Vous êtes sur la page 1sur 12

Color profile: Generic CMYK printer profile

Composite Default screen

100
209 100

95 95

75 NOTE 75

25
Prediction of impending failure of embankments 25

5 5

on soft ground
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0

Gavan Hunter and Robin Fell

Abstract: The deformation behaviour and excess pore pressure response of 13 well-monitored embankments on soft
ground that were constructed to failure have been analyzed. The analysis shows that by monitoring lateral displacement
at the toe of the embankment and vertical displacement at the toe and about 5 m beyond the toe, the onset of impend-
ing failure of embankments on soft ground can be detected while the slope is at about 70–90% of the failure height.
This equates to an actual factor of safety of around 1.25. Monitoring of borehole inclinometers at the toe of the em-
bankment, cracking of the embankment, and the pore pressure response and deformation during pauses in construction
can provide useful additional data for detection of an impending failure.

Key words: embankment on soft ground, deformation, pore pressure, failure, factor of safety.

Résumé : On a analysé le comportement en déformation et la réponse en surpression interstitielle de 13 remblais bien


instrumentés construits sur sol mou jusqu’à la rupture. Il a été montré que, en mesurant le déplacement latéral au pied
For personal use only.

du remblai, et le déplacement vertical au pied et à environ 5 m à l’arrière du pied, l’imminence de la rupture du rem-
blai sur sol mou peut être détectée lorsque la pente est d’environ 70 à 90 % de la hauteur à la rupture. Ceci corres-
pond à un coefficient réel de sécurité d’environ 1,25. Le suivi d’inclinomètres dans des forages en pied de remblai, la
fissuration du remblai, et la réponse en pression interstitielle de même que la déformation durant les pauses dans la
construction peuvent fournir des données additionnelles utiles pour détecter une rupture imminente.
Mots clés : remblai sur sol mou, déformation, pression interstitielle, rupture, coefficient de sécurité.

[Traduit par la Rédaction] Hunter and Fell 220

Introduction depth ranged from 2.5 to 13 m depth (mean, 6.3 m; standard


deviation, 2.2 m; and actual, 8 m). Most of this variation
There is considerable uncertainty in the assessment of the may relate to those involved not being directly responsible
stability of embankments on soft ground due to the difficulty for the project and the failure to differentiate between de-
in predicting the undrained shear strength of the soil in the signing a safe embankment and predicting a failure height,
foundation. This is due to the inherent variability of the soil but it does demonstrate that there is a significant element of
strata and the uncertainty in the methods of measuring the uncertainty inherent in those estimates. The state of practice
strength; modelling the strength of the upper overconsoli- has not altered significantly since then.
dated, often fissured, desiccated crust; the potential for fis-
suring in the lightly overconsolidated soils below the crust; El-Ramley (2001) showed that for the soft ground case
and the potential for cracking in the embankment. studies he studied, a factor of safety of 1.5 corresponded to a
Type A predictions for the Muar test embankment (Brand probability of unsatisfactory performance (failure) of about
and Premchitt 1989) highlight the uncertainty that is preva- 1 in 50. This reflected the large uncertainty in estimating the
lent in limit equilibrium analysis of embankments on soft undrained strength.
ground. Thirty-one predictions of fill height at failure ranged Given this uncertainty, it is desirable that methods are
from 2.8 to 9.5 m (mean, 4.7 m; standard deviation, 1.5 m; available to interpret the pore pressures developed in the
and actual, 5.4 m) and predictions of the failure surface foundation and the displacements of the embankment and
100 100
foundation for identification of an impending failure. This
95 Received 2 April 2002. Accepted 16 August 2002. Published
identification would allow measures to be undertaken to 95

on the NRC Research Press Web site at http://cgj.nrc.ca on avoid a potential failure such as by stopping construction to
75 27 January 2003. allow time for pore pressures induced by the embankment to 75

1
dissipate and the foundation to gain strength so construction
G. Hunter and R. Fell. School of Civil and Environmental can be completed. There are some methods available based
Engineering, The University of New South Wales, Sydney, on the lateral deformation behaviour under the slope and at
NSW 2052, Australia.
25 the toe of the embankment (Marsland and Powell 1977; 25
1
Corresponding author (e-mail: r.fell@unsw.edu.au). Ramalho-Ortigao et al. 1983a; Ladd 1991; Rowe et al. 1995)
5 5

0 Can. Geotech. J. 40: 209–220 (2003) doi: 10.1139/T02-081 © 2003 NRC Canada 0

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:16 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
210 Can. Geotech. J. Vol. 40, 2003 100

95 95

developed mainly for foundations of low sensitivity ductile Fig. 1. The relationship between excess pore pressure and ap-
75 soils and methods based on pore pressure response under the plied vertical stress (after Tavenas and Leroueil 1980). 75

embankment (Tavenas and Leroueil 1980). This paper de-


scribes the results of a detailed analysis of case study infor-
mation, which allows assessment of these methods and other
25 25
parameters and gives recommendations on which parameters
5
give the most reliable predictions of an impending failure. 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0
Excess pore pressure response
An important issue in the monitoring of performance of
embankments on soft ground is the excess pore pressure
(above that prior to the start of construction) generated by
the construction of the embankment. Leroueil et al. (1978)
analyzed the pore pressure response observed under the cen-
tre of 30 embankments on soft clay from case studies cover-
ing a wide variety of clays in the Americas, Europe, and
Asia. They recognized that all clays had developed some
preconsolidation by aging, past loading, and other mecha-
nisms, such as desiccation by surface drying or evapo-
transpiration of plants and fluctuations in the ground water
levels. For lightly overconsolidated clay (overconsolidation
ratio (OCR) less than 2.5) the interpretation of the observed
pore pressure response as an embankment is constructed is
shown in Fig. 1 and summarized as follows:
(1) In the early stages of loading, at effective stress levels
For personal use only.

less than the preconsolidation pressure (σ p′ ), the excess pore


pressures generated in the foundation (∆u) are influenced by
the process of consolidation. Dissipation of pore pressure
occurs during loading due to the relatively high coefficient ally initiated at a zone in the foundation below the embank-
of consolidation (cv) of the overconsolidated soil. Maximum ment slope and toe). Complete failure of the embankment
excess pore pressures developed in the clay foundation un- does not necessarily occur once a localized failure condition
der the central portion of the embankment corresponded to B is reached as the localized failure zone is supported by the
values of 0.38–0.75, where B = ∆u/ ∆σ1, ∆u is the excess surrounding unfailed soil.
pore pressure, and ∆σ1 is the change in principal stress. Equation [1] defines the threshold height.
(2) At some stage in the construction process, the excess
pore pressure generated (at a specific piezometer) is ob- (σ p′ − σ vo
′ )
served to increase to an incremental B value of approxi- [1] Hnc =
mately 1. Leroueil et al. (1978) interpreted this to be where γ I(1 − B1)
the effective vertical stress at the piezometer is equivalent to
the preconsolidation pressure (point P, Fig. 1). The embank- where at a given depth below the central portion of the em-
ment height at which this occurred they termed the “thresh- bankment σ p′ is the preconsolidation pressure; σ vo ′ is the ini-
old embankment height”, Hnc. The vertical stress at the tial effective vertical stress; γ is the saturated unit weight of
piezometer tip corresponding to the threshold embankment the embankment soil, I is the stress influence factor (for the
height they termed the “threshold stress” in terms of total applied load); and B1 is the observed initial excess pore pres-
stress, or the “effective threshold stress” in terms of effective sure response, defined in Fig. 1.
stress.
Important points to note are as follows:
(3) With further loading, the excess pore pressure re-
sponse is typically at an incremental B value of approxi- • The threshold height varies with depth as a result of varia-
mately 1 (B2 in Fig. 1), i.e., an undrained condition with the tions with depth in the preconsolidation pressure to initial
clay in a normally consolidated condition. In this stage of effective stress and pore pressure response to the applied
100
the loading, the clay is characterized by high compressibility load in the soil profile, as well as variation with depth in 100

and a low coefficient of consolidation, cv, resulting in low the applied vertical stress profile from the embankment.
95 95
rates of excess pore pressure dissipation. In most of the • B is a function of the total stress ratio (Folkes and Crooks
75 cases analyzed by Leroueil et al. (1978) the rate of construc- 1985) and will therefore vary with depth. 75
tion was sufficiently high that negligible excess pore pres- • The observation of Bf > 1 is generally only observed in
sure dissipation occurred. piezometers located within the localized failure zone.
(4) With further loading the soil reaches a localized fail- Tavenas and Leroueil (1980), Folkes and Crooks (1985,
25
ure condition, and Bf > 1 is observed (Fig. 1). A localized 1986), Leroueil and Tavenas (1986), and Muir Wood (1990) 25

failure occurs where the effective stress conditions in a por- discuss the detail of the stress paths followed by the soil as
5 5
tion of the foundation reach the failure surface (it is gener- the embankment is constructed.
0 © 2003 NRC Canada 0

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:17 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
Hunter and Fell 211 100

95 95
Table 1. Summary of case studies.
75 75
Embankment and Failure height, Threshold height, Range
location Hf (m) Hnc (m) (m)a References
Narbonne, France 9.6 ≈5.5 ≈3 to 12 Pilot (1972)
25 Rio de Janiero, Brazil 2.8 1 ≈1.5 to 4.5 Ramalho-Ortigao et al. (1983a, 1983b), Costa-Filho et al. 25

(1977), Almeida and Ramalho-Ortigao (1982)


5 5
>2.8 >≈4.5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0
King’s Lynn, U.K. 6.7 1.9 to 2.1 ≈1.2 to 6.5 Wilkes (1972), Wroth and Simpson (1972) 0
Kalix, Sweden 4.1b 2.2 to 2.9 ≈2 to > 10 Holtz and Holm (1979), Stille et al. (1976)
Thames, U.K. 6.3b 1.0 to 1.6 ≈1.5 to > 12 Marsland and Powell (1977)
St. Alban-A, Canada 4.02 2.0c 1.5c La Rochelle et al. (1974)
>4 >≈4.5
Cubzac-A, France 4.5 1.5 to 2.5 ≈2 to > 6 Magnan et al. (1982), Blondeau et al. (1977),
Josseaume et al. (1977)
James Bay, Canada 7.6 3.9 to 4.8 ≈1.5 to 11 Dascal and Tournier (1975)
Portsmouth, U.S.A. 6.55 2 to 3 ≈4 to 11 Ladd (1972)
>6.5 ≈0 to 4
Mastemyr, Norway 2.84 1 to 2 ≈2.5 to >11 Clausen (1972)
Sackville, Canada 5.7 1.5 ≈1.5 to 5 Rowe et al. (1995), Rowe et al. (1996)
>2.5 to 3 >≈5
Muar-F, Malaysia 5.4 1 ≈2 to 8 Brand and Premchitt (1989), Poulos et al. (1990),
Indraratna et al. (1992)
>3 ≈8 to 14
Muar-EC, Malaysiad 4.68 1.5 to 1.8 ≈2 to 8 Malaysian Highway Authority (1989)
>3 ≈8 to >10
For personal use only.

a
The estimated foundation depth range over which the prescribed threshold height is reached based on the excess pore pressure response from piezometers.
b
Failure height includes a 1 m deep cut at the toe of the embankment.
c
At St Alban-A the normally consolidated zone increased with increasing fill height above 2 m.
d
The foundation at Muar-EC was treated by electrochemical injection.

Case studies analyzed ment of shear strain) and the maximum lateral
deformation were analyzed.
The deformation behaviour and excess pore pressure re- Hunter et al. (2000) provide details on the individual cases
sponse of 13 well-monitored embankments on soft ground studies and an analysis of the deformation behaviour and ex-
that were constructed to failure have been analyzed. Selec- cess pore pressure response.
tion of suitable case studies was based on the level of avail-
able information on the properties of the foundation and
embankment filling, and the amount, type, location, and Excess pore pressure response
available published results of instrumentation. Table 1 lists Figure 3 shows typical plots of excess pore pressure mea-
the cases and the source of the information. sured under the centre of the embankment versus the applied
The embankments were all well instrumented rectangular embankment load and represents the response up to embank-
trial embankments (except Mastemyr, which was circular) ment failure in most cases. In these plots the excess pore
and failed at heights between 2.8 and 9.6 m (mostly 4 −7.5 m). pressure has been corrected by the stress intensity factor and
The foundation soils were silts and clays of very soft to soft the “threshold stress” calculated from the applied embank-
undrained strength, of low to high plasticity, and low to high ment load.
sensitivity, capped with a 1.2–2.5 m thick weathered, often As expected, the excess pore pressure response indicates
desiccated and fissured crust. the change from overconsolidated to normally consolidated
The type of data analyzed is shown in Fig. 2 and included: behaviour. The pore pressure response on initial loading was
• Excess pore pressures under the centre and under the toe quite variable for the case studies with B1 values ranging
of the embankment. from 0.1 to 0.8, and the response after reaching the threshold
stress was, in most cases, equivalent to B2 ≈ 1.0. The thresh-
100 • Settlement under the centre of the embankment (Point 1). 100
old stress values in Fig. 3 were assessed from the excess
• Vertical deformation at the toe of the embankment (Point 2) pore pressure response, and equivalent values of the thresh-
95
and at a point beyond the toe (Point 3), but within the old height are given in Table 1. Estimates of the effective
95

zone of the eventual failure. vertical stress at the threshold height approximate the labo-
75 75
• Lateral deformation at the toe of the embankment ratory estimated preconsolidation pressures within the accu-
(Point 2). racy of the field and laboratory measurements.
• Lateral deformation with depth, from inclinometer re- It is evident from the results that, in most cases, the ex-
25
cords, at the toe of the embankment. Both the segmental cess pore pressure response does not provide an indication 25

vertical inclination angle of the deformation profile in the of the impending failure because the piezometer is generally
5 5
vicinity of the eventual surface of rupture (as an assess- not located within the localized failure zone. In the case of
0 © 2003 NRC Canada 0

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:17 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
212 Can. Geotech. J. Vol. 40, 2003 100

95 95
Fig. 2. Idealized section of embankment indicating the types of monitored observations.
75 75

25 25

5 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0

Kalix, the piezometer (piezometer P5 in Fig. 3a) is, by tivity, ductile high plasticity clays to highly sensitive, low
chance, located in the localized failure zone, and a Bf > 1.0 plasticity clays and silts. It should be noted that for the
For personal use only.

condition is reached. highly sensitive, low plasticity clay foundations (St. Alban
and James Bay) the amount of vertical deformation is rela-
Deformation behaviour as an indicator of tively small (up to 10–15 mm) up to approximately 90% of
the eventual embankment failure height.
impending failure
The following presents information on the deformations Lateral surface displacements at the embankment toe
observed in the embankments leading up to failure and their The observed lateral displacements at the toe of the em-
value as an indicator of impending failure. The deformation bankments are shown in Fig. 6. The following observations
plots are generally presented in terms of deformation versus have been made:
the relative embankment height (embankment height H, ver- • In most cases an increase in the rate of lateral displace-
sus failure height Hf). The ideal indicator of impending fail- ment relative to the embankment height occurs at about
ure is one that shows a significant change in response at the threshold height, Hnc, determined from the pore pres-
close to the failure height and that does this consistently for sure response, which is in agreement with the interpreta-
all soil types, including brittle sensitive soils and ductile tion of the behaviour of embankments on soft ground to
soils of low sensitivity. limit state theory.
• A further increase in the rate of lateral surface displace-
Vertical deformation at and beyond the embankment toe ment occurs in most of the case studies analyzed at rela-
The vertical deformation at the toe of the embankment tive embankment heights of 70–90%. It is considered that
versus the relative embankment height for the case studies is this behaviour is an indication of an impending failure
shown in Fig. 4. Figure 5 presents the vertical deformation condition.
behaviour beyond the embankment toe (approximately 5 m • For 11 of the 12 cases analyzed, the lateral surface dis-
distance beyond the toe). In all cases the monitoring point placement is a good indicator of impending failure. These
beyond the toe was located within the eventual failure zone. cases cover a broad range of soil types from brittle to duc-
The vertical displacement at and particularly beyond the tile, and low to high plasticity. At King’s Lynn the im-
toe of the embankment is a good indicator of an impending pending failure is not clearly identifiable.
100 failure condition. A measurable change in rate or direction The results indicate that the rate of embankment construc- 100

of vertical displacement is observable in most of the case tion affects the amount of lateral surface displacement mea-
95 studies beyond the threshold height and within the range 70– sured, particularly for the high plasticity, low sensitivity 95

95% of the eventual failure height. For measurement points foundations. For embankment constructed at relatively slow
75
beyond the toe, negligible vertical deformations were usually rates (Muar-F and Muar-EC constructed at rates of 0.02– 75

observed during the initial period of embankment construc- 0.055 m/day) the lateral surface displacement was a good
tion, and the impending failure condition was identifiable by indicator of the impending failure, but for the embankments
25
heave movements or large increases in the rate of heave constructed at more rapid rates (Rio de Janiero at 0.1 m/day 25
movement with increasing embankment height. These obser- and King’s Lynn at 0.67 m/day) the lateral surface displace-
5 vations apply to a wide variety of soil types from low sensi- ment was not such a good indicator. For Muar-EC it was 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:17 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
Hunter and Fell 213 100

95 95
Fig. 3. Examples of excess pore pressure under the centre of the embankment versus applied embankment load. The designations in
75 the legend refer to the piezometer number and its depth below ground surface level. 75

25 25

5 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0
For personal use only.

evident that a significant proportion of the lateral displace- form of the vertical inclination angle (in radians) at the
100 ment occurred during rest periods in the later stages of con- eventual surface of rupture versus relative embankment 100
struction, whilst the rate during construction periods did not height (Fig. 7). It is recognized that during construction the
95 increase significantly. During these rest periods at Muar-EC eventual surface of rupture is not known. However, by mon- 95

and at embankment heights above 50% of the failure height, itoring the incremental vertical inclination angle (Fig. 8) the
75
the excess pore pressure was either steady or else increased. zone of localized failure, particularly for foundations of low 75

sensitivity, will become evident prior to the eventual failure.


Lateral displacement in inclinometers at embankment Once the localized failure zone becomes evident then the
25
toe data from this zone can be plotted in the form of Fig. 7. 25
Observed lateral displacement with depth at the embank- It would be expected, and the results indicate, that the de-
5 ment toe, from inclinometer observations, is presented in the velopment of the localized failure zone is more clearly 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:18 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
214 Can. Geotech. J. Vol. 40, 2003 100

95 95
Fig. 4. Vertical displacement at the embankment toe versus relative embankment height. Heave is negative.
75 75

25 25

5 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0
For personal use only.

developed for foundations of low sensitivity ductile soils occurs. However, it develops much more quickly and possi-
due to their significantly lower undrained brittleness index. bly not until the embankment has been constructed to its
Once a localized failure zone has developed, the amount of failure height. For these soils, peak strengths are reached at
load transferred to the surrounding soil is much less for the relatively small strains and at failure of the soil in the local-
100 low sensitivity soils due to the limited amount of post-peak ized failure zone a large reduction in the undrained strength 100
strain weakening within the localized failure zone. In addi- occurs with relatively small strain. The large loss in un-
95 tion, relatively large post-peak strains are required for strain drained strength of these sensitive soils is associated with 95

weakening of the low sensitivity soils, and therefore a pro- contraction of the soil structure on shearing and transfer of
75
gressive failure mechanism develops relatively slowly. In load onto the pore fluid, resulting in the development of 75

most of the ductile foundation cases the increase in rate of high excess pore pressures; i.e., a static liquefaction condi-
displacement occurs beyond the threshold height, and at 70– tion. Therefore, once a failure condition is reached, progres-
25
90% of the failure height (Fig. 7a). sive failure develops rapidly and failure of the embankment 25
In the case of sensitive clay foundations, particularly quickly ensues. The observation of the deformation behav-
5 those of low plasticity, a progressive failure mechanism also iour within the inclinometer at the toe of the embankment at 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:19 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
Hunter and Fell 215 100

95 95
Fig. 5. Vertical displacement beyond toe versus relative embankment height. Heave is negative.
75 75

25 25

5 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0

Portsmouth and James Bay tend to confirm this rapid pro- of the profile of liquidity index, and the median of the mea-
gression to failure. sured vane strengths. Where the liquidity index is greater
In the cases of Kalix and Mastemyr it is considered that than 1.0 the median of the vane strength is used. Above the
For personal use only.

the partial breakdown in the soil structure has reduced the level where the liquidity index (IL) equals 1.0 (i.e., where
undrained brittleness of the foundations and therefore pro- IL < 1) a constant strength equivalent to the median vane
gressive failure is more clearly evident than for the other strength at the base of the layer (i.e., where IL = 1) is
case studies with sensitive clay foundations (Fig. 7b). adopted.
(4) Embankment strength properties used were those
Other deformations as a guide to impending failure given in the published literature. If cracking in the embank-
Settlement under the centre of the embankment does not ment occurred prior to failure then it was included in the
generally provide a good indication of impending failure. analysis and modeled at the reported location.
This is because in most cases it is not in the failure zone. A (5) The model was, as far as practical, allowed to deter-
plot of the settlement under the centre of the embankment mine the surface of rupture corresponding to the minimum
versus lateral displacement at the toe will identify the factor of safety. In most cases, the surface of rupture mod-
threshold height but is not a guide to impending failure (nor elled was reasonably close to that observed.
was it intended to be by Tavenas et al. 1979). (6) Once a factor of safety of close to one was calculated
for the failure condition and a reasonable match of the ob-
Factor of safety at the time of increased deformations served surface of rupture obtained, the modelling was con-
To obtain an understanding of the relationship between tinued for lower embankment heights.
the factor of safety and relative embankment height (H/Hf) The results of the limit equilibrium analysis are presented
limit equilibrium analysis was undertaken. Published results in Fig. 9. In most cases the calculated factors of safety fall
were used for Thames, Cubzac-A, and Narbonne. For Rio de below 1.5 at relative embankment heights of 60–70% of the
Janiero, Muar-F, St. Alban A, and Portsmouth the following failure height and below 1.25 at 75–85% of the failure
approach was adopted: height. The vertical and lateral displacements at and beyond
(1) Total stress analysis using undrained strengths was de- the embankment toe (Figs. 4, 5, and 6) indicate, for all types
termined by vane shear equipment (Portsmouth, St. Alban, of foundation conditions, an increase in the rate of displace-
and Rio de Janiero) or laboratory triaxial testing (Muar-F). ment with increasing embankment height typically in the
No correction was applied to the shear strength obtained range of 70–90% of the eventual failure height. Hence, the
100 from the vane. deformation indication of embankment failure discussed 100

(2) A strength anisotropy function was used for the soft above will fall within the reasonable range of design factors
95 clay foundation where sufficient data was available from the of safety. 95

published literature (Portsmouth).


75
(3) Shear strength for the crust was estimated using the Time to failure 75

median strength analysis approach proposed by Ferkh and There is usually a time delay between end of construction
Fell (1994) to account for loss of strength of the crust asso- and failure. The case study records analyzed indicate that for
25
ciated with the presence of pre-existing fissures and the sensitive clays this is up to 24 h, and for low sensitivity 25
cracking that occurs under the settlement of the embank- clays can be up to 30 days. During this period, displace-
5 ment. The crust strength is determined from an assessment ments are ongoing and increasing excess pore pressures have 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:19 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
216 Can. Geotech. J. Vol. 40, 2003 100

95 95
Fig. 6. Lateral surface displacement at embankment toe versus relative embankment height.
75 75

25 25

5 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0
For personal use only.

been observed in piezometers located under the side slope of monitoring instrumentation provides a good indication of an
the embankments and close to the failure zone. These obser- impending failure condition:
vations indicate there is a progressive, strain weakening pro- • Lateral surface displacement at the embankment toe. In
cess in the critically stressed zones of the foundation. virtually all cases analyzed the rate of lateral displacement
increased significantly as a failure condition was ap-
100 proached, typically at embankment heights within the 100

Conclusions and guidelines on monitoring range of 70–90% of the eventual failure height.
95 • Vertical deformation at and beyond the toe of the em- 95

Analysis of the 13 case studies indicates that appropri- bankment. The monitoring point beyond the embank-
75
ately located deformation monitoring instrumentation can be ment toe must be located within the eventual failure 75

used to identify an impending failure condition during con- zone. For the case studies it was located 2–10 m (or a
struction. As a general guide, for embankments constructed distance 0.4–1.5 times the eventual embankment failure
25
on a wide variety of soft foundation conditions (covering height) beyond the embankment toe. Vertical deforma- 25
low sensitivity, ductile clays to highly sensitive, brittle clays, tion at and beyond the embankment toe is not significant
5 and silts of low to high plasticity) the following deformation until the embankment nears failure when heave, particu- 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:20 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
Hunter and Fell 217 100

95 95
Fig. 7. Vertical inclination angle on surface of rupture from inclinometer observations at the embankment toe versus relative embank-
75 ment height. 75

25 25

5 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0
For personal use only.

larly beyond the toe, can identify an impending failure levels within the foundation are relatively high beyond about
condition. Heave is observed typically at embankment 75% of the eventual failure height and are coincident with
heights of 60–90% of the eventual failure height, with the changes in undrained deformation behaviour observed
significant increases in the rate of heave movement (rela- leading up to failure.
100
tive to increase in fill height) typically at 70–95% of the Other good indicators of an impending failure condition 100
eventual failure height. that are more specific to certain foundation conditions are as
95 The changes in rate and (or) direction of the deformation follows: 95

described above occurs, in most cases, beyond the threshold • Inclinometer observations at the toe of the embankment.
75
height of the embankment, indicating that the observed For low sensitivity, ductile soil foundations an impending 75

changes in deformation behaviour leading up to failure are failure condition can be identified by large plastic shear
not associated with changes that occur at the threshold strains that occur within a localized failure zone prior to
25
height. From limit equilibrium analysis it is evident that the failure. Increases in the maximum vertical inclination an- 25
factor of safety is typically less than about 1.25 at embank- gle versus relative fill height are observed at 70–90% of
5 ment heights of 75–85% of the failure height. Shear stress the eventual failure height. 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:21 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
218 Can. Geotech. J. Vol. 40, 2003 100

95 95
Fig. 8. Incremental vertical inclination angle (from inclinometer
75 observations at the embankment toe) with fill height (and time) 75

for (a) Thames and (b) King’s Lynn.

25 25
• Cracking of the embankment is a useful guide for ductile
5 foundations. 5

• The deformation and pore pressure response during peri-


Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 ods of no construction. In several case studies the pore 0

pressure in some piezometers was observed to increase


and deformations were ongoing (i.e., creep type deforma-
tions under constant stress conditions) during periods of
no construction. With increasing embankment height (and
increasing shear stress levels), the rate of pore pressure in-
crease was more rapid and for a more prolonged period of
time, and the rate and amount of creep type deformation
increases. These observations are best identified within
the foundation under the embankment slope or toe.
It is therefore recommended that any monitoring program
include measurement of the following:
(1) Vertical displacements at, and about 5 m beyond the
toe of the embankment, and lateral displacements at the toe.
These should be surveyed at close intervals on both sides of
the embankment (e.g., every 30 m) to allow for variability in
ground conditions, which are common. Greater spacings
may be used if ground conditions are reasonably uniform or
For personal use only.

if greater risks are acceptable.


(2) At selected representative sections, install and monitor
lateral displacements at the toe, or in the embankment slope
using inclinometers. This will be particularly useful in duc-
tile soils.
(3) Piezometers under the centreline of the embankment
are considered fundamental to the understanding of the
mechanism associated with the deformation for all founda-
tion types. The pore pressure response will (a) provide an
indication of the degree of consolidation while the founda-
tion is in an overconsolidated condition; (b) allow estimation
of the initial direction of the effective stress path and its
likely intersection point with the limit state surface, which
has significant implications for the deformation behaviour;
and (c) indicate the change from partially drained over-
consolidated conditions to normally consolidated undrained
conditions (i.e., when the effective stress path intersects with
the limit state surface).
(4) Settlement under the centre of the embankment. While
this is considered optional, it is useful to monitor the amount
of settlement during construction and ongoing settlement
post construction.
(5) Piezometers under the embankment slope are consid-
ered optional. In several case studies they have been located
within the localized failure zone and the rate of pore pres-
100 sure generation observed has exceeded the rate of applied 100

loading. However, given the generally narrow band of the lo-


95 calized failure zone, particularly for the sensitive foundation 95

case studies, it is probably good fortune that a piezometer is


75
located within the localized failure zone. 75

Construction should proceed slowly in order for an im-


pending failure condition to be more clearly identified from
25
the deformation monitoring. If construction proceeds too 25
quickly it can mask the indicators of the onset to failure. Of
5 those case studies for which an impending failure was 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:21 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
Hunter and Fell 219 100

95 95
Fig. 9. Factor of safety versus relative embankment height.
75 75

25 25

5 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

0 0

clearly identified, construction rates were less than 0.15– mental ‘A’ de Cubzac-Les-Ponts. In Proceedings of the Interna-
0.2 m/day. tional Symposium on Soft Clay, Bangkok. Edited by R.P.
In at least six of the failure case studies analyzed, the base Brenner and E.W. Brand. Asian Institute of Technology.
For personal use only.

of the surface of rupture was coincident (or very close to) an pp. 419–435.
interface between layers in the sub-surface profile. Possible Brand, E.W., and Premchitt, J. 1989. Comparison of the predicted
reasons for this are considered to be due to a concentration and observed performance of the Muar test embankment. In Pro-
of shear strains at the interface of two layers with a stiffness ceedings of the International Symposium on Trial Embankments
on Malaysian Marine Clays, Kuala Lumpur. Edited by R.R.
or permeability variation, or the presence of discontinuities
Hudson, C.T. Toh, and S.F. Chan. Malaysian Highway Author-
at the interface.
ity. Vol. 2, pp. 10.1–10.29.
Clausen, C.J.F. 1972. Measurement of porewater pressures, settle-
Acknowledgements ments and lateral deformations at a test fill on soft clay brought
to failure at Mastemyr, Oslo. Technical Report No. 11. Norwe-
This work forms part of a research project on the pre- and gian Geotechnical Institute, Oslo.
post-failure deformation behaviour of soil slopes being un- Costa-Filho, L.M., Werneck, M.L.G., and Collet, H.B. 1977. The
dertaken within the School of Civil and Environmental Engi- undrained strength of a very soft clay. In Proceedings of the 9th
neering at the University of New South Wales, Australia. International Conference on Soil Mechanics and Foundation En-
The support of the Australian Research Council and industry gineering, Tokyo. Japanese Society of Soil Mechanics and Foun-
sponsors of the project (NSW Dept. of Land and Water Con- dation Engineering. Vol. 1, pp. 69–72.
servation, Snowy Mountain Engineering Corporation, Dascal, O., and Tournier, J.P. 1975. Embankments on soft and sen-
Goulburn Murray Water, Australian Water Technologies, US sitive clay foundation. Journal of the Geotechnical Engineering
Bureau of Reclamation, Dam Safety Committee of NSW, Division, ASCE, 101(GT3): 297–314.
ACT Electricity and Water Corporation, Queensland Dept. El-Ramley, H. 2001. Probabilistic analysis of landslide hazards and
of Natural Resources, Snowy Mountains Hydro-Electric Au- risks: Bridging theory and practice. Ph.D. thesis, Department of
Civil and Environmental Engineering, University of Alberta, Ed-
thority, South Australian Water Corporation, Water Corpora-
monton, Alta.
tion of Western Australia, Pells Sullivan Meynick, Roads
Ferkh, Z., and Fell, R. 1994. Design of embankments on soft clay.
and Traffic Authority NSW, NSW Dept. of Public Works In Proceedings of the 13th International Conference on Soil Me-
and Services, and Queensland Dept. of Main Roads) is ac- chanics and Foundation Engineering, New Delhi, India. A.A.
knowledged. Balkema, Rotterdam. pp. 733–738.
100 100
Folkes, D.J., and Crooks, J.H.A. 1985. Effective stress paths and
References yielding in soft clays below embankments. Canadian
95 95
Geotechnical Journal, 22: 357–374.
75
Almeida, M.S.S., and Ramalho-Ortigao, J.A. 1982. Performance of Folkes, D.J., and Crooks, J.H.A. 1986. Effective stress paths and 75
finite element analyses of a trial embankment on soft clay. In yielding in soft clays below embankments: Reply. Canadian
Proceedings of the International Symposium on Numerical Geotechnical Journal, 23: 413.
Models in Geomechanics, Zurich. Edited by R. Dungar, G.N. Holtz, R.D., and Holm, G. 1979. Test embankment on an organic
25 Pande, and J.A. Studer. A.A. Balkema, Rotterdam. pp. 548–556. silty clay. In Proceedings of the 7th European Conference on 25
Blondeau, F., Queyroi, D., Peignaud, M., Mieussens, C., Levillain, Soil Mechanics and Foundation Engineering, Brighton, England.
5 J.P., and Vogien, M. 1977. Instrumentation du remblai experi- British Geotechnical Society. Vol. 3, pp. 79–86. 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:22 PM
Color profile: Generic CMYK printer profile
Composite Default screen

100
220 Can. Geotech. J. Vol. 40, 2003 100

95 95
Hunter, G., Fell, R., and Khalili, N. 2000. The deformation behav- Pilot, G. 1972. Study of five embankment failures on soft soils. In
75 iour of embankments on soft ground. The University of New Proceedings of the ASCE Speciality Conference on Performance 75

South Wales, School of Civil and Environmental Engineering, of Earth and Earth Retaining Structures, Purdue University, La-
UNICIV Report No. R-391, October. Sydney, Australia. fayette. American Society of Civil Engineers, New York, Vol. 1
Indraratna, B., Balasubramaninam, A.S., and Balachandran, S. (1), pp. 81–100.
25 1992. Performance of test embankment constructed to failure on Poulos, H.G., Lee, C.Y., and Small, J.C. 1990. Predicted and ob- 25

soft marine clay. Journal of Geotechnical Engineering, ASCE, served behaviour of a test embankment on Malaysian soft clays.
5
118(1): 12–33. School of Civil and Mining Engineering, Research Report No. 5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by UNIVERSITY OF ARIZONA LIBRARY on 01/17/13

Josseaume, H., Blondeau, F., and Pilot, G. 1977. Etude du R620, June. The University of Sydney, N.S.W., Australia.
0 0
comportement non draine de trios argyles molles application au Ramalho-Ortigao, J.A., Lacerda, W.A., and Werneck, M.L.G.
calcul de remblais. In Proceedings of the International Sympo- 1983a. The behaviour of the instrumentation of an embankment
sium on Soft Clay. Bangkok. Edited by R.P. Brenner and E.W. on soft clay. In Proceedings of the International Symposium on
Brand. Asian Institute of Technology. pp. 487–504. Field Measurements in Geomechanics, Zurich. Edited by K.
Ladd, C.C. 1972. Test embankment on sensitive clay. In Proceed- Kovari. Vol. 1, pp. 703–717.
ings of the ASCE Speciality Conference on Performance of Ramalho-Ortigao, J.A., Werneck, M.L.G., and Lacerda, W.A.
Earth and Earth Retaining Structures, Purdue University, Lafay- 1983b. Embankment failure on clay near Rio de Janiero. Journal
ette. American Society of Civil Engineers, New York. Vol. 1 (1), of Geotechnical Engineering, ASCE, 109(11): 1460–1479.
pp. 101–128. Rowe, R.K., Gnanendran, C.T., Landva, A.O., and Valsangkar, A.J.
Ladd, C.C. 1991. Stability evaluation during staged construction 1995. Construction and performance of a full-scale geotextile
(twenty-second Karl Terzaghi lecture). Journal of Geotechnical reinforced test embankment, Sackville, New Brunswick. Cana-
Engineering, ASCE, 117(4): 540–615. dian Geotechnical Journal, 32: 512–534. [Erratum, 33: 208
La Rochelle, P., Trak, B., Tavenas, F., and Roy, M. 1974. Failure of (1996)].
a test embankment on a sensitive Champlain Clay deposit. Ca- Rowe, R.K., Gnanendran, C.T., Landva, A.O., and Valsangkar, A.J.
nadian Geotechnical Journal, 11: 142–164. 1996. Calculated and observed behaviour of a reinforced em-
Leroueil, S., and Tavenas, F. 1986. Effective stress paths and yield- bankment over soft compressible soil. Canadian Geotechnical
ing in soft clays below embankments: Discussion. Canadian Journal, 33: 324–338.
Geotechnical Journal, 23: 410–413. Stille, H., Fredriksson, A., and Broms, B.B. 1976. Analysis of a
For personal use only.

Leroueil, S., Tavenas, F., Mieussens, C., and Peignaud, M. 1978. test embankment considering the anisotropy of the soil. In Pro-
Construction pore pressures in clay foundations under embank- ceedings of the 2nd International Conference on Numerical
ments. Part II: generalized behaviour. Canadian Geotechnical Methods in Geomechanics, Virginia. Edited by C.S. Desai.
Journal, 15: 66–82. American Society of Civil Engineers, New York, Vol. 2,
Magnan, J.P., Humbert, P.,and Mouratidis, A. 1982. Finite element pp. 611–622.
analysis of soil deformations with time under an experimental Tavenas, F., and Leroueil, S. 1980. The behaviour of embankments
embankment at failure. In Proceedings of the International Sym- on clay foundations. Canadian Geotechnical Journal, 17: 236–
posium on Numerical Models in Geomechanics, Zurich. Edited 260.
by R. Dungar, G.N. Pande and J.A. Studer. A.A. Balkema, Rot- Tavenas, F., Mieussens, C., and Bourges, F. 1979. Lateral displace-
terdam. pp. 601–609. ments in clay foundations under embankments. Canadian
Malaysian Highway Authority. 1989. Factual report on perfor- Geotechnical Journal, 16: 532–550.
mance of the 13 trial embankments. In Proceedings of the Sym- Wilkes, P.F. 1972. An induced failure at a trial embankment at
posium on Trial Embankments on Malaysian Marine Clays, King’s Lynn, Norfolk, England. In Proceedings of the ASCE
Kuala Lumpur. Edited by R.R. Hudson, C.T. Toh, and S.F. Chan. Speciality Conference on Performance of Earth and Earth Re-
Malaysian Highway Authority, Vol. 1. taining Structures, Purdue University, Lafayette. ASCE, New
Marsland, A., and Powell, J.J.M. 1977. The behaviour of a trial York, Vol. 1 (1), pp. 29–63.
bank constructed to failure on soft alluvium of the River Wroth, C.P., and Simpson, B. 1972. An induced failure at a trial
Thames. In Proceedings of the International Symposium on Soft embankment: Part II finite element computations. In Proceed-
Clay, Bangkok. Edited by R.P. Brenner and E.W. Brand. Asian ings of the ASCE Speciality Conference on Performance of
Institute of Technology. pp. 505–525. Earth and Earth Retaining Structures, Purdue University, Lafay-
Muir Wood, D. 1990. Soil behaviour and critical state soil mechan- ette. American Society of Civil Engineers, New York, Vol. 1 (1),
ics. Cambridge University Press, Cambridge. pp. 65–79.

100 100

95 95

75 75

25 25

5 5

0 0
© 2003 NRC Canada

I:\cgj\CGJ40-01\T02-081.vp
Wednesday, January 22, 2003 12:02:22 PM

Vous aimerez peut-être aussi