Lateral Earth Pressures in Constrained Landslides

Friedli, B. et al. (2017). Géotechnique 67, No. 10, 890–905 [http://dx.doi.org/10.1680/jgeot.16.P.
158]
Lateral earth pressures in constrained landslides

B. FRIEDLI , D. HAUSWIRTH† and A. M. PUZRIN†
The problem of the limiting landslide pressure on an obstacle was first formulated in 1944 by Robert
Haefeli of ETH Zurich, who recognised that the kinematics of the problem does not allow for classical
active and passive earth pressure theories to be applied. He derived an approximate solution using a
limit equilibrium approach with a number of rather arbitrary assumptions and simplifications. Since
then, the Haefeli solution has been widely applied for the design and analysis of landslide retaining
structures. The paper revisits this old landslide pressure problem by means of a rigorous upper- and
lower-bound limit analysis and derives the exact landslide pressure solution for a planar landslide with a
weak slip surface parallel to the slope. Being applicable to a wide range of natural and man-made
obstacles and, unlike the classical theories, not affected by the wall friction and soil dilation, the upper-
bound solution is rather robust. The landslide pressures from this solution increase with the strength of
the sliding layer and are significantly higher than the active, but much lower than the passive, earth
pressures. Of even higher practical importance, however, is that due to their oversimplifying
assumptions, the widely used approximate solutions appear to get close to the exact solution only
over a very narrow range of slope and friction angles. It appears that for mildly inclined weak slip
surfaces and high strengths of the sliding layer, analysis and design of retaining structures based on well-
known approximate solutions can become dramatically unsafe.
KEYWORDS: creep; earth pressure; landslides; limit state design/analysis; plasticity; soil/structure
interaction
INTRODUCTION the limiting pressure in the sliding layer is in particular

Landslides are a common geohazard in mountainous and important to know when the sliding layer serves as a
alpine regions and have been studied since early times of foundation for buildings and infrastructure, as often is the
geotechnical engineering (e.g. Terzaghi, 1936; Skempton, case with slowly moving landslides in mountainous commu-
1964; Bjerrum, 1967). The sliding of the soil mass often takes nities (e.g. Puzrin & Schmid, 2011, 2012). The failure at the
place along a narrow slip surface with reduced strength in bottom of the sliding layer affects the kinematics of the
which most of the shear strains are localised (e.g. Bernander landslide displacements, both at the location of the failure
& Olofsson, 1981). In such cases, the driving component mechanism (change in the sliding direction) and upslope
(parallel to the slip surface) of the gravity force acting in the from it (landslide acceleration), which can be critical for the
sliding layer is approximately in equilibrium with the shear building and infrastructure performance.
resistance of the slip surface. The velocity of the sliding soil Earth pressures on retaining structures are traditionally
mass is controlled by fluctuation of the pore water pressures, estimated as a limiting case where the surrounding soil
rate-dependent behaviour, strain hardening or softening reaches shear failure (e.g. Coulomb, 1776; Rankine, 1857).
in the sliding layer and the slip surface, and so on Depending on the relative displacement between the struc-
(e.g. Oberender & Puzrin, 2016). The velocities may be ture and the soil, the failure takes place either in active or
very small and the process of sliding can extend over decades. passive mode. Several authors have refined the earth
Some landslides have been constrained by an obstacle, either pressure theory using limit equilibrium and limit analysis,
natural (a rock outcrop) or artificial (a retaining structure, (e.g. Müller-Breslau, 1906; Caquot & Kérisel, 1948;
designed as a mitigation measure). Puzrin & Sterba (2006) Sokolovskii, 1960; Chen, 1975; Lancellotta, 2002, etc.).
and Puzrin & Schmid (2011, 2012) showed that in such These earth pressure theories are widely used in engineering
constrained landslides the pressure acting in the sliding layer practice for the design of retaining structures. While for the
increases towards the obstacle and may reach values produ- classical kinematic conditions (wall moving away or towards
cing failure in the sliding layer, causing significant accelera- the soil, Figs 1(a) and 1(b)) the calculation of the limiting
tion of the sliding body. earth pressures follows well-established techniques, for the
This paper investigates the limiting landslide pressures special kinematic condition of a slope moving towards a
acting in the sliding layer and on the constraining obstacle. fixed retaining structure or rock outcrop, derivation of the
While the limiting pressure on the retaining structure earth pressures remains a subject of ad hoc procedures, often
provides crucial input for the design of retaining structures, using rather arbitrary assumptions.
Indeed, if the retaining wall were moving downslope faster
than the sliding layer, the active failure could take place in the
soil behind the wall. In contrast, when the sliding layer moves
faster than the retaining structure, this relative movement
Manuscript received 23 June 2016; revised manuscript accepted
seems to resemble the classical passive failure case. In reality,
11 January 2017. Published online ahead of print 20 February 2017.
Discussion on this paper closes on 1 March 2018, for further details however, because the retaining structure or rock outcrop is
see p. ii. hardly moving at all, neither of the above two cases applies
Institute for Geotechnical Engineering, ETH Zurich, Switzerland and kinematic considerations (Fig. 1(c)) imply a special case
(Orcid:0000-0003-2178-2761). for the earth pressure in constrained planar landslides,
† Institute for Geotechnical Engineering, ETH Zurich, Switzerland. denoted in this paper as the ‘landslide pressure’.
890
Downloaded by [ Indian Institute Of Technology - Kharagpur] on [25/07/19]. Copyright © ICE Publishing, all rights reserved.
LATERAL EARTH PRESSURES IN CONSTRAINED LANDSLIDES 891
v1 v1 v0
v0 v0 v1
(a) (b) (b)
Fig. 1. Kinematics of different limiting earth pressure situations: (a) active; (b) passive; (c) the landslide case
Based on his observations of snow avalanche barriers, (1967) presented a rigorous analysis of the problem of a snow
Haefeli (1944) realised that the sliding mass at the limit state sheet in an avalanche or a landslide under extension or
flows over the barrier. He proposed a solution to calculate the compression using the Tresca and the Mohr–Coulomb
landslide pressure acting on a retaining structure based on failure criterion, respectively. They found that the stress
limit equilibrium and some ad hoc assumptions (in particu- state in the moving body is limited by an active and passive
lar, with respect to the shape of the failure wedge and state and presented analytical expressions for the limit states.
inclination ϑ of the force acting on this wedge in Fig. 2(a)). Szczepinski (1972) extended their solution to non-dilative
The value of inclination ϑ has to be postulated within a range flow and also presented an analytical solution for a landslide
between zero and the angle of internal friction ϕ′, which body in compression.
results in a broad range of the landslide pressure coefficients The literature dealing with limiting pressures on
(Fig. 2(b)). It is, therefore, hardly surprising that when Brandl slope-stabilising piles has mainly focused on the flow of
& Dalmatiner (1988) compared the Haefeli solution to field cohesive soil through the piles (e.g. Ito & Matsui, 1975; De
observations, they fell within this range (Fig. 2(b)). Beer, 1977; Viggiani, 1981; Randolph & Houlsby, 1984;
Considered to have been ‘confirmed’ by Haefeli’s solution, Poulos, 1995; Kourkoulis et al., 2011 (finite-element
this measured range has since been broadly used in method)) and has not addressed the case with a continuous
engineering practice. wall. Besides these theoretical studies, field measurements of
In spite of the long history and rather broad application of earth pressures in landslides have been provided by Fukuoka
the Haefeli solution, its arbitrary assumptions remain a (1995).
subject of concern and have caused a number of authors to The review of the literature addressing the landslide
seek alternative approaches. For example, for the long-term pressure acting on retaining structures has shown that,
stability analysis of constrained landslides, Puzrin & Sterba although a number of different approaches exist – including
(2006) calculated the landslide pressure using the passive case rigorous solutions for a particular case, which are not well
of Rankine’s earth pressure theory for planar inclined ground known in the field of geotechnical engineering – engineers
(Rankine, 1857; Chu, 1991). More recently, Muraro et al. face large uncertainties when applying these approaches to
(2015) also concluded, based on numerical calculations real-life landslide problems. The goal of the present work is
(finite-element method) of the landslide pressure using one to provide a more general rigorous solution of the landslide
specific geometry and a limited parametric study on the pressure problem based on its consistent formulation within
effects of the soil–wall friction and soil dilatancy, that their the framework of the upper-bound limit analysis. This
results correspond to Rankine’s passive earth pressure. For solution will then be benchmarked against the existing
the slightly different problem of glacier flow, Nye (1951) exact solution for the particular case of a planar slope and
developed stress field solutions for extension and com- its applicability will be extended to account for important
pression of the moving ice body. Ziegler (1963) and Kupper engineering phenomena (such as wall friction, wall
4·0
φ'
Estimated by Brandl &

ϑ=
Dalmatiner (1988)
feli
3·5 Estimated limits for rigid

Hae
and flexible structures

Brandl & Dalmatiner (1988)
3·0
2
φ'/
Haefeli (1944)
=
re
ctu
iϑ
u
el
str
Ratio m = Klh/Kah
2·5
ef
gid
Ha
D
ϑ Ri
V 2·0
2 e
θ 1 ctur =0
le stru feli ϑ
Fl exib Hae
1·5
3 G
1·0
R
δ χ φ'
Eg α
0·5
0
20 25 30 35 40 45
Slope angle, α = θ = φ': degrees
(a) (b)
Fig. 2. Haefeli limit equilibrium solution: (a) assumed mechanism; (b) ratio of m = Klh/Kah plotted against slope angle for the special case of slope
angle equal to the angle of internal friction α = ϕ′ (Klh: landslide pressure coefficient; Kah: active earth pressure coefficient for a smooth wall)
892 FRIEDLI, HAUSWIRTH AND PUZRIN
inclination, soil dilation, presence of groundwater, etc.). providing, in particular cases, the exact solution for the
After comparing this solution to the approximate solutions limiting landslide pressures.
used in engineering practice, two practical examples will be
presented. (a) The landslide body is at the state of equilibrium and is
moving downslope at a constant velocity (the
gravitational shear stress is in equilibrium with the shear
PROBLEM FORMULATION resistance: ϕ2′ = α).
Consider a landslide (Fig. 3) consisting of a soil layer with (b) The shear resistance ϕ2′ is mobilised along the entire
the surface inclination θ and the total unit weight γ sliding on length of the slip surface, all the way down to the
a thin layer of weaker soil inclined by angle α with reduced obstacle. This condition is satisfied by all compliant
mechanical strength properties (either the same material with retaining structures and by those rigid structures that
residual strength, or a different material, e.g. a thin layer of were installed after the initial landslide movement took
clay). In this paper, this weak layer is denoted as the ‘slip place.
surface’. Both the sliding layer and the slip surface materials (c) The soil fails as a perfectly plastic Mohr–Coulomb
are perfectly plastic following the Mohr–Coulomb failure material with the associated flow rule. Extension of the
criterion. The sliding layer is characterised by the cohesion c′ solutions to a non-associated flow rule will be made
and the friction angle ϕ′, whereas the slip surface is assumed later in the paper.
to be cohesionless with the friction angle of ϕ2′ . Below the slip
surface, the material is assumed to have a significantly higher
strength. If the soil resistance on the slip surface is in Upper-bound solution
equilibrium with the gravitational load of the sliding layer, In order to derive an upper-bound estimate for the
that is, the friction angle of the slip surface ϕ2′ is equal to its landside pressure, a kinematically admissible failure mech-
inclination α, the sliding layer can move downhill at a anism has to be postulated, where the velocity field satisfies
constant velocity. The landslide is constrained by an obstacle the velocity boundary conditions as well as the strain and
at the bottom, which may be natural (rock outcrop) or strain rate compatibility conditions. For this failure mech-
man-made (stabilisation structure, buildings, etc.), has the anism to be activated, the rate of external work has to be
vertical height H, is inclined to the vertical by angle β and has equal (or exceed) the rate of internal dissipation. According
the friction angle δ at the interface with the soil. If the to the upper-bound theorem of limit analysis (e.g. Chen,
resistance along the slip surface drops below the driving 1975), the minimum lateral earth pressure capable of
force, the pressure on the obstacle at the bottom will increase. activating such a mechanism will always be larger than the
This can be caused, for example, by fluctuations of the pore true failure stress in the sliding body.
water pressures or due to a rate-dependent material behav- The kinematically admissible translational mechanism
iour on the slip surface (particular mechanisms bringing the proposed here consists of three rigid blocks, bounded by
slope to the failure state are outside the scope of this paper, slip discontinuities with the lengths l1 and l2 inclined to the
therefore, no groundwater table is introduced in the problem slip surface by angles ω1 and ω2, respectively (Fig. 4(a)).
formulation; the influence of water is discussed later in the According to the associated flow rule, the velocity vector v1
paper). The constraining obstacle can be compliant or rigid, of block 1 is inclined to the thin sliding surface by the friction
but cannot move upslope at failure. It is also assumed that the angle ϕ2′ . Therefore, in order to satisfy the equilibrium
landslide is sufficiently long for the difference between the condition ϕ2′ = α, block 1 has to move horizontally. Block 2 is
driving and resisting forces to bring the sliding layer close to pushed upwards with velocity v2, while block 3, similarly to
the obstacle to failure. block 1, moves horizontally with the velocity v3.
In the considered problem two different limiting landslide The relations between the velocities of blocks 1, 2 and 3 are
pressures are of practical interest: (a) the landslide pressure shown in the hodograph in Fig. 4(b). For the formulation of
in the sliding layer (for the slope stability assessment) and the work equation, of particular interest are the vertical
(b) the landslide pressure acting on the retaining structure component v2v of the velocity of block 2, as well as the
(for the design criteria). The case of a structure embedded relative velocities v12 and v23 and their components v12c′ and
within the sliding layer with no connection to the stable v23c′ parallel to the corresponding slip discontinuities
ground beneath the slip surface (e.g. by means of piles or
anchors) is outside the scope of the presented work. ðv1 v3 Þ sinðϕ′ þ ω2 αÞ
v12 ¼
sinð2ϕ′ þ ω1 þ ω2 Þ
ðv1 v3 Þ sinðϕ′ þ ω1 þ αÞ
LIMIT ANALYSIS SOLUTIONS v23 ¼
Assumptions sinð2ϕ′ þ ω1 þ ω2 Þ
Based on the following assumptions, upper- and lower- ðv1 v3 Þ sinðϕ′ þ ω2 αÞ cosðϕ′Þ
bound limit analysis solutions are derived in this section, v12c′ ¼ ð1Þ
ðv1 v3 Þ sinðϕ′ þ ω1 þ αÞ cosðϕ′Þ
v23c′ ¼
ent ðv1 v3 Þ sinðϕ′ þ ω1 þ αÞ sinðϕ′ þ ω2 αÞ
Retaining Movem
structure Sliding layer v2v ¼
θ γ, c', φ'
β The rate of external work is a sum of the work done by the

H δ urface φ'2 three external forces – the force in the sliding layer Elh, the
We ak slip s
reaction of the retaining structure Echw and the gravity force
α G – on the corresponding velocities of blocks 1, 3 and 2,
respectively (Fig. 4(a)). Note that when ϕ2′ = α, blocks 1 and 3
move horizontally and the vertical components of the forces
Fig. 3. Problem formulation, retaining structure in landslide acting on these blocks do not contribute to the rate of
v1
l1
2 Elh
l2 1
β
2c’
H v3
v1
3 ω1
ω2
v2v
Elhw v
v12
α 2
v 23
v2
3c’
φ'
ω1 φ'
v1 ω2
v2
v12 φ' α α v1 α v3
v23
v3 φ
α
(a) (b)
Fig. 4. (a) Upper-bound mechanism; (b) hodograph
external work Ẇ e , while block 2 is the only one where the The presented velocity field is only valid for 0 X , 1
gravity contributes to the external work (X = v3/v1 1 is anyway not relevant for constrained land-
slides). In order to quantify both forces Elh and Elhw,
Ẇ e ¼ v1 Elh v3 Elhw v2v G2 ð2Þ equation (6) alone is not sufficient and additional consider-
with the weight of block 2 given by ations are required. One such consideration is that for
0 , X , 1, that is, in the case of a compliant retaining
H 2 γ cos2 ðα βÞ sinðω2 α þ θÞ sinðω1 þ ω2 Þ structure with v3 . 0, the shear resistance ϕ2′ will be mobilised
G2 ¼ ð3Þ along the entire length of the slip surface, all the way down to
2 cos2 ðβÞ sin2 ðω2 Þ sinðω1 þ α θÞ
the structure. Michalowski (1989) and later Drescher &
The rate of internal dissipation along the slip discontinuity Detournay (1993) demonstrated that, if a translational
Ḋi is calculated as kinematic mechanism forms a statically determinate
problem, the limit equilibrium method applied to this
Ḋi ¼ v12c′ c′l1 þ v23c′ c′l2 ð4Þ mechanism should provide the same upper-bound solution
as the work equation. When the shear resistance ϕ2′ is
where
mobilised along the entire length of the slip surface
H cosðα βÞ sinðω2 þ θ αÞ (i.e. also below block 3 in Fig. 4), the kinematic mechanism
l1 ¼ becomes statically determinate and the same upper-bound
cosðβÞ sinðω2 Þ sinðω1 θ þ αÞ
ð5Þ landslide pressure can also be found by formulating limit
H cosðα βÞ equilibrium of blocks 1, 2 and 3 (Fig. 5). It follows that,
l2 ¼
cosðβÞ sinðω2 Þ because out of all the external forces acting on the three
blocks (G1, G2, G3, R1, R3, Ec, Ecw) only the latter two have
By equating the rate of external work (2) and the rate of horizontal components, these two components have to be
internal dissipation (4), the limiting force in the sliding layer equal to satisfy the equilibrium: Elh = Elhw. When substituted
Elh and the corresponding force acting on the retaining into equation (6), this relationship gives
structure Elhw can be related depending on the two variable
angles of the slip lines ω1 and ω2 Elh ¼ Elhw ¼ f ðH; γ; α; β; θ; ϕ′; c′; ω1 ; ω2 Þ ð8Þ
In other words, the upper bounds of horizontal components
Elh X Elhw ¼ ð1 X Þ f ðH; γ; α; β; θ; ϕ′; c′; ω1 ; ω2 Þ of both landslide forces are equal and independent of X for
ð6Þ any 0 , X , 1. Note that the horizontal component of the
landslide pressure acting on the wall is independent from the
where X = v3/v1 and wall–soil interface friction angle δ, unlike in the classical
( active and passive cases. This results from the fact that the
H 2 γ cos2 ðα βÞ sinðω2 α þ θÞ sinðω1 þ ω2 Þ
f ¼ only external horizontal forces in the system are Elh and Elhw.
2 cos2 ðβÞ sin2 ðω2 Þ sinðω1 θ þ αÞ From the above considerations it follows that the special
case of X = 0, which corresponds to a rigid obstacle with
sinðϕ′ þ ω1 þ αÞ sinðϕ′ þ ω2 αÞ v3 = 0, can also be described by equation (8), if this obstacle

sinð2ϕ′ þ ω1 þ ω2 Þ was installed after the landslide had moved sufficiently far to
mobilise the shear resistance ϕ2′ along the entire length of the
Hc′ cosðα βÞ cosðϕ′Þ sinðϕ′ þ ω1 þ αÞ
þ slip surface. If, however, the rigid obstacle existed before the
cosðβÞ sinð2ϕ′ þ ω1 þ ω2 Þ sinðω2 Þ landslide moved, the slip surface resistance in its vicinity will
not be mobilised and equation (8) can only be used for
sinðϕ′ þ ω2 αÞ sinðω2 þ θ αÞ
þ calculating the landslide force in the sliding layer Elh, while
sinðω2 Þ sinðω1 θ þ αÞ
the force on the obstacle Elhw . Elh remains undetermined.
ð7Þ The effect of the not fully mobilised resistance ϕ2′ can be
El
G1
Sc' 1 φ'
2 R12
G2 R 12
φ' Sc' 1
φ' Sc' 2
R23 R23
Sc' 2 φ' R1 φ'
Elw 2
G3 3
δ
R3
φ'2
3 2 1 El
Elhw α
Elh
ω1
δ
Elw G3 R12
R1
β φ'
Sc' 1
π/2 – φ' – ω2 + α G2
R3 R23
G2
R12
R23 Sc' 2 Sc' 1
Sc' 2
Fig. 5. Limit equilibrium and corresponding force polygons
evaluated approximately by changing the angle of the force Assuming constant stresses in the sliding layer in the
R3 in Fig. 5. This evaluation is, however, not within the scope direction of the slip surface ∂τnt/∂t = ∂σt/∂t = 0, integrating the
of the present paper. While for natural rock outcrops Elhw is equilibrium equations in the normal direction and trans-
of limited practical significance, for man-made structures it forming the normal coordinate n = z cos α into the depth of
represents an important design factor. However, the fact that the sliding layer z yields
they will have to withstand higher pressures than the failure
pressure in the sliding layer becomes a strong argument σ n ¼ γz cos2 α
ð10Þ
against the use of pre-installed rigid structures as a protection τ nt ¼ γz sin α cos α
measure against future landslides.
Finally, since in equation (2) the term v1Elh is positive, the which satisfy the boundary condition τnt = σntanα on the slip
minimum work principle requires Elh to be minimised with surface (defined by ϕ2′ = α). The slip surface acts as a stress
respect to the angles ω1 and ω2 of the slip discontinuities and discontinuity and the stress state below the slip surface has to
Elh obtained from the work equation represents an upper pass the stress point (σn, τnt) in Fig. 6(b) and must not violate
bound for the landslide force which is required to fail the the failure condition.
landslide body in the vicinity of the obstacle. The question Combined with the failure criterion for the soil in the
remains how good is this upper bound; that is, how close it is sliding layer (Fig. 6(a)), equation (10) allows definition of the
to the exact solution. stress field from the Mohr circle shown in Fig. 6(b). The large
circle represents the stress state in the sliding layer and the
small circle represents the stress state in the slip surface.
Substituting the stress state defined by equation (10) into the
Lower-bound solution equation of the Mohr circle gives
A lower-bound solution requires a statically admissible
stress field, namely, the stress field that stays in equilibrium ðσ m σ n Þ2 þτ 2nt ¼ r2 ¼ σ 2m sin2 ϕ′ ð11Þ
internally, satisfies external stress boundary conditions and
does not violate the failure criterion (e.g. Chen, 1975). A which can be resolved with respect to the mean stress σm
statically admissible stress field in the present problem with pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
arbitrary slope angle θ, slip surface angle α and inclination of 1 + 1 cos2 ϕ′ð1 þ tan2 αÞ
σm ¼ σn ð12Þ
the retaining structure β with frictional–cohesive material is cos2 ϕ′
difficult to find. Hence, only the specific case with α = θ = ϕ2′ ,
β = c′ = 0 is considered below, which corresponds to the The horizontal normal stress σx and the corresponding
particular case investigated by Muraro et al. (2015). The shear stress τxz acting on a vertical plane are also related by
boundary between the sliding layer and the slip surface is way of the Mohr circle equation
introduced as a static discontinuity (i.e. the normal σn and ðσ x σ m Þ2 þ τ 2xz ¼ r2 ¼ σ 2m sin2 ϕ′ ð13Þ
shear stresses τnt are continuous but the tangential stress σt
may be discontinuous). The plane strain equilibrium Because on the slip surface ϕ2′ = α and τnt = σntanα, the pole
equations (bearing in mind τnt = τtn) are P of the Mohr circle in Fig. 6(b) is located on the straight line
@σ t @τ nt with inclination α passing through the origin, so that
þ γ sin α ¼ 0
@t @n τ xz ¼ σ x tan α ð14Þ
ð9Þ
@σ n @τ nt that is, the force Elw acting on the vertical wall has the same
þ γ cos α ¼ 0
@n @t inclination as the slope.
exact solution. In particular for steep slopes, where the true
t α passive earth pressures should increase with increasing slope
x inclination, the values given by this lower-bound solution
z n ξ σI decrease. In contrast, as is shown in the following section, for
α σ II the problem of landslide pressures this not optimal lower
α ξ ω1
ω2 bound from a different problem appears to provide the exact
Elw
α
solution!
(a)
THE EXACT SOLUTION
τ
For the special case of α = θ and c′ = β = 0, it can be shown
φ' that the above presented upper- and lower-bound solutions
become identical, providing the exact solution for the
problem of landslide pressure, as was demonstrated by
ω1 Kupper (1967) and Szczepinski (1972). Indeed, for this
σz,τxz ω2 P α = φ'2 case the upper-bound solution (8) simplifies to
ξ
σn,τnt ξ H 2 γ cos2 ðαÞ sinðω1 þ ω2 Þ
Elhw ¼
σII σm σI 2 sinðω2 Þ sinðω1 Þ
ð17Þ
sinðϕ′ þ ω1 þ αÞ sinðϕ′ þ ω2 αÞ

σ x,τzx sinð2ϕ′ þ ω1 þ ω2 Þ
In order to find the lowest upper bound, in general, critical
inclinations of the slip discontinuities ω1 and ω2 in Fig. 4
have to be found. Consider a case where these two
(b)
inclinations are not independent and related by way of the
same relationship
Fig. 6. (a) Statically admissible stress field for slope parallel slip π
surface and vertical wall. (b) Mohr circle of the stress state in the ω1 þ ω2 ¼ ϕ′ ð18Þ
2
weaker layer (slip surface, dashed circle) and in the sliding layer (solid
circle) as the inclinations of the failure planes in the lower-bound
solution (see Mohr’s circle in Fig. 6(b)). This relation further
simplifies the upper-bound solution to
Solving equations (13) and (14) results in the horizontal
normal stress acting on a vertical plane H 2 γ cos2 ðαÞ sinðω1 þ ϕ′ þ αÞ cosðω1 þ αÞ
Elhw ¼ ð19Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 sinðω1 Þ cosðω1 þ ϕ′Þ
1 + 1 cos2 ϕ′ð1 þ tan2 αÞ
σx ¼ σm ð15Þ Solving equation dEchw(ω1)/dω1 = 0 with respect to ω1
1 þ tan2 α
gives the critical inclination
where only the case with higher horizontal stress is of interest.
The total force on the wall is found by substituting the mean 1 sinðαÞ
ω1 ¼ ½arccosðtÞ ϕ′ α; where t ¼ ð20Þ
stress from equation (12) into equation (15) and integrating 2 sinðϕ′Þ
this stress over the thickness of the sliding layer. A lower
bound for the landslide pressure coefficient Klhw is then providing the lowest upper bound for the landslide force
given by under assumption (18)
σ x 2Elhw cos4 α H 2 γ cos2 ðαÞ sin½arccosðtÞ þ α þ sinðϕ′Þ

Klhw ¼ ¼ ¼ Elhw ¼
γz γH 2 cos2 ϕ′ 2 sin½arccosðtÞ α sinðϕ′Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 ð16Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
H γ cos ðαÞ
2 4
1 þ 1 cos2 ϕ′ð1 þ tan2 αÞ ¼ 1 þ 1 cos2 ðϕ′Þ½1 þ tan2 ðαÞ
2 cos2 ðϕ′Þ
ð21Þ
Algebraically, this solution corresponds exactly to
another lower-bound solution, which Rankine (1857) pro- which is identical to the lower-bound solution (16), therefore,
posed for a rather different problem of the classical passive representing the rigorous exact solution of the landslide
earth pressure in inclined planar ground (Fig. 1(b)). This is pressure problem for the particular case of α = θ and a vertical
due to the fact that in both problems the same assumption retaining structure in cohesionless soil. This result is
(10) has been adopted for the planes parallel to the slope consistent with the findings of Muraro et al. (2015), who,
surface, although the implications of this assumption are based on the results of the finite-element analysis of a
very different for the two problems. Indeed, whereas in the particular case of the wall and slope geometry (using a linear
Rankine passive pressure formulation (with homogeneous elastic–perfectly plastic constitutive model), found that the
soil of strength ϕ′) this assumption does not produce the true numerically calculated earth pressures were close to those
stress field, in the landslide pressure problem, the same obtained using Rankine’s solution for passive pressure.
assumption (10) allows for the true failure state to be Assuming linear pressure distribution in the sliding layer,
achieved at the slope parallel slip surface with a lower the exact landslide pressure coefficients Klhw for five different
strength ϕ2′ = α. It is, therefore, hardly surprising that for the friction angles are plotted in Fig. 7 as a function of the slope
problem it was originally supposed to solve (i.e. the wall angle α, together with active and passive earth pressure
moving towards the soil), this Rankine lower bound coefficients for inclined planar ground and frictionless
appeared to be only useful for special cases (e.g. α = 0), vertical wall (δ = β = 0), derived from Coulomb (1776) and
providing in general values that are significantly below the Müller-Breslau (1906) solutions.
6 depend on the wall friction δ, in the lower-bound solution the
Passive
stress state at the vertical wall (14) results in the resultant
θ force Elw parallel to the slope. This implies that the wall–soil
5 interface friction angle should be higher than the slope
Horizontal earth pressure coefficient, Kh
inclination (δ α), limiting the applicability range of the

α exact solution (21). Note, however, that the results from
4 Muraro et al. (2015) support the conclusion from the upper-
bound solution (8) that the landslide force is likely to be
Exact solution independent of the wall–soil friction even if δ , α.
3
Landslide
pressure
2
φ' = φ' φ' φ' φ' Extension to arbitrary wall inclination
=2 = = =
20
° 5° 30
°
35
°
40
°
The upper-bound solution for arbitrary wall inclination β
cos2(α) is given by equation (8). If it was possible to find a
1 lower-bound solution, identical to equation (8), it would
provide the exact solution for arbitrary wall inclination.
Active From a stress field in Fig. 8 it follows that the only horizontal
0 components of stresses act on the wall with inclination β (left)
0 5 10 15 20 25 30 35 40 and the vertical cut (right). Hence, the horizontal force
Slope angle, α = θ : degrees components Elh and Elhw have to be equal and the
corresponding landslide pressure coefficients Klh(β = 0) and
Fig. 7. Earth pressure coefficients on retaining structures: solid lines Klhw(β) are related by way of the squared ratio between the
correspond to the exact solution for the landslide pressure, plotted heights H and Hz
together with the active and the passive earth pressure coefficients for
planar inclined ground and a smooth vertical wall, according to Hz2
Coulomb (1776) and Müller-Breslau (1906) (dashed and dash-dotted Klhw ðβÞ ¼ Klhw ðβ ¼ 0Þ ð22Þ
lines, respectively) H2
where
Although it is of little practical significance to calculate cosðα βÞ

Hz ¼ H ð23Þ
landslide pressures for very mild slope angles (only very few cos α cos β
terrestrial landslides have slope angles below 15°), it is
interesting to compare their theoretical pressure coefficients It follows that the lower-bound landslide pressure coeffi-
to the classical earth pressure theory. It appears that for the cient Klhw(β) can be expressed using the lower-bound
mildest (horizontal) slope, the landslide pressure coefficient is solution (16) for Klh(β = 0) in the case of the vertical wall.
identical to that for the passive earth pressure, while for cos2 ðα βÞ
another extreme case, of the steepest (unstable) slope with Klhw ðβÞ ¼ Kchw ðβ ¼ 0Þ
α = ϕ′, the landslide pressure is equal to the active earth cos2 α cos2 β
pressure, given by Kah(ϕ′ = α) = cos2α. However, as is seen in cos2 ðα βÞ cos2 α
¼ ð24Þ
Fig. 7, in between these two extreme cases there is a significant cos2 β cos2 ϕ′
difference between the landslide pressure and both the active qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
and passive pressures. While passive pressure coefficients are 1 þ 1 cos2 ϕ′ð1 þ tan2 αÞ
overly conservative, using active pressure values as an estimate
of the landslide pressure is in particular unsafe, since they
predict decrease in the earth pressure coefficients with On the other hand, in the upper-bound solution (8) for
increasing strength of the sliding layer ϕ′, whereas the true c′ = 0, the wall inclination β is only present inside the
landslide pressures are seen to increase with strength. proportionality coefficient for the landslide force.
It follows that the minimised upper bound from Therefore, the optimal inclination of the slip lines ωi is not
equation (8) degenerates, in a particular case, to the known affected by the wall inclination, resulting in the same
exact solution, providing some confidence for the application optimisation as in equations (17)–(21). The resulting upper
of this upper bound to more general cases. It appears, bound appears to be identical to the lower bound given by
however, that the application of the limit analysis solutions equation (24), providing the exact solution for the case of
can also be extended to account for more realistic soil and β = 0.
wall conditions. If a large inclination of the wall β is accompanied by a low
wall–soil friction it is possible that the landslide pressure is
different from the presented exact solution. In such a case it
EXTENSION OF THE SOLUTION TO MORE
GENERAL CASES β
In fact, the exact solution (21) derived for the special case
of the vertical wall, with the limitations imposed on the flow
rule and the wall–soil friction, can be extended to more γ σx,τzx
HZ
general cases in cohesionless soil c′ = 0 and α = θ. Note that H
each extension is only valid for the whole range of α
σw,τwβ
corresponding parameters when it is considered individually.
δ
σn,τnt
Extension to arbitrary wall friction
Note, that while in the upper-bound solution (8) the Fig. 8. Extension of the lower-bound solution to arbitrary wall
horizontal component of the landslide force Elhw does not inclination, β
might be ‘easier’ for the landslide body to fail directly on the Fig. 5. Another argument is that the finite-element para-
soil–wall interface. A limiting condition is shown after which metric study by Muraro et al. (2015) has also shown
the exact solution will no longer be valid. The expression is practically no influence of dilation on the landslide pressures.
derived by comparison of the landslide force of the exact Finally, in Appendix 1 it is demonstrated that, for the
solution (21) with a modified landslide force formulated assumption of linear slip discontinuities and the correspond-
using a mechanism involving the wall–soil interface as
second slip discontinuity.
8 9
>
> H 2 γ cosðα βÞ cosðβ δÞ sin 2 tan1 ðΓÞ α þ β sinðϕ′ þ βÞ >
>
>
> Elh2 ¼ >
>
< 2 cos ðβÞ
2 sinðϕ′ þ α þ δ βÞ þ sin½2 tan ðΓÞ δ þ β = 1
Elh ¼ min ð25Þ
>
> qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 >
>
>
> H 2 γ cos2 ðα βÞ cos2 α >
>
: Elh ¼ 1 þ 1 cos 2 ϕ′ð1 þ tan2 αÞ
;
2 cos β
2 cos ϕ′
2
With
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sinðϕ′ þ αÞ sinðα þ δ 2βÞ þ sin2 ðϕ′ þ αÞ þ 2 sinðϕ′ þ αÞ cosðα þ δ 2βÞ sinðϕ′ þ δÞ þ sin2 ðϕ′ þ δÞ sin2 ðα δÞ
Γ¼
sinðα δÞ sinðϕ′ þ αÞ cosðα þ δ 2βÞ sinðϕ′ þ δÞ
ð26Þ
For cases where the exact solution is no longer valid, the

modified landslide force Elh2 can be used as an upper bound ing linear stress field, the non-associated solution is equiv-
for the force acting on the retaining structure. alent to the exact solution with the associated flow rule.
Extension to arbitrary soil dilation Cases with variable thickness of the sliding layer
In order to fulfil the conditions of limit theorems, the and cohesive soils
associated flow rule has been assumed, which for real soils is For θ . α and c′ . 0, the upper-bound solution presented
not a good approximation of the constitutive behaviour at here does not necessarily represent the exact solution.
failure, in particular, on the slip surface where a significant However, as the general upper-bound solution (8) over-
amount of shearing leads to the critical state condition. It is, estimates the landslide pressures, it can still be used for the
however, very likely that the present problem is not affected design of retaining structures. Although this is a safe solution
by soil dilation significantly and that the proposed exact for the retaining structure, it is an unsafe solution if used as
solution is also valid for a non-associated flow rule. One way the limiting stabilisation force in a slope stability calculation.
to support this statement is to follow the general argument of For a fixed friction angle of ϕ′ = 30°, Fig. 9 shows the
Drescher & Detournay (1993) by using the fact that the same landslide earth pressure coefficient Klhw for sliding layers
force values can also be calculated, without accounting for with (a) variable thickness (also see Table 2 in Appendix 3)
the flow rule, from the limit equilibrium analysis of the and (b) cohesion. Compared to the special case of α = θ, the
statically determinate, translational kinematic mechanism in landslide pressure increases significantly with increasing
5·0 ° 4·0 °
5·0 5·0
Angle of internal friction φ' = 30° 1 θ = 1 ·5°
4·5 cohesion c' = 0 kPa α= α = θ = 17 0°
·
Horizontal earth pressure coefficient, Klh
3·5
Horizontal earth pressure coefficient, Klh
° α = = 20 °
7·5 θ 5
4·0 1 α = 2 ·
2
α= θ= 0°
3·0 α = = 25·
3·5 0° = θ · 5°
0· α 7
α =2 θ =2 °
3·0 2·5 α = = 30·0
5° θ
=2
2· α=
2·5 α 2·0
°
2 5·0
2·0 α=
1·5
Exa ° θ
1·5 ct so 2 7·5
lutio α=
θ nα= 1·0
θ
1·0
α
0·5 0·5
Angle of internal friction, φ' = 30°
α
0 0
15 20 25 30 0 0·05 0·10 0·15 0·20 0·25 0·30
Slope surface angle, θ : degrees c'/(H·γ )
(a) (b)
Fig. 9. (a) Influence of variable thickness of the sliding layer θ α shown for the case of ϕ′ = 30° in the upper-bound solution. (b) Influence of the
cohesion c′ for the case of ϕ′ = 30° in the upper-bound solution
slope surface angles θ. Soil cohesion in the sliding layer is condition. Therefore, the use of the presented exact solution
also seen to increase the landslide pressure coefficient. for dry condition (21) represents a safe estimate for the
design of a retaining structure. However, if used as the
limiting stabilisation force in a slope stability calculation this
Accounting for groundwater solution is not safe, and a more complex lower-bound
Planar landslides normally exhibit a slope parallel ground- solution for the slope parallel flow should be used as a safe
water flow (Fig. 10(a) case i). However, in the vicinity of the estimate.
obstacle the flow may deviate from the slope parallel case, for
example, owing to the drainage of the water at a retaining
wall (Fig. 10(a), case ii). Since the permeability of the soil is COMPARISON WITH EXISTING SOLUTIONS
often anisotropic and may be spatially variable, estimating The problem of lateral landslide pressures was first studied
the exact transient flow field in the vicinity of the obstacle by Haefeli (1944). He identified the special kinematical
may prove difficult. The real phreatic surface and the situation to which the classical active and passive pressure
corresponding water pressures will often lie between the dry solutions are not applicable and tried to derive a solution for
case (Fig. 10(a) case iii), and the case with slope parallel flow the landslide pressures using limit equilibrium. This solution
(Fig. 10(a) case i). Therefore, the landslide pressures will also is widely applied in the practice of geotechnical engineering
lie between the dry and the slope parallel flow cases and in order to assess the landslide pressure acting on retaining
provide practically applicable bounds for the design of structures. Brandl & Dalmatiner (1988) used Haefeli’s
retaining structures and the assessment of constrained land- solution for the special case of a slope angle equal to the
slides. For the dry case, the pressures are provided by the angle of internal friction α = ϕ′ and compared it to field
exact solution (21). For the slope parallel case, derivation of observations. They found out that the landslide pressure
the upper- and the lower-bound solutions is presented in depends on the stiffness of the retaining structure and
Appendix 2. compares well to the solution of Haefeli. Their findings are
Note that for the slope parallel case the presence of the published in textbooks (Brandl, 1987, 2001) and are
groundwater reduces the effective stresses and therefore the frequently used in engineering practice. In this section these
strength of the slip surface ϕ2′ has to be larger than its solutions and their assumptions are compared to the
inclination to fulfil the condition of constant velocity of the presented exact solution.
sliding layer Haefeli (1944) assumed the failure mechanism shown
γH in Fig. 2(a). A sliding layer on a rock surface was considered.
tan ϕ′2 ¼ tan α ð27Þ Although the soil mass has to slide towards the retaining
γH γw Hw
structure, no localised weak slip surface was introduced
In other words, the landslide is only activated when the (it is not clear why the slide would occur if the strength of soil
phreatic surface reaches the height Hw. The results of the is higher than the slope angle). The solution is based on the
upper- and lower-bound limit analysis solutions from formulation of limit equilibrium for block 2 and block 3 in
Appendix 2 are shown in Fig. 10(b) for the total normalised Fig. 11(a), introducing some arbitrary assumptions.
landslide force Klh,tot = 2Elh,tot/(H 2γ) depending on the ratio
Hw/H for six different slope angles α. The two solutions • The slip surface between blocks 2 and 3 is fixed as
are no longer identical for the whole range of parameters horizontal, which does not allow the solution to be
and, therefore, they do not represent the exact solution. optimised.
However, it can be concluded that the presence of water • On the vertical boundary of block 2, inclination ϑ of the
will always weaken the sliding body and the force needed to force D is a free parameter, while the proper application of
fail the soil in the vicinity of a constraining obstacle will limit equilibrium would require failure on this slip
always be smaller than the one of a landslide in dry surface, that is, ϑ = ϕ′.
3·5
Angle of internal friction φ' = 30°
Normalised total horizontal landslide force, Klh,tot
Soil unit weight γ = 20 kN/m3

3·0 α = 0°
α = 5°
α = 10°
2·5
α = 15°
2·0 α = 20°
1·5
α = 25°
i
H ii
Hw
1·0
= φ
φ' 2 θ
iii
0·5 H
Lower bound Hw
Upper bound α
0
0 0·25 0·50 0·75 1·0
Ratio Hw/H
(a) (b)
Fig. 10. (a) Schematic sketch of possible groundwater flow in a landslide: (i) slope parallel flow; (ii) drained flow; (iii) no groundwater in the sliding
layer. (b) Normalised total horizontal landslide force acting on a vertical wall with slope parallel groundwater flow for ϕ′ = 30° and γ = 20 kN/m3
3·5
Angle of internal friction φ' = 30°
α=θ θ
3·0
Horizontal earth pressure coefficient, Kh

Ex
D 2·5 lim act α
ϑ it a sol
V na utio
2 lys n
is
2·0
φ' K
φ'
K
1·5
30°
φ '= °
G δ= = 20
3 /3 φ'
1·0 cos2(α) δ=2 0
δ=
δ Haefeli’s solution
Eg R φ' 0·5
Active
0
0 5 10 15 20 25 30
Slope angle, α = θ: degrees
(a) (b)
Fig. 11. Haefeli limit equilibrium solution: (a) blocks with edge forces; (b) comparison to exact solution
Equilibrium on blocks 2 and 3 gives the following landslide pressure coefficient Klhw is normalised by the
expression for the Haefeli landslide force acting on the horizontal active earth pressure for a smooth wall
retaining structure Kah(δ = 0) = cos2α. Although the comparison looks rather
favourable for Haefeli’s solution (subject of the correct guess
h2 tan θ 1 1 þ ðtan χ= tan θÞ of the inclination ϑ), the same cannot, unfortunately, be said
Eg ¼ γ þ
2 tan2 ϕ′ cos δ cot ϕ′ tan δ tan δ þ cotðχ ϕ′Þ about the exact solution m = Klhw/Kah = 1, which unlike
ð28Þ Brandl’s observations and Haefeli’s solution does not
exhibit in this special case any increase with increasing
slope (and friction) angle.
The subscript g is chosen according to the original How is it possible that an approximate solution fits the
publication by Haefeli (1944), it refers to the word ‘gleit’, field observations better than the exact solution derived
which is German for ‘slide’. The force depends on the angle χ within the framework of limit analysis? The most likely
and has to be maximised with respect to χ, which is explanation of this paradox is the inaccuracy of the extreme
constrained by the rock surface angle χ α. In the case of a ϕ′ = α assumption. In fact, by relaxing the α = ϕ′ constraint, a
slope parallel rock surface θ = α with an inclination close to good fit of the exact solution to Brandl’s estimate for flexible
the angle of internal friction ϕ′, the angle χ is usually taken as structures (Fig. 2(b)) can be achieved by the slight increase of
equal to α. 5–10% in friction angle of the sliding layer ϕ′ over the slope
In Fig. 11(b) Haefeli’s solution is shown for the case of a angle α. This implies that the sliding most likely took place
slope parallel rock surface for three different wall–soil on a weak slope parallel slip surface, which seems to be more
frictions δ. For comparison, the horizontal component of realistic than Brandl’s assumption of the global failure in the
the landslide force Egh is normalised assuming a linear sliding layer. A more detailed analysis requires the exact
pressure distribution at the wall. In contrast to the exact knowledge of boundary conditions of every measured land-
solution (21) from limit analysis, where the earth pressure slide, which at this stage is outside the scope of the presented
drops significantly with increasing slope angle α, Haefeli’s work.
solution is close to the active earth pressure at low slope
angles and grows with increasing slope. Another difference is
the significant influence of the wall friction δ: while the exact
solution is, at least within certain bounds (e.g. δ α for EXAMPLES
β ¼ 0), independent of δ, the Haefeli solution at steep slope This section demonstrates the application of the derived
angles may vary by as much as 50%. Comparison with the solutions to two practical examples, see Fig. 12: (a) case 1 of
exact solution shows that for realistic slope and friction a constrained creeping landslide with a slope parallel weak
angles in the sliding layer, of around α = 20° and ϕ′ = 30°, layer and (b) case 2 of a retaining wall with a mildly inclined
respectively, Haefeli’s solution underestimates landslide weak layer.
pressures significantly and has to be treated with care.
It is also seen in Fig. 11(b) that, as the slope angles get
closer to the friction angle, Haefeli’s solution approaches the Constrained creeping landslide with a slope parallel weak layer
exact one, but for the special case of a slope angle α equal to First a constrained creeping landslide with a slope parallel
the angle of internal friction ϕ′, the landslide pressures of weak layer is investigated (Fig. 12(a)). The thickness of the
Haefeli’s solution can become significantly higher than those sliding layer is 20 m, which corresponds to a vertical height
of the exact limit analysis solution. Although this case is Hz = 21·3 m. At the bottom of the slope the landslide meets a
rather hypothetical, because it implies that every single soil rock outcrop. Owing to this constraint, the pressure in the
element in the sliding layer has reached the failure state sliding layer increases towards the toe and might at some
(rarely the case in planar slides), for some reason Brandl & point in time reach failure. Such a situation is common in
Dalmatiner (1988) have chosen this special case of Haefeli’s mountainous areas and there are also cases where buildings
solution to compare to their field observations. This are constructed on the landslide (e.g. Puzrin & Schmid,
comparison is shown in Fig. 2(b), where the horizontal 2012). For this particular case the landslide pressure
77·6 m
Slop 34·6 m
Buil eb
ding θ= 29·6 m
s 20°
φ' = 30° Slope a

θ = 7°
41 c' = 0 kPa φ' = 30°
·6 c' = 0 kP
° γ = 20 kN/m3 a
Hz = 21·3 m
γ = 20 kN 3
H = 10 m
Planned
20° /m
18·4° highway
We
ak l Weak laye
aye 7° r
r Rock
20° outcrop
(a) (b)
Fig. 12. (a) Case 1 constrained creeping landslide. (b) Case 2 retaining wall for highway (slope a: θ = 7°; slope b: θ = 20°)
coefficient can be calculated using the exact solution (16) as application of the exact solution and a possibly more
Klh = 2·0, whereas the landslide pressure acting on the rock conservative upper-bound solution (8) has to be used
outcrop Klhw is higher and cannot be determined from the (Table 2 in Appendix 3). In this case, due to the mild
presented solutions. Note that the limiting landslide force inclination of the slip surface, the lowest landslide pressure
acting in the sliding layer is independent from the inclination coefficient calculated from the general upper-bound solution
of the rock outcrop β and can be calculated using the vertical (8) is Klhw = 5·4, which is significantly higher than the one in
height of the sliding layer, Hz = 21·3 m the case 2a with the slope parallel slip surface. The resulting
total horizontal landslide force acting on the retaining wall is
γHz2 Klh
Elh ¼ ¼ 9073 kN=m ð29Þ γH 2 Klhw
2 Elhw ¼ ¼ 5394 kN=m ð31Þ
2
The critical mechanism for this geometry is shown in
Fig. 12(a). Its length measured in the horizontal direction Although this upper-bound result is conservative, it
amounts to 77·6 m, which is significantly higher than the provides an insight into how enormous the landslide forces
thickness of the sliding layer and can involve a significant can become for variable thickness of the sliding layer with
number of structures within the landslide compression zone low inclination of the slip surface and emphasises a need for
that could be more damaged by differential displacements more detailed analysis.
than those located further uphill on the practically uniformly
sliding layer.
Comparison
Comparison of the two case studies shows that there is a
significant influence of the geometry on the landslide
Retaining wall with a mildly inclined weak layer pressure coefficient. A low inclination of the weak slip
The second case is more unusual but certainly possible. It surface increases the landslide pressure coefficient signifi-
is chosen to show the influence of a mildly inclined weak cantly, in particular with variable thickness of the sliding
layer, first in the case 2a with constant thickness of the sliding layer (see Table 1, also showing Haefeli’s solution and the
layer (Fig. 12(b), slope a) and then in the case 2b with active and passive coefficients according to Müller-Breslau,
variable thickness of the sliding layer (Fig. 12(b), slope b). A 1906). This comparison reveals that in both cases Haefeli’s
planned highway requires a retaining wall at the foot of a solution underestimates the earth pressures significantly.
slope where a weak bentonite layer with an inclination of While for case 1 and case 2a the classical passive earth
α = 7° is located at the depth of H = 10 m, measured at the pressure is too high compared to the exact limit analysis
wall. The landslide can take place if the friction angle in the solution, it is quite close to the upper-bound solution for
weak layer drops to the angle of the layer inclination, ϕ case 2b, at least when a frictionless wall is assumed.
′ = α = 7°, which is close to the residual friction angle of a
2
pure bentonite (e.g. Lupini et al., 1981; Di Maio & Fenellif,
2015). The parameters of the sliding layer are taken to be the CONCLUSIONS
same as in case 1 above. The problem of the limiting landslide pressure on an
In the case 2a of constant thickness of the sliding layer, obstacle was first formulated by Haefeli (1944), who
θ = α = 7° (slope a), the exact solution can be used to recognised that the kinematics of the problem does not
calculate the landslide pressure and the corresponding force allow for classical active and passive earth pressure theories
acting on the retaining wall. Owing to the milder inclination to be applied. He derived an approximate solution using a
of the weak layer, the landslide pressure coefficient Klhw = 2·9 limiting equilibrium approach with a number of rather
increases by 45% compared to case 1 and the resulting arbitrary assumptions and simplifications. Since then, the
landslide force acting on the retaining wall becomes Haefeli solution has been widely applied for design and
γH 2 Klhw analysis of landslide retaining structures.
Elhw ¼ ¼ 2867 kN=m ð30Þ This paper revisits the old landslide pressure problem by
2
means of a rigorous upper-bound limit analysis, which has
Higher strength of the sliding layer allows for steeper slope been shown for a special case of a planar landslide to produce
inclinations than those of the weak layer. To illustrate the the known exact landslide pressure solution. The proposed
effect of the variable thickness on the landslide pressure, case approach also allows extending the limit analysis solutions to
2b with a slope inclination of θ = 20° (Fig. 12(b), slope b) has account for arbitrary wall friction and inclination, as well as
been investigated. The fact that θ = α excludes the the soil dilatancy and groundwater effects, significantly
Table 1. Comparison of the horizontal earth pressure coefficients for the two case studies for different soil–wall frictions (note that the sign of the
soil–wall friction angles δ corresponds to the kinematic conditions chosen, i.e. negative for the passive case)
Soil–wall friction Exact solution Upper bound Haefeli θ = 20° Active θ = 20° Passive θ = 20°
angle, δ
Case 1: Case 2:
α = θ = 20° Slope a: α = 7°, θ = 7° Slope b: α = 7°, θ = 20°
δ = 0° 0·63 0·44 5·74

2·0 2·9 5·4
δ = 2/3 ϕ′ = 20° 0·97 0·39 21·96
broadening its field of application. Algebraically, this APPENDIX 1

solution corresponds exactly to another lower-bound sol- In the case of non-associated flow the failure load will always be
ution, which William Rankine proposed in 1857 for a rather equal to or lower than the one derived using associated flow
(e.g. Chen, 1975). To derive a solution using the non-associated flow
different problem of the classical passive earth pressure in
rule the work dissipated along the slip discontinuities has to be
inclined planar ground assumed to be homogeneous (i.e. calculated. In contrast to the associated flow this dissipation is
without a weaker slip surface). Ironically, although for the non-zero even for purely frictional material. For this reason, in such
problem it was originally supposed to solve (i.e. the wall a case, both a kinematically admissible velocity field and a
moving towards the soil) this Rankine lower bound appeared corresponding statically admissible stress field have to be postulated
to be useful only for special cases (e.g. α = 0), providing in (e.g. Drescher & Detournay, 1993; Smith, 2012). In the special case
general values significantly below the exact solution, for the of slope parallel slip surface, the stress field from the lower-bound
problem of landslide pressures this suboptimal lower bound solution Fig. 6 and the mechanism shown in Fig. 13 are used to
from a different problem appears to provide the exact derive the limiting landslide pressure for the case of a non-associated
solution! This finding explains the ‘paradox’ of why flow in the sliding layer and a non-dilative flow on the slip surface.
Consequently, block 1 in Fig. 13(a) moves parallel to the slip surface
some numerical finite-element solutions of this problem and the hodograph changes (Fig. 13(b); note also that the vertical
(e.g. Muraro et al., 2015) happen to be close to the Rankine component of the landslide force contributes to the rate of external
solution. work).
If the full shear resistance is mobilised along the entire From the hodograph in Fig. 13(b), using the constraint on the
length of the slip surface, all the way down to the obstacle, the angles of the slip discontinuities ω1 þ ω2 = π/2 ϕ′ the velocities are
exact solution allows for derivation of both the landslide related by
pressure acting on the obstacle and the (identical to it)
v1 sinðω1 þ ψ Þ
landslide failure pressure in the sliding layer. This condition v2 ¼
is satisfied by all compliant retaining structures and by cosð2ψ ϕÞ
those rigid structures that were installed after the initial v1 cosðω1 þ ϕ′ ψ Þ
landslide movement took place. If, however, the rigid v12 ¼ ð32Þ
cosð2ψ ϕÞ
obstacle existed before the landslide moves, the exact solution
provides only the landslide failure pressure in the sliding v1 sinðω1 þ ψ Þ cosðω1 þ α þ ϕ′ ψ Þ
v2v ¼
layer, with the (higher) landslide force on the obstacle cosð2ψ ϕÞ
remaining undetermined. For natural rock outcrops the and the lengths of the slip discontinuities are
latter force is of limited practical significance, but for
man-made rigid structures it represents an important H cosðαÞ
l1 ¼
design factor. However, the fact that they will have to sinðω1 Þ
withstand higher pressures than the failure pressure in the
sliding layer becomes a strong argument against the use of H cosðαÞ
l2 ¼ ð33Þ
such pre-installed rigid structures as a protection measure cosðω1 þ ϕ′Þ
against future landslides.
H cosðω1 þ αÞ
The landslide pressures from the exact limit analysis ld ¼
solution increase with the strength of the sliding layer and sinðω1 Þ
appear to be significantly higher than the classical active but The rate of external work is calculated as
much lower than the classical passive earth pressures. Of even
greater practical importance is that, due to their over- v1 Elh
Ẇ e ¼ þ v1 sin αG1 v2v G2 ð34Þ
simplifying assumptions, the well-established approximate cos α
limit equilibrium solutions (e.g. Haefeli, 1944; Brandl & where
Dalmatiner, 1988) seem to get close to the exact limit
H cosðαÞ cosðω1 þ αÞ
analysis solution only over a very narrow range of slope and G1 ¼ Hγ cosðαÞlinf þ
friction angles of the sliding layer. It follows that for mildly 2 sinðω1 Þ
ð35Þ
inclined weak slip surfaces and high strengths of the sliding
H 2 γ cos2 ðαÞ cosðϕ′Þ
layer, analysis and design of retaining structures based on G2 ¼
2 cosðω1 þ ϕ′Þ sinðω1 Þ
these approximate solutions can become dramatically unsafe.
The rate of internal dissipation is calculated from the scalar
product of the relative velocities and the resulting failure stress on the
ACKNOWLEDGEMENTS slip discontinuities in the sliding layer and the slip surface,
The authors would like to thank Pascal Minder, Andreas respectively
ð l1 ð l2
Stöcklin, Philipp Oberender, Luca Pizzetti, David Perozzi,
Manuel Stocker, Johannes Schindler, Kallivokas Gerasimos Ḋi ¼ ½tanðϕ′Þ cosðψ Þ sinðψ Þ σ f v12 dl þ σ f v2 dl
0 0 ð36Þ
and Athanasios Kontis (all ETH Zurich, Switzerland) for
valuable discussions on the topic. þ σ n;ss tanðαÞv1 ðlinf þ ld Þ
v1
2
1 El
ω1
H
3 l inf
ω2
ld
α
v
v2v
2
2
v1
ψ α
v2 v12 v1
ω2 ψ
ψ v2
ω1 ψ
α v1
v3 = 0
(a) (b)
Fig. 13. (a) Upper-bound mechanism for non-associated flow in the sliding layer and non-dilative flow on the slip surface; (b) the corresponding
hodograph
Since the normal stress acting on the slip discontinuities in the to be taken into account (e.g. Michalowski, 1995). The assumed
sliding layer σf is distributed linearly, the rate of internal dissipation mechanism corresponds to the one without water and consists of two
simplifies to slip discontinuities, Fig. 14(a). For the sake of simplicity, block 3 is
assumed to remain rigid and at rest in front of the obstacle. The
ðv12 l1 þ v2 l2 Þ velocities v2v, v12u, v23u are derived from the hodograph, Fig. 14(b),
Ḋi ¼ σ f;b ½tanðϕ′Þ cosðψ Þ sinðψ Þ and the water forces U12, U23 are calculated by integration of the
2 ð37Þ
water pressure along the slip discontinuities, Fig. 14(a).
þ σ n;ss tanðαÞv1 ðlinf þ ld Þ
v1 sinðϕ′ þ ω1 þ ϕ′2 Þ sinðϕ′ þ ω2 αÞ
where σf,b denotes the normal stress in the slip discontinuities at the v2v ¼
bottom of the sliding layer and σn,ss is the normal stress acting on the
slip surface. v1 sinðϕ′ þ ω2 ϕ′2 Þ sinðϕ′Þ
v12u ¼
σ f;b ¼ σ m;b cos2 ðϕ′Þ ¼ Hγ cos2 ðαÞ cos2 ðϕ′Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi v1 sinðϕ′ þ ω1 þ ϕ′2 Þ sinðϕ′Þ
1 þ 1 cos2 ϕð1 þ tan2 αÞ v23u ¼ ð40Þ
ð38Þ sinð2ϕ′ þ ω1 þ ω2 Þ
cos2 ϕ
Hw2 γw cos3 α
σ n;ss ¼ Hγ cos2 ðαÞ U12 ¼
2 sin ω1
Equating the rates of external work and internal dissipation Hw2 γw cos3 α
Ẇ e ¼ Ḋi and solving for the landslide force leads to U23 ¼
2 sin ω2
(
cos α ðv12 l1 þ v2 l2 Þ The weight of block 2, G2, is not affected by the water and
Elh ¼ σ f;b ½tanðϕ′Þ cosðψ Þ sinðψ Þ
v1 2 remains as in equation (3). Note that, owing to ϕ2′ α, block 1 does
) ð39Þ not move horizontally. However, because the work done by the infinite
þ σ n;ss tanðαÞv1 ðlinf þ ld Þ v1 sin αG1 þ v2v G2 part of block 1 exactly equals the work done by the pore pressure on
the slip surface, only the work done against the weight of the
triangular block Gd between the first slip discontinuity and the vertical
Since the dissipation rate along the infinite length of the landslide dashed line in Fig. 14(a), as well as the corresponding pore pressure
linf and the corresponding rate of external work of this part are equal, Ud on the slip surface enters the equation for the landslide force.
the landslide pressure is independent of this unknown landslide
length. Inserting the stresses, velocities, weights and geometric Hw Hγw cosðα þ ω1 Þ
parameters (equations (32)–(35)) and rearranging equation (39) Ud ¼
sin ω1
leads to the same expression for the landslide force as the exact ð41Þ
solution (21) derived assuming associated flow. H 2 γ cos2 ðαÞ
Gd ¼
tan ω1
APPENDIX 2 The resulting total horizontal force required to fail the soil in the
Consider a slope parallel (θ = α) cohesionless (c′ = 0) landslide vicinity of the constraining obstacle is calculated from the work
with slope parallel groundwater flow and the total saturated soil unit equation to
weight equal to the total moist soil unit weight. It is assumed that the
landslide is only activated when the phreatic level of the groundwater ½v2v G2 þ v1 sinðϕ′2 αÞGd v12u U12
reaches a certain height, Hw. In this case the strength of the v23u U23 v1 sinðϕ′2 ÞUd ð42Þ
slip surface is higher than the slope inclination and has to fulfil Elh ¼ cosðαÞ
v1 cosðϕ′2 Þ
condition (27).
The formulation of an upper-bound limit analysis solution follows
the same procedure as without water, except that the work done Similarly to the dry solution, it depends on the two unknown
against the pore water pressures caused by expansion of a body (due angles of the slip discontinuities ω1 and ω2. The best upper-bound
to dilatancy) as well as the work done by external water pressures has solution is again found by minimising the total landslide force, Elh.
1 El
2 v1
2u
v2
φ'
v12
v2
3
3u
H φ' v1
v2
ω1
ω2 φ'2
Hw v
v2v
2
α s2 (α)
v12
u = H wγ wco
G2 φ'
El ω1
φ'
φ'2 v1 ω2
Gd φ'2
U12 α
U23
Ud
(a) (b)
Fig. 14. (a) Upper-bound mechanism with water; (b) corresponding hodograph
As the equations are rather cumbersome, this minimisation is only where the water pressure is given by
done numerically and the results for some special cases are shown in
Fig. 10(b). u ¼ γw cos α½n ðH Hw Þ cos α ð44Þ
A lower-bound limit analysis solution can be found by dividing
the stress field in the landslide body into two regions. The upper Assuming constant stresses in the sliding layer in the direction of
region 1 is dry and the lower region 2 is wet, see Fig. 15(a). In the dry the slip surface ∂τnt/∂t = ∂σt′/∂t = 0, integrating the equilibrium
part the solution (16) is still valid. In the wet part a stress field equations in the normal direction and transforming the normal
solution is derived below. coordinate n = z cos α into the depth of the sliding layer z yields
The equilibrium conditions for the effective stresses depend on the σ′n ¼ cos2 α½ðγ γw Þz þ ðH Hw Þγw ð45Þ
water pressure, u
@σ′t @τ nt @u Since for the total stresses the equilibrium conditions (9) are still
þ γ sin α þ ¼0 valid, their expression (10) also does not change.
@t @n @t
ð43Þ
@σ′n @τ nt @u σ n ¼ γz cos2 α
þ γ cos α þ ¼0 ð46Þ
@n @t @n τ nt ¼ γz sin α cos α
On the slip surface (z = H ), the ratio τnt/σn′ = tanα[γH/(γH γwHw)]

fulfils condition (27). Substitution of equations (45) and (46) into the
t Mohr circle of the effective stress state gives
x 1
z n
ðσ′m σ′n Þ2 þτ 2nt ¼ r2 ¼ σ′m2 sin2 ϕ′ ð47Þ
H
2
which can be resolved with respect to the mean effective stress
Hw
α qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ffi
(a) σ′n + σ′n2 cos2 ϕ′ σ′n2 þ τ 2nt
σ′m ¼ ð48Þ
cos2 ϕ′
τ
The total horizontal normal stress σx and the corresponding shear
φ' stress τxz acting on a vertical plane are also related by way of the
Mohr circle equation
φ'2
ðσ x σ m Þ2 þ τ 2xz ¼ r2 ¼ σ′m2 sin2 ϕ′ ð49Þ
P
α
σ 'n,τnt Because the total stresses are still related by equation (14), the pole
u
σn,τnt P of the Mohr circle describing the total stress state in Fig. 15(b) is
located on the straight line with inclination α passing through the
σ 'm σm σ origin. Solving equation (49), using equation (14), results in the total
horizontal normal stress acting on a vertical plane
σ 'x ,τzx
u σx ,τzx qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ffi
σ m + σ 2m ð1 þ tan2 αÞ σ 2m σ′m2 sin2 ϕ′
σx ¼ ð50Þ
1 þ tan2 α
′ þ u. The effective stress acting
where the total mean stress is σm = σm
(b) on a vertical plane is calculated from the total stress using
σx′ = σx u. The total landslide force acting on a vertical plane is
Fig. 15. (a) Statically admissible stress field for slope parallel slip determined by integrating equation (50) over the thickness of the
surface and vertical wall with water. (b) Mohr circle of the stress state sliding layer. The resulting normalised total landslide forces are
in the wet layer (solid line: effective stress state; dashed line: total stress shown in Fig. 10(b) depending on the ratio Hw/H for six slope
state) inclinations α.
Table 2. Landslide pressure coefficients Klh for vertical walls (β = 0) APPENDIX 3
in cohesionless soil (c′ = 0), varying slope angle θ, slip surface In Table 2 the landslide pressure coefficients for cohesionless soils
inclination α and angle of internal friction in the sliding layer ϕ′ (in and variable thickness of the sliding layer are presented for five
general upper-bound solution, values corresponding to the exact different values of internal friction.
solution α = θ marked in bold)
ϕ′ = 20°
NOTATION
α: degrees θ: degrees c′ cohesion of sliding layer
D force acting on mechanism in Haefeli’s solution
5·0 8·0 11·0 14·0 17·0 20·0 Ḋi rate of internal dissipation
Eg landslide force in Haefeli’s solution (Fig. 2)
5·0 1·98 2·27 2·56 2·85 3·15 3·47 Egh horizontal landslide force in Haefeli’s solution
8·0 1·88 2·17 2·45 2·73 3·01 Elh limiting horizontal landslide force in sliding layer
11·0 1·75 2·03 2·29 2·55 Ell limiting landslide force in sliding layer
14·0 1·57 1·84 2·10 Elw limiting landslide force acting on retaining structure
17·0 1·33 1·63 Elhw horizontal landslide pressure force acting on retaining
20·0 0·88 structure
G weight of block 1 in Haefeli’s solution
Gi weight of blocks in upper-bound solution
ϕ′ = 25° H vertical height of retaining structure above slip surface
Hw vertical height of water table in sliding layer
α: degrees θ: degrees Hz vertical height of sliding layer
Kah active earth pressure coefficient for a smooth wall
5·0 9·0 13·0 17·0 21·0 25·0 Kl landslide pressure coefficient
Klh horizontal landslide pressure coefficient
5·0 2·40 2·93 3·47 4·06 4·71 5·42 li length of slip discontinuity
9·0 2·26 2·77 3·28 3·83 4·43 m ratio of landslide pressure and active earth pressure
13·0 2·05 2·53 3·00 3·50 coefficients m = Klh/Kah
17·0 1·77 2·21 2·66 Ri resulting frictional resistance force
21·0 1·43 1·87 Si cohesive resistant force
25·0 0·82 u water pressure
V weight of block 2 in Haefeli’s solution
vi virtual velocity
ϕ′ = 30°
Ẇ e rate of external work
X ratio v3/v1
α: degrees θ: degrees
α inclination of slip surface
β inclination of retaining structure
5·0 10·0 15·0 20·0 25·0 30·0
γ unit weight of soil
γw unit weight of water
5·0 2·93 3·84 4·85 6·01 7·39 9·05
δ soil–wall interface friction angle (= δmob)
10·0 2·73 3·59 4·52 5·58 6·82
θ inclination of slope surface
15·0 2·42 3·19 4·02 4·95
ϑ inclination of force D in Haefeli’s solution
20·0 2·00 2·69 3·39
σi total normal stress
25·0 1·51 2·10
σi′ effective normal stress
30·0 0·75
τij shear stress
ϕ′ angle of internal friction in sliding layer
ϕ′ = 35° ϕ2′ angle of internal friction in slip surface
χ inclination of slip discontinuity in Haefeli’s solution
α: degrees θ: degrees ψ dilation angle in sliding layer
ωi angle of slip discontinuity, in the upper-bound solution the
5·0 11·0 17·0 23·0 29·0 35·0 following constraints have to be fulfilled:
π ω2 2ϕ . ω1 . max{0, θ α} and 0 , ω2 , max
5·0 3·61 5·16 7·02 9·38 12·45 16·54 {π/2 β þ α, π/2 ϕ′ þ α}, the lower-bound solution
11·0 3·33 4·75 6·44 8·53 11·22 constrains the two angles by ω1 þ ω2 = π/2 ϕ′
17·0 2·86 4·12 5·55 7·32
23·0 2·27 3·31 4·46
29·0 1·59 2·41
35·0 0·67 REFERENCES
Bernander, S. & Olofsson, I. (1981). On formation of progressive
failures in slopes. In Proceedings of the 10th international
ϕ′ = 40° conference on soil mechanics and foundation engineering,
Stockholm, Sweden, vol. 3, pp. 357–362, Rotterdam, the
α: degrees θ: degrees Netherlands: Balkema.
Bjerrum, L. (1967). Progressive failure in slopes of overconsolidated
5·0 12·0 19·0 26·0 33·0 40·0 plastic clay and clay shales. ASCE J. Soil Mech. Found. Engng
Div. 93, No. 5, 3–49.
5·0 4·51 7·13 10·65 15·67 23·08 34·49 Brandl, H. (1987). Retaining walls and other restraining structures.
12·0 4·10 6·48 9·60 13·95 20·26 In Ground engineer’s reference book (ed. F. G. Bell), Ch. 47,
19·0 3·43 5·44 8·00 11·49 pp. 1–34. London, UK: Butterworths.
26·0 2·58 4·15 6·08 Brandl, H. (2001). Stützbauwerke und konstruktive Hangsicherungen.
33·0 1·67 2·79 In Grundbau-Taschenbuch, Teil. 3 Gründungen, 6th edn (ed.
40·0 0·59 U. Smoltzyk), pp. 495–651. Berlin, Germany: Ernst & Sohn (in
German).
Brandl, H. & Dalmatiner, J. (1988). Brunnenfundierungen von Michalowski, R. L. (1995). Slope stability analysis: a kinematical
Bauwerken in Hängen (insbesondere Brücken), Strassen- approach. Géotechnique 45, No. 2, 283–293, http://dx.doi.org/
forschung Heft 352. Vienna, Austria: Bundesministeriums für 10.1680/geot.1995.45.2.283.
Wirtschaftliche Angelegenheiten (in German). Müller-Breslau, H. F. B. (1906). Erddruck auf Stützmauern.
Caquot, A. & Kérisel, J. (1948). Tableau de poussée et butée et de Stuttgart, Germany: Alfred Kröner Verlag (in German).
force portante des fondation. Paris, France: Gauthiers-Villars Muraro, S., Madaschi, A. & Gajo, A. (2015). Passive soil pressure on
(in French). sloping ground and design of retaining structures for slope
Chen, W. F. (1975). Limit analysis and soil plasticity. Amsterdam, stabilisation. Géotechnique 65, No. 6, 507–516, http://dx.doi.
the Netherlands: Elsevier Scientific Publishing Co. org/10.1680/geot.14.P.211.
Chu, S. C. (1991). Rankine’s analysis of active and passive pressures Nye, J. F. (1951). The flow of glaciers and ice-sheets as a problem in
in dry sands. Soils Found. 31, No. 4, 115–120. plasticity. Proc. R. Soc. A: Math., Phys. Engng Sci. 207, No.
Coulomb, C. A. (1776). Essai sur une application des règles de 1091, 554–572.
maximis & minimis à quelques problèmes de statique, relatifs à Oberender, P. W. & Puzrin, A. M. (2016). Observation-guided
l’architecture. Mémoires de Mathématiques et de Physique constitutive modelling for creeping landslides. Géotechnique 66,
Présentés à l’Académie Royale des Sciences par Divers Savants, No. 3, 232–247, http://dx.doi.org/10.1680/jgeot.15.LM.003.
et Lus sans ses Assemblées VII, 343–382 (in French). Poulos, H. G. (1995). Design of reinforcing piles to increase slope
De Beer, E. (1977). Piles subjected to static lateral loads. stability. Can. Geotech. J. 32, No. 5, 808–818.
Proceedings of the 9th international conference on soil mechanics Puzrin, A. M. & Schmid, A. (2011). Progressive failure of a
and foundation engineering, speciality session 10, Tokyo, Japan, constrained creeping landslide. Proc. R. Soc. A: Math., Phys.
pp. 1–14. Engng Sci. 467, No. 2133, 2444–2461.
Di Maio, C. & Fenellif, G. B. (2015). Residual strength of kaolin and Puzrin, A. M. & Schmid, A. (2012). Evolution of stabilised creeping
bentonite: the influence of their constituent pore fluid. landslides. Géotechnique 62, No. 6, 491–501, http://dx.doi.
Géotechnique 44, No. 2, 217–226, http://dx.doi.org/10.1680/ org/10.1680/geot.11.P.041.
geot.1994.44.2.217. Puzrin, A. M. & Sterba, I. (2006). Inverse long-term stability
Drescher, A. & Detournay, E. (1993). Limit load in translational analysis of a constrained landslide. Geotechnique 56, No. 7,
failure mechanisms for associative and non-associative 483–489, http://dx.doi.org/10.1680/geot.56.7.483.
materials. Géotechnique 43, No. 3, 443–456, http://dx.doi. Randolph, M. F. & Houlsby, G. T. (1984). The limiting pressure on a
org/10.1680/geot.1993.43.3.443. circular pile loaded laterally in cohesive soil. Géotechnique 34,
Fukuoka, M. (1995). Earth pressure in moving soil mass. No. 4, 613–623, http://dx.doi.org/10.1680/geot.1984.34.4.613.
Proceedings of the Bengt B Broms symposium on geotechnical Rankine, W. J. M. (1857). On the stability of loose earth. Phil.
engineering, Singapore, pp. 135–147. Trans. R. Soc. London 147, 9–27.
Haefeli, R. (1944). Zur Erd- und Kriechdruck-Theorie. Skempton, A. W. (1964). Long-term stability of clay slopes.
Schweizerische Bauzeitung 124, No. 20, 256–260 (in German). Géotechnique 14, No. 2, 77–102, http://dx.doi.org/10.1680/
Ito, T. & Matsui, T. (1975). Methods to estimate lateral force acting geot.1964.14.2.77.
on stabilizing piles. Soils Found. 15, No. 4, 43–59. Smith, C. (2012). Limit loads for a shallow anchor/trapdoor
Kourkoulis, R., Gelagoti, F., Anastasopoulos, I. & Gazetas, G. embedded in a non-associative Coulomb soil. Géotechnique
(2011). Slope stabilizing piles and pile-groups: parametric study 62, No. 7, 563–571, http://dx.doi.org/10.1680/geot.10.P.136.
and design insights. J. Geotech. Geoenviron. Engng 137, No. 7, Sokolovskii, V. V. (1960). Statics of soil media. London, UK:
663–677. Butterworths Scientific Publications.
Kupper, W. (1967). Der plastische Grenzzustand in der schiefen Szczepinski, W. (1972). On the motion of flat landslides and
ebenen Erd- oder Schneeschicht. Zeitschrift für Angewandte avalanches treated as a problem in plasticity. Archs Mech. 24,
Mathematik und Physik ZAMP 18, No. 5, 705–736 (in No. 5–6, 919–930.
German). Terzaghi, K. (1936). Stability of slopes of natural clay. Proceedings of
Lancellotta, R. (2002). Analytical solution of passive earth pressure. the 1st international conference on soil mechanics and foundation
Géotechnique 52, No. 8, 617–619, http://dx.doi.org/10.1680/ engineering, Cambridge, UK, vol. 1, pp. 161–165.
geot.2002.52.8.617. Viggiani, C. (1981). Ultimate lateral load on piles used to stabilise
Lupini, J. F., Skinner, A. E. & Vaughan, P. R. (1981). The drained landslides. In Proceedings of the 10th international conference on
residual strength of cohesive soils. Géotechnique 31, No. 2, soil mechanics and foundation engineering, Stockholm, Sweden,
181–213, http://dx.doi.org/10.1680/geot.1981.31.2.181. vol. 3, pp. 555–560, Rotterdam, the Netherlands: Balkema.
Michalowski, R. L. (1989). Three-dimensional analysis of locally Ziegler, H. (1963). Methoden der Plastizitätstheorie in der
loaded slopes. Géotechnique 39, No. 1, 27–38, http://dx.doi. Schneemechanik. Zeitschrift für Angewandte Mathematik und
org/10.1680/geot.1989.39.1.27. Physik ZAMP 14, No. 6, 713–737 (in German).

Lateral Earth Pressures in Constrained Landslides

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Lateral Earth Pressures in Constrained Landslides

Transféré par

Droits d'auteur :

Formats disponibles

Friedli, B. et al. (2017). Géotechnique 67, No. 10, 890–905 [http://dx.doi.org/10.1680/jgeot.16.P.