You are on page 1of 12

1330

Radial consolidation of clay using compressibility


indices and varying horizontal permeability
Buddhima Indraratna, Cholachat Rujikiatkamjorn, and Iyathurai Sathananthan

Abstract: A system of vertical drains, with surcharge load to accelerate consolidation by shortening the drainage path,
is one of the most popular methods of soft ground improvement. The conventional radial consolidation theory (includ-
ing smear and well resistance) has been commonly used to predict the behaviour of vertical drains in soft clay. Its
mathematical formulation is based on the small strain theory; and for a given stress range, a constant volume com-
pressibility (mv) and a constant coefficient of horizontal permeability (kh) are assumed. However, the value of mv varies
along the consolidation curve over a wide range of applied pressure (∆p). In the same manner, kh also changes with the
void ratio (e). In this paper, the writers have replaced mv with the compressibility indices (Cc and Cr), which define the
slopes of the e – log σ ′ relationship. Moreover, the variation of the horizontal permeability coefficient (kh) with the
void ratio (e) during consolidation is represented by the e – log kh relationship that has a slope of Ck. In contrast to
the conventional analysis, the current study highlights the influence of the Cc /Ck (or Cr /Ck) ratio and the preloading in-
crement ratio (∆p/σi) on the consolidation process. The analytical predictions are compared with the experimental re-
sults when a large-scale consolidation chamber was used, and these predictions show good agreement with the
measured data. Finally, an embankment case history taken from Muar Plains, Malaysia, is analysed on the basis of the
current solution and compared with field measurements.
Key words: compressibility, embankments, permeability, soft soils, soil consolidation, vertical drains.
Résumé : Un système de drains verticaux avec surcharge pour accélérer la consolidation en diminuant la longueur de
cheminement du drainage est une des méthodes les plus populaires d’amélioration des sols mous. La théorie conven-
tionnelle de consolidation radiale (incluant la résistance du puits et des souillures) a été utilisée couramment pour pré-
dire le comportement des drains verticaux dans l’argile molle. Sa formulation mathématique est basée sur la théorie des
petites déformations, et on suppose pour une plage donnée de contraintes, une compressibilité à volume constant (mv)
et un coefficient constant de perméabilité (kh). Cependant, la valeur de mv varie le long de la courbe de consolidation
pour une plage étendue de pression appliquée (∆p). De la même manière, kh change également avec le rapport de vide
(e). Dans cet article, les Auteurs ont remplacé mv par les indices de compressibilité (Cc et Cr) qui définissent les pentes
de la relation e – log σ ′. De plus, la variation du coefficient de perméabilité (kh) avec le rapport de vide (e) durant la
consolidation est représentée par la relation e – log kh qui a une pente Ck. Contrairement à l’analyse conventionnelle,
l’étude courante met en valeur l’influence du rapport Cc /Ck (ou Cr /Ck) et du rapport d’accroissement de préchargement
(∆p/ σ i) sur le processus de consolidation. Les prédictions des analyses sont comparées avec les résultats expérimen-
taux au moyen d’une chambre de consolidation à grande échelle, et ces prédictions montrent une bonne concordance
avec les données des mesures. Finalement, une histoire de cas d’un remblai provenant des Muar Plains, Malaisie, est
analysée sur la base de la présente solution, et comparées avec les mesures sur le terrain.

Mots clés : compressibilité, remblais, perméabilité, sols mous, consolidation du sol, drains verticaux.
[Traduit par la Rédaction] Indraratna et al. 1341

Introduction times be too much to meet the need for rapid construction
(Johnson 1970). With the use of vertical drains, the drainage
Preloading of soft clay over vertical drains is one of the length is considerably shortened from the equivalent of the
most commonly used methods to increase the shear strength thickness of the soft soil layer (e.g., 10–30 m in the vertical
of soil and to reduce its postconstruction settlement. Because direction, depending on the geological characteristics and
most compressible soils are characterized by very low per- the thickness of the compressible layer, as well as the extent
meability and considerable thickness, the time required to of loaded area) to half of the drain spacing in the horizontal
achieve the desired settlement or shear strength can some- direction (typical drain spacing is 1.0–2.0 m (Indraratna et

Received 27 September 2004. Accepted 11 April 2005. Published on the NRC Research Press Web site at http://cgj.nrc.ca on
27 September 2005.
B. Indraratna,1 C. Rujikiatkamjorn, and I. Sathananthan. Faculty of Engineering, University of Wollongong, Wollongong,
NSW 2522, Australia.
1
Corresponding author (e-mail: indra@uow.edu.au).
Can. Geotech. J. 42: 1330–1341 (2005) doi: 10.1139/T05-052 © 2005 NRC Canada
Indraratna et al. 1331

Fig. 1. Vertical drain and its layout: (a) unit cell; (b) triangular grid pattern; and (c) square grid pattern.

al. 1997)). It is important to note that for most soft clay de- tions of settlements and excess pore pressures are compared
posits, the horizontal permeability is higher than the vertical with field measurements at the centerline of embankments,
permeability; hence, the rapid radial drainage accelerates the obtained readily from settlement plates and piezometers.
consolidation process (Jamiolkowski et al. 1983). This sys-
tem has been used successfully to improve foundation soils
for embankments, airports, and highways (Indraratna and Vertical drain theory
Redana 2000; Li and Rowe 2002).
Barron (1948) introduced an analytical solution for radial Barron (1948) presented a comprehensive solution to the
consolidation of soil without the smear effect. Subsequently, problem of radial consolidation by drain wells. He studied
Hansbo (1981) incorporated the smear effect and well resis- two extreme cases: free strain and equal strain. The free
tance into Barron’s formulation. Because a small strain the- strain hypothesis assumes that the load is uniform over a cir-
ory is used in Hansbo’s theory, a constant coefficient of cular zone of influence for each vertical drain and that the
volume compressibility (mv) and a constant coefficient of differential settlements occurring over this zone have no ef-
horizontal permeability (kh) were assumed for a given stress fect on the redistribution of stresses by the arching of the fill
range. In contrast, for a relatively large applied stress range, load. The equal vertical strain hypothesis, in contrast, as-
it is known that both soil permeability and soil volume com- sumes that arching occurs in the upper layer during the con-
pressibility coefficients decrease as a result of physical re- solidation process without any differential settlement in the
duction in void ratio during the consolidation process clay layer. The arching effect implies that a more or less
(Tavenas et al. 1983; Seah et al. 2004). The stress state in re- rigid boundary at the surface of the soil layer is consolidated
lation to preloading (surcharge) and preconsolidation pres- with vertical drains, on the assumption that the vertical
sure is essential for predicting the actual settlement (Casagrande strain is uniform in the horizontal plane of the soil. Never-
1932; Holtz and Kovacs 1981; Burland 1990; Indraratna and theless, for all practical purposes, the average consolidation
Balasubramaniam 1993). obtained in both these cases is nearly the same, and the solu-
In this paper, the e – log σ ′ relationship is used to deter- tion obtained from the second assumption is simpler than in
mine the compressibility indices (Cc and Cr), and the e – the first case (Barron 1948). Therefore, it has been common
log kh relationship is used to represent permeability varia- to use equal vertical strain in most radial drainage consolida-
tion. In contrast, in the conventional radial consolidation tion analyses.
(Barron 1948), the parameters mv and kh were not changed The key assumptions for the conventional equal vertical
as a function of the void ratio for a given stress range. The strain solution (Hansbo 1981) are also applicable to a unit
smear effect is considered, but the well resistance is ne- cell defined by any two adjacent centerlines between drains
glected. It can be noted that for the most common length of and are as follows:
drains (<20 m), the well resistance is not significant (Holtz (i) Soil is fully saturated and homogeneous, and laminar
et al. 1991; Indraratna et al. 1994; Hansbo 1997). The ef- flow thorough the soil (Darcy’s law) is adopted. At the
fects of the compressibility indices, the variation of soil per- outer boundary of the unit cell, flow is not allowed to
meability, and the magnitude of preloading are examined occur (Fig. 1), and for relatively long vertical drains,
through the consolidation process. Subsequently, the writers’ only the radial (horizontal) flow is permitted to occur
model is verified by laboratory testing. Finally, the predic- (i.e., no vertical flow).
© 2005 NRC Canada
1332 Can. Geotech. J. Vol. 42, 2005

(ii) Soil strain is uniform at the upper boundary of the unit Fig. 2. (a) Compression during preloading; and (b) semi-log
cell. The small strain theory is valid. permeability – void ratio relationship.
In this paper, the discharge capacity (qw) of the drain is
assumed to be high enough for well resistance to be ne-
glected. Holtz et al. (1989) suggested that as long as the
working discharge capacity of a prefabricated vertical drain
(PVD) exceeds, say, 150 m3/year after installation, the effect
on consolidation due to well resistance (e.g., folding, in-
creased lateral pressure, or siltation) may not be significant.
Indraratna and Redana (2000) described how well resistance
over the long term becomes significant for PVD with qw of
<40–60 m3/year. Hansbo (1981) considered the smear effect
that occurs during the vertical drain installation. The average
degree of consolidation for vertical drains can be given by
⎛ −8Th ⎞
[1a] Ur = 1 − exp ⎜ ⎟
⎝ µ ⎠

in which
n kh
[1b] µ = ln + ln s − 0.75
s ks

where Ur is the average degree of consolidation due to radial


drainage; Th is the dimensionless time factor for consolida-
tion due to radial drainage; µ is a group of parameters repre- (ii) In radial drainage, the horizontal permeability of the
senting the geometry of the vertical drain system and the soil decreases with the average void ratio (Fig. 2b). The
smear effect; kh is the average horizontal permeability in the relationship between these two parameters can be com-
undisturbed zone (m/s); ks is the average horizontal perme- monly found by e = e0 + Ck log (kh /khi ) (Tavenas et al.
ability in the smear zone (m/s); and n = de /dw and s = ds /dw, 1983). The permeability index (Ck) is generally consid-
in which de is the equivalent diameter of the cylinder of soil ered to be independent of stress history (pc′) (e.g.,
around the drain, dw is the diameter of the drain well, and ds Nagaraj et al. 1994).
is the diameter of the smear zone.
Figure 1 shows the unit cell adopted for the analysis and Proposed analytical solution
the patterns for the vertical drain. The equivalent drain
diameter (de) is a function of drain spacing (d) and its con- Laboratory testing has shown that during the consolida-
figuration. It is equal to 1.05d and 1.13d for square and tri- tion process, a variation of soil volume compressibility and
angular patterns, respectively (Fig. 1). Rixner et al. (1986) soil permeability can be found that imparts a direct influence
indicated that the triangular spacing gives more uniform set- on the shape of e – log σ ′ and e – log kh (Lekha et al. 2003).
tlement than the square pattern, although the square pattern In addition, in the field, the different subsoil stress histories
is more convenient to control in the field. The installed drain (normally consolidated or lightly overconsolidated soil) result
pattern also depends on how the drains are installed in the in different consolidation responses (Seah and Juirnarongrit
field, what equipment is used, etc. The symbol dw is the 2003). Therefore, to predict the behaviour of a vertical drain
equivalent band drain diameter, or the actual diameter of the system more accurately, it is necessary to incorporate the
sand drain (m). In the case of PVD, dw can be determined relationships of e – log σ ′ and e – log kh with radial consoli-
from 2(a + b)/ π, where a and b are the width and thickness dation and then find a new solution for the radial consolida-
of the PVD, respectively (Hansbo 1979). tion.
The flow rate in the unit cell can be expressed by Darcy’s
law as
Other assumptions for the proposed
∂Q kh ∂u
analytical solution [2] = Acs
∂t γ w ∂r
Apart from the two assumptions stated earlier for the
Hansbo (1981) theory, the additional assumptions made in where Q is the flow in soil mass; Acs is the cross-sectional
the writers’ analysis are summarized below: area of the flow at distance r and is equal to 2 πr(dz) for an
(i) During the consolidation process, at a given depth, the element thickness of dz.
relationship between the average void ratio and the log- The rate of changing volume of soil mass is given by
arithm of average effective stress in the normally ∂V ∂ε
consolidated range (Fig. 2a) can be expressed by e = [3] = π (r 2e − r 2) dz
∂t ∂t
e0 – Cc log (σ ′/σ ′i ). If the current vertical effective
stress (σ ′) is smaller than pc′, the recompression index The flow rate in the unit cell is equal to the rate of volume
(Cr) is used instead of Cc for the overconsolidated range. change of soil mass; therefore,
© 2005 NRC Canada
Indraratna et al. 1333

kh ⎛ ∂u ⎞ ∂ε ∂ e d 2e µ
[4] ⎜ ⎟ 2 π r dz = π (r 2e − r 2) dz [12] ut = γw
γ w ⎝ ∂r ⎠ ∂t ∂t 8kh(1 + e0)

where kh is the average coefficient of permeability in the un- Substituting Ru = ut /∆p in eq. [12] and further modifying the
disturbed zone. Rearranging eq. [4] gives the following equation results in
equation for the pore pressure gradient in the undisturbed
∂ e ∂ (σ − ut) khi d 2e µ
soil domain outside the smear zone: [13] Ru = γw
∂σ ′ ∂t kh 8khi (1 + e0) ∆ p
∂u γ ∂ε ⎛ r 2e − r 2 ⎞
[5] = w ⎜ ⎟ re ≥ r ≥ rs The dimensionless parameter Ru is useful when the role of
∂r 2kh ∂t ⎝ r ⎠
applied surcharge in pore-water pressure is considered and
In the smear (disturbed) zone, the corresponding pore pres- when comparisons are made between two or more embank-
sure gradient is then given by ments during construction.
Because preloading pressure ( ∆p) is assumed to be an in-
∂us γ ∂ε ⎛ r 2e − r 2 ⎞ stantaneous loading on the top of the unit cell and because
[6] = w ⎜ ⎟ rs ≥ r ≥ rw total stress (σ) is constant throughout the consolidation pro-
∂r 2ks ∂t ⎝ r ⎠
cess, ∂σ/ ∂t = 0. Simplifying eq. [13] gives
where us is the excess pore pressure in the smear zone; ks is ∂Ru 8 m vi kh
the average permeability in the smear zone; and γ w is the [14] =− Ru
∂Th µ m v khi
unit weight of water.
The excess pore pressure in the smear zone can now be and
obtained by integrating eq. [6] in the r direction and using
chi t
the boundary condition u = 0 at r = rw: [15] Th =
d 2e
γ w ∂ε ⎛ 2 r r 2 − r 2w ⎞
[7] us = ⎜ r e ln − ⎟ rs ≥ r ≥ rw
2ks ∂t ⎝ rw 2 ⎠ where mv = (∂e / ∂ σ ′)/(1 + e0); mvi = (∂e /∂ σ ′t = 0)/(1 + e0);
and chi = khi(1 + e0)/(γ w ∂e / ∂ σ ′t = 0).
Meanwhile, us at r = rs can be determined from The void ratio – effective stress and void ratio – perme-
ability relations for normally consolidated clays can be ex-
γ w ∂ε ⎛ 2 rs r 2s − r 2w ⎞ press as follows (Tavenas et al. 1983):
[8] us,r=rs = ⎜ r e ln − ⎟
2ks ∂t ⎝ rw 2 ⎠ ⎛ σ′ ⎞
[16] e = e0 − C c log ⎜⎜ ⎟⎟
Integrating eq. [5] in the r direction with the boundary con- ⎝ σ ′i ⎠
dition u = ur=rs at r = rs yields
⎛k ⎞
γ ∂ε ⎛ 2 r r 2 − r 2s ⎞ [17] e = e0 + C k log ⎜ h ⎟
[9] u − ur=rs = w ⎜ r e ln − ⎟ re ≥ r ≥ rs ⎝ khi ⎠
2kh ∂t ⎝ rs 2 ⎠
Differentiating eq. [16] with respect to the effective stress
If it is assumed that the excess pore pressure at the outer (σ ′) gives
boundary of the smear zone (ur=rs ) is equal to the excess
pore pressure at the inner boundary of the undisturbed zone m vi ⎛ ∆ p ⎞ ⎛ Ru ∆ p ⎞
(us,r=rs ), as shown in Fig. 1, then [18] = 1 + ⎜⎜ ⎟ −⎜
⎟ ⎜


mv ⎝ σ ′i ⎠ ⎝ σ ′i ⎠
⎡ γ ∂ε ⎛ r r 2 − r 2s ⎞
[10] u=⎢ w ⎜ r 2e ln − ⎟ Combining eqs. [16] and [17] gives
⎢⎣ 2kh ∂t ⎝ rs 2 ⎠
−Cc / C k −Cc / C k
kh ⎛ σ′ ⎞ ⎛m ⎞
γ w ∂ε ⎛ 2 rs r 2s − r 2w ⎞ ⎤ [19] = ⎜⎜ ⎟⎟ = ⎜ vi ⎟
+ ⎜ r e ln − ⎟⎥ re ≥ r ≥ rs khi ⎝ σ ′i ⎠ ⎝ mv ⎠
2kh,s ∂t ⎝ rw 2 ⎠ ⎥⎦
Therefore,
Let ut be the average excess pore pressure of the smear and
−Cc / C k
intact zones at depth z and for a given time, t; hence, ⎡ ⎛ ⎤
kh ∆ p ⎞⎟ ⎛⎜ Ru ∆ p ⎞⎟ ⎥
[20] = ⎢1 + ⎜⎜ −
khi ⎢ ⎝ σ ′i ⎟⎠ ⎜⎝ σ ′i ⎟⎠ ⎥
l re l rs

∫ ∫ u 2 π rdrdz + ∫ ∫ us 2πrdrdz ⎣ ⎦
0 rs 0 rw
[11] ut = Substituting eqs. [18] and [20] into [14] yields
π (r 2e − r 2w) l
∂Ru 8
Substituting eqs. [7] and [10] into eq. [11] gives the follow- [21] = − PRu
∂Th µ
ing expression for the average excess pore pressure, ut , at
any time, given that de = 2re and ∂ε = ∂e /(1 + e0) : where
© 2005 NRC Canada
1334 Can. Geotech. J. Vol. 42, 2005

1 − (Cc / Ck )
⎡ ⎛ ⎤ When σ ′ > pc′ along the slope of the normal consolidation
∆ p ⎞⎟ ⎛⎜ Ru∆ p ⎞⎟ ⎥
P = ⎢1 + ⎜⎜ − curve (Cc), the expression for excess pore pressure ratio
⎢ ⎝ σ ′i ⎟⎠ ⎜⎝ σ ′i ⎟⎠ ⎥ when Th > Th, pc is given by
⎣ ⎦
⎛ σ ′ + ∆ p − p′ ⎞
It can be seen that eq. [21] is a nonlinear partial differential [26d] Ru = ⎜⎜ i c⎟

equation for radial consolidation under instantaneous load- ⎝ ∆ p ⎠
ing, incorporating the e – log σ ′ and e – log kh relations. The
nonlinear differential eq. [21] with variable Ru does not have ⎧ ⎡ 1 − (Cc / C k ) ⎤ ⎫
a general solution, and P varies from [1 + (∆p/σ ′i )]1− (Cc / Ck ) ⎪ ⎢ ⎛⎜ σ ′i + ∆ p − pc′ ⎞⎟ ⎥ (Th − Th, pc ) ⎪
× exp ⎨− 4 1 + ⎜1 + ⎟ ⎬
to 1. Hence, it can be assumed to have an average value ⎢ pc′ ⎥ µ
⎪ ⎢ ⎝ ⎠ ⎥ ⎪
given by ⎩ ⎣ ⎦ ⎭
⎡ ⎛ ⎞
1 − (Cc / C k ) ⎤
⎢ ∆ p ⎥
[22] P = Pav = 0.5 1 + ⎜⎜1 + ⎟

where
⎢ σ ′i ⎠ ⎥
⎢⎣ ⎝ ⎥⎦ ch, pc t ⎛ p′ ⎞
1 −C r / C k

Th = and ch, pc = chi ⎜⎜ c ⎟⎟


Incorporating the above assumption for Pav, eq. [21] can be d 2e ⎝ σ ′i ⎠
written as
When eqs. [9] and [12] are combined, the normalized excess
∂ Ru 8 pore pressure at any point within the smear zone can be
[23] =− Ru
∂T h* µ found as

where T h* is the modified time factor, defined by u′ 4k R ⎡ ⎛ r ⎞ (r 2 − r 2w) ⎤


[27a] = h u 2 ⎢ r 2e ln ⎜ ⎟ − ⎥
∆p ks µ d e ⎢⎣ ⎝ rw ⎠ 2 ⎥⎦
⎡ ⎛ ⎞
1 − (Cc / C k ) ⎤
⎢ ∆ p ⎥T
[24] T h* = PavTh = 0.5 1 + ⎜⎜1 + ⎟
⎟ When eqs. [10] and [12] are combined, the normalized ex-
⎢ σ ′i ⎠ ⎥ h
⎢⎣ ⎝ ⎥⎦ cess pore pressure at any point outside the smear zone is
given by
Integrating eq. [23], subject to the boundary condition that ⎧⎪ ⎛ r ⎞ (r 2 − r 2s)
ut = ∆p at T h* = 0, gives the following expression, which is u 4Ru
[27b] = ⎨r e ln ⎜ ⎟ −
2
similar to the original equation (eq. [1]): ∆p µ d 2e ⎪⎩ ⎝ rs ⎠ 2

⎛ − 8 T h* ⎞ ⎫
[25] Ru = exp ⎜⎜ ⎟ kh ⎡ 2 ⎛ rs ⎞ (r 2s − r 2w) ⎤ ⎪
⎟ + ⎢ r e ln ⎜ ⎟ − ⎥⎬
⎝ µ ⎠ ks ⎢⎣ ⎝ rw ⎠ 2 ⎥⎦ ⎪⎭
With the substitution of T h* in eq. [25], the expression for ex-
where the value of Cc /Ck approaches unity, the writers’ solu-
cess pore pressure ratio for normally consolidated clay be-
tion converges to that of Hansbo (1981) described earlier
comes
(see eq. [1]); hence,
⎧ ⎡ 1 − (Cc / C k ) ⎤ ⎫ ⎛ −8Th ⎞
⎪ ⎢ ⎛⎜ ∆ p ⎞⎟ ⎥ Th ⎪ [28] Ru = exp ⎜ ⎟
[26a] Ru = exp ⎨− 4 1 + ⎜1 +
⎢ ⎟
σ ′i ⎠ ⎥ µ ⎬ ⎝ µ ⎠
⎪ ⎢ ⎝ ⎥⎦ ⎪⎭
⎩ ⎣
Because the relationship between effective stress and strain
For overconsolidated soil, σ ′ < pc′ (e.g., topmost layer is not linear, the average degree of consolidation can be de-
close to surface): fined on the basis of excess pore pressure (stress) (Up) or of
strain (settlement at the top surface) (Us). The variable Up
⎧ ⎡ 1 − (C r / C k ) ⎤ ⎫ indicates the rate of dissipation of excess pore pressure,
⎪ ⎢ ⎛⎜ ∆ p ⎞⎟ ⎥ Th ⎪ whereas Us shows the rate of development of the surface set-
[26b] Ru = exp ⎨− 4 1 + ⎜1 +
⎢ ⎟
σ ′i ⎠ ⎥ µ ⎬ tlement. Normally, Up ≠ Us, except when the relationship be-
⎪ ⎢ ⎝ ⎥⎦ ⎪⎭
⎩ ⎣ tween effective stress and strain is linear, which is Terzaghi’s
one-dimensional (1D) theory. Therefore, the average degree
When the effective stress equals the preconsolidation pres- of consolidation based on excess pore pressure can be ob-
sure (σ ′ = pc′ ), the corresponding time factor, Th, pc , can then tained as follows:
be determined by
[29] Up = 1 – Ru
µ ⎛ ∆p ⎞
[26c] Th, pc = ln ⎜⎜ ⎟

The average degree of consolidation based on settlement
⎡ ⎛ ⎞
1 − (C r k ⎤
/ C )
⎝ σ ′i + ∆ p − pc′ ⎠ (strain) can be given by:
⎢ ∆ p ⎥
4 1 + ⎜⎜1 + ⎟
⎟ ρ
⎢ σ ′i ⎠ ⎥ [30a] Us =
⎢⎣ ⎝ ⎥⎦ ρ∞
© 2005 NRC Canada
Indraratna et al. 1335

The associated settlements (ρ) are evaluated by the fol- Table 1. Parameters used to compare the effects of
lowing equations: loading increment ratio and Cc /Ck.

HC r ⎛ σ′ ⎞ Parameter Value
[30b] ρ = log ⎜⎜ ⎟⎟ σ ′i ≤ σ ′ ≤ p ′c n = de /dw 15
1 + e0 ⎝ σ ′i ⎠
s = ds /dw 4
Horizontal permeability, kh (×10–9 m/s) 1.0
⎡ ⎛ p′ ⎞ ⎛ σ′ ⎞⎤
H ⎢ kh /ks 1.5
[30c] ρ = C r log ⎜⎜ i ⎟⎟ + C c log ⎜ ⎟⎥
⎜ ⎟⎥
1 + e0 ⎢ ⎝ σ ′i ⎠ ⎝ p ′c ⎠ ⎦ Initial void ratio, e0 2

In situ effective stress, σ ′i (kPa) 20
p ′c ≤ σ ′ ≤ σ ′i + ∆ p Cc 0.25

HC c ⎛ σ′ ⎞
[30d] ρ = log ⎜⎜ ⎟⎟ for normally consolidated clay process is slower than that predicted by Hansbo’s solution.
1 + e0 ⎝ σ ′i ⎠ For Cc /Ck < 1, when the load increment ratio increases, the
rate of settlement increases (Fig. 3). However, for Cc /Ck >
It should be noted that ρ∞ can be obtained by substituting
1, the reverse is true: when the load increment ratio in-
σ ′ = σ ′i + ∆p into the above equations, where ρ is the settle-
creases, the rate of settlement decreases. It is important to
ment at a given time; ρc is the total primary consolidation
note that the degree of consolidation based on strain (Us) is
settlement; Cc is the compression index; Cr is the
greater than that based on excess pore pressure (Up). The
recompression index; and H is the compressible soil thick-
magnitude of the difference between them depends on the
ness.
soil properties and the applied stress levels. To correctly
Depending on the location of the initial and final effective
evaluate the actual consolidation, it is necessary to consider
stresses with respect to the normally consolidated and over-
the variations in permeability and compressibility, stress his-
consolidated domains, the following is a summary of the rel-
tory, and the magnitude of preloading pressure. More signif-
evant computational steps:
icantly, the roles of Cc /Ck and ∆p/σi are found to be
(i) If both the initial and final effective stresses are in the important, as demonstrated here.
normally consolidated range, eqs. [26a] and [29] are
used to calculate Up, and eqs. [30a] and [30d] are used
to compute Us. Verification of the proposed model
(ii) If both the initial and final effective stresses are in the The proposed model was validated by comparing the set-
overconsolidated range, eqs. [26b] and [29] are used to tlement predictions with laboratory data. The laboratory test-
calculate Up, and eqs. [30a] and [30b] are used to deter- ing was conducted by using a large-scale, 450 mm diameter
mine Us. consolidation apparatus with a height of 950 mm. Reconsti-
(iii) If the initial effective stress falls in the overconsolidated tuted alluvial clay from Moruya (New South Wales) was
domain, and the final effective stress is in the normally used in the apparatus. To reduce friction between the side
consolidated domain, eqs. [26b]–[26d] and [29] are used wall of the cylinder and the soil, a Teflon sheet was laid
to calculate Up, and eqs. [30a]–[30c] are used to calcu- around the inner periphery of the cell. Table 2 summarizes
late Us. the properties of this reconstituted clay. The e – log σ ′ and
e – log kh relationships were obtained by the four conven-
Comparisons of Hansbo (1981) with proposed solutions tional oedometer tests (Fig. 4). From the e – log σ ′ and e –
In this section, the effects of the values of Cc /Ck and load log kh plots, the slope of the e – log σ ′ line (Cc) and the
increment ratio (∆p/σ i) are examined in the post-p ′c region. slope of the e – log kh line (Ck) were found to be 0.29 and
Relevant parameters used in the analysis are given in Ta- 0.45, respectively. Therefore, the corresponding Cc /Ck could
ble 1. According to Berry and Wilkinson (1969), the typical be calculated as 0.64. The detailed testing procedure was ex-
values of Cc /Ck for soil in the range of 0.5–2.0 are used in plained elsewhere by Indraratna and Redana (1998). Follow-
the analysis. The load increment ratio is either 1or 2. Fig- ing Burland (1990), the clay specimen was prepared with a
ure 3 shows the comparison of the degree of consolidation water content slightly greater than the liquid limit. The clay
based on the excess pore pressure (Fig. 3a) and strain was placed and compacted in layers in the apparatus. In two
(Fig. 3b) approaches, both using the compression index and different tests, initial preconsolidation pressures of 20 and
the variation in horizontal permeability (eqs. [29] and [30a]). 50 kPa, respectively, were applied for 5 days before the in-
The maximum difference between Up and Us is around 10%. stallation of the vertical drain. The 100 mm × 4 mm band
It is evident that the values of Cc /Ck and ∆p/σi with the drain was then installed vertically in the center of the cell by
same initial coefficient of consolidation for horizontal drain- using a steel mandrel. After drain installation, the mandrel
age (chi) control the rate of the consolidation process. For a was withdrawn with a hoist system, and subsequently, the
given ∆p/σ i , when Cc /Ck < 1 (i.e., the rate of decrease in preconsolidation pressures of 20 and 50 kPa were main-
void ratio with an increase in effective stress is smaller than tained in the two tests. The two large clay samples were then
the rate of decrease of void ratio with a decrease in perme- further subjected to different loading sequences, that is,
ability), the actual consolidation process is faster than the loading increments (∆p) of 30 and 50 kPa, to give the final
corresponding result from Hansbo’s theory. In contrast, total pressures of 50 and 100 kPa, respectively. Both samples
when Cc /Ck > 1 for a given ∆p/σ i , the actual consolidation were then loaded in the normally consolidated range. The
© 2005 NRC Canada
1336 Can. Geotech. J. Vol. 42, 2005

Fig. 3. Average degree of radial consolidation U plotted against time factor Th for varying compressibility and permeability relation-
ships (a) based on excess pore pressure, (b) based on strain.

Table 2. Soil properties of the reconstituted Fig. 4. Typical e – log σ ′ and e – log kh plots for Moruya clay.
clay sample.
Soil property Reconstituted clay
Water content (%) 45
Plastic limit, wp (%) 17
Plasticity index, PI (%) 25
Unit weight, γt (kN/m3) 17
Cc 0.29
Ck 0.45

© 2005 NRC Canada


Indraratna et al. 1337

Fig. 5. Comparison between measured and predicted results from proposed model and Hansbo’s solution: (a) pc′ = 20 kPa, ∆p =
30 kPa; and (b) pc′ = 50 kPa, ∆p = 50 kPa.

corresponding settlement behaviour was recorded, as shown is the initial void ratio; and mvi is the initial coefficient of
in Fig. 5. volume compressibility.
The initial coefficient of horizontal permeability in the un- As shown in Fig. 5, the settlement predictions from
disturbed zone, khi, was determined from the e – log kh plots Hansbo’s solution slightly underestimate the laboratory re-
(Fig. 4) and the initial void ratio (e0). The diameter of the sults, whereas the predictions incorporating the compression
smear zone (ds) and the initial ratio kh /ks were evaluated to indices and the variation of soil lateral permeability agree
be 200 mm and 1.5, respectively, on the basis of previous very well with the laboratory results. It is noted that the rate
testing (Indraratna and Redana 1998). Table 3 shows the pa- of consolidation from the modified solution is greater than
rameters used in the analytical model for settlement predic- that found by the conventional analysis for Cc /Ck < 1, as ex-
tion. plained earlier in the previous section. This verifies that in
In Figs. 5a and 5b, the degree of consolidation based on addition to depending on the permeability and drain configu-
strain is shown for the writers’s model prediction (eq. [30a]), ration, the consolidation behavior depends on the magnitude
the calculation based on laboratory data, and Hansbo’s solu- of the loading increment ratio and the values of Cc /Ck.
tion. The initial coefficient of horizontal consolidation (chi),
which is related to a time factor (Th), can be calculated by
Application of the model to a selected case
khi
[31] chi = history
γ w m vi
The Malaysian Highway Authority was responsible for
where khi is the initial horizontal permeability coefficient; e0 constructing a number of test embankments on the Muar
© 2005 NRC Canada
1338 Can. Geotech. J. Vol. 42, 2005

Table 3. Parameters used in the analysis.


Test 1 (final total Test 2 (final total
Parameter pressure = 50 kPa) pressure = 100 kPa)
Diameter of influence zone, de (m) 0.45 0.45
Equivalent diameter of drain, dw (m) 0.066 0.066
Diameter of smear zone, ds (m) 0.2 0.2
n = de /dw 6.79 6.79
s = ds /dw 3.02 3.02
Initial horizontal permeability, khi (×10–10 m/s) 4.6 4.0
kh /ks 1.5 1.5
Initial void ratio, e0 1.00 0.95
chi (×10–3 m2/day) 1.58 3.02
Initial height (m) 0.925 0.870
Preconsolidation pressure, pc′ (kPa) 20 50
∆p (kPa) 30 50

Fig. 6. Cross section of embankments with soil profile, Muar clay, Malaysia.

Plain and using various forms of ground improvement tech- Table 4. Geometric parameters of the vertical drain
niques, including vertical drains. Figure 6 shows a vertical system for a Muar clay embankment.
cross section of one such embankment, together with the
subsoil profile and PVDs installed in a triangular pattern at a Geometric parameter Value
spacing of 1.3 m. The details of the test embankment were Installation pattern Triangular
explained elsewhere by Indraratna et al. (1994). Table 4 Drain spacing, d (m) 1.3
gives the details of the drain geometry. The soil parameters, Diameter of influence zone, de (m) 1.365
the in situ effective stress, and the soil permeability for
Equivalent drain diameter, dw (m) 0.07
Muar clay subsoils are summarized in Table 5. The relevant
soil properties, including compressibility indices, soil unit s = ds /dw 4
weights, initial void ratios, preconsolidation pressures, and
permeability coefficients, were obtained from CKoU triaxial
tests (Ratnayake 1991). As suggested by Tavenas et al. (1983), tion with depth. The value of soil compressibility index (Cc
the slope of e – log kh (Ck) can be calculated by or Cr) is associated with the actual stress state within a given
[32] Ck = 0.5e0 region of the foundation, where the working stress range
must be considered in relation to the preconsolidation pres-
In the analysis, each subsoil layer was divided into smaller sure of soil at that particular depth (Indraratna et al. 1994).
sublayers to derive a more accurate effective stress distribu- According to laboratory experiments conducted by Indraratna

© 2005 NRC Canada


Indraratna et al. 1339

Table 5. Soil parameters for Muar clay embankments (data from Ratnayake 1991).

γt σ ′ at middle pc′ khi (in situ value)


Depth (m) Cr Cc (kN/m3) e0 of layer (kPa) (kPa) (×10–9 m/s)
0.00–1.75 0.35 0.71 16.5 3.10 4.88 60 6.4
1.50–2.50 0.37 0.71 15.0 3.10 12.25 55 5.2
2.50–5.50 1.38 15.0 3.00 22.25 50 5.2
5.50–6.50 1.38 15.5 3.00 32.50 44 3.1
6.50–8.00 0.71 15.5 1.95 39.38 51 3.1
8.00–10.00 0.71 16.0 1.82 49.50 60 1.3
10.00–12.00 0.83 16.0 1.86 61.50 73 0.6
12.00–14.00 0.83 16.0 1.89 73.50 86 0.6
14.00–16.00 0.83 16.0 1.86 85.50 97 0.6
16.00–18.00 0.83 16.0 1.86 97.50 110 0.6

Fig. 7. Muar clay embankment in Malaysia: (a) stages of load- a rest period of 105 days, an additional fill layer (with com-
ing; (b) surface settlements under the embankment centerline; pacted unit weight of 20.5 kN/m3) was placed during stage
and (c) excess pore pressures at a depth of 11.2 m below ground 2, until the embankment reached the height of 4.74 m, in
surface, 0.65 m away from the centerline. 24 days. The settlements at the centerline were monitored
for about 400 days.
The embankment loading was simulated by assuming an
instantaneous loading at the upper boundary. Settlement pre-
dictions were carried out at the embankment centerline by
using the writers’ analytical model (eqs. [26]–[30a]). At the
beginning of stage 2, the initial in situ effective stress and
initial coefficient of horizontal consolidation (chi) were cal-
culated from the stage 1 final degree of consolidation. As
the computation of consolidation settlement at the centerline
(zero lateral displacements) is straightforward and follows
the 1D consolidation theory, the use of an Excel spreadsheet
formulation for this purpose was most convenient. The value
of soil compressibility (Cc or Cr), in association with the
correct working effective stress, plays a very important role
in predicting settlement. For stage 1 loading, where the ef-
fective preconsolidation pressure (pc′ ) was not exceeded, the
value of recompression index (Cr) could be used. In particu-
lar, the surface crust was heavily overconsolidated (up to
about 3 m depth). Once pc′ was exceeded, the value of the
compression index (Cc) followed the normally consolidated
line, as indicated by the values in Table 5. The measured ini-
tial horizontal permeability coefficients of undisturbed soil
(khi) are also given in Table 5. The predicted settlement
agrees well with the measured values at the embankment
centerline (Fig. 7b). In contrast, Hansbo’s solution under-
predicts before 170 days and overpredicts after 170 days.
The analytically predicted and the measured excess pore
pressures beneath the embankment at a depth of 11.2 m be-
low ground surface and at a location of 0.65 m away from
the centerline are shown in Fig. 7c. Acceptable agreement
between the predictions and measurements is found. It
should be noted that the writers’ and Hansbo’s (1981) solu-
tions are close to each other, because the ratio Cc /Ck is al-
most unity at the measurement location.
and Redana (1998), the ratio of kh /ks is approximately 1.5–
2.0. However, this ratio in the field can vary from 1.5 to 5, Conclusions
depending on the type of drain and installation procedures
(Saye 2003). The value of kh /ks for this case study was taken A system of vertical drains is an effective method for ac-
to be 3. celerating soil consolidation. A revised analytical model for
The embankment load was applied in two stages. As soft clay stabilized by vertical drains, incorporating the com-
shown in Fig. 7a, during stage 1 construction, the embank- pressibility indices (Cc and Cr), was proposed, and the varia-
ment was raised to a height of 2.57 m in 14 days. Following tion of horizontal permeability coefficient (kh) was represented

© 2005 NRC Canada


1340 Can. Geotech. J. Vol. 42, 2005

by the e – log kh relationship. The parameters, such as the Hansbo, S. 1997. Aspects of vertical drain design: Darcian or non-
slopes of the e – log σ ′ relationship (Cc and Cr), the slope of Darcian flow. Géotechnique, 47(5): 983–992.
e – log kh relationship (Ck), and the loading increment ratio Holtz, R.D., and Kovacs, W.D. 1981. An introduction to geotech-
( ∆p/σ i), were explicitly included in the analytical model to nical engineering. Prentice-Hall, Inc., Englewood Cliffs, N.J.
predict consolidation behaviour. The writers’ analysis partic- Holtz, R.D., Jamiolkowski, M.B., Lancellotta, R., and Pedroni, R.
ularly elaborates on the important role of Cc /Ck and ∆p/σi 1989. Behavior of bent prefabricated vertical drains. In Proceed-
ratios, which are found to govern the rate of consolidation. ings of the 12th International Conference on Soil Mechanics and
When Cc /Ck < 1, the actual rate of consolidation is higher Foundation Engineering, Rio de Janeiro, August 1986. Vol. 3,
than that determined by using the conventional solution of pp. 1657–1660.
Hansbo (1981). Also, the rate of consolidation decreases Holtz, R.D., Jamiolkowski, M.B., Lancellotta, R., and Pedroni, R.
with the reduction of the load increment ratio ( ∆p/σi). When 1991. Prefabricated vertical drains: design and performance.
Cc /Ck > 1, the consolidation process takes place at a slower Construction Industry Research and Information Association,
rate than the conventional solution. In these circumstances London, UK. CIRIA Report.
(Cc /Ck > 1), the rate of consolidation increases with the de- Indraratna, B., and Balasubramaniam, A.S. 1993. Authors’ closure:
performance of test embankment constructed to failure on soft
creasing ratio ( ∆p/σi). This also shows that settlement oc-
clay. Journal of Geotechnical Engineering, ASCE, 119(9): 1321–
curs faster than excess pore pressure dissipation.
1329.
The predictions from the writers’ analytical model agree
Indraratna, B., and Redana, I.W. 1998. Laboratory determination of
well with the laboratory results based on large-scale consoli- smear zone due to vertical drain installation. Journal of Geo-
dation testing. The proposed solution gives a more accurate technical and Geoenvironmental Engineering, ASCE, 124(2):
settlement prediction when applied to a selected case history 180–184.
(Muar clay, Malaysia). In this analysis, the smear effect (due Indraratna, B., and Redana, I.W. 2000. Numerical modeling of ver-
to mandrel-driven prefabricated drains) was included, but the tical drains with smear and well resistance installed in soft clay.
well resistance was ignored (drain length < 20 m). For a Canadian Geotechnical Journal, 37(1): 133–145.
given drain pattern, the findings of this study confirm that Indraratna, B., Balasubramaniam, A.S., and Ratnayake, A.M.P. 1994.
the soil compressibility indices based on the e – log σ ′ rela- Performance of embankment stabilised with vertical drains on
tionship, as well as the load increment ratio ( ∆p/σi), are ma- soft clay. Journal of Geotechnical Engineering, ASCE, 120(2):
jor factors influencing embankment settlement, apart from 257–273.
the change in the lateral soil permeability (kh) with the void Indraratna, B., Balasubramaniam, A.S., and Sivaneswaran, N. 1997.
ratio. Analysis of settlement and lateral deformation of soft clay foun-
dation beneath two full-scale embankments. International Jour-
Acknowledgements nal for Numerical and Analytical Methods in Geomechanics, 21:
599–618.
The authors acknowledge the support of the Malaysian Jamiolkowski, M., Lancellotta, R., and Wolski, W. 1983. Precom-
Highway Authority and the Airports Authority of Thailand. pression and speeding up consolidation. In Proceedings of the
The research funding from the Rail Infrastructure Corpora- 8th European Conference on Soil Mechanics and Foundations,
tion (Sydney) and from Queensland Department of Main Helsinki, Finland, May 1983. Edited by H.G. Rathmayer and
Roads (Brisbane) is acknowledged. The writers also appreci- K.H.O. Saari. Vol. 3, pp. 1201–1206.
ate the efforts of John Koumakis (former undergraduate stu- Johnson, S.J. 1970. Precompression for improving foundation soils.
dent) and Alan Grant (technical staff member) at University Journal of Soil Mechanics and Foundation Engineering Divi-
of Wollongong for their assistance during laboratory testing. sion, 96(1): 111–114.
The second and third authors gratefully acknowledge finan- Lekha, K.R., Krishnaswamy, N.R., and Basak, P. 2003. Consolida-
cial support (international Ph.D. scholarships) from the Uni- tion of clays for variable permeability and compressibility. Jour-
versity of Wollongong. nal of Geotechnical and Geoenvironmental Engineering, ASCE,
129(11): 1001–1009.
Li, A.L., and Rowe, R.K. 2002. Combined effect of reinforcement
References and prefabricated vertical drains on embankment performance.
Barron, R.A. 1948. Consolidation of fine-grained soils by drain Canadian Geotechnical Journal, 38: 1266–1282.
wells. Transactions of the American Society of Civil Engineers, Nagaraj, T.S., Pandian, N.S., and Narashima Raju, P.S.R. 1994.
113: 718–754. Stress-state–permeability relations for overconsolidated clays.
Berry, P.L., and Wilkinson, W.B. 1969. The radial consolidation of Géotechnique, 44(2): 349–352.
clay soils. Géotechnique, 19(2): 253–284. Ratnayake, A.M.P. 1991. Performance of test embankments with
Burland, J.B. 1990. On the compressibility and shear strength of and without vertical drains at Muar Flats site. M.Eng. thesis,
natural clays. Géotechnique, 40(3): 329–378. Asian Institute of Technology, Bangkok, Thailand.
Casagrande, A. 1932. Research on the Atterberg limits of soils. Rixner, J.J., Kraener, S.R., and Smith, A.D. 1986. Prefabricated
Public Roads, 13(8): 121–136. vertical drains. Summary of research effort—final report. Fed-
Hansbo, S. 1979. Consolidation of clay by band-shaped prefabri- eral Highway Administration, US Department of Commerce,
cated vertical drains. Ground Engineering, 12(5): 16–25. Washington, D.C. Report FHWA-RD-86-169, Vol. 2.
Hansbo, S. 1981. Consolidation of fine-grained soils by prefabri- Saye, S.R. 2003. Assessment of soil disturbance by the installation
cated drains. In Proceedings of the 10th International Confer- of displacement sand drains and prefabricated vertical drains. In
ence on Soil Mechanics and Foundation Engineering, Stockholm, Soil Behavior and Soft Ground Construction: Proceedings of an
June 1981. A.A. Balkema, Rotterdam, The Netherlands. Vol. 3, ASCE Symposium, Cambridge, Mass., 5–6 October 2001.
pp. 677–682. Edited by J.T. Germaine, T.C. Sheahan, and R.V. Whitman.

© 2005 NRC Canada


Indraratna et al. 1341

ASCE Geo-Institute, Reston, Va. Geotechnical Special Publica- Us average degree of radial consolidation based on strain
tion 120, pp. 325–362. u excess pore-water pressure (kN/m2)
Seah, T.H., and Juirnarongrit, T. 2003. Constant rate of strain con- ur excess pore pressure at radius r
solidation with radial drainage. Geotechnical Testing Journal, us excess pore-water pressure in smear zone (kN/m2)
ASTM, 26(4): 1–12. ut average excess pore-water pressure at a given time t
Seah, T.H., Tangthansup, B., and Wongsatian, P. 2004. Horizontal (kN/m2)
coefficient of consolidation of soft Bangkok clay. Geotechnical V volume of soil mass (m3)
Testing Journal, ASTM, 27(5): 1–11. vr velocity of flow (at radius r) (m/s)
Tavenas, P., Jean, P., Leblond, P., and Leroueil, S. 1983. The per- Wp plastic limit
meability of natural soft clays. Part II: Permeability characteris- z depth (m)
tics. Canadian Geotechnical Journal, 20: 645–659.
γt total unit weight of soil
γw unit weight of water (kN/m3)
List of symbols ε volumetric strain
Acs cross-sectional area corresponding to flow (m2) ∆p preloading pressure (kN/m2)
a width of the prefabricated vertical drain (m) µ a function of n, s, kh, ks
b thickness of the prefabricated vertical drain (m) ρ settlement at a given time
Cc compression index ρc total primary consolidation settlement
Ck permeability index ρ∞ ultimate consolidation settlement
Cr recompression index σ vertical total stress (kN/m2)
chi initial coefficient of consolidation for horizontal drain- σi in situ effective stress (kPa)
age (m2/s) σ′ vertical effective stress (kN/m2)
ch , p c coefficient of consolidation for horizontal drainage at σ ′i in situ vertical effective stress or initial vertical effective
effective preconsolidation pressure (m2/s) stress (kN/m2)
d drain spacing (m)
de diameter of influenced zone (m)
ds diameter of smear zone (m) Appendix A. Details of eq. [18] derivation
dw diameter of drain well (m) Differentiating eq. [16] with respect to the effective stress
e void ratio (σ ′) gives
e0 average void ratio at initial in situ effective stress
H compressible soil thickness ∂e Cc
k average permeability coefficient of soil (m/s) [A1] =−
∂σ ′ 2.303σ ′
kh average horizontal permeability in the undisturbed zone
(m/s)
khi initial horizontal permeability coefficient in the undis- and
turbed zone (m/s)
ks average horizontal permeability coefficient in the smear [A2] m v = (∂ e / ∂σ ′ ) / (1 + e0)
zone (m/s)
mv coefficient of volume compressibility [A3] m vi = (∂ e / ∂σ ′t = 0 ) / (1 + e0)
mvi initial coefficient of volume compressibility
n ratio re/rw or de /dw Therefore,
P modification factor for pore pressure
Pav average modification factor for pore pressure
m vi (∂ e / ∂σ ′t = 0 )
PI plasticity index [A4] =
pc′ preconsolidation pressure (kPa) mv ∂ e / ∂σ ′
Q flow in unit cell (m3)
qw discharge capacity (m3/year) Substituting eq. [A1] into eq. [A4] gives
Ru average excess pore-water pressure ratio (ut /∆p)
r distance from center of the drain in axisymmetrical unit m vi σ′
[A5] =
cell (m) mv σ ′t = 0
re radius of influenced zone (m)
rs radius of smear zone (m) but
rw radius of drain well (m)
s ratio rs /rw or ds /dw
Th dimensionless time factor for horizontal drainage, Th =
[A6] σ ′ = σ ′i + ∆ p − ut
chit /de2
T *h modified dimensionless time factor Substituting eq. [A6] into eq. [A5] gives
Th , p c dimensionless time factor at σ ′ = pc′
m vi
t time (s, day) [A7] = 1 + ∆ p / σ ′i − Ru ∆ p / σ ′i
Up average degree of radial consolidation based on excess mv
pore pressure
Ur average degree of consolidation due to radial drainage where Ru = ut / ∆ p

© 2005 NRC Canada

You might also like