the back analysis results. The use of the
technique has been demonstrated through application
to a case history.
REFERENCES
Atkinson, J.H. & Farrar, D.M. 1985. Stress path tests to
measure soils strength parameters for shallow landslips.
Proc.11th Int. Conf. Soil Mech. Foundation Eng., San
Francisco, 2:983986.
Bromhead, E.N. 1992. Slope Stability. 2nd Edition, Blackie
Academic & Professional, London, 411 pp.
De Mello, V.F.B. (1977). Reflections on design decisions
of practical significance to embankment dams. Gotech-
nique, 27 (3):281354.
Duncan, J.M. 1999. The use of back analysis to reduce slope
failure risk. J. Boston Soc. Civil Eng., 14:1:7591.
Interactive Software Designs, Inc. 1994. XSTABL An Inte-
grated Slope Stability Analysis Program for Personal
Computers. Reference Manual, Version 5, Moscow, ID.
Ito, T. & Matsui, T. 1975. Methods to estimate landslide
forces acting on stabilizing piles. Soils and Foundations,
15:4:4359.
Liang, R.Y. 2002. Drilled shaft foundations for noise barrier
walls and slope stabilization. University of Akron, Akron,
Ohio, Report prepared for the Ohio DOT and FHWA.
Leroueil, S. & Tavenas, F. 1981. Pitfalls of back-analyses.
Proc. 10th Int. Conf. Soil Mech. Found. Eng., 1:185190.
Popescu, M.E. 1991. Landslide control by means of a row of
piles. Keynote paper. Proc. Int. Conf. on Slope Stability
Engineering, Isle of Wight, Thomas Telford, 389394.
Popescu, M.E. 1995. Keynote Lecture: Back analysis of
slope failures to design stabilizing piles. 2nd Turkish
Symposium on Landslides, 1526 October, Adapazari,
Turkey.
Popescu, M.E. 1996. From Landslide Causes to Landslide
Remediation, Special Lecture. Proc. 7th Int. Symp. on
Landslides, Trondheim, 1:7596.
Popescu, M.E. 2001. A Suggested Method for Report-
ing Landslide Remedial Measures. IAEG Bulletin, 60,
1:6974.
Popescu, M.E. & Yamagami, T. 1994. Back analysis of slope
failuresa possibility or a challenge? Proc. 7th Congress
Int. Assoc. Eng. Geology, Lisbon, (6), p. 47374744.
1793
Popescu, M.E., Yamagami, T. & Stefanescu, S. 1995. Non-
linear strength envelope parameters from slope failures.
Proc.11th ECSMFE, Copenhagen, (1), p. 211216.
Morgenstern, N.R. 1992. Keynote Paper: The role of analysis
in the evaluation of slope stability. Proc. 6th International
Symposium on Landslides, Christchurch, 3:16151629.
Saito, M. 1980. Reverse calculation method to obtain c and
on a slip surface. Proc. Int. Symp. Landslides, New Delhi,
1:281284.
Terzaghi, K. 1950. Mechanisms of Landslides, Geological
Society of America, Berkley, 83123.