You are on page 1of 37

16.

EMBANKMENTS OVER SOFT


SOIL AND PEAT

S. Leroueil
R.K. Rowe

16.1 General Behavior of Clay servations. Figure 16.1 shows the excess pore
Foundations Under Embankments pressures, ~u, observed beneath the center-
line during the construction of embankment
In the analysis of the behavior of embank- B at Saint-Alban (Leroueil et al. 1978a). At a
ments on clay foundations it has commonly depth of 2.5 m, the pore pressure increase is
been assumed that the behavior is perfectly small up to a yH value of about 25 kPa and
undrained during construction and that then rises roughly at the same rate as the total
drainage and consolidation start only after the applied vertical stress. The behavior is essen-
end of construction. This approach has been tially the same at a depth of 5 m, but with a
widely used and has generally performed well change at yH = 48 kPa. At greater depths,
for conventional design situations. However, the pore pressure increase remains small in
observations during construction have shown comparison with the applied stress. In gen-
that while this approach may often provide eral, two phases of behavior can be identified:
reasonable designs, the actual behavior may the first during which the pore pressure in-
be more complicated and that conventional crease is low; the second during which the
undrained analyses may overpredict pore increase in pore pressure is approximately
pressures and lateral displacements. Thus, if equal to the increase in total vertical stress.
one wishes to predict the actual behavior of Significant partial consolidation during
an embankment on clay, it is essential to have construction has been reported by a number
a good knowledge of the mechanical behavior of investigators (e.g. Law & Bozozuk 1979;
of natural clays, as described in Chapter 2, Tavenas & Leroueil1980; Ortigao 1980; Kab-
and to understand what may happen under baj 1985; Rowe et al. 1995d). The pore pres-
an embankment during construction as de- sure increase observed during the first phase
scribed in Section 16.l.l. of loading under more than 30 embankments
(characterized by B1 = ~ul ~(Jv) is plotted in
16.1.1 BEHAVIOR Fig. 16.2 as a function of the normalized
DURING CONSTRUCTION depth, zlD, with D being the thickness of the
As in most geotechnical problems, it becomes clay layer. Two observations w~re made by
possible to understand soil response only Leroueil et al. (1978b): first, Bl is smaller
when the corresponding stress path is known. than predicted when perfectly undr~ned be-
Under embankments, the effective stress havior is assumed; and second, the B 1 versus
paths can be deduced from pore pressure ob- zlD relationship has the shape of a consolida-
R. K. Rowe (ed.), Geotechnical and Geoenvironmental Engineering Handbook
© Kluwer Academic Publishers 2001
464 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

40r--------r------~--------~
• T1 = 2.5 m
o T2 =5.0m
a G3=9.3m
o T4=12.0m
30~-------r--~---1--------~

(ij'
~20~------.r-----~--~--~~
Silty clay 0 ::l
-.::]

----.A

OO~~~::~2~O=------J40~------6JO
"(H(kPa)
(a) (b)

FIGURE 16.1. Trial embankment (section B) at Saint-Alban: (a) location of piezometers; (b) pore pres-
sures measured under the centre of the section, as a function of the embankment load (after Leroueil
et al. 1978a; reproduced with permission, Canadian Geotechnical Journal).
tion isochrone, indicating that in these cases
there is significant consolidation during the
o 0.2 1.0
early stages of construction when the soil is
o
overconsolidated. The amount of consolida-
• tion that occurs during the early stages of
0.21-----t=-='-----"-..r construction will vary with the consolidation
• characteristics of the clay, the thickness of the
clay deposit, the drainage conditions and the
0.4 t----+-----,J---lJ-lJ-:::~--+_-__j
o rate of loading of the embankment and this
can explain the scatter of data evident in Fig.
0.6t----+----~......--at-~--~+_-__j 16.2. This said, however, there does appear
to be a general tendency that can be approxi-
mated by an isochrone of the form:
0.81-1/~4--;::::::!:::t:==t=~
• Tavena. and Leroueil. 1980
o Ortigao, 1960 (16.1)
A Kabbaj. 1965
1.0~~...L_ _.L=:::::I==:::C=:::.J

where z = the distance from the upper drain-


FIGURE 16.2. Compilation of observed excess pore
age boundary, A = the drainage path (A =
pressures in clay foundation in the first phase of em-
bankment construction. Isochrone shown is given by O.5D in Fig. 16.2), and Bm = the maximum
Eq.16.1 for Bm == 0.6. Data from (e) Tavenas & Le- pore pressure ratio. The value of Bm can be
roueil (1980), 0 Ortigao (1980), (L.) Kabbaj (1985). expected to vary from case to case as can be
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 465

(b)

FIGURE 16.3. (a) Effective and total stress paths, and (b) pore pressure under the centerline of an
embankment.

seen from the data in Fig. 16.2. This figure compared with the preconsolidation pressure
shows an isochrone for Bm == 0.6. measured in conventional 24-h oedometer
If the behavior of the clay foundation un- tests, (cr~conv) (Morin et ai. 1983; Leroueil
der an embankment were perfectly un- 1996). The results can be summarized as fol-
drained, the effective stress path for a point lows: for overconsolidation ratios (OCRs, es-
at or near the centerline would be as 0' -U' timated on the basis of laboratory tests) be-
in Fig. 16.3a. However, as a consequence of tween 1.2 and 2, there is a good agreement
the rapid consolidation during early stages of between the two parameters; at lower OCRs,
construction, the effective stress path may be laboratory tests generally slightly underesti-
0' -P', reaching the limit state curve at P', at mate in situ values, typically by 10%; for
a vertical effective stress, cr~, close to the pre- OCRs larger than 2, laboratory tests generally
consolidation pressure, cr~, of the clay. As the overestimate in situ values, as indicated in
clay becomes normally consolidated, its coef- Fig. 16.4. Based on data for clays with 2 :::;
ficient of consolidation is reduced by a sig- OCR ::s; 6 (Fig. 16.4), Leroueil (1996) pro-
nificant amount and the behavior becomes posed:
essentially undrained. Due to the shape of
the limit state curve of natural clays (see Sec- , cr;conv
tion 2.10), further loading is associated with O'vy = 0.64 + 0.26 OCR (16.2)
a stress path such as P' -A', under an essen-
tially constant vertical effective stress equal The overestimation of cr~ by laboratory tests
to cr~. Such a stress path corresponds to an in overconsolidated clays with OCR > 2 can
increase in pore pressure equal to the in- be explained by the shape of the limit state
crease in total stress (B 2 = !!.ul !!.cr v = 1.0), as curve of natural clays and the fact that coef-
shown in Fig. 16.1 during the second phase ficient of earth pressure at rest, Ko, is high in
of loading. these materials (Leroueil et ai. 1978b). If the
The change in pore pressure generation embankment is built to a height in excess of
during construction is thus associated with that corresponding to point A in Fig. 16.3,
the yielding of the soil when the effective the effective stress path will continue up to
stress path reaches the limit state curve. This F', on the strength envelope of the normally
in situ vertical yield stress, (cr~), has been consolidated clay, where there is local failure
466 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

2.5 ....---r--.....,..--.,--~Ir-r-----r---,
I i /
/
?
:
A I~ ~ 50
2.01---t--r- --i---I1
I
I .... ~/-...J
I!!
::l

~ 40
l/
/'
Ii
l!I3=2.5
,f1.5 ~.. ,. d peony. Q. /
8
); • I'rI
UII
dvy= 0.64 + 0.26 OCR I!?
8.30
/ ~
d..e - d vo = 22 kP~ Q
""bQ.1.0 I---T-Ili" ~Rupert - 7 }
Q Aries Leroueil et al., 1978 ~ 20
an /' A V
82=0.98

• Interstate 95
0.51----+-1 • Saint-Esprit Bouclin, 1990
/ /a'
/"c lf1 =0.63
E 10

o
IS] Olga-C
::.' Olga-B
St-Amaud et al., 1992
SEBJ, 1983
~
1::
Gl
?
// V
P' F'

---
1~---2~~~3~--~4----~5----~6~---7 E 0
~
-y.
OCR = dpconvldvo ~ -.,
Q.
~ ~
=6

""
10
r-.....
FIGURE 16.4. Variation of ratio cr' onJ(J~ with ~ ~~
overconsolidation ratio of clay (onG sites with 2
·c 20
1\
Pause in construction
OCRs 2 :5 OCR :5 6 are presented) (after Ler- ,g for 2 days
oueil 1996; reproduced with permission ASCE, ~
~ 30
Journal of Geotechnical Engineering). E
'1;1
E 40
"2as
and then possibly to the critical state C'. Be-
C/) ,--
//-:./:.~:.~:.;....}.)~.. 1\
\
1:: 50
Gl L1u 0 Ym----.,
tween F' and C', the increase in excess pore E
Gl
pressure is larger than the increase in total ien 60
0 10 20 30 40 50 60 70
stress (8 3 = l!u//::'JJ y > 1.0) as shown in Fig. yH(kPa)
16.3b. It should be noted that Br , B2 and B3
discussed above are incremental values dur- FIGURE 16.5. Behavior of trial embankment at
ing different stages ofloading and do not cor- Kalix, Sweden, during construction (modified
respond directly to the conventional 8 = from Holtz & Holm 1979).
(l!u/ l!0') under the entire loading (where l!0'
= yH). Hence a high value of 8 3 does not 16.1.2 CLAY BEHAVIOR AFTER THE END
necessarily mean that the embankment is un- OF CONSTRUCTION
stable. As discussed later, pore pressure may During construction the effective stress path
develop even after construction is completed, is such as 0' -P' -A' (Fig. 16.3a). The total
i.e. when there is no increase in total stress, stress state at the end of construction is at A
but 8 may still be less than unity. and the pore pressure is defined as the hori-
The pore pressure generated during the zontal distance between A and A'. During the
construction of an embankment and the cor- consolidation stage, after the end of construc-
responding stress path have a direct influence tion, the effective stress condition moves
on settlements and lateral displacements. along A' -B', whereas the total stress state
The observations of the Kalix test embank- moves from A to B under an essentially con-
ment (Holtz & Holm 1979) shown in Fig. stant vertical total stress (allOwing here for
16.5 (with p' and F' conditions correspond- some stress redistribution and hence some
ing to those in Fig. 16.3a) illustrate this fact. change in O'r - 0'3)'
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 467

16.1.3 STAGE CONSTRUCTION can reach local failure at F~, and will then
As the vertical effective stress increases be- move towards the critical state of the soil,
yond O'~, there is a decrease in void ratio and C~, corresponding to the new void ratio of
an increase in undrained shear strength. Con- the soil. The corresponding pore pressures
sequently, if the initial undrained shear are as illustrated in Fig. 16.6b, with pore
strength is too small to ensure the stability of pressure increases during the reloading
the embankment under its final height, the stages approximately equal to the increase in
strength can be progressively increased by total vertical stress near the embankment
constructing the embankment in stages, with centerline.
sufficiently long periods of consolidation be-
tween the stages to allow pore pressure to de- 16.1.4 PORE PRESSURES
crease and strength to increase. In some Pore pressures generated during the con-
cases, there may be little or no measurable struction of an embankment are generally
strength gain in a reasonable time and con- as illustrated in Fig. 16.3b, with B1 during
ventional stage construction may not be use- the early stages of loading as described by
ful (e.g. see Lo & Stermac 1965; Stermac et Fig. 16.2 and Eq. 16.1. At the end of con-
al. 1967). In some cases, relatively large struction, the vertical effective stress is gen-
strains (thus consolidation) are necessary be- erally close to the preconsolidation pressure
fore observing a measurable strength in- of the clay and in this case the excess pore
crease. The consolidation process can be ac- pressure that will dissipate after construction
celerated by appropriate use of vertical drains is given by:
(see Section 15.2.3).
Figure 16.6a presents a typical stress path Llu = Ll(Jv - ((J; - (J:). (16.3)
at or near the embankment centerline during
stage construction. During the second and When the embankment is loaded to such a
follOwing stages, loading is associated with height that local failure is reached at point F'
stress paths such as B'-D' and E'-G' on the (Fig. 16.3a), there may be softening of the
new limit state curves. If loading is continued clay associated with an increase in pore pres-
beyond the point G', then this soil element sure larger than the increase in total vertical

/
/
/
/
/
/
/ A'
/
/
/.
0'
o Total applied vertical stress ..10'v = 1"1 H
(a) (b)

FIGURE 16.6. (a) Effective stress path, and (b) pore pressure under the centerline during stage-
construction of an embankment.
468 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

stress (Fig. 16.3b) and the stress path may be 16.2 General Behavior of
along F'-C'. As indicated by Folkes and Reinforced Embankments
Crooks (1985) and Leroueil and Tavenas
(1986), there are some field cases where the
on Clay Foundations
behavior is essentially undrained, many cases The earth pressure that develops within the
like those discussed here where there is some fill placed over a soft clay foundation causes
yielding after partial dissipation of pore pres- outward shear stresses on the clay, which
sure, and some cases in which yielding was serve to reduce the bearing capacity of the
not reached during construction. In this latter underlying foundation and hence reduce the
situation, the pore pressures rapidly decrease stability of the embankment (Fig. 16.7a-c).
after the end of construction. Thus the stability can be increased by the in-

(a) (d)

Similar case to :
I
*
~..li----l
I I 2
cr I I

-+ . •-+ ...-1 :
1.. : 1...J..,J ...-I
(b) (e)

Nc Reduction in Nc
bearing capacity
from outward
shear stress Increase in bearing
5 5 capacity from
inward shear stress

00 1.0 00 3
P b*
_c_
't Is u
Su
0
(c) (f)

FIGURE 16.7. Summary of mechanics in a reinforced embankment on soft soil (modified from Jewell
1996): (a) unreinforced; (b) footing subject to outward shear stress; (c) reduction in bearing capacity,
N c , from outward shear stress; (d) reinforced; (e) rough footing; and (f) increase in bearing capacity,
Nc , from inward shear stress.
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 469

clusion of one or more layers of reinforce- ness of most readily available reinforcement
ment in the fill at (or close to) the fill-clay (typically woven geotextiles or geogrids; see
interface. The reinforcement serves two pri- Chapter 7 for a discussion of these materials;
mary functions. First, it resists some or all of but steel strips or grids may also be used),
the earth pressure that develops within the reinforcement tends to have the greatest ef-
embankment, thereby reducing the dis- fect on soft foundations (su < 30 kPa).
turbing forces. The extent to which it attracts The behavior of reinforced embankments
the horizontal thrust depends on the stiffness can best be explained in the context of the
of the reinforcement relative to the underly- analYSis of observed field behavior. Rowe &
ing soil and the shear strength, su, of the Soderman (1984) examined the behavior of
foundation. The movement of the reinforce- an unreinforced and a reinforced embank-
ment and adjacent soil will be the same until ment constructed on a relatively thin (3.3 m)
slip occurs at the interface. Thus, in response layer of organic clay (su == 8 kPa) underlain by
to the development of earth pressure as the dense sand. The unreinforced embankment
embankment is constructed, there will be failed quickly when the fill height reached
some lateral movement away from the cen- about 1.8 m. In contrast to the rapid failure
terline and a sharing of the horizontal thrust of the unreinforced embankment, the rein-
between the reinforcement and the underly- forced embankment experienced relatively
ing foundation. The degree of load sharing ductile failure at a height of 2.75 m. Both the
will be controlled by considerations such as observed and calculated reinforcement
strain compatibility and the relative stiffness strains exhibited the same general trends as
of the reinforcement and foundation. shown in Fig. 16.8. The analysis indicated
The second primary function of the rein- that for fill heights less than 1 m, the mobi-
forcement is to resist the lateral deformations lized shear stress was less than the shear
of the foundation that would otherwise occur strength of the soil and the strains in the rein-
in response to the applied vertical load. If the forcement were small. As the fill increased
soil below the reinforcement is trying to from 1 to 2 m, there was an increase in the
move more than the movement of the rein- strain in the reinforcement as a region of
forcement, then the reinforcement will in- plastic failure developed within the clay.
duce an inward shear stress on the foundation Once the soil had reached its shear strength
(see Fig. 16.7d-f). The magnitude of this along a potential failure surface (at a fill
shear stress depends on the axial stiffness of height of about 2 m here), the embankment
the reinforcement, the strength and stiffness was completely dependent upon the geosyn-
of the foundation and, in particular, the shear thetic reinforcement for the support of addi-
strength at or near the fill-foundation inter- tional fill and, as a consequence, the strains
face. This shear stress serves to increase the increased rapidly as more fill was added. In
embankment stability if the thickness of this case, the geosynthetic-soil interface
the soft soil layer, D, is less than 0.7 times strength was reached over a significant length
the width of the reinforced embankment, b" of the geosynthetic at a height of about 2.7
(details of how to calculate b" will be given m and, since the geosynthetic could not sus-
in Section 16.8.3.2) or if the undrained shear tain any additional load, this corresponded to
strength increases with depth. the limiting strain shown in Fig. 16.8 and fail-
The effectiveness of reinforcing embank- ure of the embankment.
ments on soft foundations depends on the In the case examined above, even with this
relative stiffness of the reinforcement and very soft clay (s u == 8 kPa) and relatively stiff
underlying foundation soil. Given the stiff- geotextile reinforcement (strength Tf = 200
470 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

brittle soils that experience significant strain


softening (see Mylleville & Rowe 1991).
3 • As noted in Section 16.1.1, for most em-
bankment cases the soil will be initially over-
consolidated and there will be relatively rapid
dissipation of pore pressures and small lateral
displacements until the soil becomes nor-
mally consolidated. This has been examined
2
theoretically by Rowe and Li (1999) and
Embankment shown to have significant implications with
Thickness respect to increasing stability. However, it is
H
(m) generally (and conservatively) not considered
in the design of reinforced embankments,
and undrained conditions are normally as-
Observed •
sumed from the commencement of construc-
(SCW, 1981)
Calculated - - tion. Partly as a consequence of this, the ob-
(Rowe a Soderman, 1984)
served strains in reinforcement are often less
than those expected based on the design as-
sumption of undrained conditions.

OL.......,......._.....1-_ _.L..-_.....1-_ _~
Soil reinforcement may be used in conjunc-
o 2 3 4 tion with vertical drains (see Sections 15.2.3.4
Increase in Strain at "A" (%) and 15.2.3.5) and stage construction (Section
16.1.3) to accelerate the rate at which an em-
FIGURE 16.8. Comparison of predicted - (SCW bankment can be constructed to a desired
1981) and obseIVed (e) reinforcement strains at grade. The reinforcement allows construction
strain gage location A (modified from Rowe & Sod-
erman 1984).
of thicker lifts in each stage and hence fewer
stages are needed (see Li & Rowe 1999).

kN m -1 and tensile stiffness ] = 2000 kN


m- 1 ), the reinforcement was only signifi-
16.3 General Behavior of Peat
cantly mobilized after extensive shear failure Deposits Under Embankments
had developed within the foundation soil. 16.3.1 PEAT CHARACTERISTICS
Taking the collapse height to be 2.75 m and There are many different classification sys-
applying a factor of safety of 1.3, the working tems with different definitions of what con-
height for a reinforced embankment with this stitutes "peat" (e.g. see Landva et al. 1983;
reinforcement and soil profile would have Hartlen & Wolski 1996). Unfortunately,
been about 2.1 m. At this height, the strain these systems are not very consistent and the
in the reinforcement would only have been inorganic component (ash content) of peat
about 1%, even though this height is a little may range from less than 20 up to 80%. Thus
above the height at which the unreinforced the term "peat" is often used to include or-
embankment collapsed. Thus the reinforce- ganic soils that may range from jelly-like or-
ment usually does not play a significant role ganic silts and very soft organic clay "muds",
in providing a stabilizing force until the soil to extremely coarse-fibrous meshes of woody
has reached its shear strength along a poten- remains and fibers.
tial failure surface. For this reason, one must The wide range of classification for "peat"
be careful in assessing the shear strength of has led to much confusion, since a "peat"
the soil and particular care is required on with 80% ash content and a Significant clay
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 471

component will respond to embankment paratus. Table 16.1 summarizes the strength
loads in a manner similar to a clay (as dis- properties obtained for some Canadian peats.
cussed in Section 16.1) whereas a fibrous peat Good predictions of embankment behavior
with less than 20% ash content may behave have been obtained using these strength pa-
more like a frictional material than a cohesive rameters evaluated in this manner (see Rowe
material. For the purposes of this book, the et al. 1984b; Rowe & Mylleville 1996).
term "peat" is reserved for soils with less than
20% ash content. Even with this restriction, 16.3.2 PORE PRESSURES
there can be a variety of peats ranging from Because of the high initial hydraulic conduc-
amorphous granular peats to coarse fibrous tivity of peat, some pore pressure dissipation
peats (see Section 3.2.11 and Table 3.7), will occur during construction; however, ex-
Peats typically have a high natural water cess pore pressures will be developed and
content (50-2000%), high void ratio (usually they normally reach a maximum value at the
5-15, but may be up to 25) and high com- end of construction. In a similar manner to
pressibility. The hydraulic conductivity of that already described in Section 16.1 for the
peat is typically high at high void ratios, but early stages of loading on clay, the observed
reduces significantly as the peat compresses increase in pore pressure in peat due to em-
(Lefebvre et al. 1984). Due to the high void bankment loading can be characterized in
ratio and compressibility, it is usually not terms of the pore pressure parameter B =
practical to obtain realistic strength parame- !1ul!1O'v and Eq. 16.1, as shown in Fig. 16.9.
ters from conventional triaxial tests (these The range of Bm values that can be de-
yield "friction angles" of 45-55° based on en- duced from published field cases is typically
velopes drawn when the tests were termi- 0.1-0.35, although values as high as 0.8-0.9
nated due to excessive deformation rather have been reported (Lupien et al. 1983; Ad-
than shear failure (see Adams 1961; Edil & ams 1965).
Dhowian 1981; Rowe et al. 1984b). Also, due It may be expected that the magnitude of
to the fibrosity of many peats, it may be difficult the maximum excess pore pressures and,
obtaining reliable strength parameters from hence the likelihood of failure, will depend
direct shear tests for these peats. Landva on the rate ofloading. Weber (1969) reported
(1980) has had success using ring shear appa- that in his experience with peat deposits in
ratus and Rowe (Rowe et al. 1984b; Rowe & the San Francisco Bay area, fills in excess of
Mylleville 1996; and various unpublished re- 1.5 m above original ground level were sub-
ports) has obtained clearly defined failure en- ject to collapse. The exact height at which in-
velopes using the Norwegian simple shear ap- stability occurred depended upon the foun-

TABLE 16.1. Strength properties reported for peats


Description and
classification category Natural water
(see Table 3.7) content (%) c'(kPa) <\>' (0) Reference
Fine fibrous peat
Cll 527 1 26 Rowe et al. (1984b)
C9 656 1.2 26 Personal Files
Coarse fibrous peat
C13 680 2.5 28 Rowe & Mylleville (1996)
C12 1450 0.5 29 Personal Files
Sphagnum peat 1200-1500 2.5 27 Landva (1980)
1200-1500 3.5 33 Landva (1980)
472 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

tensile strength of any root mat, which acts


__ ____ __ ____ __
like reinforcement. Thus when loaded
o~ ~ ~~ ~ ~ ~

quickly, peats are likely to fail. For example,


0.2 Flaate and Rygg (1964), Ripley and Leonoff
(1961) and Lupien et aZ. (1983) report "shear
failures" that, on the basis of the published
0.4 data, appear to involve failure in the peat and
z not in the soil underlying peat. These failures
7\
LEGEND
do not usually involve the formation of a
0.6 • Raymond (1969)
J;. Rowe et al. (1984a)
definite sliding surface. Rather, the collapse
1\ = Drainage Path involves rapid and excessive shear deforma-
0.8 z = Depth Below Peat·Fili Interface
tions that give rise to large embankment set-
tlement, lateral movement and "mud" waves.
1.0 L...-_ _---L_ _ _ _----L._ _ _ _--L-_ _ _ _-L-_ _---'
In this context, collapse will be distinguished
o 0.2 0.4 0.6 0.8 1.0 by the fact that additional fill will not increase
the height of the embankment unless steps
are taken to stabilize the embankment, i.e.
allowing pore pressure diSSipation during
FIGURE 16.9. Observed excess pore pressure dis-
tribution at end of construction for two embank- construction, use of berms, etc.
ments constructed over peat (modified from The behavior of a number of unreinforced
Rowe & Soderman 1985b). Data from: • (Ray- embankments constructed on peat underlain
mond 1969), .. (Rowe et aZ. 1984). A = drainage by a weak layer has been reported in the liter-
path, z = depth below peat-fill interface. ature (MacFarlane & Rutka 1959; Lea &
Brawner 1963; Weber 1969; Raymond 1969;
dation condition and rate of loading. At a Rowe & Mylleville 1996). Collapse in these
construction rate of 0.9 m of fill per week, he cases usually involves formation of a crack in
reported a collapse of a fill at a height of 2.4 the embankment and a block-like slide mov-
m. At an even slower construction rate (0.15- ing on or just below the boundary between
0.3 m per week), fills to a height of about 3 m the peat and the underlying soft soil.
could be constructed without failure. Field observations have established that
embankments constructed on peat settle rap-
16.3.3 BEHAVIOR DURING AND idly during construction (loading) and then
AFTER CONSTRUCTION continue to settle at a reduced rate after load-
The peats examined in the investigations ing (e.g. Hanrahan 1964; Adams 1965; Weber
summarized in Table 16.1 all exhibited an es- 1969; Rowe et aZ. 1984a,b). An examination
sentially frictional behavior, with a friction of the time-settlement curves indicates that
angle typically between 26 and 29°. The ini- "primary" consolidation occurs relatively
tial effective stresses both within the peat and qUickly (typically 5-200 days after construc-
any very soft clay or silt layer below the peat tion) for typical thicknesses of peat (2-10 m),
are initially very low. If loaded too quickly, after which the settlement continues to in-
i.e. without significant pore pressure dissipa- crease with time due to secondary compres-
tion during loading, the only shear strength sion. However, significant excess pore pres-
available to carry the load is a small apparent sures have also been observed at times long
cohesion (only a few kPa for the peats exam- after the time-settlement curve indicates
ined in Table 16.1 and this is due to the ten- that "primary" consolidation is complete.
sile strength of the fibers in the peat) and the These excess pore pressures appear to re-
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 473

main at a relatively constant value for consid-


erable periods of time (e.g. Adams 1965;
Rowe et al. 1984a). The reasons for this be-
havior have not generally been agreed upon
(e.g. see Adams 1965; Landva & La Rochelle l--!--++tttnT:g-~;::'l::: .... ~ ~'7
1983). It could be that these excess pore pres- :t n
'iI=:: 1---i--+--t-t"nI ..v.v V'
, "",;' 0
rr I .
VI
sures are "creep induced pore pressures" re-
VI
.....

sulting from time-dependent compression of 6 "<>,,


v.vv ",,'
V
II o.{
.. /
both the peat fibers and of the organic struc- ... I)._v V
ture itself, they could also be due to a rise
_
~ "'>,, ...0.3' / H- r,
v.",2vot==::j::Ff+ttm-J~V11 stress at base
of the water table within the embankment. g . .0.2 /
of embankment
Clearly, the possibility of the presence of ~ 0.1... ... V
these excess pore pressures for long periods E 0.1ul~~r-----:f--t-++tttltP,nu. ./V I 1.~!Jnll%
of time may have important consequences Ir,H ~
il
O'v=
0.05 ...0·",,+-+-+-+1
when considering stage loading of a peat de- O'v

posit and, in such cases, it would be advisable -I-


10
to provide instrumentation to measure the aJz
pore pressure and to use this information to
monitor the construction. FIGURE 16.10. Chart for calculation of vertical
From the available data, it may be con- stresses caused by embankment (modified from
cluded that where drainage of the peat is pro- Osterberg 1957).
vided and a slow rate of construction is used,
there will be little difficulty in constructing For long and trapeZOidal embankments,
low (2 m or less) embankments on peat, al- the solution proposed by Osterberg (1957),
though settlements may be large. However, Fig. 16.10, is generally adequate. This solu-
when an impervious fill is used or where the tion gives the change in stress at depth z due
peat is underlain by soft marl or clay, it may to the trapeZOidal applied stress shown in the
not be possible to construct even low em- insert. To obtain the stress at any point, one
bankments on the peat unless speCial mea- must use the principle of superposition. Thus
sures are taken to ensure stability, e.g. pro- at the centerline one would multiply the
vide drainage, berms, geotextiles, use of value deduced by two, i.e. an equal contribu-
lightweight fill, etc. tion from each side. For a point away from
the centerline, one would need to use Fig.
16.10 twice (once for the geometry on the left
16.4 Settlements of the point of interest, and once for the ge-
16.4.1 STRESS DISTRIBUTION
ometry on the right). Similarly, berms could
UNDER EMBANKMENTS also be approximately considered by ex-
The vertical stress increase due to an em- tending the use of superposition to include
bankment is generally calculated assuming the contribution of the berm to the change in
isotropy and elasticity of the soil, and by: stress at the point of interest. Poulos & Davis
(1974) give a compendium of elastic solutions
(16.4) that may be used for more complex cases. Al-
ternatively, numerous computer programs
in which I = the influence factor, 'Yf = the are available for calculating stresses in elastic
unit weight of the fill material, and H = the media.
thickness of embankment fill. It is important to note that stress distribu-
474 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

tion changes during consolidation and this


(16.9)
can affect the final settlement. Considering - Yw(Sae + 0.5hw) + IYrH.
a point a on the surface of the compressible
deposit, a point b of interest within the de-
Although the changes in effective stress aris-
posit and a point e at the bottom of the de-
ing from Eqs 16.5-16.9 have often been ne-
posit, different aspects to be considered in-
glected, they do need to be considered be-
clude:
cause the change in effective stress, !:10'~, can
• The total vertical stress applied at a given be quite significant and the use of these equa-
point, b, in the foundation decreases with tions is quite straightforward.
the expulsion of water associated with the
settlement (change in soil thickness) that 16.4.2 CONSTRUCTION SETTLEMENT
has occurred above this point (Sab):
The settlement Scat the end of construction
results from the sum of the recompression
settlement, Sr (0' towards P' in Fig. 16.3),
• On the other hand, the total vertical stress and a settlement, Su, resulting from un-
increases due to the fact that the upper drained shearing (P' -A' in Fig. 16.3):
part of the natural soil deposit, and possibly
the lower part of the embankment, has (16.10)
moved below the water table. If !:1y is the
increase in unit weight of the soil when it
becomes submerged and if hw is the rise of Consider a layer, i, of thickness D i' If a rep-
water table above its initial level, and Sae is resentative point in the layer with an initial
the surface settlement, then: effective stress, O'~o;, and void ratio, eo;, expe-
riences an increase in stress, !:1O'y, due to the
I1q = l1y(Sae + hw). (16.6) embankment, then the recompression settle-
ment, Sri, of this layer can be calculated as:
• The reference pore pressure changes due
to the settling of the point, b, and a possi- If (O'~i + !:1O'y) is less than the preconsoli-
ble change of the water table. If the entire dation pressure, 0';, then:
deposit is influenced by this latter change,
then: Sr-i - Di
- - C siDg
I (O'~ + 110')
, . (16.11a)
1 + eoi O'voi

in which She = settlement of point b. Rec- Otherwise, if O'~ + !:1O'y ;::: 0';
ognizing that Sab = Sae - She and combin-
ing Eqs 16.4-16.7, the change in effective
stress due to embankment construction is
Sri = __D i-Csi log ..::.;-
1 + eoi O'voi
(0") (16.11b)

given by:
where C si is the swelling index for the layer
110'~ = (l1y - Yw)(Sae + hw) + IYrH. (16.8)
(see Sections 2.7.3 and 3.5.2 and Fig. 2.7).
When the change in water table affects Now the total recompression settlement, S"
only the upper boundary of the consolidat- can be obtained by summing the values for
ing layer and not the water pressure at the each layer, giving:
bottom, the change in stress applied by the
I
n
embankment varies with depth, but is close Sr = Sri' (16.12)
to: ;=1
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 475

The undrained shearing settlement, Su, asso- The total post-construction settlement, Sd,
ciated to the stress path P' -A' in Fig. 16.3 is the sum of the settlements, Sdi, of the dif-
results from plastic flow of the normally con- ferent soil layers:
solidated clay. This could be evaluated using
I
n
a suitable finite element program and repre-
Sd = Sdi (16.15)
sentative soil parameters. Alternatively, the i=l
following relation established by Tavenas and
Leroueil (1980) on the observations ofl2 em-
Thus the total settlement, S, is given by add-
bankments on clay from different countries
ing Eqs 16.lO and 16.15, namely:
could be used as an approximation:

Su = (0.07 ± 0.03)(H - Hnc) (16.13) (16.16)

where H ne = the height of the embankment 16.4.3.2 Contribution of the Viscosity of Clay
when the foundation clay becomes normally to Total Settlement
consolidated (point P' in Fig. 16.3). Hnc can It has been shown (Leroueil et al. 1985a; Imai
be estimated from the concepts described in & Tang 1992) that, during one-dimensional
Section 16.1.1. compression and in laboratory conditions,
the behavior of clays is controlled by a
16.4.3 LONG-TERM SETTLEMENT
unique effective stress-strain-strain rate
16.4.3.1 Primary Consolidation Settlement (cr~, Ev, tv) [or (cr~, e, ev)] relationship. At a
At the end of construction, the effective given stress, the lower the strain rate, the
stress conditions below the centerline of an higher the strain. This applies during both
embankment often correspond to a vertical primary and secondary consolidation. In par-
effective stress close to the pre consolidation ticular and as a first approximation, it can be
pressure of the clay. The long-term settle- considered that the logarithm of the pre-
ment is thus associated with an effective consolidation pressure varies linearly with
stress increase from cr~ to (cr~o + Llcr) and can the logarithm of the strain rate. It can also
be calculated as follows for a homogeneous be shown that the slope m' = Ll log tv!Ll
layer of thickness D i : log cr; is in fact equal to the ratio CeiC ae ,
where Cae is the secondary compression in-
Sdi = ~Cci log(O'~oi ~ AO'). (16.14) dex as defined in Section 2.7.5. Table 3.14
1 + eoi O'pi gives typical values of Cae/C c and m' for dif-
ferent types of soils. For inorganic clays, Cael
At the end of primary consolidation, when C e is typically about 0.04 (Mesri & Godlewski
the excess pore pressures have dissipated, the 1977).
settlement of embankments can usually be Temperature also influences the behavior
calculated on the basis of parameters from of clay. At a given strain rate and at a given
the conventional 24-hour oedometer test (see effective stress, the higher the temperature,
also Section 16.4.3.2). Due to the fact that the higher the strain (Leroueil & Marques
the normally consolidated branch of the com- 1996). Since the temperature beneath an em-
pression curve is not always linear, C c has to bankment may be different from that of the
be taken as a secant compression index be- laboratory and the field strain rate is generally
tween the preconsolidation pressure and the smaller than that in the laboratory by a few
final vertical effective stress. Typical correla- orders of magnitude, the settlements in situ
tions for C e are given in Section 3.5.2. could be different from those expected for
476 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

the same stress increase in laboratory condi-


tions. 12.51"'""'7~--r---r---,...----'
1 +eo =1.0
U sing the viscous model described above, 10.01----'~-+---1___--_+--__I

Leroueil (1988, 1996) evaluated the strain 0.8


component, ,1.£: , that should be added to the
strain, estimated on the basis of the conven-
tional 24-h oedometer test, to evaluate the
strain at the end of the in situ primary consol-
idation in the field. Considering the end-of-
primary consolidation (EOP) to correspond
to a degree of consolidation of 95% accord-
ing to one-dimensional consolidation theory
(Section 2.7.4), the strain rate at this time can
be described by (Leroueil 1988, 1996): FIGURE 16.11. Estimation of the end-of-primary
in situ strain, LlE", in excess of that determined
(16.17) from conventional 24-h laboratory oedometer
test, when strain rate and temperature effects are
considered (after Leroueil1996; reproduced with
where k = hydraulic conductivity of the clay,
permission ASCE, Journal of Geotechnical Engi-
u~ = initial excess pore pressure, and A = neering).
maximum drainage length. Assuming that the
temperature in situ is about 12°C below that
in the laboratory, the viscous strain, ,1.£:, and neglects the effects of possible disturbance of
settlement Svi can be deduced from the strain the specimens tested in laboratory, and the
rate model to be given by: fact that during consolidation in situ struc-
ture may develop when there is a combina-
tion of low strain rate and significant elapsed
(16.18) time and this would reduce the effect of the
strain rate (Leroueil et al. 1996a).
Mesri & Choi (1985) suggest that the set-
tlement at the end of primary consolidation
be calculated from Eq. 16.14, with compres-
where D j = the layer thickness, C c = com- sion parameters deduced from incremental
pression index under the final effective stress, loading oedometer tests with reloading at the
eo = initial void ratio, m' = ,1. log £'/,1. log end of primary consolidation (when the ex-
O'~ = C /C ae = a strain rate factor having typ- cess pore pressure becomes negligible). This
ical values as given in Table 3.14. Figure approach leads to predicted settlements
16.11 shows ,1.£: as a function of C/(l + eo) smaller than those estimated from conven-
and £EOP for an m' value of 32. tional24-h oedometer tests since these latter
In the previously described "viscous ap- incorporate some secondary compression. It
proach", the settlement of layer i at the end is thus thought that the Mesri & Choi (1985)
of primary consolidation would be the sum approach (also called EOP approach) pro-
of the settlement, Sm, calculated from Eq. vides a lower limit of settlements.
16.14, using the compression parameters ob- Mesri & Choi (1985) and Mesri et al.
tained in 24-h oedometer tests, and the vis- (1994) found that settlement predicted using
cous settlement Svi defined as Eq. 16.18. This the EOP approach gives results representa-
gives an upper limit of settlements since it tive of observed field settlements. On the
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 477

other hand, detailed studies performed by cated in Section 16.4.3.1 without the
Kabbaj et al. (1988) on four different sites correction for viscous effects described in
from Canada and Sweden show that the EOP this section.
approach significantly underestimates in situ • In highly compressible clays [C/(l + eo)
settlements. Leroueil (1988) and Leroueil values larger than 0.25] or for embank-
ments of special importance (test embank-
(1996) indicates that the observed settle-
ments, embankments for which the
ments are in rather good agreement with the magnitude of the settlements may have im-
viscous model described above. Other de- portant consequences), viscous effects
tailed studies and experience gained in Swe- could be significant and should be consid-
den (Larsson 1986), France (Magnan 1992) ered as described above.
and elsewhere (Leroueil 1988) confirm the
importance of rate effects during primary After the end of primary consolidation, the
consolidation. settlement continues with time due to sec-
The apparently contradictory observations ondary consolidation. The secondary settle-
noted above are thought to be due to the ef- ments can be estimated as:
fects of: (a) sample disturbance, which in-
creases laboratory strains; and (b) structuring
effects, which can decrease in situ strains.
(t)
S5 = DC(J£ log -
tp
= -DCIle
-log -
1 + ep
(t)
tp
(16.19)

Unfortunately, there is no validated method


for evaluating the ability of a soil to develop in which CaE and Cae are measured in oedo-
structure in in situ conditions. On the basis of meter tests (see Section 2.7.5) or estimated
all these remarks and observations, Leroueil from C c (see Table 3.14); D = the thickness
(1996) recommended the following ap- of the clay layer; tp = the time at the end-
proach: of-primary consolidation, which can be prac-
tically estimated as the time necessary to
• For typical projects in clays of low com- obtain a degree of consolidation of 95% (as
pressibility [C/(l + eo) values less than
discussed in Section 16.6); t = the time of
0.25], D.E: is smaller than 2.5% and is often
compensated by some disturbance of the interest; and e p = the void ratio at time tp.
clay and some rise in the water table (see
Section 16.4.1), which is often neglected in 16.4.4 EFFECT OF EMBANKMENT
analyses for these soils. In these cases con- REINFORCEMENT ON SETTLEMENT
ventional 24-hour oedometer test results, The inclusion of suitable reinforcement at or
which already include some secondary near the base of the embankment may reduce
compression, can be used, as indicated in undrained shear deformations relative to
Section 16.4.3.1, without any additional what would have occurred at the same fill
viscous correction (described in this sec- height without reinforcement. However, in
tion). most practical cases one will not be compar-
• When vertical drains are used to accelerate ing settlement at the same fill height. The fact
the consolidation process (see Sections
that reinforcement is being used usually
15.2.3.4 and 15.2.3.5) or when the clay de-
means that the embankment will be con-
posit is thin or stratified, the in situ strain
rate at the end of primary consolidation is structed to a greater height than an unrein-
larger than 3 X 10- 10 S-l in most cases. forced embankment for the same soil profile.
Consequently, D.E: remains relatively small This will mean a larger zone of shear failure
and, for the same reasons as those indi- in the soil for the same factor of safety and,
cated above, the conventional 24-hour oe- consequently, potentially larger undrained
dometer test results can be used as indi- shear deformations than would have been ex-
478 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

j, of 1000, 2000 and 4000 kN m- I . In all


I cases, the calculated collapse fill thickness is
COllapse~
essentially the same (6 m), however with j =
I
:J(kN/m)
1000 kN m -1 the embankment could only be
I
I
constructed to about 3.6 m above original
, 4000 grade before the undrained shear deforma-
Net I
tion became excessive. In contrast, for j =
/~......,.--.
I
Fill 4
Height
Eu/su = 125 4000 kN m -1, the maximum net fill height
hnet 3
(m) was about 4.7 m. Thus when dealing with re-
inforced embankments one could achieve a
2
practical failure height that is less than the
collapse height predicted from limit equilib-
rium or bearing capacity considerations. This
2 3 4 5 6 7 necessitates limiting the allowable strain in
Fill Thickness, H (m) the reinforcement as discussed in Section
16.8.3.
FIGURE 16.12. Net fill height versus fill thickness The inclusion of reinforcement in an em-
for various reinforcement tensile stiffness, J, val- bankment has very little effect on long-term
ues (modified from Rowe & Soderman 1987): (e) settlements, Sd, that would be calculated for
continuous plasticity, (i) maximum net fill height. an unreinforced embankment with the same
fill height as described in Section 16.4.3.
pected for an unreinforced embankment at However, reinforcement may provide the ad-
the same factor of safety against collapse. The ditional stability required to allow the placing
magnitude of the undrained deformations of a surcharge (see Section 16.8.3).
will depend significantly on the stiffness of
reinforcement used near the base of the em- 16.4.5 SETTLEMENT OF EMBANKMENTS
bankment. ON PEAT
To illustrate the point made above, con- Peat deposits typically experience significant
sider an embankment and soil profile as settlement by the end of construction, addi-
shown in the insert to Fig. 16.12. For this soil tional consolidation settlement and second-
profile an unreinforced embankment will col- ary compression. Due to the low initial
lapse at a fill thickness H = 3 m. Provided stresses, equations such as Eqs 16.11 and
there is sufficient reinforcement to maintain 16.14 generally cannot be used to establish
equilibrium, the embankment can be con- the construction settlement, Sri, and consoli-
structed to a maximum fill thickness of 6 m. dation settlement, Sdi, of a layer. Rather, total
This limit is imposed by bearing capacity con- settlement, S, of a peat deposit of thickness,
siderations (to be discussed in Section 16.8.3) D must either be established based on data
and no amount of reinforcement allows a from consolidation tests (oedometer) giving
higher embankment under these foundation !1e versus !10'~:
conditions unless other methods of soil im-
provement (see Chapter 15 and Section 16.8) s = (D Ae(z) dz = (D Ev(z)dz (16.20)
are also introduced. Figure 16.12 shows a Jo 1 + eo Jo
plot of the calculated net fill height, h neb be-
neath the shoulder versus the fill thickness, where !1e(z) and Ev(Z) will depend on !1O'~(z),
H, (where h net = H-Su and it is assumed Sr which may vary with depth, z. Alternatively,
== 0) for a reinforcement with tensile stiffness, one can estimate the settlement, S:
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 479

(16.21) values of Bm for peat typically range between


0.1 and 0.35 (see Section 16.3.2). This equa-
where tv = a function of the change in verti- tion can be evaluated using a hand calculator.
cal effective stress, LlO'~. In the simplest case, For typical combinations of parameters (Bm
assuming a linear relationship between strain :::; 0.35), the settlement at the end of con-
and effective stress increase, tv is given by: struction, Sf> ranges between 80 and 100% of
the "final" consolidation settlement, i.e. Sr ==
tv = Q(Acr~)
pa
(16.22) 0.9S ± O.IS. The "final" settlement corre-
sponds to Bm = 0 and Eq. 16.24 reduces to:
where Pa = atmospheric pressure and, based
S= QYrHD (16.25)
on data from a large number of field cases
(Pa + QYw D )
(see Rowe 1984), n = 0.24 represents a mini-
mum settlement, n = 0.66 represents a typi-
Equation 16.22 suffers from the presumption
cal situation (recognizing wide scatter of
that the strain is linearly proportional to ef-
data), n = 1.6 (for LlO'~ :::; 40 kPa) represents
fective stress where, in fact, the peat becomes
an upper bound to most field data.
stiffer with increasing applied stress. A gener-
When considering embankments on peat
ally conservative, i.e. tending to overpredict
the change in effective stress will usually de-
settlement, bi-linear approximation is given by:
pend on the settlement and will, as a first ap-
proximation, be given by:
Q = 1.33 Acr') :::; 0.4
for ( -;: (16.26)
Acr~(z) = YrH - YwS - Au(z) (16.23)

where H = the fill thickness and S = the set- Q = 0.33 + 0.4 palAcr~ for ( ~~~) > 0.4. (16.27)
tlement (assuming here that the water table
is at or very close to the original ground level,
as is often the case) and that the unit weight For values of L1O'~/pa :::; 0.4, Eqs 16.24 and
of the fill does not increase Significantly, and 16.25 can be used directly to get an upper
b.u (z) is the change in pore pressure at depth bound on end of construction, Sf> and final
z (e.g., due to a rise in water table into the settlement, S (including some secondary
embankment). It also assumes that the em- compression). For L1O'~/pa :2: 0.4, an iterative
bankment is wide compared to the peat de- solution can be obtained by estimating S rand
posit thickness. S, calculating n from Eq. 16.27, calculating
To estimate the end of construction settle- an improved estimate of Sr and S from Eqs
ment, Sf> one can use the end of construction 16.24 and 16.25 and repeating the procedure
pore pressure isochrone given by Eq. 16.1 to until convergence is attained.
calculate the end of construction change in Settlements due to secondary compression
stress, b.O'~(z), and then combining Eqs can be particularly significant for embank-
16.20, 16.22 and 16.23, integrating and solv- ments on peat, especially if the embankment
ing for Sr gives: was not surcharged. There is evidence that
this settlement can be calculated using Eq.
Sf = QYrHD(l - BmD/A + BmD2/3A2)1 16.19 and values of CaelC c for peat given in
[pa + QYw D(1 - BmD/A + BmD2/3N)] (16.24) Table 3.14 (Cae/C c = 0.06 ± 0.01). There has
been some questioning of the C ae/ C c concept
and A = 0.5D for two-way drainage and A = for peat (Fox et al. 1992), however more re-
D for one-way drainage. End of construction cent research on the same peat (Mesri et al.
480 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

1997; Fox et ai. 1999) provides additional


a'v y(cm)

.. ,
support for the concept and suggests that de- (kPa)
o 0 20 40 60 800 246 8
partures from the concept in the laboratory
test may be due to biodegradation of the peat
in the laboratory.
2
\

1\
/' V
a'v ""~~
When dealing with embankments where t"\
the surface settlement and especially differ- \ /
ential settlement of the embankment with "' \
\ ./
V
time can be problematic, e.g. road embank- 8
ments, surcharging of the embankment, has 10
a'vo\
\
a',\ /
been successfully used as a means of reducing (.)
or delaying the onset of secondary compres-
o\
sion after removal of the surcharge (e.g. see
Mesri et ai. 1997). As noted by Rowe et ai.
(1984a), the surcharge should be applied to
2 \
\
\
(c-lIL consolidated
state /
~
/
4
\ \ V
...L-
the embankment over the entire peat deposit
and not just over the thicker portions of the J
deposit. \ ~ I
f
8
\ \

16.5 Lateral Displacements


10 " \
\
a'vo
\
,
a'v
\

IJ
Field observations of embankments on clay 12 (b)
show that the lateral displacements during
and at the end of construction can generally FIGURE 16.13. Effective stresses and lateral dis-
be described as a function of the settlement placements at end of construction: (a) Cubzac-Ies-
under the centerline of the embankment. In Ponts, and (b) Saint-Alban B (after Tavenas et al.
the overconsolidated region (from 0' to P', 1979; reproduced with permission, Canadian
Fig. 16.3), the maximum lateral displacement Geotechnical Journal).
under the toe of the embankment is linearly
related to the surface settlement under the clay has become normally consolidated over
center of the embankment. Tavenas et ai. the full depth of the deposit (Fig. 16.13a), the
(1979) found: displacement curve, y = f(z), reflects this ho-
Ym, = (0.18 ± 0.09)S,. (16.28a) mogeneous condition and roughly corre-
sponds to the classical theoretical solution
During the follOwing phase, when the soil is (Poulos 1972). If only a part of the foundation
normally consolidated (P'A'F' in Fig. 16.3), has become normally consolidated as for the
the increase in maximum horizontal displace- Saint-Alban B test fill (Fig. 16.13b), the y =
ment is very close to the increase in settle- f(z) curve reflects the resulting non-uniform
ment and Tavenas et ai. (1979) found that on condition with small deformations developing
average: in the lower, still overconsolidated clay, and
large deformations occurring in the upper nor-
Ymu = (0.91 ± 0.2)Su' (16.28b)
mally consolidated clay.
The distribution of lateral displacements with Tavenas et ai. (1979) noted two points con-
depth at the end of construction is directly cerning the long-term lateral displacements:
dependent on the effective stress condition in first, the distribution of lateral displacement
the foundation, as shown in Fig. 16.13. If the with depth does not Significantly change fur-
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 481

ther after the end of construction; second, the 7. There is a linear relation between the ver-
increase in lateral displacement varies lin- tical effective stress and the void ratio.
early with the increase in settlement. For em- 8. The soil is not viscous, thus not influenced
bankments with a factor of safety, F, a full by time or strain rate.
width b and crest width B on a clay layer of
thickness D, where 0.9 ::5 biD ::5 2.3, 0.35 ::5 In practice, the strain will not be purely one-
(b - D)/D ::5 0.9, 1.25 ::5 F ::5 1.5 and side dimensional and the flow of water is generally
slopes of 1.5 : 1, the relationship between the two-dimensional. To take this into account,
increase in maximum horizontal displace- Biot (1941, 1955) proposed a general theory
ment, l1y md, and the increase in settlement, for predicting pore pressure dissipation and
I1S d , was given by: consolidation settlement, which is imple-
mented in many modem finite element codes.
t1Ymd = (0.16 ± 0.02)t1Sd' (16.28c) Also, Davis & Poulos (1972) provided general
Thus the total horizontal displacement is solutions to the two- and three-dimensional
given by summing Eq. 16.28a-c to give: consolidation equations in a graphical form
that can be readily used in practice, as shown
(16.29) in Fig. 16.14 (note: Davis & Poulos also give
solutions for additional cases).
For different geometries, the ratio Ym/S var- Variances from hypotheses 2-5 and 7 are
ies, having a tendency to decrease when the likely to result in excess pore pressures that
factor of safety rises and/or at lower embank- are smaller than the theoretical values and
ment side slopes. the settlements at a given time, which are
larger than the theoretical values.
16.6 Consolidation Contrary to hypothesis 6, the hydraulic
conductivity k is not constant, but decreases
16.6.1 PRIMARY CONSOLIDATION with the void ratio, e. As a first approxima-
The settlements that occur following con- tion, it follows the law:
struction (see Section 16.4.3) develop with
time due to consolidation of the soil as the k = kr exp[2.3(e - er)/Cd (16.30)
excess pore pressures dissipate. Traditionally,
consolidation has been predicted based on where kr = the hydraulic conductivity at the
Terzaghi's (1925) theory for one-dimensional reference void ratio, en and C k is close to
consolidation as discussed in Section 2.7.4 0.5eo (Tavenas et al. 1983b).
and Fig. 2.8. This theory is based on several Finally, as indicated in Sections 16.4.3.2
hypotheses, some of which may not be fully and 16.6.2, clays are viscous and this influ-
realized in practice. Specifically, the hypothe- ences pore pressure generation and the con-
ses are: solidation process. Thus, the coefficient of
1. The strains in the clay layer are one- consolidation is not a basic property of a soil.
dimensional and small. As a consequence, it depends on the way it
2. The flow of water is one-dimensional and is estimated (see Section 2.7.4). Experience
is controlled by Darcy's law. shows that:
3. The clay layer is homogeneous.
4. The soil particles and the pore liquid are dCasagrande; Eq. 2.33) < C v
incompressible.
5. The soil is saturated. (Taylor; Eq. 2.34) < C v = kMc (Eq. 2.29). (16.31)
6. The hydraulic conductivity is constant. Yw
482

0r---~~~==~.------.-------r----.
(a) Values of ~
0.2 o (one dimensional)
1
0.4
u
0.6
Permeable Top

0.8 o
Permeable Base
1.0 --5--------~----------~--------~----------~-1----~
10- 10-4 10-3 10-2 10- 5 X 10-1

O~--------~-----------r----------,-----------~~

Values of ~
0.2 o (one dimensional)
0.5
1
0.4 2
u
0.6

Permeable Top
0.8 Impermeable Base
(b)

1.0 ~-4~--------L-=3~--------"'--::-2-----------'~1---------"--~
10 10- 10- 10- 1
cvt
Tv= 0 2

FIGURE 16.14. Degree of settlement/consolidation curves for two-dimensional How conditions: (a) per-
meable top, permeable base; (b) permeable top, impermeable base (modified from Davis & Poulos
1972).
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 483

Furthermore, experience also shows that the or by Fig. 16.14 for Tv = c vnJlD2 (where D
coefficient of consolidation in the field is = layer thickness irrespective of drainage
larger than that calculated from laboratory path). Figure 16.14a is for a layer with a per-
data using the method of Casagrande. Com- meable top and bottom, and Fig. 16.14b is
piling data from 16 embankments, Leroueil for a layer with a permeable top and imper-
(1988) found an average ratio of 20 between meable bottom.
these two parameters. Thus, when possible,
it is desirable to use a numerical model to 16.6.2 CONSIDERATION OF VISCOUS
take into account the variation in consolida- AND STRUCTURING EFFECTS
tion characteristics with depth using repre- As indicated in Sections 2.7.5 and 16.4.3.2,
sentative laws for the compression curve and clays (and also peats) are viscous and this in-
the permeability-void ratio relation. fluences the settlement during primary con-
As a consequence of hypothesis 7, Ter- solidation. As schematically shown in Fig.
zaghi's theory also considers the degree of 16.15, there are two extreme possibilities (Jam-
consolidation U, defined in terms of pore iolkowski et al. 1985): curve A assumes that
pressure, to be equal to that defined in terms creep occurs only after the end-of-primary
of settlement. This is often not the case, espe- consolidation, and consequently, the strain at
cially for soils that are initially overconsoli- the end-of-primary consolidation would be the
dated and become normally consolidated same in situ (EvA on field curve A in Fig. 16.15)
during loading. In such a case, separate con- and in the laboratory. Curve B assumes that
sideration can be given to the settlement and viscous strains develop during primary consoli-
pore pressure diSSipation in the overconsoli- dation and, consequently, the strain at the end-
dated and normally consolidated stages, e.g. of-primary consolidation is larger in situ (E vB on
see discussion in Section 16.1.1 recognizing field curve B in Fig. 16.15) than in the labora-
that the rate of settlement in the normally tory. The viscous model described in Section
consolidated state will depend on the excess 16.4.3.2 corresponds to curve B. However, if
pore pressures remaining when the soil be- structure develops under small strain rates for
comes normally consolidated. The excess in situ conditions, the in situ strain at the end-
pore pressures at the end of construction are of-primary consolidation could be smaller
best established by field measurement. Alter- than EvB (see Section 16.4.3.2).
natively, they can be calculated either from A consequence of the viscous behavior of
current numerical models or, based on the
discussion in Section 16.1.2, approximated as
the difference between the final expected ef- log t
fective stress and the preconsolidation pres-
sure.
The settlement at time t after the end of ...
" Field curve A
construction can then be expressed as:
Field curve ~"\
(16.32)

where Sd is given by Eqs 16.14 and 16.15, and


U ne the degree of consolidation in the nor-
mally consolidated range. U ne is given approx- FIGURE 16.15. Strain-time relations in the labo-
imately by Fig. 2.8b for a dimensionless time, ratory and in the field (modified from Ladd et al.
Tv = cvnP N (where A is the drainage path) 1977).
484 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

clays is that, in some cases involving high manner during the early stages of construc-
compressibility clays and a long drainage tion. Also, for other reasons such as sample
path, the pore pressures continue to increase disturbance, different stress paths, strain rate
after the end of construction (Crooks et al. effects, influence of the intermediate princi-
1984; Kabbaj et al. 1988; Leroueil 1996; pal stress, etc., the undrained shear strengths,
Hinchberger & Rowe 1998). s u> measured in undrained laboratory or in
situ tests may not be directly applicable un-
der an embankment and are used on an em-
16.7 Stability pirical or semi-empirical basis. Several ap-
proaches have been proposed and will be
16.7.1 GENERAL REMARKS
described below. To gain confidence with the
Ladd (1991) defined three types of stability
results of stability analyses, it is recom-
analysis: (a) a total stress analysis (TSA); (b)
mended that at least two of these approaches
an undrained strength analysis (USA), and (c)
be considered in practical applications. Since
an effective stress analysis (ESA). A total
vane shear strength and preconsolidation
stress analysis is often used in single-stage
pressure profiles are available in most proj-
construction analysis and is usually based on
ects, the corresponding approaches can be
the undrained strength profile prior to con-
used, but other approaches may also be used,
struction. In undrained strength analysis, the
based on the regional experience.
in situ undrained shear strength is computed
as a function of the pre-shear effective stress.
This type of analysis is often used in evaluat- l. Field vane test approach. Back-analyses
ing the stability of stage-constructed embank- of numerous failures have shown that the
ments. vane shear strength often overestimates the
Several authors have examined effective average strength mobilized under embank-
stress analyses (Tavenas et al. 1980; Ladd ments. Bjerrum (1972) suggested that a cor-
1991) and noted several inherent difficulties: rection factor, /-1, which decreases from 1.0 to
(a) the evaluation of pore pressures and thus 0.6 when the plasticity index increases from
of the effective stress normal to the failure 20 to 100 (Fig. 3.15), be used to adjust vane
surface; (b) the assumption that, at any time, shear strength for design purposes. Similarly,
the overall factor of safety is equal to the local correction factors as a function of the liquid
factor of safety at different points along the limit were proposed by Pilot (1972) and
potential failure surface; (c) the choice of Helenelund (1977).
representative effective stress strength pa-
2. Preconsolidation pressure approach.
rameters; and (d) the definition of the factor
Mesri (1975) suggested that the ratio, sJcr;,
of safety.
between the strength available at failure and
Numerical methods that can be used for
the preconsolidation pressure of the clay is
stability analyses are described in Section
essentially constant and equal to 0.22. Lars-
14.5.
son (1980) and Jardine & Hight (1987)
computed ratios between the strength mobi-
16.7.2 EMBANKMENTS CONSTRUCTED
lized at failure and the preconsolidation pres-
IN ONE STAGE
sure, generally obtained in conventional 24-
16.7.2.1 Evaluation of Mobilized Shear hour oedometer tests. It was generally found
Strength, Su that the ratio sJcr; slightly increases from
As indicated in Section 16.1, clay foundations about 0.19 at a plasticity index of ten to about
generally do not behave in an undrained 0.28 at a plasticity index of 80 (see Fig.
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 485

in which m is a parameter approximately


0.4 equal to 0.8. By consolidating the specimens

.- ·t' ... ---' ....


~. beyond their preconsolidation pressure, the
0.3
~ SHANSEP approach neglects the influence
o. I~ ~
- -- •
-4~
of structure on the strength of clays. For
.,.- •
~
:; 0.2
CI) this reason, Ladd (1991) specifies that
0.1
1.After Larsson, 1980 and Jardine & Hight, 1987~ "SHANSEP is strictly applicable only to me-
• After Trak et al., 1980 chanically overconsolidated and truly nor-
mally consolidated soils exhibiting normal-
oo 20 40 60 80 100 ized behavior." According to Leroueil &
Plasticity index, Ip (%)
Jamiolkowski (1991), this approach could be
conservative.
FIGURE 16.16. Strength ratios deduced by differ-
ent approaches: (e) after Larsson 1980; Jardine & 4. Direct simple shear (DSS) test approach.
Hight 1987; • after Trak et al. 1980. "For preliminary design and for final design
ofless important projects involving 'ordinary'
16.16). As noted by Leroueil et al. (1985b) soils with low to moderate anisotropy," Ladd
and Ladd (1991), the upper values often cor- (1991) recommends the use of CKoU direct
respond to organic clays. simple shear test results for stability analyses.
3. Recompression and SHANSEP ap- He observed average strength ratios, sj(J~c of
proaches. Ladd (1969) proposed a methodol- 0.225 for 16 non -varved normally consolidated
ogy in which the undrained shear strength is clays (i.e. very close to 0.22 reported by Mesri
obtained from a laboratory program of C Ko U 1975) and of 0.26 for nine normally consoli-
shear tests on samples consolidated under Ko dated silts and organic soils.
conditions (see Sections 2.6.5 and 3.4) and
5. Unconfined (UC) and unconsolidated
sheared under undrained conditions (called
undrained (UUC) compression test ap-
CKoU tests) in different modes of failure:
proach. Due to swelling of the specimens fol-
compression under the central part of the
lowing sampling and the influence of small
embankment, direct simple shear where
defects in the specimens, the test results are
failure is close to horizontal, and extension
often smaller and more scattered than those
near the toe. The Norwegian Geotechnical
obtained from tests where the sample has
Institute (Bjerrum 1973) uses the "recom-
been reconsolidated. As a consequence, this
pression technique" in which the specimens
approach is likely to underestimate stability
are reconsolidated to in situ stresses prior to
(other things being equal) and is generally
shearing. Ladd & Foott (1974) proposed
not recommended when there are significant
the SHANSEP technique in which the speci-
economic consequences to underestimating
mens are first consolidated well beyond
the stability. However, this approach has been
their preconsolidation pressure, and main-
successfully and extensively used in some
tained in these conditions or rebounded to
countries (especially Japan, see Nagase 1967).
various OCRs before being subjected to un-
drained shear tests. Ladd et al. (1977) 6. USALS approach. Trak et al. (1980) in-
showed that the results can generally be writ- dicated that the large strain strength mea-
ten as: sured in triaxial compression tests, called
USALS, may be representative of the
( s~)
ave DC
= (s~)ave nc
OCRm (16.33) strength mobilized at failure under embank-
ments for some clays from eastern Canada.
486 III. SLo.PE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

As shown in Fig. 16.16, the USALS values


reported by the authors are close to the field Suggestions for the de-
J:: termination of sue:
values reported by Larsson (1980) and Jar- ~..:::;.:...:,r~~_.-....::.....' Tavenas and Leroueil, 1980
dine & Hight (1987). suo - sum
suv measured sue Sum + = 3
7. CPT tip resistance approach. The cone with vane
Su to be used in • Lefebvre et al., 1987
penetration tip resistance, q c (or the value
corrected for pore pressure behind the tip,
=(I)Q. the calculations
Sue = sum + 0.25
2
r, H
Cl
qT), is related to the undrained shear strength
of the clay (see Section 4.4.6), and has been
used to evaluate design strength (Lunne et al. FIGURE 16.17. Correction to obtain operational
strength of a crust beneath an embankment.
1976; Hanzawa 1989). The advantage of this
test is that it provides a continuous strength
profile, which may allow detection of weak tion is often between 1.3 and 1.5, with the
layers that might be missed with other ap- actual target value depending on the project
proaches. Because it has not been calibrated and the consequences of a failure.
yet against failure, the CPT tip resistance is The stability is generally evaluated using
invariably used together with other measure- limit equilibrium analysis incorporated into
ments when establishing the design strength. computer programs (see Section 14.5). How-
ever, charts corresponding to Simplified con-
8. Effective stress analysis approach. Back- ditions (e.g. full-fill strength, no tension
analyses performed in effective stress, on the cracks) can be used for preliminary design.
basis of pore pressures measured at failure in- Figure 16.18 presents charts for the case of
dicate that the mobilized strength parameters
are close to those of the normally consolidated
clay (Parry 1968; Rivard & Lu 1978; Hanzawa
3.0 SIOO8: 2H11 V
et al. 1982; Pilot et al. 1982). However, due to 2.8
2.6
the difficulty of predicting pore pressures in 2.4
the entire foundation of the embankment, the

--
N-O.4 - -
2.2
effective stress analysis is not commonly used 2.0
in simple limit equilibrium analyses.
=
N 0.3 ""'--
1.8
1.6
16.7.2.2 Strength of the Weathered Crust N=0.2
1.4
Because it is generally weathered and fis- ...........
sured, the mobilized strength of the surface 1.1.. 1 .2
""- ~

crust of clay deposits is smaller than that


measured in vane shear tests (Lo 1970). Cor-
rections have thus been proposed by a number
1.0
0.9

0.8
N=0.1

\
I~ ;:-:<H
IP= 35°
N=.!JL

of investigators. For example, Tavenas and =constant


Leroueil (1980) and Lefebvreetal. (1987) sug- 0.7 "'r-...
---
D Su

"-..... *
gested that the operational strength in the 0.6
crust is given as shown in Fig. 16.17.
0.5 1.0 1.5
16.7.2.3 Estimation of Stability DIH
The geometry of an embankment is chosen
on a case-by-case basis. The desired calcu- FIGURE 16.18. Chart for calculation of factor of
lated factor of safety at the end of construc- safety (modified from Pilot & Moreau 1973).
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 487

an embankment on a clay deposit of constant Leroueil et ai. (198Sb) proposed an estima-


undrained shear strength. tion of the shear strength on the basis of the
vertical effective stress, using a strength ratio
16.7.3 CHOICE OF STRENGTH FOR suja~c = 0.25. Such a value would take into
EMBANKMENTS CONSTRUCTED IN account the fact that the largest strength in-
SEVERAL STAGES crease develops in compression, under the
Although there are many reported failures of center of the embankment.
single-stage embankments, which have al-
lowed the calibration of the stability methods 4. SHANSEP approach. In staged con-
mentioned in Section 16.7.2, there is a pau- struction, where the soil has become nor-
city of well-documented failures of stage- mally consolidated along most of the failure
constructed embankments. Nevertheless, sev- surface, the recompression and SHANSEP
eral approaches have been proposed that are techniques are similar. They also seem more
based on the observation that, for stresses ex- appropriate since the main criticisms formu-
ceeding the preconsolidation pressure, the un- lated for single-stage constructions (the fact
drained shear strength increases with decreas- that the effects of structure and partial drain-
ing void ratio and increasing effective stress. age are not considered) are not relevant to
the second and subsequent stages.
1. Field vane test approach. The vane
shear strength measured under embank-
ments has often been used to evaluate the 16.8 Solutions to Problems
stability of stage-constructed embankments. of Stability and Settlement
However, observing ratios suja~c under em-
bankments between 0.18 and 0.25, and 16.8.1 GENERAL
smaller than the ratio s uja~ of the intact clay, Whenever pOSSible, embankments are con-
Tavenas et ai. (1978) suggested that the Bjer- structed with conventional fill materials, in
rum's correction considered in single-stage one single stage, on the natural soil deposit.
construction (Section 16.7.2.1) should not be However, to obtain satisfactory factor of
applied to the vane shear strength measured safety, to have acceptable post-construction
under embankments for stability analyses of settlements, and to satisfy construction con-
stage-constructed embankments. Law (1985) straints such as limited site area or maximum
came to the same conclusion. They explained duration of construction, special solutions
the decrease in suja; by a loss of clay struc- may be required. Table 15.1 summarizes the
ture when it becomes normally consolidated. solutions most commonly considered and
Chapter 15 discusses many of these solutions.
2. CPT tip resistance approach. There is These solutions can be classified into five dif-
evidence (e.g. Leroueil et ai. 1995) that the ferent groups:
cone penetration test can be used for the
evaluation of the strength increase under 1. Eliminate the problem by replacing the
stage-constructed embankments in terms of soils of unfavorable characteristics or by
the change in qT - a v • Research is in progress using lightweight fill materials (see Sec-
to establish the validity of this approach for tion 15.2.1 and Rowe & Soderman 1986).
more general application. 2. Improve the stability of the embankment
by using lateral berms (Section 16.8.2) or
3. Vertical effective stress approach. Using reinforcement (Section 16.8.3 and 16.8.4),
an analogy with the suja; ratio suggested in and/or by building the embankment in
Mesri (1975) for single-stage construction, several stages (Section 16.7.3).
488 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

3. Reduce the post-construction settlements would be possible under Single-stage loading


by using a temporary surcharge (Section for the same soil profile without the use of rein-
15.2.3.2). forcement. When one cannot achieve the de-
4. Accelerate consolidation with vertical sired grade with an adequate factor of safety
drains or drainage trenches (Sections by single-stage loading without reinforcement,
15.2.3.4 and 15.2.3.5).
the following steps should be followed.
5. Improve the soft soil foundation with lime
columns, electro-osmosis (Section 15.2.4), • Step 1. Establish whether it is possible to
preloading or compaction (Section construct a reinforced embankment to the
15.2.5), etc. desired fill thickness. This involves the ap-
plication of bearing capacity theory as de-
In some cases it may be necessary or desir- scribed in Section 16.8.3.2. If the desired
able to combine solutions from a number of thickness can be achieved, proceed to Step
groups. The following will focus on the com- 2 below. If it cannot be achieved using
mon solutions not discussed in detail in reinforcement alone then an alternative
Chapter 15. method of addressing stability is required,
e.g. use of stage construction in conjunc-
16.8.2 BERMS tion with provision of vertical drains, or the
The stability of embankments can be im- use of lightweight fill. The stability calcula-
proved by either flattening the side slopes (in tion already performed in Step 1 indicates
some cases to as much as 10 H: IV; see Rowe the maximum fill thickness that can be
1982) or by the use of berms. The dimensions placed in the first stage of construction. Al-
of the berms are usually established based on ternatively, this stability calculation may be
stability considerations (see Section 16.7), used to assess the thickness/density of
lightweight fill that would be required.
but the height of a berm is typically 40-50%
• Step 2. Establish the reinforcement force
of the height ofthe embankment. Due to com- required to achieve the desired fill thick-
mon limitations on the space available for em- ness at the desired factor of safety using
bankment construction (either for economic limit equilibrium methods as discussed in
or environmental reasons) berms are not as Section 16.8.3.3.
widely used these days as in the past, since the • Step 3. Check the potential for sliding of
same effect often can be achieved by the use the embankment on top of the upper layer
of alternative techniques (e.g. basal reinforce- of reinforcement (embankment splitting)
ment). For example, Rowe and Soderman due to the earth pressure that will develop
(1985b) illustrate that either the use of basal in the fill. Referring to Fig. 16.19, this can
reinforcement or lightweight fill can have a be achieved by considering horizontal
equilibrium and checking the factor of
similar effect to berms in terms of improving
safety from:
the performance of embankments on peat.
F = O.5nYrH2 tan <Pd./
16.8.3 REINFORCED EMBANKMENTS
ON SOFT CLAY [Ka(O.5YrH2 + qH)] (16.34)

16.8.3.1 General Considerations where K. = the Rankine active earth pres-


As discussed in Section 16.2, under appro- sure coefficient corresponding to the fric-
priate conditions, the provision of basal rein- tion angle of the fill (see Section 17.2.2),
forcement, i.e. reinforcement located at or <l>ds = the soil-reinforcement friction angle
near the fill-clay interface, can substantially in- in direct shear (see Sections 7.2.2 and
crease embankment stability and hence allow 7.3.2), q = the surcharge (if any) acting on
the construction of a higher embankment than the embankment, and all other terms are as
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 489

Rowe 1991; Rowe 1997a). The reasons for


q
this include: (a) most soils would have been
slightly overconsolidated before loading (see
Section 16.1); (b) there is usually a significant
increase in undrained strength with depth in
these deposits and the effect is not correctly
considered by using an "average strength";
and (c) often the very soft layer is underlain
FIGURE 16.19. Direct sliding failure of the fill or by a firm stratum at relatively shallow depth,
reinforcement. i.e. depths less than 0.7 times the width of the
embankment, and the presence of this firm
defined in Fig. 16.19 or previously defined.
stratum can Significantly increase the bearing
While this mechanism should be checked,
capacity. This has been recognized by a num-
for typical values of embankment side
ber of investigators (Humphrey & Holtz
slope, n, with n ~ 1.5 (see Fig. 16.19), it
1986; Rowe & Soderman 1987; Houlsby &
rarely controls the design.
Jewell 1988). Rowe and Soderman (1987)
In addition to the foregoing, consideration synthesized the bearing capacity factors of
should also be given to long-term settlement Davis & Booker (1973) and Matar & Salen-
and rate of consolidation in the same manner ~on (1977) and proposed a design methodol-
to that discussed in Sections 16.4.3 and 16.6, ogy, involving only hand calculations, that can
respectively. be used to allow for the effects of both
In the preceding and following discussions strength increase with depth and the pres-
it is assumed that the fill is a free-draining ence of a stifflayer at a depth ofless than 0.7
frictional material. Lower quality fills, e.g. co- times the embankment width.
hesive fills, have been used and the analysis Since an embankment will generally be
presented herein can be readily extended to trapezoidal in shape and the plasticity solu-
deal with a cohesive fill, however, these fills tions are for a rigid footing, an approximation
introduce a number of additional problems must be made to obtain the equivalent width
such as tension cracking in the fill, Significant of the embankment b<>. Since the vertical
pore pressure changes due to rainfall events, pressure at the edge of a rigid footing at the
potentially low soil-reinforcement interface pOint of collapse is (2 + n)suo (where suo is
strength for some types of reinforcement, the undrained shear strength directly be-
creep of the fill, etc., which are beyond the neath the footing), the equivalent width of a
scope of this handbook. rigidly reinforced embankment, b<>, can be
taken to be the distance between the points
16.8.3.2 Bearing Capacity Limits on Reinforced on either side of the embankment when the
Embankment Height applied pressure, 'Yfh<>, is equal to (2 + n)suo
Historically, bearing capacity calculations as shown in Fig. 16.20. Thus:
have been based on the Prandtl (1920) solu-
tion (qu = 5.14su), which was developed for h" = B + 2n(H - h")
(16.35)
strip loading on a deep homogeneous layer. = B + 2n[H - (2 + n)suo/'yrJ
However, it has been found in numerous
cases that reinforced embankments on very The bearing capacity, q u> of the equivalent
soft to soft clays may be stable at heights up rigid footing of width b<> is given by:
to 2 m above the theoretical collapse height
based on Prandtl's theory (Humphrey & (16.36)
490

,.. - - - - -•.....j·~1 •
~14n(H-h )~14nh~1
Embankment

! Unit ~~i~ht T
I 1'f

FIGURE 16.20. Definition of variables used to estimate collapse height for a perfectly reinforced em-
bankment (modified from Rowe & Soderman 1987).

35~--------------------------------------------~

30

25

20
I qu = suoNc + qs I
15
For 50 < ~~~* S 100 and biD < 10
10
Nc::=: 11.3 +0.384 Pc D/suo

5.14 - 5
3

5 10 15 20 25 30 35 40 45 50
AD
c
suo

FIGURE 16.21. Bearing capacity factor for non-homogeneous soil (modified from Rowe & Soderman
1987).
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 491

where q s is a uniform surcharge pressure ap- Thus, distributing the applied pressure
plied to the foundation soil surface outside of due to the triangular distribution over a dis-
the footing width. The bearing capacity factor tance x gives:
Nc is obtained from Fig. 16.2l. Inspection of
Fig. 16.20 shows that the triangular edge of qs = 0.5ny~h")2Ix for x > nh" (16.37a)
the embankment (beyond the equivalent
rigid footing) provides a surcharge that would and
increase stability, and hence an estimate of q s
in terms of the pressure applied by this trian- qs = 0.5(2 nh" - x)Yrh"/(nh")
gular distribution is required. Figure 16.22 for x ::5 nh". (16.37b)
shows the depth, d, to which the failure
mechanism is expected to extend. The lateral The bearing capacity of an equivalent rigid
extent of the plastic region involved in the footing can then be calculated from Eqs
collapse of a rigid footing extends a distance 16.36-16.37, and this may be compared with
x from the footing, where x is approximately the average applied pressure, q., due to the
equal to the minimum of: d as obtained from embankment over the width b":
Fig. 16.22 and the actual thickness of the de-
posit, D. qa = Yr[BH + n[H2 - (h")2]]1b". (16.38)

I. 0 .1
I

d
tr
0.5

o
0.1 10 100 1000
Pet>
suo

FIGURE 16.22. Effect of non-homogeneity on depth of the failure zone beneath a rough rigid footing
(modified from Matar & Salen90n 1977).
492 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

For the purposes of estimating the maximum some form of limit equilibrium analysis (see
possible factor of safety (defined here as F Section 14.5).
= quiqa) that could be achieved for a given Most commonly, this involves some form
embankment geometry and soil profile, q u of slip circle analYSiS, wherein for any trial slip
and q a can be calculated directly from Eqs circle an additional restoring moment is in-
16,35-16,38. cluded in the calculation of moment equilib-
The maximum possible factor of safety, F, rium and this restoring moment is equal to
calculated above was obtained without con- the force in the reinforcement, T, multiplied
sidering the amount or properties of the by a lever arm between the center of the trial
reinforcement and implicitly assumes that circle and the point of action of the tensile
the reinforcement has sufficient strength and force in the reinforcement. There has been
stiffness to develop the required reinforce- considerable debate regarding the appro-
ment forces without breaking or excessive de- priate selection of the lever arm, with esti-
formation. Having established that it is possi- mates ranging from the radius of the circle,
ble to build the embankment to the desired R (see Fig. 16.23, this assumes the reinforce-
factor of safety, the next step (to be discussed ment deforms to a tangent to the slip circle
in Section 16.8.3.3) is to select reinforcement at the point of interaction), as a maximum, to
that has sufficient strength and stiffness to the vertical distance from the center of the
achieve the desired factor of safety. circle to a horizontal line drawn through the
reinforcement, YT, (see Fig. 16.23, this as-
16.8.3.3 Selecting Reinforcement Properties sumes that the tensile force in the reinforce-
The reinforcement needs to be selected such ment acts in the horizontal direction). Using
that it develops the force required to main- limit analysis, Michalowski (1998) has theo-
tain equilibrium with the desired factor of retically established that the reinforcement
safety at a strain that will not permit excessive should be considered to be horizontal. Fur-
undrained shear deformation (see Section thermore, back-analysis of a number of cases
16.2). There are a number of different ap- by Palmeira et ai. (1998) indicates that this
proaches, but they are generally based on approach also gives a good prediction of the

r&W ,I ' - . - 1_______

8
H
Sum Sue suo

b
o
Soft Soil

//~
Firm Stratum
z

FIGURE 16.23. Rotational failure of embankments on soft soil.


16. EMBANKMENTS OVER SOFT SOIL AND PEAT 493

factor of safety for embankments that have expected and adopt a friction angle for the
failed. fill equal to the critical state value, <1>~v (in
There are two approaches to establishing recognition that it is difficult to gain good
the required reinforcement force, T, to main- compaction of fill on very soft foundations
tain stability at a given factor of safety. The that require basal reinforcement, espe-
ciallynear the bottom of the embankment).
first uses the expected strength profile in the
3. Assume a trial slip circle as shown in Fig.
soil (see Section 16.7.2) and the expected fill 16.23, such that a vertical line from the
properties, assumes trial value of Tr, and per- point of intersection of the circle and the
forms a series of limit equilibrium analyses reinforcement passes through the crest of
to establish the minimum factor of safety, F, the embankment (note the slip circle is
associated with that value of T t . By trial and not carried through the fill; rather, the fill
error, one can adjust the value of T t until the is modeled as described in 4. below).
minimum value of F corresponds to the de- 4. Separate the consideration of the over-
sired factor of safety, F d, (typically 1.3-1.5) turning moments of the fill from that
thereby yielding the required force, T req . of the foundation (as recommended by
However, Treq must not exceed the maximum Jewell 1996) by replacing the fill by a hori-
force, T], that can be developed in the rein- zontal thrust, P r, given by Eq. 16.40, with
forcement and which is given by the sum of Ka evaluated for <1>' = <1>~v acting at the
centroid of the horizontal pressure (lever
the horizontal thrust in the fill, P r, and the arm Yp) and a vertical force equal to the
resisting shear force that can be transferred weight of the fill to the left (in Fig. 16.23)
to the soil, P s , e.g. see Fig. 16.23, thus: of the point of intersection between a
given trial slip circle and the reinforce-
Tl = P r + Ps (16.39)
ment, acting at the centroid of this block
of fill (lever arm Xw).
where
5. Consider a trial slip circle with a restoring
moment, m s , due to the shear strength

r
(16.40)
along the surface in the clay given by:
and the earth pressure coefficient, Ka, is
based on the factored mobilized fill friction ms = R2 s~de. (16.42)
angle, <1>£ = tan-1[(tan <1>')/F d J, and
6. For a given trial slip circle, evaluate the
(16.41)
force required to maintain equilibrium
where Sue = the operational shear strength at from:
the soil interface, ex = a reduction factor to
T = (WX w + PrYp - m,)/Y T . (16.43)
reflect any reduction in shear strength that
may occur at the reinforcement-foundation 7. Check that the force calculated in step 6
interface, andXT = the distance from the toe of can be mobilized, namely T ::::; T], with Tl
the embankment to the point where the circle given in Eq. 16.39. (Note: If T > T] then
intersects the reinforcement (see Fig. 16.23). the desired factor of safety cannot be
Referring to Fig. 16.23, an alternative ap- achieved irrespective of how much rein-
proach, and the one recommended here is to: forcement is provided and additional
means of soil improvement will be re-
1. Obtain design shear strength parameters quired, see Section 16.8.1).
for the foundation, S ~, by dividing by the 8. Repeat steps 3-7 to find the maximum
desired factor of safety, F d , (s~ = SulFd)' value of T and this becomes the required
2. Take the fill unit weight as the maximum force, T req .
494 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

Once the desired design force, T req , has been ommendations regarding what maximum
established (by either method described strain in reinforcement should be allowed for
above) it remains to select the reinforcement a range of different soil conditions.
with the required strength and stiffness. With It has also been found that for soft plastic
respect to the strength, the reinforcement is cohesive soils where there is a significant in-
required to have an allowable strength, Ta, or crease in strength with depth, i.e. a rate of
design strength, T D, larger than T req . increase in undrained strength with depth Pc
Rowe & Mylleville (1993) discuss the use 2::: 1 kPa m -1, a maximum allowable strain of
of steel strip reinforcement and the selection 5% will often provide a conservative estimate
of T D for steel. If geosynthetic reinforcement of the failure height of the embankment.
is used, then T D is selected to support the re- However, it should be emphasized that spe-
quired load fully until consolidation is 90% cial care is required in designing embank-
complete (see Section 16.6). Two approaches ments on soils that are susceptible to strain
are commonly adopted for establishing the softening at the anticipated rate of straining
allowable and design strength, T a or T D. The associated with the proposed construction
simplest and most commonly used method in schedule and that the allowable strain in
North America is to calculate the allowable these cases may be substantially less than 5%.
strength, Ta, by reducing the short-term ulti- The reader is referred to Mylleville & Rowe
mate strength by a number of reduction fac- (1988, 1991) and Rowe & Mylleville (1989,
tors as discussed in Section 7.9.1.1. The alter- 1990) for a more detailed discussion of the
native limit state approach (mostly used in relationship between strain in the reinforce-
Europe) is to establish the load leading to ment and that in the underlying soil.
rupture or limiting strain (based on the geo- Having established a reasonable allowable
synthetic properties) at the end of the design strain, Ea, (e.g. if it is 5%, Ea = 0.05) then
life and then use these partial factors as dis- required tensile stiffness of the reinforce-
cussed in Section 7.9.1.2 to obtain the design ment is given by:
strength, T D' Both approaches have proven
successful in the past. ] req =~ (16.44)
lOa
The failure height and required reinforce-
ment calculated as described above using where Treq and Ea are obtained as described
limit equilibrium, e.g. based on a slip circle above.
failure mechanism, relate to the ultimate The foregoing has dealt with a single layer
limit state (collapse) and provide no indica- of reinforcement. The single layer may be re-
tion regarding the serviceability limit state placed by a few layers provided that they are
(deformations prior to or at failure). In some all located relatively close to the bottom of
instances, the embankment may have failed the embankment with a layer of granular soil
due to excessive displacements (serviceability between each layer. Thus if there are m layers
limit state) prior to reaching the collapse (provided that m :::; 3 in these applications)
height (ultimate limit state). In some in- each with allowable strength, T., or design
stances, excessively large displacements may strength, T D, one requires that:
be required to cause failure of reinforced em-
bankments. This has been recognized by (16.45)
Jewell (1982), Rowe & Soderman (1985a),
and
Bonaparte & Christopher (1987), Mylle-
ville & Rowe (1988, 1991) and Rowe & Myl- ] req = Treg
. (16.46)
leville (1989,1990), who have each made rec- mEa
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 495

16.8.4 REINFORCED EMBANKMENTS cluding a layer of suitable reinforcement at


ON PEAT or near the interface between the peat and
16.8.4.1 Problem Definition the fill. Figure 16.24 gives design curves re-
Due to the pore pressure dissipation that oc- lating the design height, h (above original
curs during construction (see Section 16.3), ground level; h = H - S., as noted above)
there is no simple limit equilibrium method to the reinforcement tensile stiffness, J, for
that can be used to assess the stability of em- embankments on peat deposits 3, 5, and 8 m
bankments constructed on peat (see Section thick underlain by a firm base; interpolation
16.3.1 for qualifications regarding what is can be used for intermediate thicknesses.
meant by the term "peat"). Referring to Fig. 16.24, for the conditions
Rowe & Soderman (1985b, 1986) have re- examined a desired end of construction
ported the results of a series of finite element height of 1.5 m above original ground level
analyses that examine the effect of geosyn- and Yf = 20 kN m -3, no reinforcement would
thetic reinforcement on the stability and de- be required if the peat deposit was 3 m deep,
formation of embankments on peat. The but reinforcement with a tensile stiffness] >
method of analysis has been validated against 800 kN m -1 would be required for a peat de-
case histories (e.g. Rowe et al. 1984b; posit 8 m deep. A major reason for the differ-
Rowe & Mylleville 1996). ence is the fact that, other things being equal,
The peat parameters (c' = 1.8 kPa, <1>' = for an 8 m thick deposit the settlement during
27°, v' = 0.15, Ko = 0.18, eo = 9) were se- construction will be between two-three
lected based on a review of the literature. It times that for a 3 m deposit and the reinforce-
would be considered unusual for a fibrous ment is required to carry the load due to the
peat to have a combination of parameters sig- additional fill below the ground surface that
nificantly worse than those used, although would be present for the 8 m deposit. Settle-
this possibility cannot be excluded. The cases ment can be estimated as described in Sec-
examined involved low granular embank- tion 16.4.5.
ments with heights above original ground
surface of 2.5 m or less, a crest width of 13.4
m and 2 : 1 side slopes resting on peat deposits 50r-----.---~_r-----.~--~

with depths of 3, 5 and 8 m. The granular fill


was assumed to have c' = 0, <1>' = 32°, Y = 40
20 kN m -3. The end of construction pore
pressure distribution was assumed to be as
sho~ in Fig. 16.9 for a maximum pore pres-
'Yfh
sure B m = 0.34. The solution presented in (kPa)
the following takes into account the fact that 20
some settlement, S., does occur during con- 8m ' 0.34
struction and is represented in terms of the 10 h ' Height of Embankment Above
end of construction fill height, h, above the Original Ground Level (h= H-S r )
'Yf = Unit Weight of Fi II
original ground level (i.e. where h = H-S.,
and H = the fill thickness; see Section o 500 1500 2000
16.4.5). Geosynthetic Tensile Stiffness J (kN/m)

16.8.4.2 Peat Underlain by a Firm Base FIGURE 16.24. Design chart for peat underlain by
The shear deformations and stability of em- a firm base; see text for assumptions and limita-
bankments on peat can be improved by in- tions (modified from Rowe & Soderman 1985b).
496 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

16.8.4.3 Peat Underlain by a Soft Clay/Marl the embankment and a block-like slide mov-
Layer ing on or just below the boundary between
Peat deposits are often encountered in re- the peat and the soft soil.
gions that have been subjected to "recent" Figure 16.25 shows an embankment on a
glaciation. Frequently, the depositional his- peat layer underlain by very soft clay. The ac-
tory of these deposits involves the sedimenta- tive earth pressure in the fill creates a lateral
tion of clay, silts or lake marl followed by the thrust. The applied pressure due to the em-
formation of a peat deposit that will usually bankment resting on the peat also induces a
be covered by some form of living vegetation. lateral thrust in the peat. This lateral thrust
Because the peat has low unit weight, the must be resisted by the passive resistance that
clay, silt or marl in a recent normally consoli- can be developed in the peat, the tensile
dated deposit will be very weak just below the force in the reinforcement and the shear
peat, although the increase in effective stress strength that can be mobilized at (or near)
with increasing depth will generally result in the peat/very soft clay interface. Generally,
a significant increase in strength with depth due to the low effective stress in the peat out-
below the peat. For the cases reported in the side the embankment, the passive resistance
literature, the vane shear strength in a very is negligible (unless berms are used; see Sec-
soft stratum below the peat is typically in the tion 16.8.2). Thus the reinforcement and the
range 5-15 kPa, although shear strengths as shear strength at the peat-very soft clay in-
low as 2 kPa have been reported. terface must resist the lateral force. Initially,
It has generally been observed (e.g. Mac- the contribution of the very soft clay is small
Farlane & Rutka 1959; Lea & Brawner 1963; and hence even with very high strength rein-
Weber 1969; Raymond 1969) that construc- forcement there is a severe limit to the em-
tion and maintenance problems and shear bankment height that can be achieved in
failure are far more likely to occur when the single-stage construction. However, the rein-
peat is underlain by a weak soil than when it forcement (either geosynthetic or natural
rests on a firm stratum. Collapse in these root mat or both) does allow some fill to be
cases usually involves formation of a crack in placed (while maintaining stability) and with

Active Earth Pressure


in Fill due to Fill
H
Treq = Force Developed
in Reinforcement

Positve Earth Active Earth Pressure


Pressure in Peat in Peat (mostly due
due to Peat to vertical stress
self weight imposed by the fill)

~su~~
Shear Strength at (or near)
Very Soft Clay Peat I Very Soft Clay Interface

FIGURE 16.25. Potential failure mechanism due to lateral thrust and sliding on a deeper low strength
layer, e.g. very soft clay beneath a peat layer (modified from Rowe 1997).
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 497

time the consequent increase in effective friction angle <\>' == <\>~ for the fill and <\>' == tan- 1
stress in the very soft soil will give rise to an [(tan <\>~eat)/F], respectively; <\>~eat = the fric-
increase in strength and hence more fill can tion angle of the peat; and if = Yf - Yw'
be placed in a second stage of construction. In order to evaluate the required force, T req
To assess stability, one can perform slip cir- (Eq. 16.47), it is necessary to estimate the set-
cle calculations to identifY the critical slip cir- tlement, S" that will occur by the end of con-
cle through the peat and/or through the peat struction. This may be evaluated as discussed
and clay, but to do so one must make some in Section 16.4.5. Once the Treq has been es-
assumptions about the distribution of excess tablished, the required strength and tensile
pore pressure both vertically, e.g. based on stiffness can be established using Eqs 16.44-
Eq. 16.1 and horizontally. This approach may 16.46 as discussed in Section 16.8.3.3 with a
not always capture the worst case and consid- typical allowable strain for peat, Ca, of 5-8%
eration should also be given to non-circular (provided that these strains do not exceed the
failure mechanisms such as that shown in Fig. allowable strain in reinforcement).
16.25. In this case, the fill is modeled as earth Equation 16.47 represents a simple limit
pressures (as discussed in Section 16.8.3.3) equilibrium approach that can be adopted in
and the objective is to deduce the reinforce- assessing reinforcement requirements. A
ment force required to achieve the desired more sophisticated alternative approach is to
factor of safety (as discussed in Section perform finite element analyses. For exam-
16.8.3.3). The required reinforcement force, ple, using the parameters discussed in Sec-
T req , can be calculated by assuming that at the tion 16.8.4.1, Rowe & Soderman (1986) ana-
end of construction: (a) there will be a settle- lyzed a number of cases where peat was
ment, S" of the fill into the peat so the total underlain by soft clay. Based on these analy-
thickness of fill material will be H = h + S r ses and those already discussed for peat un-
and peat thickness will be D' = D - Sr; (b) derlain by a firm base they deduced approxi-
the water table remains constant; (c) the ex- mate values of the reinforcement force, T req ,
cess pore pressures are as given by Eq. 16.1; required to maintain stability with a factor of
(d) the submerged unit weight of the peat is safety of unity as indicated in Table 16.2. The
negligible; and (e) the apparent cohesion of values can be approximately used by finding
the peat can be conservatively neglected. a result in the table corresponding to the fac-
One can write an approximate equation for tored strength s~o = suo/F, where Suo = the
horizontal equilibrium: expected undrained shear strength of the clay
near the peat-clay interface. Intermediate
Treq = 0.5Yfh2Kaf + KafYfhSr values of peat thickness and clay strength can
+ O.5Ka((rS~ + (Yfh + irSr) be estimated by interpolation.
(16.47)
X [ Kap ~
+ B mD'(l - Kap)]
16.9 Practical Considerations:
Construction, Instrumentation
where in addition to the terms defined in Fig. and Observation Analysis
16.25 and above: 16.9.1 DESIGN AND CONSTRUCTION
Ka = tan 2(45 - <1>'/2) (16.48) The upper part of soft clay and peat deposits
(crust and root mat, respectively) is generally
and Kaf and Kap = the values of the active stronger than the underlying soil and should
earth pressure coefficient, K a , evaluated for be left in place to allow the passage of vehi-
498 III. SLOPE, EMBANKMENT AND WALL STABILITY, AND SOIL IMPROVEMENT

TABLE 16.2. Approximate required geosynthetic force,


Treq (kN m- I ) (F = 1) values for embankments on fibrous peat Bm = 0.34
Maximum height of fill above original
Peat Strength of ground level (m)b.c
thickness Underlying underlying strata,
(m) strata s~o (kPa) 1.0 1.5 2.0 2.5
3 Firm NRR NRR 25 60
Clay 15 NRR NRR 25 60
10 NRR NRR 30 65
7.5 NRR 15 40 70
5 NRR 40 80 PF
2.5-5 15 no PF PF
5 Firm NRR 30 50 75
Clay 15 NRR 30 50 75
10 10 60 no PF
7.5 25 100 PF PF
8 Firm 65 75 95 120
'See Rowe and Soderman 1985b, 1986 for limitations.
h NRR, no reinforcing required.
, PF, potential failure for the assumed conditions: do not construct.

cles and protect the underlying soil from dis- to estimate when the following stage could be
turbance. This crust is often covered by a per- constructed and to what level. When there
meable fill layer that, in addition to improving are services, structures or piles close to the
the bearing capacity of the upper layer, facili- embankment, it can be important to monitor
tates the consolidation of the clay or peat the lateral displacements using inclinome-
layer. In cases where the crust is non-exis- ters.
tent, or rather soft or thin, a geosynthetic can
be placed over the natural soil as a separator 16.9.3 OBSERVATION ANALYSIS
and/or reinforcement layer (see Chapter 7).
The monitoring data obtained during and
In order to avoid disturbance of the clay de-
after construction should be used to verify
posit, the lower part of the embankment
how the observed behavior compares with
should not be heavily compacted. When
the parameters used in design, and where ap-
clayey soils are used as fill material, the com-
propriate the calculations can be upgraded.
paction conditions must be selected to avoid
As indicated in Fig. 16.1 and Section
their collapse when wetted and to limit long-
16.1.1, the observation of pore pressures gen-
term creep movements. The use of an appro-
erated during construction under the center-
priate construction procedure is crucial to the
line of an embankment can generally be used
success of the project (see Holtz et al. 1997).
to calculate the vertical yield stress of the clay
at the level of the piezometer. It can also give
16.9.2 INSTRUMENTATION an indication of local shear failure when the
In order to evaluate the degree of consolida- ratio /1u/ /1cr becomes larger than 1.0.
tion and the strength of the clay, it is desir- The strain between two deep settlement
able to instrument the clay foundation with gages can be used in conjunction with pore
piezometers and settlement gages. This is pressure measurements to define an in situ
particularly important in stage construction effective stress-strain curve, that can then be
16. EMBANKMENTS OVER SOFT SOIL AND PEAT 499

I I I
1_ _.<1t al. <it---.1
I I I
I I I
I I I
I I I
I I I
5(1) ------Sl- I I
I I

5
(a,

FIGURE 16.26. Stages in analysis of settlements using Asaoka's (1978) method: (a) settlement curve for
soil layer, and (b) Asaoka's construction.

compared with the compression curve as- 5N (16.49)


sumed for the sublayer considered.
Cv = -12M ln~

When the clay deposit and the consolida-


tion conditions are relatively simple, Asaoka's in which A = the drainage path (A = D,
(1978) method can be used during the con- the thickness of the layer for one-way
solidation process to evaluate the magnitude drainage and A = 0.5D for two-way
drainage).
of the final settlement as well as a representa-
tive coefficient of consolidation. This method 2. For the case where vertical drains are used
involves the following procedure: (a) plot the (Magnan & Deroy 1980):
settlement, S(t), of the compressible layer
with time (Fig. 16.26a); (b) choose a time in- c = - d;F(n) InR (16.50)
v 8At p
terval, 13.t, and read from the curve the settle-
ment, Sj, at times (to + iLlt) for i = 0, 1,2 .... in which de and n =, respectively, the di-
(Fig. 16.26a); (c) plot Sj-1 versus Sj on a spe- ameter of the zone of influence of each
cial diagram and draw the line A (Fig. drain, and the drain spacing ratio (n = del
16.26b); and (d) the point at which Sj-1 = Sj d w , where d w = the equivalent diameter
corresponds to the final settlement, Sr. The of the drains); and F(n) is defined as dis-
slope of the line A can be used to calculate cussed in Section 15.2.3.4.
the average coefficient of consolidation of the
clay, using one of the follOwing equations: The Asaoka method can be used for embank-
ments constructed in one stage or in several
l. For the case of a vertically drained com- stages separated by periods of consolidation
pressible layer (Asaoka 1978): at constant load.

You might also like