kChapter 24

Guy LEFEBVRE

SOFT SENSITIVE CLAYS

1. INTRODUCTION highly sensitive clays, therefore, it is not just

the risk of a slope failure that is of concern, but
ensitivity is defined as the ratio of the undis- also the area that can be affected by retrogressive
S turbed to the remolded shear strength of a
soil and thus expresses the loss of strength when a
sliding.
Their high sensitivity combined with the fluid-
soil is remolded. A variety of clay deposits may ity of the remolded materials makes quick clays
exhibit sensitivity, but some marine clay deposits very susceptible to retrogressive landsliding. Rapid
exhibit very high sensitivity and very low re- and dramatic geotechnical failures are the result
molded shear strength. Such deposits are termed (Figure 24-1). This type of retrogressive failure has
quick clays, which is a direct translation of the been referred to as a flow slide or earth flow. The
term kvikkleira used by Norwegian investigators to latter term is preferred, since it is in line with cur-
describe these deposits, common in their country. rent international landslide terminology (see
The Norwegians define quick clays as those soils Cruden and Varnes, Chap. 3 in this report).
exhibiting a sensitivity greater than 30 and hav- Several cases have been documented in which
ing a remolded shear strength less than 0.5 kPa such retrogressive failures affected areas larger
(Norsk Geoteknisk Forening 1974). This defini- than 20 ha and extended more than 500 in be-
tion requires the remolded soil to behave like a yondthe crest of the initial slopes (Eden and
fluid. It should be noted that this definition of Jarrett 1971; Eden et al. 1971; Tavenas et al. 1971;
quick clay considers soil behavior only, ignoring Brooks et al. 1994; Evans and Brooks 1994).
geographic location, depositional environment, or At present, the risk and extent of retrogression
any other factors related to origin or material. are difficult to evaluate. However, on the basis of
In nonsensitive soils, landslide masses generally studies conducted in eastern Canada, retrogres-
come to rest at the toe of the slope and act like a sion may be considered to be a consequence of an
stabilizing mass in the new slope geometry. For initial slope failure. In terms of analysis, the first
sensitive clays, however, the remolding involved step is to quantitatively examine the initial slope
in the mass movement results in a drastic reduc- stability. The second step is to evaluate, in a semi-
tion in shear resistance, causing the remolded clay quantitative manner, the risk of retrogression that
to behave like a thick liquid so that the slide mass could develop following initial slope failure. In ad-
moves away and leaves the new slope unsup- dition to numerical analysis, a certain compre-
ported. A new instability may result and a series of hension of the formation of slopes in sensitive clay
retrogressive failures will be triggered that can ex- deposits is helpful in order to put the processes
tend far beyond the crest of the initial slope. In into a rational framework.

607
608 Landslides:Investigation and Mitigation

FIGIJRE 24-1
Aerial view of
I emieijx landslide.
Ontario, Canada,
late mnrnlng,
Wednesday, June
23, 1993, involving
failure of 2.5 to
3.5 million m3,
much of which
flowed into South
Nation valley
causing
impoundment of
river (foreground).
Failure took place
within 1 hour on
June 20, temporarily
damming river with
debris. Approximate
maximum height of
river impoundment
is shown during
overtopping of dam.
Ground surface
within failure is
about 12 m below
2. DESCRIPTION OF SENSITIVE CLAY 2.2 Geologic Origin
original ground
surface and slopes
DEPOSITS
Sensitive clay deposits are generally young marine
gently toward river.
The existence of sensitive clay deposits is related to clays deposited in preglacial and postglacial bod-
Debris flowed about
1.6 km upstream
ies of water that existed during the retreat of the
Mineral sources and depositional conditions re- last Wisconsin ice sheet between 18,000 and
and 1.7 km
sponsible for an open soil fabric and a relatively 6,000 years before the present (BP). Many of the
downstream of
landslide
high water content and topographic depressions left by the glaciers fonned
(Brooks et al. 1994; Geochemical factors responsible for a reduction freshwater lakes, but some were connected to the
Evans and Brooks of the liquid limit and the remolded shear oceans at some stage.
1994). strength. The weight of the continental ice sheets caused
i'I-IOTOGRAPII 193.254F
COURTESY OE GEOLOGICAl.
These factors naturally restrict the geographic dis- depression of the land surface up to several hun-
SURVEY OF CANADA
tribution and geologic characteristics (including dred meters. Isostatic rebounding of the land sur-
composition, stratigraphy, groundwater geochem- faces occurred after the ice retreated, but this is a
istry, and groundwater volumes) of soft sensitive relatively siow process that is still continuing in
clay deposits. some regions. The enormous amounts of water in-
corporated into the continental ice sheets caused
the ocean volumes to be somewhat less than at
2.1 Geographic Distribution
present, and the ocean levels dropped worldwide
Although soft clay deposits are found in many by at least 120 m (Kenney 1964). However, the
parts of the worki, the most extensive areas of rate of ice-sheet melting and the consequent rise in
highly sensitive clays occur in Scandinavia and sea level were much more rapid than the rebound-
eastern Canada. They have also been reported in ing of the land surface following ice-sheet removal.
Alaska, Japan, the former Soviet Union, and New Isostatic rebound and eustatic sea-level changes
Zealand (Torrance 1987). Small areas of similar after the last deglaciation have been well docu-
deposits may still be discovered, but it appears un- mented (Kenney 1964; Andrews 1972; llillaire-
likely that any large areas of such deposits remain Marcel and Fairbridge 1978; Quigley 1980). The
undiscovered. isostatic rebound was much larger than the in-
 Soft Sensitive Clays

crease in sea level. The net result of the interac- mineralogy. The silt fraction is most often high,
tion between changing land elevations and sea typically in the range of 30 to 70 percent. The
levels was a period of inundation and subsequent fraction with grain sizes smaller than 2 ,um gener-
reemergence of certain portions of the continental ally contains more rock flour than clay minerals,
land masses in the time between the melting of and the clay minerals are predominantly illite
the continental ice sheets and the present day. (Quigley 1980; Locat et al. 1984).
Sensitive clays are considered to have been de-
posited in marine, or at least brackish, bodies of 2.5 Stratigraphy
water. In North America the last glaciation covered
most of Canada, parts of the conterminous north- In large depositional basins, marine clay deposits
ern United States, and part of Alaska. The largest are generally massive and notably uniform. In
deposits of postglacial marine clays were formed in small basins, or in the proximal areas of large
the Champlain Sea, which occupied the St. Law- basins, marine clay deposits can be interlayered
rence lowlands from the Gulf of St. Lawrence to the with sand and gravel and even with glacial till
region around the city of Ottawa during the period (Gadd 1975). For most slope stability analyses, the
approximately 12,500 to 10,000 years BP (Elson stratigraphy of typical postglacial soft clay deposits
1969; Gadd 1975; Hillaire-Marcel 1979). can generally be simplified (Figure 24-2). The
massive clay formation is confined between two
2.3 Groundwater Geochemistry relatively pervious boundary layers. The upper
boundary consists of either a weathered fissured FIGURE 24-2
The emergence of these postglacial marine clay crust or sand and silt derived from the final depo- Simplified
formations above sea level altered their ground- sitional stages. The lower boundary is formed ei- down-valley cross
water regimes, and because of hydraulic or ther by till deposited on the bedrock by glaciers or
section of typical
concentration gradients, their pore-water salt con- soft marine clay
by fissured bedrock. The overall coefficient of per- deposits of eastern
centration, which was initially as high as 32 gIL,
meability in the boundary layers is generally at Canada
was reduced. Present-day pore-water salinity in
least two orders of magnitude higher than that in (modified from
many sensitive clay deposits has been reduced to
the massive clay deposit. Lefebvre 1986).
values below 1 g/L. Reduction of pore-water
cations (salinity) is known to reduce the liquid
limits (Rosenqvist 1966) and the remolded shear
strength of clays.
Because of an open fabric and high water con-
tent retained since deposition, some lacustrine SURFICIAL LAYER
clays and unleached marine clays can also be fairly
a, fussured
sensitive. The soft Barlow-Oj ibway lacustrine clay
"' ,clay crust
in northern Quebec Province, for example, has a
water content of as much as 100 percent, a plas-
ticity index of 40, a liquidity index of 1.5, and a INTACT CLAY
sensitivity of about 20. However, geochemical fac-
tors, such as salinity reduction, are considered
necessary for the development of extrasensitive or
quick clays, which have a liquidity index greater
than 2 and flow when remolded. However, it
should be noted that geochemical factors other
than salinity reduction, such as the introduction TILL
of dispersants, can also lead to an increased li-
quidity index and sensitivity.

BEDROCK
2.4 Composition and Mineralogy

Because of their geologic origin, sensitive clays fre-

quently have a characteristic composition and
.M Landslides: Investigation and Mitigation

3. VALLEY FORMATION AND reflects either cementation (Locat and Lefebvre

GROUNDWATER REGIME 1986; Lefebvre et al. 1988) or secondary consolida-
tion (Bjerrum 1967). Stages of valley formation are
The process of clay deposition in shallow marine or illustrated in a simplified manner in Figure 24-3,
brackish environments usually produced an initial where the valley bottom is seen to gradually ap-
deposit with little or no topographic relief and gen- proach the lower pervious boundary layer.
tle slopes. Subsequent emergence of these deposits Valleys in soft sensitive clays can be found today
FIGURE 24-3 subjected them to subaerial erosion. Today the in different stages of formation, and the ground-
Stages of valley slopes of geotechnical concern in these sensitive water regime evolves according to the valley devel-
formation in marine clay deposits are the result of erosion by rivers or opment. This evolution of the groundwater regime
clay deposits streams that cut valleys through these deposits. In
(modified from plays an important role in the general state of slope
the Champlain Sea region of eastern Canada, the stability and in the morphology of the landsliding
Lefebvre 1986):
preconsolidation pressure is, in general, in agree- that does occur. The groundwater regime in the
early phase,
intermediate ment with the maximum filling level of the valley vicinity of a slope not only determines the pore-
phase, (Eden and Crawford 1957; Crawford and Eden pressure distribution in the clay deposits, but also is
advanced phase. 1965). In other areas the apparent preconsolidation a prime factor in the general slope stability along
the valley sides. In sensitive clay areas the critical
location for the water table is often close to ground
urficial Layer surface. The groundwater regime can vary widely
from one site to another, ranging from a general un-
derdrainage to high artesian conditions. It is diffi-
cult to interpret and to extrapolate piezometric
Intact Clay readings without knowing the pattern of the
groundwater regime. Simple considerations of typi-
cal present-day topographic profiles and of the
valley erosion process, as shown inFigure 24-3, are
helpful in predicting and understanding the
groundwater regime.
Once the boundary conditions have been as-
Surficial Layer sumed, it is indeed a simple matter to produce a
b)
flow net because for practical purposes the perme-
'S S
ability of marine clay deposits can be considered
isotropic (Tavenas et al. 1983). Figure 24-4 shows
groundwater flow patterns for the geometric con-
ditions corresponding to the three stages of valley
Intact Clay - - - - - development shown in Figure 24-3. These flow
nets have been determined numerically assuming
a permeability in the lower boundary layer of 10
times the permeability in the clay.
Bedrock In the early stage of valley formation, the
groundwater regime is not significantly influenced
by the lower pervious boundary [Figure 24-4(a)]
because the bottom of the valley is well above it.
However, as the bottom of the valley is eroded
closer to the lower boundary [Figure 24-4(b)], high
Intact Clay artesian pressures develop. At this stage, because
pore-pressure discharge or dissipation is restricted
by the relatively impervious clay, the por pressure
in the boundary layer under the bottom of the val-
ley and the lower part of the slope tends to equili-
Bedrock brate with the head existing behind the slope
 Soft Sensitive Clays 611

crest. Such a condition is obviously detrimental to eral sites, especially in the Tyrrell Sea basin of
stability and favors a deep failure surface. In fact, northern Québec Province where valleys appear to
the artesian pressures evaluated from the flow net have evolved more rapidly because of a clay that is
of Figure 24-4(b) are sufficient to blow up the bot- more erodible than that in the St. Lawrence valley
tom of the valley. (Lefebvre 1986). The groundwater regime deduced
Groundwater within the lower boundary layer from a large number of piezometers at one site in
discharges into the river once the valley forma- the Broadback River valley is presented in Figure
tion reaches a sufficiently advanced stage [Figure 24-5(a) along with the stratigraphy obtained from
24-401. In the vicinity of the slope, the piezo- several borings. Figure 24-5(b) presents the results
metric head is slightly higher than the river level of a numerical analysis by the finite-element
and creates a downward gradient in the clay for- method. The permeability coefficients used in
mation. The surface infiltration into the upper these analyses were obtained by permeability tests
pervious boundary materials is much larger than in open-tube piezometers.
the flow that can percolate through the intact clay Both the numerical analysis and the piezomet-
formation. Thus, the water table is essentially un- nc data confirm that strong underdrainage devel-
affected by the downward drainage and generally oped in the clay formation when the underlying
remains close to the surface inside the upper till formation discharged into the river. The
boundary layer, fluctuating with the season. In the groundwater table remained close to the ground
upper boundary layer, the flow can be assumed to surface in the wet season so that essentially hydro-
be parallel to the ground surface and under almost static conditions prevailed in the surficial crust.
hydrostatic conditions. Under the crest of the slope, the pore pressure in-
Groundwater regimes along valleys in their final creased with depth in the upper part of the intact
stages of formation have been documented at sev- clay formation and then decreased toward the

N
N
N

permeability = k1

permeability = k1 FIGURE 24-4

Flow nets showing
groundwater
regimes in vicinity
: çcççcç : kkk\ \ of clay slopes
corresponding to
stages of valley
formation shown
in Figure 24-3
permeability kj (modified from
Lafleur and Lefebvre
1980):
PA
early phase,
intermediate
\ %\\\\\\\ \ N phase,
advanced phase.
612 Landslides: Investigation and Mitigation

FIGURE 24-5
Equipotential lines
a)
obtained from (a) so
piezometer readings
November 1980 and 40
(b) finite-element
analyses at I
Broadback River site,
northern Québec
Province, Canada 10 Mif Woe
(modified from 7/f //v
Lefebvre 1986).
100 150 200 250
HORIZONTAL DISTANCE (m)

-
so k=1.7 x 10 5 cm/s

Uj !:

100 150 200 250

HORIZONTAL DISTANCE (m)

lower boundary value. Such general underdrain- 4. STABILITY ANALYSIS

age is obviously beneficial for stability.
In contrast, the artesian pore pressure that is As noted in Chapter 13, slope stability methods
characteristic of the intermediate stage of valley evaluate the state of stress on a potential failure
formation [Figure 24-4(b)] creates conditions that surface and compare the shear stresses with the
may lead to deep landslides involving large masses available shear strength in the form of a safety fac-
of sensitive clays and to subsequent retrogressive tor. The stress evaluation for effective stress analy-
failures. sis is more complex because the effective normal
When the lower-boundary material discharges stress on the potential failure surface must be cal-
into the river, at the later stages of valley formation culated in order to evaluate the shear strength. In
[Figure 24-4(c)], the valley walls have reached a such analyses, realistic pore-pressure distribution
phase of equilibrium and stability. Because of im- and shear-strength parameters must be used to ob-
portant underdrainage at this stage, landsliding in tain a correct evaluation of the available shear
response to bank erosion is generally shallow and strength.
does not penetrate deeply into the sensitive intact
clay. Risk of retrogression is then fairly low. At this 4.1 Shear Strength
stage, slopes as high as 30 m and as steep as 30 de-
grees have been found to stand in normally con- The shear-strength parameters must correctly
solidated deposits (Lefebvre et al. 1991). Cases of model the shear strength that can be mobilized on
retrogressive failure have been identified in regions the failure surface of a slope that has been formed
reflecting this final stage of valley formation, but by the erosion processes described in the previous
only under conditions of exceptionally severe toe sections. For that purpose, one must distinguish
erosion and undercutting (Lefebvre et al. 1991). between those conditions causing an initial failure
 Soft Sensitive Clays 613

of a slope that has existed for some time and those

conditions causing the almost instantaneous ret-
rogressive failures taking place after a first slide. 1•
As shown in Figure 24-6, starting from the max-
imum level of sediment filling, the slope that exists
today is the result of a slow process of erosion that
has decreased the confining stress and increased
the shear stress. The stress state on a soil element
lying on a potential failure surface can thus be con-
sidered the end point of a drained stress path (A to
B, Figure 24-6). 'Xfhen stress state B approaches
the strength envelope closely enough, a slight re-
A a-'
duction in effective stress or increase in shear stress
caused by any number of reasons is sufficient to When tested under low confining stress in the
make the slope susceptible to incipient failure. FIGURE 24-6
overconsolidated range, sensitive clays exhibit Approximate stress
Actual failure may never occur, or it may occur
brittle behavior (Figure 24-7). A sharp peak path of clay
following a slight additional change to the stress strength is mobilized at relatively small axial de- element lying on
state resulting from such causes as a higher water formation on a triaxial stress-strain curve; the potential failure
table during a wet season or erosion at the toe of surface in slope.
strength rapidly decreases after the peak and sta-
the slope. This failure will obviously be undrained. bilizes on a plateau. The shear strength remains
Because of the overconsolidated state of the soil, if approximately constant beyond an axial deforma-
the failure is triggered by an undrained increase of
tion of 4 to 5 percent. The strength of these east-
shear stress due to rapid toe erosion, the undrained ern Canadian sensitive clays at large deformations
stress path (BO) will not deviate much from the is routinely determined at an axial deformation of
drained stress path (BN). Thus, an effective stress 8 percent. Study of many case records for locations
analysis that implies a drained stress path remains in eastern Canada has shown that it is the large-
valid until the incipience of failure and even then deformation, or postpeak, strength that is mobi-
does not introduce a significant error into the
lized during the initial failure of a natural slope
analysis of an initial landslide. or long-term excavation slope (Lefebvre and
However, the situation is completely different
for retrogressive failures. The shear stresses are
applied almost instantaneously and create an
100
undrained loading. Stability consideration for the STANDARD TEST
role of retrogression must therefore be based on 90
the undrained shear strength. FAILURE AT 35 DAYS
80

4.2 Effective Stress Parameters 70 CREEP TESTS

The effective stress parameters for analysis of an 0.

.19
60
initial failure must describe the shear strength for
the range of stresses existing on a potential failure 50 STABLE AT 45 DAYS
surface; these parameters are generally determined
40
by drained triaxial compression tests. The recon-
solidation, or cell, pressure must be fairly low in 30
order to have a state of stress at failure in the range FIGURE 24-7
20 Creep and standard
of those stresses existing in the field on the po-
drained triaxial tests
tential failure surface. For this purpose, triaxial 10 on sensitive Nicolet
specimens are commonly reconsolidated at cell clay, eastern
pressures ranging from 5 to 30 kPa in order to Canada
have, at failure, a state of stress that is in the over- 2 3 4 5 6 7 (modified from
consolidated range. AXIAL DEFORMATION % Philibert 1976).
614 Landslides: Investigation and Mitigation

LaRochelle 1974; Lefebvre 1981). Skempton all these sites, located over a region about 700 km
(1964) reached a similar conclusion, namely, that long, plot in a surprisingly narrow band. They rep-
for stiff overconsolidated clays, it is the strength at resent a range of natural water content from 40 to
large deformations that is mobilized during a slope 90 percent, plasticity index from 5 to 40, and pre-
failure. consolidation pressure from 100 to 600 kPa.
The use of the postpeak strength in performing The normalized postpeak envelope (Figure 24-8)
stability analyses of sensitive clays can be justified exhibits some curvature, with the shear resistance
by progressive failure along the failure surface or initially increasing rapidly with effective stress and
by the fact that the peak strength in sensitive clay then flattening toward the normally consolidated
cannot be mobilized under sustained load. Triaxial envelope. Because this normalized envelope is
drained creep tests performed on sensitive St. common to all Champlain Sea clay sites, it can be
Lawrence clays (Figure 24-7) have shown that used as a first approximation to derive the strength
creep failure develops for any loading above the parameters from a. As indicated by the curvature
strength level corresponding to large deformations of the envelope, for the same confinement stress
(Lefebvre 1981). In any event, the strength at range, ' increases with aç . If the parameters are
large deformations may be considered to form a determined for an identical confinement pressure
stability threshold in terms of either deformation (as') range of 5 to 20 kPa, the normalized enve-
or time. The postpeak strength is generally deter- lope of Figure 24-8 provides an estimate of 4' that
mined by means of an isotropically consolidated increases from 29 to 43 degrees, whereas G in-
triaxial drained test with a rate of loading of 1 to 2 creases from 100 to 400 kPa and c' remains ap-
percent per day to allow drainage; the test is con- proximately constant at 7 kPa. These values are
tinued to an axial deformation of 10 to 12 percent. summarized in Table 24-1. Test results from tube
Sampling disturbance is generally a significant samples generally correlate with the envelope de-
factor with sensitive clay in the overconsolidated fined in Figure 24-8 but exhibit a larger scatter.
state. The peak strength is, for example, much af- The postpeak strength defined for the soft sen-
fected by sampling disturbance. However, good sitive Champlain clays is surprisingly high and ap-
commercial tube sampling is sufficient to deter- pears somewhat paradoxical when compared with
mine the strength at large deformations. When the residual strength determined for stiff overcon-
tube sampling methods are used, the scatter in the solidated clays, which is defined by a zero cohesion
results may be larger than when block samples are intercept and a value for 4' that sometimes is as
used. In principle, the large-deformation strength low as 10 to 15 degrees. The high postpeak c' and
envelope can be determined for overconsolidated ' values for the Champlain clay are primarily re-
samples by using undrained tests with pore- lated to its mineralogy. The high silt and rock flour
pressure measurement. However, difficulties arise content and the relatively low fraction of illitic
in obtaining this determination accurately by such clay minerals result in a fairly large friction angle.
methods because the stress paths all tend to reach In addition, as shown by the correlations estab-
the failure envelope at the same stress state. lished with a, the shear strength at large defor-
mations is still influenced by the stress history and
4.3 Normalized Shear Strength may be different from a true residual strength.
The correlations established for Champlain
A large data bank has accumulated over the years clay should be checked, before they are adopted to
on the postpeak shear strengths of the sensitive describe the characteristics of sensitive clays from
Champlain Sea clays of eastern Canada. For spec- other regions of the world. From limited data ob-
imens reconsolidated in the laboratory to the same tained from analyses of block samples of sensitive
confining pressure, the strength at large deforma- clays from four sites in the Tyrrell Sea basin in
tions appears to be influenced by the stress history; northern Québec Province, the postpeak parame-
that is, it increases with preconsolidation pressure. ters for the Tyrrell Sea clays appear to be slightly
Figure 24-8 presents test results on block sample lower than those for Champlain clays. Figure 24-9
specimens of Champlain Sea clays from nearly 20 shows a comparison of the normalized postpeak
sites (Lefebvre 1981). The data are normalized by strength envelopes for the Tyrrell Sea and Cham-
the preconsolidation pressure, . The data from plain Sea clays.
 Soft Sensitive Clays 615

Marine clay deposits are generally massive and Table 24-1

fairly homogeneous; thus the use of a circular fail- Postpeak Strength Parameters for Sensitive Clays from Champlain Sea
ure surface is generally appropriate. Bishop's mod- Deposits of Eastern Canada (modified from Lefebvre 1981)
ified method is most often used for stability RANGE OF CONSOLIDATION a' cm'
analyses of soft clay slopes at sites in eastern Can- STRESS (kPa) (kPa) (kPa) (degrees)
ada. The relatively high ' that is used for the sta-
5-20 100 7.4 28.7
bility analysis of an initial failure in sensitive clay 5-20 200 7.7 34.7
slopes points to the importance of a good under- 5-20 300 7.7 39.8
standing of the pore-pressure distribution. 5-20 400 7.5 43.6

5. ESTABLISHING RISK OF RETROGRESSION

Although the analysis of an initial landslide fail-
ure on a slope composed of sensitive clay appears 0.40
fairly straightforward, there is much more uncer-
tainty in the evaluation of the risk of retrogression
following this initial failure. The potential retro-
gression distance is indeed an important parame-
ter in assessing the risk of damage.
The risk of retrogression is broadly related to
two conditions. First, the clay has to be sensitive
enough that the debris of the initial landslide will
move away, or flow Out, instead of creating a stabi-
lizing mass at the toe of the newly created slope.
Second, the newly created slope must have an un-
stable geometry, or a geometry such that the in- 0.00 0.20 0.40 0.60
duced shear stresses exceed the undrained shear (O"i +C7'3)/20' p
strength of the clay.
FIGURE 24-8
Normalized postpeak strength envelope from block samples of Champlain
5.1 Role of Sensitivity and Remolded Sea sensitive clay (Lefebvre 1981).
Shear Strength
Because factors other than clay sensitivity, such as
slope geometry and stratigraphy, play a role, it is 0.40
difficult to establish a sensitivity threshold below
which the debris from the initial landslide will not CL 1 i Champlain Sea Clays
move away and prevent retrogression. However, b - - - Tyrell Sea Clays
large retrogressive failures, for example, the typical
"bottleneck" earth flow, develop generally in very I
sensitive or quick clays. In fact, the remolded shear _ 0.20
strength controls the susceptibility of the earth
mass to flow. Figure 24-10 presents values of sensi-
tivity as a function of the remolded shear strength
for a dozen sites of slope failure in Champlain clays
(Grondin 1978). Those sites where retrogression 'I,
has or has not developed are also identified. The
0.00 0.20 0.40 0.60
remolded shear strengths and sensitivities shown
in Figure 24-10 were determined in the laboratory. (cr'1 +CT3)/2CYp
At sites where retrogressive failures developed, the FIGURE 24-9
remolded shear strength was lower than 1 kPa and Comparison of normalized postpeak strength envelopes for Champlain Sea
the sensitivity was generally higher than 40. Figure and Tyrell Sea sensitive clays.
616 Landslides: Investigation and Mitigation

FIGURE 24-10
Relation among type
of landslide failure, 800- 0
sensitivity, and
remolded shear 700 - 0
strength of sensitive
clays (modified from Landslide (non retrogressive)
600 - 0 0 Flowslide (retrogressive)
Grondin 1978).
>1
500 -
>
400 -
U) 0
C
FIGURE 24-11 300-o
(below) 0
Risk of retrogression 200 -
in terms of
0
sensitivity and 100- 00
liquidity index; data
c90
points are those 0 •.
shown in
Figure 24-10 0 1 2 3 4 5
(modified from Remolded Shear Strength (kPa)
Grondin 1978).

24-10 clearly shows that the quick clays with re-

400- molded shear strengths lower than 0.2 kPa are
/ 0
300- 0 more prone to flow than those with remolded
0 shear strengths between 0.5 and 1.0 kPa.
200-
/
At its liquid limit, the remolded shear strength
/ of a typical clay is about 1.7 kPa (Wroth and Wood
0,
1978). The data shown in Figure 24-10 indicate
7l
100- that the liquidity index has to be above unity for a
80 / 0 site to have retrogression potential. Relationships
/ between sensitivity and liquidity index have in-
0
60-
>.. - p deed been proposed by many authors (Skempton
4J /0
40-
Mitchell and Houston, 1969 and Northey 1953; Bjerrum 1954; Mitchell and
I Houston 1969). Some of these relations are repro-
4-I / Bjerrum, 1954
U,
duced in Figure 24-11 together with the data for
C the sites shown in Fig-ure 24-10. The criteria of
(D20.
U) Mitchell and Houston (1969) suggest that a sen-
UI.
I. sitivity of 40 corresponds to a liquidity index of
I about 1.5 to 2.0. Clays at sites where retrogression
10- / developed (from Figure 24-10) generally have a
sensitivity higher than 40 and a liquidity index
II
- / • Landslide (non retrogressive) exceeding 1.5. Therefore, the liquidity index,
5- , /o Flowslide (retrogressive) which is routinely available, is a good indicator of
the susceptibility of these clays to flow.

5.2 Stability Number

0 1 2 3 4
The susceptibility of the clay to flow is not suffi-
Liquidity Index (Ii) cient to induce retrogressive earth-flow failures.
The slope geometry after the initial failure must

11
 Soft Sensitive Clays 617

be such that the shear stress in the new slope ex- sometimes to human intervention. In slopes in
ceeds the undrained shear strength of the clay. which stability has already become marginal, some
Therefore, the potential distance of retrogression slight seasonal variations in the groundwater
is difficult to evaluate. Nevertheless, the risk of regime may trigger a landslide. For the long term,
retrogression after the occurrence of an initial however, the groundwater regime in the vicinity
failure can be assessed at least qualitatively. From of slopes evolves with the development of valleys
analytical studies and examination of more than in a way that affects the general stability of the
40 case records, Mitchell and Markell (1974) pro- valley walls and the morphology of the landslides.
posed that retrogression can occur only if the sta- Detailed piezometric readings are required to per-
bility number, N, = 'yHICu, is larger than 6 within mit a rational interpretation of the groundwater
the depth of potential failure. regime. Piezometric data should not be considered
This stability number is similar to the coeffi- isolated data values but a reflection of a ground-
cient used in the evaluation of bearing capacity, as water regime controlled by boundary conditions.
if the stress yH imposed by the new slope were cre- Approaches to stability analyses of sensitive
ating an undrained extrusion of the clay below. clay slopes are based largely on experience and the
The larger the value of 1H/Cu, the larger will be study of many case records. However, considera-
the retrogression distance before equilibrium. tion of the genesis of clay slopes justifies the type
Mitchell and Markell (1974) found a general rela- of analysis as well as the approach for determina-
tionship between yHICu and the distance of retro- tion of the shear-strength parameters. Landslides
gression, although other factors such as sensitivity, in sensitive clays are notorious for large and very
stratigraphy, and topography also appear to play a rapidly enlarging retrogressive earth flows that can
significant role in limiting retrogression. be triggered as a result of the initial failure. Exper-
The groundwater regime influences the effective- tise has been developed in the evaluation of sta-
stress profile and consequently the undrained bility against an initial failure using effective stress
shear strength profile, especially in normally con- analysis, proper pore-pressure data, and postpeak
solidated deposits. In the final stage of valley strength parameters. However, it remains difficult
formation, deposits may be affected by strong un- to assess whether an initial failure will generate a
sequence of retrogressive earth flows. Slopes com-
derdrainage, as described earlier, which causes the
posed of sensitive clays with a stability number
undrained shear strength to increase rapidly with
yH/Cu larger than 6 and a liquidity index higher
depth in the lower half of the clay deposit. Under
than 1.5 are generally considered as presenting
these circumstances, the most critical position for
a risk of retrogressive failure following an initial
a retrogression surface will be much higher on the
failure.
slope than would otherwise be the case because
of the unique relationships between the yI-I/Cu
REFERENCES
values and elevation (or depth).
The liquidity index and the ratio ?H/Cu are Andrews, J.T. 1972. Post-Glacial Rebound. In The
probably the most useful parameters for assessing National Atlas of Canada, Canada Department of
the risk of occurrence of retrogressive failures. Energy, Mines, and Resources, Ottawa, Ontario,
However, because retrogressive landslides can pp. 35-36.
only be triggered after the initial failure has oc- Bjerrum, L. 1954. Geotechnical Properties of Norwe-
curred, securing the stability of a slope against an gian Marine Clays. Geotechnique, Vol.4, pp. 49-69.
initial failure is the logical approach in preventing Bjerrum, L. 1967. Engineering Geology of Nor-
wegian Normally Consolidated Marine Clays as
large retrogressive earth flows.
Related to Settlements of Buildings. Ceo-
technique, Vol. 17, pp. 81-1 18.
6. CONCLUSIONS Brooks, G.R., J.M. Aylsworth, S.C. Evans, and D.E.
Lawrence. 1994. The Lemieux Landslide of June 20,
In sensitive clay deposits, landslides are an impor- 1993, South Nation Valley, Southeastern Ontario—
tant component in the development of young A Photographic Record. Miscellaneous Report 56.
river valleys. The primary cause of instability is Geological Survey of Canada, Ottawa, Ontario,
changes of geometry related to natural erosion or 18 pp.
618 Landslides: Investigation and Mitigation

Crawford, C.G., and W.J. Eden. 1965. A Comparison Lefebvre, G. 1986. Slope Instability and Valley
of Laboratory Results with In Situ Properties of Formation in Canadian Soft Clay Deposits.
Leda Clay. In Proc., Sixth International Conference Canadian Geotechnical Journal; Vol. 23, No. 3,
on Soil Mechanics and Foundation Engineering, pp. 261-270.
Montreal, Canada, University of Toronto Press, Lefebvre, G., and P. La Rochelle. 1974. The Analysis
Vol. 1, pp. 3 1-35. of Two Slope Failures in Cemented Champlain
Eden, W.J., and C.B. Crawford. 1957. Geotechnical Clays. Canadian Geotechnical Journal, Vol. 11,
Properties of Leda Clay in the Ottawa Area. In No. 1, pp. 89—lO8.
Proc., Fourth International Conference on Soil Lefebvre, 0., C.C. Ladd, and J.J. Pare. 1988. Com-
Mechanics and Foundation Engineering, London, parison of Field Vane and Laboratory Undrained
England, Butterworths Scientific Publications, Shear Strength in Soft Sensitive Clays. In
London, Vol. 1, pp. 22-57. Laboratory and Field Vane Shear Strength Testing,
Eden, W.J., and P.M. Jarrett. 1971. Landslide at
Special Technical Publication, ASTM, Philadel-
Orleans, Ontario. Technical Paper 321. Division
phia, Pa., pp. 233-246.
of Building Research, National Research Council
Lefebvre, 0., P. Rosenberg, J. Paquette, and J.-G.
of Canada, Ottawa, Ontario.
Lavallée. 1991. The September 5, 1988, Land-
Eden, W.J., E.D. Fletcher, and R.J. Mitchell. 1971.
South Nation River Landslide. Canadian Geotech- slide on the La Grande River, James Bay, Québec,
nicalJourrtal, Vol. 8, No. 3, pp. 446-451. Canada. Canadian Geotechnical Journal, Vol. 28,
Elson, J.A. 1969. Late Quatemary Marine Submer- No. 2, pp. 263-275.
gence of Quebec. Revue de Géographie de Mon- Locat, J., and G. Lefebvre. 1986. The Origin of
tréal, Vol. 23, pp. 247-259. Structuration of the Grande-Baleine Marine
Evans, S.G., and G.R. Brooks. 1994. An Earthflow Sediments Quebec, Canada. Quarterly Journal of
in Sensitive Champlain Sea Sediments at Engineering Geology, Vol. 19, pp. 365-3 74.
Lemieux, Ontario, June 20, 1993, and Its Impact Locat, J., G. Lefebvre, and G. Ballivy. 1984. Mine-
on the South Nation River. Canadian Geotechni- ralogy, Chemistry and Physical Properties Inter-
cal Journal, Vol. 31, No. 3, pp. 384-394. relationships of Some Sensitive Clays from
Gadd, N.R. 1975. Geology of Leda Clay. In Mass Eastern Canada. Canadian Geotechnical Journal,
Wasting: Proc., Fourth 0uelph Symposium on Geo- Vol. 21, No. 3, pp. 530-540.
morphology, Geo Abstracts Ltd., Norwich, Eng- Mitchell, J.K., and W.N. Houston. 1969. The Causes
land, pp. 137-15 1. of Clay Sensitivity. Journal of the Soil Mechanics
Grondin, G. 1978. Etude de Ia susceptibilité aux coulees and Foundations Division, ASCE, Vol. 95, No.
d'argile. M.Sc. thesis. Department of Civil En- SM3, Paper 6568, pp. 845-871.
gineering, Université de Sherbrooke, Québec, Mitchell, R.J., and A.R. Markell. 1974. Flowslides in
Canada. Sensitive Soils. Canadian Geotechnical Journal,
Hillaire-Marcel, C. 1979. Les mers post-glaciaires du Vol. 11,No. "pp. 11-31.
Québec: quelques aspects. These de doctorat Norsk Geoteknisk Fbrening. 1974. Retiiingslinjer for
d'Etats Sciences Naturelles, Université Pierre
Presentasjon av Geotekniske Undersokelser. Oslo,
et Marie Curie, Pans, France (2 vols).
16 pp.
Hillaire-Marcel, C., and R.W. Fairbridge. 1978. Iso-
Philibert, A. 1976. Etude de Ia résistance au cisaille-
stacy and Eustacy of Hudson Bay. Geology (Geo.
logical Society of America), Vol.6, pp. 117-122. mont d'une argue Champlain. M.Sc. thesis. Faculté
Kenney, T.C. 1964. Sea Level Movements and the des Sciences Appliquees, Université de Sher-
Geologic Histories of the Post-glacial Marine Soils brooke, Québec, Canada.
at Boston, Nicolet, Ottawa and Oslo. Geotechnique, Quigley, R.M. 1980. Geology, Mineralogy and
Vol. 14, pp. 203-230. Geochemistry of Canadian Soft Soils: A Geotech-
Lafleur, J., and G. Lefebvre. 1980. Groundwater nical Perspective. Canadian Geotechnical Journal,
Regime Associated with Slope Stability in Vol. 17, No. 2, pp. 261-285.
Champlain Clay Deposits. Canadian Geotechnical Rosenqvist, I. Th. 1966. Norwegian Research into
Journal, Vol. 17, No. 1, pp. 44-53. the Properties of Quick Clay—A Review. Engi-
Lefebvre, G. 1981. Strength and Slope Stability in neering Geology (Elsevier), Vol. 1, pp. 445-450.
Canadian Soft Clay Deposits. Canadian Geotech- Skempton, A.W. 1964. Long-term Stability of Clay
nicalJournal, Vol. 18, No. 3, pp. 420-442. Slopes. Geotechnique, Vol. 14, pp. 75-102.
 Soft Sensitive Clays

Skempton, A.W., and R.D. Northey. 1953. The Part II: Permeability Characteristics. Canadian
Sensitivity of Clays. Geotechnique, Vol. 3, pp. Geotechnical Journal, Vol. 20, pp. 645-660.
30-53. Torrance, J.K. 1987. Quick Clays. In Slope Stability
Tavenas, F.,J.-Y. Chagnon, and P. La Rochelle. 1971. (M.G. Anderson and K.S. Richards, eds.), John
The Saint-Jean-Vianney Landslide: Observations Wiley and Sons, New York, pp. 447-473.
and Eye-Witness Accounts. Canadian Geotechni- Wroth, C.P., and D.M. Wood. 1978. The Correla-
cal Journal, Vol. 8, No. 3, pp. 463-478. tion of Index Properties with Some Basic Engi-
Tavenas, F., P. Jean, P. Leblond, and S. Leroueuil. neering Properties of Soils. Canadian Geotechnical
1983. The Permeability of Natural Soft Clays, Journal, Vol. 15, pp. 137-145.