You are on page 1of 17

Analytical Solutions for a Single Vertical Drain with

Vacuum and Time-Dependent Surcharge Preloading


in Membrane and Membraneless Systems
Xueyu Geng1; Buddhima Indraratna, F.ASCE2; and Cholachat Rujikiatkamjorn3

Abstract: A system of vertical drains with combined vacuum and surcharge preloading is an effective method for promoting radial flow,
which accelerates soil consolidation. This study presents the analytical solutions of vertical drains with vacuum preloading for membrane and
membraneless systems under time-dependent surcharge preloading. Both vertical and horizontal drainage were considered in this analysis
because they reflect realistic in situ conditions. According to the field and laboratory observations, the vacuum in the membraneless system
was assumed to be decreasing along the drain; in the membrane system, it was maintained at a constant level. This model was verified by
using the measured settlements and excess pore pressures obtained from large-scale laboratory testing and case histories in Australia, Korea,
and China. The analytical solutions improved the accuracy of predicting the dissipation of pore water pressure and the associated settlement.
The effect of the permeability of the sand blanket in a membrane system, and the possible loss of vacuum, were also discussed. DOI: 10.1061/
(ASCE)GM.1943-5622.0000106. © 2012 American Society of Civil Engineers.
CE Database subject headings: Membranes; Time dependence; Vacuum; Prefabrication; Drainage; Analytical techniques; Preloading.
Author keywords: Membranes; Membraneless; Time-dependent loading; Vacuum pressure; Prefabricated vertical drain: Analytical
solution; Laplace transform.

Introduction Currently there are two types of vacuum preloading systems


commercially available: (a) a membrane system with an airtight
Vertical drains with combined vacuum and surcharge preloading membrane over the drainage layer, and (b) a membraneless system
have been widely used to improve soft clay in coastal areas. The in which a vacuum system is connected to individual PVDs (Fig. 1).
design, construction, and stability problems in these areas are eco- Their effectiveness varies from site to site depending on the types
nomically significant in many countries. When prefabricated ver- of soil treated and the characteristics of the vacuum and drain.
tical drains (PVDs) are installed, the length of the drainage path Several previous studies have highlighted difficulties associated
(radial flow) is shortened, which reduces the time for consolidation with vacuum preloading in the field, with respect to prediction
(Atkinson and Eldred 1981; Hansbo 1981; Runesson et al. 1985; (Bo et al. 2004; Indraratna et al. 2004; Wijeyakulasuriya 1999).
Holtz et al. 1991; Hird et al. 1992; Mesri et al. 1994; Indraratna and Mohamedelhassan and Shang (2002) initially developed a com-
Redana 2000; Zhu and Yin 2000; Fox et al. 2003; Walker and bined one-dimensional vacuum and surcharge consolidation model
Indraratna 2006; Ghandeharioon et al. 2010). This system also based on Terzaghi’s consolidation theory, but subsequent prelimi-
helps to dissipate the pore water pressure in the soil (Shang et al. nary analytical and numerical models capturing the vacuum
1998; Bergado et al. 2002; Bo et al. 2004). Negative suction along consolidation mechanisms were introduced by Indraratna et al.
(2005b).
the drain increases the radial hydraulic gradient toward the drain,
All the aforementioned solutions assume that the surcharge
which in turn prevents the build-up of excess pore water pressure in
loading is applied instantaneously and kept constant during con-
the soil and reduces the risk of failure (Indraratna et al. 2004).
solidation, or just simple step loading. In reality, loads are gradually
applied over time, which means that the soil begins to consolidate
1
Research Fellow, Centre for Geomechanics and Railway Engineering; almost simultaneously. As a result, the dissipation rates of excess
Faculty of Engineering, Univ. of Wollongong, Wollongong City, NSW pore water pressure and settlement may be significantly affected by
2522, Australia.
2 the load variations over time. Ramp loading for ideal drains was
Professor of Civil Engineering, Head, School of Civil, Mining and
Environmental Engineering; and Director, Centre for Geomechanics & addressed by Olson (1977), and Tang and Onitsuka (2000), while
Railway Engineering; Faculty of Engineering, Univ. of Wollongong, Leo (2004) addressed drains with smear and well resistance, under
Wollongong City, NSW 2522, Australia (corresponding author). E-mail: the assumption of equal strain conditions. Conte et al. (2009) pro-
indra@uow.edu.au posed an analytical solution that considered vertical drains with
3
Senior Lecturer, Centre for Geomechanics and Railway Engineering; general time-dependent surcharge loading. However, none of these
and Faculty of Engineering, Univ. of Wollongong, Wollongong City, NSW solutions considered the effect of boundary conditions as a result
2522, Australia. of different vacuum systems with time-dependent surcharge
Note. This manuscript was submitted on January 12, 2010; approved on
preloading.
December 15, 2010; published online on December 17, 2010. Discussion
period open until July 1, 2012; separate discussions must be submitted for In this paper, analytical solutions are presented that reflect
individual papers. This paper is part of the International Journal of Geo- the difference between the vacuum preloading systems (membrane
mechanics, Vol. 12, No. 1, February 1, 2012. ©ASCE, ISSN 1532-3641/ and membraneless system) under time-dependent surcharge pre-
2012/1-27–42/$25.00. loading. The smear zone and well resistance of the drain were also

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 27


Fig. 1. Schematic diagram of PVDs incorporating preloading system: (a) membrane system; (b) membraneless system [adapted from Indraratna et al.
(2004)]

 
considered. The accuracy of the proposed solutions was verified kh1 1 ∂u1 ∂ 2 u1 k ∂ 2 u ∂ε
using existing laboratory and field results.  þ 2  v1 21 ¼ v1 rw ≤ r ≤ r e ð2Þ
γw r ∂r ∂r γw ∂z ∂t

∂ 2 uw1 2k h1 ∂u1 
Governing Equations and Solutions ¼  ð3Þ
∂z2 r w k v1 ∂r r¼rw
To analyze the behavior of vertical drains, the unit cell theory, rep-
Z
resenting a single drain surrounded by a soil annulus in axisymmet- 1 re
ric conditions (Fig. 2) using an equal strain approach, was u1 ¼ 2πru1 dr ð4Þ
πðr 2e  r 2w Þ rw
considered. As described earlier, there are two distinct vacuum sys-
tems used in current practice, and their governing equations and The governing equations for the underlying soil
boundary conditions may be described as follows: (Lw ≤ z ≤ H), may be written as
1. Membrane system  
In a membrane system, a sand blanket is placed on top of ∂εv2 ∂u2 dq
¼ mv2  ð5Þ
the PVDs, a membrane is placed on it, and a vacuum is then ∂t ∂t dt
applied through the horizontal drainage pipe located on top of  
the sand blanket. The vacuum propagates from the horizontal k s2 1 ∂us2 ∂ 2 us2 kv2 ∂ 2 u2 ∂εv2
drain through the layer of sand, PVDs, and layer of clay  þ  ¼ rw ≤ r ≤ rs
γw r ∂r ∂r2 γw ∂z2 ∂t
[Figs. 1(a) and 2(a)]. This three-dimensional flow in the sand
ð6Þ
blanket beneath the membrane (0 ≤ z ≤ Lw ) can be expressed
as  
k h2 1 ∂un2 ∂ 2 un2 k ∂ 2 u ∂ε
   þ  v2 22 ¼ v2 rs ≤ r ≤ re
∂εv1 ∂u1 dq γw r ∂r ∂r 2 γw ∂z ∂t
¼ mv1  ð1Þ
∂t ∂t dt ð7Þ

Fig. 2. Analysis schemes of unit cell with vertical drain: (a) membrane system; (b) membraneless system

28 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012



∂ 2 uw2 2ks2 ∂us2  at the top of the drain to ηp at the bottom, where η is a ratio
¼ ð8Þ
∂z2 r w k w ∂r r¼rw between the vacuum at the top and bottom of the drain. The
value of η varies between 0 and 1. If there is no vacuum loss at
Z Z  the bottom of the PVDs, η ¼ 1, and if vacuum pressure is 0 at
1 rs re
the bottom of the drain, η ¼ 0.
u2 ¼ 2πrus2 dr þ 2πrun2 dr ð9Þ
πðr 2e  r 2w Þ rw rs The boundary conditions for a membraneless system are
The boundary conditions for both the radial and vertical ∂ u
z ¼ 0: uw ¼ p; ¼0 ð10l Þ
directions are as follows: ∂z
∂un2 ∂u1
r ¼ re : ¼ 0; ¼0 ð10a Þ ∂uw η  1 ∂ u
∂r ∂r z ¼ H: ¼ p; ¼0 ð10m Þ
∂z H ∂z
∂us2 ∂u
r ¼ r s : k s2 ¼ kh2 n2 ð10b Þ
∂r ∂r
Proposed Analytical Solution
r ¼ r s : us2 ¼ un2 ð10c Þ
The pore water pressure within the vertical drain and the average
pore water pressure for the membrane system, which can be solved
r ¼ r w : us2 ¼ uw2 ; u1 ¼ uw1 ð10d Þ
by considering the applicable boundary conditions and loading
pattern (detailed derivations can be found in Appendix I), in the
z ¼ 0: uw1 ¼ p; u1 ¼ p ð10e Þ
Laplace frequency domain are
∂uw2 ∂u2
z ¼ H: ¼ 0; ¼0 ð10f Þ ^uw1 ðZ; SÞ ¼ X 1 ea1 Z þ X 2 ea1 Z þ X 3 ea2 Z þ X 4 ea2 Z þ QðSÞ
^ ð11Þ
∂z ∂z
Continuity at the interface between the sand blanket and ^uw2 ðZ; SÞ ¼ Y 1 eb1 Z þ Y 2 eb1 Z þ Y 3 eb2 Z þ Y 4 eb2 Z þ QðSÞ
^ ð12Þ
underlying layer of soil (z ¼ Lw ) may be then expressed by
z ¼ Lw : uw1 ¼ uw2 ð10 g Þ    
a2 a2
^u1 ðZ; SÞ ¼ X 1 1  1 ea1 Z þ X 2 1  1 ea1 Z
B2 B2
z ¼ Lw : u1 ¼ u2 ð10h Þ    
a22 a2 Z a2
∂uw1 ∂u þ X3 1  e þ X 4 1  2 ea2 Z þ QðSÞ
^ ð13Þ
z ¼ Lw : k v1 ¼ kw w2 ð10i Þ B2 B2
∂z ∂z
   
b2 b2
∂u1 ∂u ^u2 ðZ; SÞ ¼ Y 1 1  1 eb1 Z þ Y 2 1  1 eb1 Z
z ¼ Lw : k v1 ¼ kv2 2 ð10 j Þ B4 B4
∂z ∂z    
2
b2 b 2 Z b22 b2 Z ^
The initial condition is þ Y3 1  e þ Y4 1  e þ QðSÞ ð14Þ
B4 B4
At t ¼ 0; u1 ¼ u2 ¼ u0 ðzÞ ¼ q0 ð10k Þ
where
where i = index number of arbitrary layer, (i ¼ 1, 2); r si = sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
radius of smear zone; r e = radius of influence zone; r = radial ðΘ=C þ B1 þ B2 Þ þ ðΘ=C þ B1 þ B2 Þ2  4ðΘ=CÞB2
coordinate; z = vertical coordinate; t = time; εvi = vertical a1 ¼
2
strain; mvi = coefficient of volume compressibility of soil; sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k hi = horizontal coefficient of permeability of soil; k vi = vertical ðΘ=C þ B1 þ B2 Þ  ðΘ=C þ B1 þ B2 Þ2  4ðΘ=CÞB2
coefficient of permeability of the soil; k w = coefficient of a2 ¼
2
permeability of the vertical drain; ui = average pore pressure; sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
usi = pore pressure at any point in the smear zone; uni = pore ðΘ þ B3 þ B4 Þ þ ðΘ þ B3 þ B4 Þ2  4ΘB4
pressure at any point in the natural soil zone; uwi = excess pore b1 ¼
2
water pressure within the vertical drain; q = time-dependent sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
surcharge preloading; q0 = initial value of preloading; Lw = pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðΘ þ B3 þ B4 Þ  ðΘ þ B3 þ B4 Þ2  4ΘB4
thickness of the sand layer; H = thickness of the whole layer b2 ¼
(i.e., for the membrane system, both the sand blanket and clay 2
layer, and for the membraneless system, only the clay layer); SK 3 h22
and p = vacuum pressure. Θ¼
n2
2. Membraneless system
The main difference between a membrane and mem- By considering the boundary conditions [Eqs. (10a)–(10f)], the
braneless system are the boundary conditions. In the mem- continuity conditions at the interface between the underlying soil
braneless system, a vacuum pump is connected directly and the sand blanket [Eqs. (10g)–(10j)], and the initial condition
to individual PVDs through a system of horizontal pipes [Eq. (10k)], the following matrix can be obtained to derive X i
[Figs. 1(b) and 2(b)]. The governing equations and initial con- and Y i (i ¼ 1, 2, 3, 4):
ditions of the underlying soil improved by PVDs are the same
as for the membrane system [Eqs. (10a)–(10d) and (10k)]. In ξ 8×8 ψT ¼ PT ð15Þ
order to study the loss of vacuum, the vacuum pressure along
the boundary of the drain is considered to vary linearly from p where ξ 8×8 as Appendix I shows

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 29


ψ ¼ ½ X1 X2 X3 X4 Y1 Y2 Y3 Y 4 ; Also, ^uw1 ðZ; SÞ, ^uw2 ðZ; SÞ, ^u1 ðZ; SÞ, u^2 ðZ; SÞ, QðSÞ,
^ S = Laplace
transform of uw1 ðZ; T h2 Þ, uw2 ðZ; T h2 Þ, u1 ðZ; T v1 Þ, u2 ðZ; T v1 Þ,
P ¼ ½P
^  Q;
^ 0; 0; 0; 0; 0; 0; 0 qðT h1 Þ, T h2 .
8K 1 h22 8h22 ðn2  1ÞK 2 The solutions to the excess pore water pressure, uwi , and average
B1 ¼ ; B2 ¼ ; pore water pressure, ui (i ¼ 1, 2), were obtained using the inverse
F a1 n2 F a1 n2
Laplace transform of Eqs. (11)–(13), hence
8K 3 h22 8h2 ðn2  1ÞK 4
B3 ¼ 2
; B4 ¼ 2 ; Z aþI∞
F a2 n F a2 n2 1
    uwi ðZ; SÞ ¼ ^u ðZ; SÞeST dS ði ¼ 1; 2Þ ð16Þ
3 n2 1 1 2πI aI∞ wi
F a1 ¼ ln n  þ 1  ;
4 n2  1 n2  1 4n2 Z
  1 aþI∞
n 3 n2 ui ðZ; SÞ ¼ ^ui ðZ; SÞeST dS ði ¼ 1; 2Þ ð17Þ
F a2 ¼ ln þ K 5 ln m  2πI
m 4 n 1
2 aI∞
    pffiffiffiffiffiffiffi
m2 m2 1 1 where I ¼ 1. The analytical solutions of Eqs. (16) and (17)
þ 2 ð1  K 5 Þ 1  2 þ K 5 2 1 2 ;
n 1 4n n 1 4n were obtained using the numerical inversion of the Laplace trans-
form (Durbin 1974).
Similar to the membrane system, the pore water pressure within
re r k the vertical drain and the average pore water pressure for the mem-
n¼ ; m¼ s; s ¼ r s =r w ; cvi ¼ vi ;
rw rw mvi γw braneless system, which can be solved by considering the appli-
k k k k cable boundary conditions and loading pattern (Appendix I), in
chi ¼ hi ; K 1 ¼ h1 ; K 2 ¼ h1 ; K 3 ¼ h2 ; the Laplace frequency domain are:
mvi γw k v1 kw kv2
kh2 kh2 H ch2 · t ^uw ðZ; SÞ ¼ χ1 eb1 Z þ χ2 eb1 Z þ χ3 eb2 Z þ χ4 eb2 Z þ QðSÞ
^ ð18Þ
K4 ¼ ; K5 ¼ ; h2 ¼ ; T h2 ¼ ;
kw ks2 dw de2
c z    
C ¼ v1 ; Z¼ : b2 b2
cv2 H ^uðZ; SÞ ¼ χ1 1  1 eb1 Z þ χ2 1  1 eb1 Z
B6 B6
 2
  
b2 b 2 Z b22 b2 Z ^
þ χ3 1  e þ χ4 1  e þ QðSÞ ð19Þ
Table 1. Soil Properties of the Reconstituted Moruya Clay Sample B6 B6
[Adapted from Indraratna et al. (2005a)]
Property Value Table 2. Summary of the Large-Scale Tests
Clay content (%) 40–50 Test number Applied Applied
Silt content (%) 45–60 (1,2 = test vacuum surcharge Preconsolidation
Water content, w (%) 38 Series 1 or test 2) pressure (kPa) pressure (kPa) pressure (kPa)
Liquid limit, wL (%) 57 1 VP1 20 0 20
Plastic limit, wp (%) 17 VP2 40 0 20
Unit weight, (t=m3 ) 1.81 2 SV1 20 30 20
Specific gravity, Gs 2.56 SV2 40 30 20
Undrained shear strength (kPa) 7.3

Fig. 3. Predicted and measured settlement at the top of the consolid- Fig. 4. Schematic of the large-scale, radial-drainage consolidometer
ometer cell [adapted from Indraratna et al. (2004)]: (a) vacuum load showing the central drain, associated smear zone, and typical locations
model; (b) predicted and measured settlement of pore-pressure transducers

30 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012


The matrix from the Eqs. (18) and (19) are determined by The average degree of consolidation may be defined in terms of
effective stress (i.e., dissipation of excess pore water pressure) as
ξ 04×4 ψ0T ¼ P0T ð20Þ R ρ 1 R
0 L ð
^u1 ÞdZ þ ρ1 L1 ð^u2 ÞdZ
Up ¼ 1  R ρ
 R1 ð23Þ
where ξ 0 , as Appendix I shows 0 ðqu  pÞdZ þ ρ ðqu  pÞdZ

Because the consolidation of the sand layer was insignificant


ψ 0 ¼ ½ χ1 χ2 χ3 χ4 ; and could be ignored, the degree of consolidation of the single layer
P0 ¼ ½ P
^  Q;
^ 0; ðη  1ÞP;^ 0 of soil, using the constant average soil compressibility of the clay
    improved by PVDs, will give U p ¼ U s.
n 3 n2 m2 m2
F a ¼ ln þ K 5 ln m  þ ð1  K 5 Þ 1 
m 4 n2  1 n2  1 4n2
  Verification of the Proposed Model
1 1
þ K5 2 1 2
n 1 4n The proposed model was validated by comparing the predictions
8K h2 8h2 ðn2  1ÞK 4 for settlement and excess pore pressure with the laboratory data
B5 ¼ 2 3 2 ; B6 ¼ 2 : and analytical solutions proposed by Olsen (1977) and Indraratna
n Fa F a2 n2
et al. (2005a). Durbin (1974) extensively investigated the numerical
inversion of the Laplace transform (NILT). In the following analy-
sis, Durbin’s method is adopted.
Using the inverse Laplace transform, the excess pore water pres-
sure, uw , and average pore water pressure, u, can be obtained. Membrane System
The settlement of the soil is given by
Indraratna et al (2004) conducted a large-scale, radial-drainage con-
Z H
solidation test to study the soil improvement using a vacuum with a
sðtÞ ¼ ε2 dz ð21Þ membrane system, and to examine what occurs when the vacuum
Lw was reapplied. To simulate a membrane system, a large-scale con-
solidometer was modified and used to examine vacuum preloading
Theoretically, the average degree of consolidation may be de- in conjunction with conventional surcharge loading. Several series
fined either in terms of strain or pore pressure. While the former of tests were performed to examine the effect of vacuum and
describes the rate of settlement, the latter indicates the dissipation surcharge preloading and vacuum distribution along the drain. Ap-
rate of excess pore water pressure. proximately 0:14 m3 of soil was required for each sample. Because
The average degree of consolidation, in terms of settlement, can obtaining an undisturbed sample of this size was impractical, recon-
be expressed as stituted alluvial clay from Moruya (New South Wales) was used
instead. Moruya clay consists of approximately 40–50% clay par-
Rρ R1 ticles (< 2 μm), and approximately 60% particles smaller than silt
mv1 L1 ð^u1 ÞdZ þ mv2 L1 ð^u2 ÞdZ
s ¼ 1 
U R 0ρ Rρ1 ð22Þ (< 6 mm). The cell used for this test was 450 mm (inside diameter)
mv1 0 ðqu  pÞdZ þ mv1 ρ ðqu  pÞdZ by 850 mm high. A 1.5-mm thick sheet of Teflon was placed on the

Fig. 5. Distribution of measured negative pore water pressure along drain boundary in laboratory testing (details of VP and SV are given in Table 2):
(a) 20 kPa vacuum pressure; (b) 40 kPa vacuum pressure

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 31


bottom of the cell and around the internal periphery to reduce fric- two stages to investigate the effects of unloading and reloading.
tion, and a 50-mm thick layer of sand was placed at the top to The suction at the piezometers close to the PVD decreased to ap-
represent the membrane system explained previously. Selected proximately 80 kPa at the bottom of the drain.
geotechnical properties of a typical specimen are shown in Table 1. The settlement of the soil predicted by the membrane system
An average vacuum of 100 kPa was applied to the PVD and the model [based on Eq. (21)] and the laboratory data are plotted in
surface of the soil through a hole in the center of the rigid piston. A Fig. 3, in which it can be seen that the model predicted the settle-
subsequent surcharge load of 50 and 100 kPa was applied instanta- ment quite well.
neously in two stages, with a 14-day gap between each stage. Over
Membraneless System
this 28-day period, the vacuum was released for short periods in
In order to simulate a membraneless system, another large-scale
consolidometer was modified. Apart from no sand at the top drain-
age layer, every other aspect was the same as the membrane system
test described previously. A series of three different large-scale
tests were conducted to investigate vacuum preloading, including
vacuum preloading (VP), and combined surcharge and vacuum (SV).
After placing the soft saturated clay (unit weight 18:1 kN=m3 )
in the chamber, an initial consolidation pressure (pc ) was applied,
and then a PVD was installed. The sample preparations and testing
procedures were previously described by Indraratna et al. (2005a),
and Table 2 summarizes them in detail. Six saturated pore pressure
transducers were manually inserted at various locations, as shown
in Fig. 4. Fig. 5 shows the pore pressures measured along the drain,
where it was noted that the vacuum propagated immediately before
decreasing along its length. The loss of vacuum at the bottom of the
drain was approximately 15–20% of that which was applied to the
surface. Based on the oedometer testing of soil samples obtained
from different radii, the permeability was 3:6 × 1011 m=s in the
smear zone and 9:1 × 1011 m=s in the undisturbed zone.
To investigate the extent of the smear zone, small specimens
were collected along Sections A-A and B-B within the unit cell,
both vertically and horizontally, for tests SV1 and VP1 [Fig. 6(b)].
Indraratna and Redana (1998) and Sathananthan et al. (2008) sug-
gested that a dimensionless ratio k h =k v is a proper method for de-
termining the smear zone because it minimizes the error that occurs
when determining the coefficient of permeability in any direction.
In other words, the method of obtaining kh and kv is not critical as
long as the same laboratory approach is made consistent for both kh
and kv , thus the kh =kv ratio. Fig. 7 illustrates the ratio kh =kv along
the radii of the unit cell and smear zone boundary, determined at
the preconsolidation pressure of 40 kPa, by a square-root fitting
method. Observing a blatant drop of k h =k v approaching the
Fig. 6. Schematic diagrams: (a) section of test equipment showing
PVD, the radius of the smear zone for both tests is assumed to
central drain and associated smear; (b) locations of specimens obtained
be 80 and 120 mm for Sections A-A and B-B, respectively. The
to determine the smear zone by employing permeability characteristics
variation of kh =kv from both tests, along the radial distance from

Fig. 7. Ratio of kh =kv along the radial distance from central drain (details of VP1 and SV1 are given in Table 2)

32 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012


the drain, were similar and independent of the applied vacuum pres- pore pressure than those with only a vacuum (VP1 and VP2). With
sure. These tests also indicate that the smear boundary is elliptical a vacuum pressure surcharge, the final excess pore pressure ap-
rather than circular. However, for simplicity, a circular smear boun- proached the vacuum applied to the boundary of the drain at a given
dary was assumed in this study. The average radius of the smear height. The dissipation rate of excess pore water pressure depends
zone based on an equivalent area was 100 mm, or approximately on the magnitude of the vacuum applied. The PVD allowed
3 times the value of r w , which agrees with the observation of negative pore pressure to generate along the boundary of the drain
Indraratna and Redana (1998) for the same soil. The ratio of aver- but this maximum excess pore pressure can be reduced with the
age permeability in the undisturbed zone to the smear zone (kh =ks ) application of vacuum.
of 1.3 was obtained from both tests. The surface settlement time curves associated with the vacuum
Figs. 8–11 compare the calculated [Eq. (19)] and measured and surcharge load are shown in Fig. 12 [Eq. (21)]. The final degree
average excess pore water pressure. The excess pore pressures were of consolidation for each test was approximately 96%, clearly
determined by excluding the hydrostatic pore water pressure from showing that when the vacuum and surcharge load are combined,
the pore pressure at the measurement heights, whereas the calcu- the rate of settlement and ultimate settlement increase. The appli-
lated excess pore water was predicted by Eq. (19). These excess cation of a vacuum increases the lateral pore pressure gradient and
pore water pressures agree with the predicted results. As expected, promotes radial flow. This accelerated consolidation increases the
the tests simulating surcharge loading (SV1 and SV2; the number 1 rate of settlement and ultimate settlement, which is analogous to
and 2 refer to test 1 and test 2) experienced a much higher excess increasing the applied surcharge load.

Fig. 8. Comparison between measured and calculated excess pore Fig. 9. Comparison between measured and calculated excess pore
water pressure dissipation for SV1 (details in Table 2): (a) 0.79 m from water pressure dissipation for SV2 (details in Table 2): (a) 0.79 m from
bottom; (b) 0.47 m from bottom; (c) 0.15 m from bottom bottom; (b) 0.47 m from bottom; (c) 0.15 m from bottom

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 33


Comparison with Previous Study predict the negative excess pore pressure distributions attributed to
vacuum application (i.e., plots 1 and 2 are always positive). If an
The proposed model has been compared with the consolidation
equivalent surcharge load technique is used to represent vacuum
settlement, and associated excess pore pressure with solutions pro- pressure, it can be seen that the Olson (1977) solution overpredicts
posed by Olsen (1977) and Indraratna et al. (2005a). The param- the excess pore water pressure (plot 2 in Fig. 13).
eters were assumed as follows: d e ¼ 1:5 m, d w ¼ 0:1 m, ch ¼ Indraratna et al. (2005a) presents an analytical solution for the
0:01 m2 =day, cv ¼ 0:5ch , tc ¼ 30 day, H ¼ 10 m, d s =d w ¼ 2:0, radial consolidation of a single vertical drain with vacuum pressure
and kh =kw ¼ 106 . In Olsen (1977), the vertical flow and radial and instantaneous surcharge loading. This solution is therefore
flow were calculated separately, whereas the current method representative of only the membraneless system, in which there
couples both the vertical and the radial flow in a simultaneous is no sand blanket and no surface distribution of suction.
analysis. Fig. 14 compares Indraratna et al. (2005a)’s results with the
Fig. 13 compares the degree of consolidation and the normal- current model results. For this study, a ch value of 0:32 m2 =year
ized excess pore pressure obtained from Olsen (1977) and the cur- obtained from a 1D consolidation test, and a radius of drain influ-
rent solution. The authors’ results agree with Olsen (1977) when ence zone of 750 mm, were employed with an assumed 25% vac-
both the vacuum pressure and smear zone effects are ignored uum pressure loss for the membraneless system. For the membrane
(Fig. 13; plots 1 and 3). It can be seen that the Olson (1977) cannot system, the thickness of the sand layer is assumed to be 1 m; the

Fig. 10. Comparison between measured and calculated excess pore Fig. 11. Comparison between the measured and calculated excess pore
water pressure dissipation for VP1 (details in Table 2): (a) 0.79 m from water pressure dissipation for VP2 (details in Table 2): (a) 0.79 m from
bottom; (b) 0.47 m from bottom; (c) 0.15 m from bottom bottom; (b) 0.47 m from bottom; (c) 0.15 m from bottom

34 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012


permeability of the sand layer is assumed to be 100 times less than Application to Case Histories
the PVD’s permeability. No vacuum loss is considered along the
drain. The current results for the membraneless system considering Membrane System
instantaneous loading (i.e., plot 2 in Fig. 14) are in agreement with
the previous analysis (i.e., plot 1 of Fig. 14) by Indraratna et al. Tianjin Port is approximately 100 km from Beijing, as reported by
Chu and Yan (2005). At this site, the soft, muddy clay beneath the
(2005a). However, if a commonly employed ramp loading for a
reclaimed soil was approximately 5 m deep, below which was a
membrane system is considered (i.e., plot 5 of Fig. 14), realistic
soft, muddy layer of clay from 8.5–16 m deep. A 6-m thick layer
results can only be obtained from the current authors’ solution.
of stiff, silty clay underlay the soft, muddy clay. A profile of the soil
and its related properties is in Fig. 15.
The storage facility was approximately 7;500 m2 and, since the
undrained shear strength of the topsoil was very low, a combined
vacuum and fill surcharge preloading method was chosen to improve
the soil. A preloading pressure of approximately 140 kPa, and a
nominal vacuum of 80 kPa, was required to achieve the desired
settlement. Settlement gauges were placed at various depths to mea-
sure differential subsurface settlements. Pore water-pressure trans-
ducers were installed under the test embankment to a maximum
depth of 16 m at 3-m deep intervals. Twenty meter long PVDs
(100 mm × 3 mm) were placed in a square pattern, 1 m apart, in
all three sections. A 0.3-m sand blanket beneath the membrane
served as a platform for installing the PVDs and placing the hori-
zontal perforated pipes required to distribute the vacuum. The prop-
erties of the drain are shown in Table 3.
The site for the oil storage station covered a total area of approx-
imately 50,000 m2 . For the purpose of soil improvement, the site
was divided into two sections: Area I of 30,000 m2 and Area II of
20,000 m2 . Using the method introduced previously, Fig. 16 com-
Fig. 12. Comparison between measured and calculated settlement
(details in Table 2)
pares the predicted and field data of the settlement, together
with the loading history for Area I. The predicted results were

Fig. 13. Comparison between Olsen (1977) and the authors’ Fig. 14. Comparison between Indraratna et al. (2005a) and the authors’
solution: (a) normalized excess pore pressure; (b) average degree of solution: (a) normalized excess pore pressure; (b) average
consolidation degree of consolidation

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 35


Fig. 15. Soil properties and profile at Tianjin port [adapted from Yan and Chu (2003)]

determined based on Eq. (21). The predicted settlements are shown Table 3. Vertical Drain Parameters
to agree with the field data. A comparison between the predicted Parameters Value
and measured excess pore water pressure variation, over time and at
different depths of Section II is illustrated in Fig. 17 based on Spacing, S 1.0 m (square)
Eq. (21), using the same method. Again, agreement verified the Length of vertical drain 20 m
accuracy of the model. Dimension of drain 100 × 3 mm2
Discharge capacity, qw 100 mm3 =year (per drain)
Membraneless System Dimension of mandrel 120 × 50 mm2
An application of the membraneless system with PVDs on re-
claimed land near Hwaseong-si, Gyeonggi-do, Korea, is presented
in this study (Kim et al. 2009). Most of the soft, reclaimed clay at
this site consists of marine (estuarine) deposits dredged from the
coast line. The soft ground was approximately 10-m thick, so only
a vacuum was applied. The applied suction was in 3 stages at 20 kPa
intervals, from 20 to 80 kPa. The vacuum at each stage is shown
in Fig. 18; it was 90 and 70% of the vacuum applied at depths of
1.5 m and 9.5 m, respectively, along the length of the drain. Fig. 19
compares the field data at each depth and settlement using the ana-
lytical method [based on Eq. (23)], according to changes at each
stage of vacuum. The settlement predicted by this analytical method
was similar to the actual measurement of settlement.

Influence of the Loss of Vacuum and Permeability


of the Sand Blanket on Vacuum Preloading
The assumed parameters were: d w ¼ 0:1 m; d e ¼ 1:0 m; d s ¼
0:3 m; kh1 =k v1 ¼ 1:0; k h1 =k w ¼ 101106 ; k h2 =kv2 ¼ 2:0;
k h2 =k w ¼ 106 ; kh2 =k s2 ¼ 4; Lw ¼ 2 m; H clay ¼ 10–40 m; mv1 ¼
mv2 ¼ 0:25 m2 =MN; qu ¼ 80 kPa; and p ¼ 80 kPa. They are
similar to the typical values obtained from previous case studies
[i.e., Tianjin Port, China (Rujikiatkamjorn et al. 2008)] and the lab-
oratory experiment results (Indraratna et al. 2005b).
Fig. 20 presents a variation of the time-normalized settlement
Fig. 16. Section I (Tianjin Port): (a) loading history; (b) settlements
curves. Normalized settlement may be defined as “settlement for

36 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012


a given system at a given time over ultimate settlement as a result of assumed to be 2 m. Fig. 20 illustrates the effect of the permeability
surcharge preloading alone (st =s∞ )”. If there is no loss of vacuum, of a sand blanket in a membrane system. As expected, when per-
normalized settlement between a membrane system and mem- meability decreases, consolidation takes longer. When the PVDs
braneless system is negligible with increasing PVD length. are short, e.g., 10 m long [Fig. 20(a)], the permeability of the sand
The permeability of a sand blanket in a membrane system con- blanket should not be less than 0.01 times the permeability of the
trols the speed at which a vacuum can propagate from a horizontal PVD and 104 times the permeability of the clay in order to maintain
drainage system to PVDs, and the interface between the bottom of the consolidation time to achieve 90% consolidation. With longer
the sand blanket and the upper layer of clay. A sand blanket gen-
erally varies from 0.3–2 m in thickness, so in this analysis Lw was

Fig. 19. Comparison of measured settlements with analytical results


(Korea)

Fig. 17. Section II (Tianjin Port): (a) loading history; (b) excess pore
water pressure; (c) settlements

Fig. 20. Parametric analysis of normalized settlement-time factor


curves for varying K 2 (for membrane system) and η (for membraneless
Fig. 18. Efficiency of vacuum pressure by elapsed time (Korea) system): (a) clay thickness of 10 m; (b) clay thickness of 40 m

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 37


drains [Fig. 20(b)], the ratios between the permeability of the sand water pressure and qu ¼ the value of the surcharge preloading.
blanket and the PVD, and between the permeability of the sand Excess pore pressure for the membrane system became an applied
blanket and the clay layer, should be more than 0.1 and 105 , vacuum pressure (p) as soon as the sand blanket was assumed to be
respectively. highly permeable [Fig. 21(a)]. Dissipation of excess pore pressure
Any loss of vacuum along the drain in a membraneless system in the clay layer of both systems was similar if there was no loss
directly affects the performance of vacuum preloading. If the vac- of vacuum, but when this was considered, there were significant
uum continues to decrease, so too does the rate of settlement and differences in excess pore pressure as the time factor increased.
final settlement. Varying the vacuum along the drain in a mem-
braneless system was more realistic because the pressure usually
diminishes as the depth increases. These observations were verified Conclusions
in the laboratory and the field. In other words, it is possible that the
vacuum applied at the top of long vertical drains may not propagate PVDs increase the rate of consolidation in soft soils by assisting
toward the drain bottom. faster dissipation of excess pore water pressure through short radial
Three possible scenarios of vacuum observed from the field drainage paths. There are two types of vacuum preloading systems
were analyzed, namely: (a) a membrane system with no loss of vac- commercially available: (a) membrane system with an airtight
uum, (b) a membraneless system with 25% loss of vacuum at the membrane over the drainage layer; and (b) membraneless system
bottom of the drain, and (c) a membraneless system with no loss of in which a vacuum system is connected to an individual PVD. Their
vacuum. The variation of average pore water pressure with depth, effectiveness varies from site to site depending on the types of soil
the time factor, and the degree of consolidation are shown in Fig. 21. treated and the characteristics of the vacuum and drain. The ana-
The vertical axis shows this normalized depth (z=H), where z ¼ lytical solutions of vertical drains incorporating the vacuum pre-
depth and H ¼ thickness of the whole layer. The horizontal loading of both systems (membrane and membraneless) under
axis shows the normalized excess pore pressure, where u ¼ pore
time-dependent surcharge loading were presented in this paper.
The smear zone and well resistance were also considered in both
models. Laboratory and field testing demonstrate that the proposed
solution for the membraneless system also includes a loss of vac-
uum along the length of the drain. The general solutions for pore
water pressure, settlement, and the degree of consolidation are
derived by applying the Laplace transform technique. Using the
Laplace transform, the time-dependent surcharge preloading can
simply be considered instead of using instantaneously applied con-
stant loading, which cannot simulate the construction time of the
embankment and the time changing of the applied vacuum pres-
sure. The formulations were applied to several case histories, from
the muddy clay region in China, from marine (estuarine) deposits
coast line in Korea, and laboratory tests results. The results verify
the proposed model for determining the permeability of the sand
blanket in a membrane system and the vacuum loss in a membrane-
less system. The permeability of the sand blanket was found to sig-
nificantly affect consolidation in a membrane system. The
permeability ratio of the PVD and sand blanket of short PVDs
(< 10 m) should be less than 100 in order to minimize its effect
on consolidation (the degree of consolidation, U, up to 90%),
but this ratio for deeper, soft clay (longer PVDs) should be less
than 10. If there is no vacuum loss, there will be negligible time
differences for the two systems to reach 90% consolidation. How-
ever, it has been verified in the laboratory and the field that the
vacuum applied at the top of long vertical drains may not propagate
toward the drain bottom in the membraneless system. For deep, soft
clay, a membrane system with a permeable sand blanket layer, with
permeability in the range shown previously, will better achieve the
designed settlement and pore water pressure in the given time.

Appendix I. Detailed Derivation of Excess Pore


Water Pressure

Membrane System
Fig. 21. Parametric analysis of average pore water pressure change After integrating Eqs. (2), (6), and (7) with respect to the radial
with depth: (a) membrane system; (b) membraneless system without distance (r), and considering the boundary conditions, the continu-
vacuum loss; (c) membraneless system with 25% vacuum loss ity equation may be expressed by

38 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012


  2
∂ 4 uwi ∂ 3 uwi 2 kvi 2 ∂ uwi The solutions for Eqs. (26)–(29) are:
cvi   chi 2 1 þ ðn  1Þ
∂z 4
∂z ∂t
2
re F ai kw ∂z2
 
2 k hi ∂uwi dq ^uw1 ðZ; SÞ ¼ X 1 ea1 Z þ X 2 ea1 Z þ X 3 ea2 Z þ X 4 ea2 Z þ QðSÞ
^ ð30Þ
þ ðn2  1Þ 2  ¼0
r e F ai k vi ∂t dt
ði ¼ 1; 2Þ ð24Þ

and ^uw2 ðZ; SÞ ¼ Y 1 eb1 Z þ Y 2 eb1 Z þ Y 3 eb2 Z þ Y 4 eb2 Z þ QðSÞ


^ ð31Þ

∂ 2 uwi 2 khi
¼ ðn2  1Þ 2 ðui  uwi Þ ð25Þ
∂z2 r e F ai k w
   
a2 a2
^u1 ðZ; SÞ ¼ X 1 1  1 ea1 Z þ X 2 1  1 ea1 Z
where F a1 ¼ ðln n  3 n2
4Þ n2 1
þ  1
n2 1
ð1 1
4n2
Þ B2 B2
     
n k 3 n2 a22 a2 Z a2
F a2 ¼ ln þ h2 ln m  þ X3 1  e þ X 4 1  2 ea2 Z þ QðSÞ
^ ð32Þ
m ks2 4 n 1
2 B2 B2
2
    
m kh2 m2 kh2 1 1
þ 2 1 1 2 þ 1 2 ;
n 1 k s2 4n k s2 n2  1 4n
   
r r k b2 b2
n¼ em¼ s; s ¼ r s =rw ; cvi ¼ vi ; ^u2 ðZ; SÞ ¼ Y 1 1  1 eb1 Z þ Y 2 1  1 eb1 Z
rw rw mvi γw B4 B4
k  2
  
chi ¼ hi : b2 b 2 Z b22 b2 Z ^
mvi γw þ Y3 1  e þ Y4 1  e þ QðSÞ ð33Þ
B4 B4
Let K 1 ¼ kkh1v1 ; K 2 ¼ kkh1w ; K 3 ¼ kkh2v2 ; K 4 ¼ kkh2w ; K 5 ¼ kkh2s2 ; h2 ¼ dHw ;
where
T h2 ¼ cde 2 ; C ¼ c ; Z ¼ H; ρ ¼ H .
h2 ·t cv1 z Lw
v2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 H 2
8K 1 h22 ðΘ=C þ B1 þ B2 Þ þ ðΘ=C þ B1 þ B2 Þ2  4ðΘ=CÞB2
B1 ¼ ch1 ¼ ; a1 ¼ ;
r 2e F a1 cv1 F a1 n2 2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kv 2 8h2 ðn2  1ÞK 2
2 H2 ðΘ=C þ B1 þ B2 Þ  ðΘ=C þ B1 þ B2 Þ2  4ðΘ=CÞB2
B2 ¼ ðn  1Þch 2 ¼ 2 ; a2 ¼ ;
kw r e F a1 cv F a1 n2 2
2 H 2
8K 3 h22 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B3 ¼ ch2 ¼ ; ðΘ þ B3 þ B4 Þ þ ðΘ þ B3 þ B4 Þ2  4ΘB4
r 2e F a2 cv2 F a2 n2 b1 ¼ ;
2
2 H 8h2 ðn2  1ÞK 4
2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B4 ¼ ðn2  1Þch2 2 ¼ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r e F a2 cv2 F a2 n2 ðΘ þ B3 þ B4 Þ  ðΘ þ B3 þ B4 Þ2  4ΘB4
b2 ¼ ;
By using the Laplace transform technique, the solution for 2
Eqs. (24) and (25) are given by SK 3 h22
Θ¼ :
  2 n2
∂ 4 ^uw1 SK 3 h22 ∂ ^uw1
 þ B1 þ B2
∂Z 4 2
n C ∂Z 2 By considering the boundary conditions [Eqs. (10a)–(10f)], the
SK h2 ^ continuity conditions at the interface between the underlying soil
þ B2 23 2 ^uw1  B2 QðSÞ ¼0 ð26Þ and the sand blanket [Eqs. (10g)–(10j)], and the initial condition
n C
[Eq. (10k)], the following matrix obtained from Eqs. (30) and
(33) may be written as:
∂ 2 ^uw1
¼ B2 ð^u1  ^uw1 Þ ð27Þ
∂Z 2 ξ 8×8 ψT ¼ PT ð34Þ

  2 where
∂ 4 ^uw2 SK 3 h22 ∂ ^uw2
 þ B3 þ B4 2 3
∂Z 4 n2 ∂Z 2 1 1 1 1 0 0 0 0
6 2 7
SK 3 h22 ^ 6 a1 a21 a22 a22 0 0 0 0 7
þ B4 ^uw2  B4 QðSÞ ¼0 ð28Þ 6 7
n2 6 0 0 0 0 ξ35 ξ36 ξ37 ξ38 7
6 7
6 7
6 0 0 0 0 ξ45 ξ46 ξ47 ξ48 7
6 7
∂ 2 ^uw2 ξ¼6 7
6 ξ51 ξ52 ξ53 ξ54 ξ55 ξ56 ξ57 ξ58 7
¼ B4 ð^u2  ^uw2 Þ ð29Þ 6 7
∂Z 2 6 7
6 ξ61 ξ62 ξ63 ξ64 ξ65 ξ66 ξ67 ξ68 7
6 7
where ^uw1 ðZ; SÞ, u^w2 ðZ; SÞ, ^u1 ðZ; SÞ, ^u2 ðZ; SÞ, QðSÞ,
^ S = Laplace 6 7
4 ξ71 ξ72 ξ73 ξ74 ξ75 ξ76 ξ77 ξ78 5
transform of uw1 ðZ; T v1 Þ, uw2 ðZ; T v1 Þ, u1 ðZ; T v1 Þ, u2 ðZ; T v1 Þ,
qðT v1 Þ, T h2 . ξ81 ξ82 ξ83 ξ84 ξ85 ξ86 ξ87 ξ88

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 39


ξ35 ¼ b1 eb1 ; ξ36 ¼ b1 eb1 ; ξ37 ¼ b2 eb2 ; The matrix from the Eqs. (37) and (38) are determined by
ξ38 ¼ b2 eb2 ; ξ45 ¼ b31 eb1 ; ξ46 ¼ b31 eb1 ;
ξ 04×4 ψ0T ¼ P0T ð39Þ
ξ47 ¼ b32 eb2 ; ξ48 ¼ b32 eb2 ; ξ51 ¼ ea1 ρ ;
ξ52 ¼ ea1 ρ ; ξ53 ¼ ea2 ρ ; ξ54 ¼ ea2 ρ ; where
ξ55 ¼ eb1 ρ ; ξ56 ¼ eb1 ρ ; ξ57 ¼ eb2 ρ ; 2 3
1 1 1 1
ξ58 ¼ eb2 ρ ; ξ61 ¼ a21 ea1 ρ =B2 ; ξ62 ¼ a21 ea1 ρ =B2 ; 6 2
 
b2
 
b2
 
b2
 7
6 1  Bb1 b1  1  B16 b1 1  B26 b2  1  B26 b2 7
6 7
ξ0 ¼ 6 7
6
ξ63 ¼ a22 ea2 ρ =B2 ; ξ64 ¼ a22 ea2 ρ =B2 ; 6 b eb1 7
4 1 b1 eb1 b2 e b 2 b2 e b2
5
ξ65 ¼ b21 eb1 ρ =B4 ; ξ66 ¼ b21 eb1 ρ =B4 ; b31 eb1 b31 eb1 b31 eb2 b31 eb2
ξ67 ¼ b22 eb2 ρ =B4 ; ξ68 ¼ b22 eb2 ρ =B4 ;
ξ71 ¼ k 2 a1 ea1 ρ ; ξ72 ¼ k2 a1 ea1 ρ ; ξ73 ¼ k 2 a2 ea2 ρ ; ψ 0 ¼ ½ χ1 χ2 χ3 χ4  P0 ¼ ½ P
^  Q;
^ 0; ðη  1ÞP;
^ 0
a2 ρ b1 ρ
ξ74 ¼ k 2 a2 e ; ξ75 ¼ k 1 b1 e ;  
n k 3 n2
ξ76 ¼ k 1 b1 eb1 ρ ; ξ77 ¼ k 1 b2 eb2 ρ ; ξ78 ¼ k 1 b2 eb2 ρ ; Fa ¼ ln þ h2 ln m 
m k s2 4 n2  1
K K     
ξ81 ¼ 2 3 a1 ea1 ρ ð1  a21 =B2 Þ; m2 k m2 k 1 1
K1K4 þ 2 1  h2 1  2 þ h2 2 1 2
n 1 k s2 4n k s2 n  1 4n
K K
ξ82 ¼  2 3 a1 ea1 ρ ð1  a21 =B2 Þ; 2 H2 2 H2
K 1K4 B5 ¼ ch2 ; B6 ¼ ðn2  1Þch2 :
r 2e F a cv2 r 2e F a cv2
K K
ξ83 ¼ 2 3 a2 ea2 ρ ð1  a22 =B2 Þ;
K1K4
K K
ξ84 ¼  2 3 a2 ea2 ρ ð1  a22 =B2 Þ;
K 1K4 Appendix II. Dimensionless Parameters
ξ85 ¼ b1 eb1 ρ ð1  b21 =B4 Þ; ξ86 ¼ b1 eb1 ρ ð1  b21 =B4 Þ;
ξ87 ¼ b2 eb2 ρ ð1  b22 =B4 Þ; re

rw
ξ88 ¼ b2 eb2 ρ ð1  b22 =B4 Þ:
ψ ¼ ½ X1 X2 X3 X4 Y1 Y2 Y3 Y4  m¼
rs
rw
P ¼ ½P
^  Q;
^ 0; 0; 0; 0; 0; 0; 0

The solutions to the excess pore water pressure, uwi (i ¼ 1, 2), s ¼ r s =r w


were obtained using the inverse Laplace transform of Eqs. (30) and
(34) k vi
cvi ¼
Z aþI∞ mvi γw
1
uwi ðZ; SÞ ¼ ^u ðZ; SÞeST dS ði ¼ 1; 2Þ ð35Þ
2πI aI∞ wi k hi
chi ¼
mvi γw
Z aþI∞    
1 3 n2 1 1
ui ðZ; SÞ ¼ ^u ðZ; SÞeST dS ði ¼ 1; 2Þ ð36Þ F a1 ¼ ln n  þ 1 2
2πI aI∞ i 4 n2  1 n2  1 4n
pffiffiffiffiffiffiffi  
where I ¼ 1. The analytical solutions of Eq. (2), (35) and (36) n 3 n2
F a2 ¼ ln þ K 5 ln m 
are obtained using the numerical inversion of the Laplace transform m 4 n 1
2
(Durbin 1974). 2
   
m m2 1 1
þ 2 ð1  K 5 Þ 1  2 þ K 5 2 1 2
Membrane System n 1 4n n 1 4n
The solution to the membraneless system based on Eqs. (5)–(9),    
n 3 n2 m2 m2
which are similar to Eqs. (30)–(33), may be expressed as F a ¼ ln þ K 5 ln m  þ ð1  K 5 Þ 1  2
m 4 n2  1 n2  1 4n
 
^uw ðZ; SÞ ¼ χ1 eb1 Z þ χ2 eb1 Z þ χ3 eb2 Z þ χ4 eb2 Z þ QðSÞ
^ ð37Þ 1 1
þ K5 2 1 2
n 1 4n
    k h1
b2 b2 K1 ¼
^uðZ; SÞ ¼ χ1 1  1 eb1 Z þ χ2 1  1 eb1 Z k v1
B6 B6
 2
  
b b2 k h1
þ χ3 1  2 eb2 Z þ χ4 1  2 eb2 Z þ QðSÞ
^ ð38Þ K2 ¼
B6 B6 kw

40 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012


k h2 Acknowledgments
K3 ¼
kv2
The authors wish to thank the Australia Research Council
k (Australia) for its continuous support.
K 4 ¼ h2
kw

k h2 Notation
K5 ¼
ks2
The following symbols are used in this paper:
L H = thickness of the whole layer;
ρ¼ w i = layer, (i ¼ 1, 2);
H
k hi = horizontal coefficient of permeability of soil;
H kvi = vertical coefficient of permeability of soil;
h2 ¼ kwi = coefficient of permeability of vertical drain;
dw
Lw = thickness of the sand layer;
cv2 · t mvi = coefficient of volume compressibility of soil;
Tv ¼ p = vacuum pressure;
H2
q = time-dependent surcharge preloading;
ch2 · t q0 = initial preloading;
Th ¼
d 2e r = radial coordinate;
r e = radius of influence zone;
cv1
C¼ r s = radius of smear zone;
cv2 r w = radius of vertical drain;
t = time;
z
Z¼  p = degree of consolidation defined by pore water pressure;
U
H  s = degree of consolidation defined by settlement.
U
8K 1 h22 ui = average pore pressure at the same depth;
B1 ¼ usi = pore pressure at any point in the smear zone;
F a1 n2 uni = pore pressure at any point in the natural soil zone;
uwi = excess pore water pressure within vertical drain;
8h22 ðn2  1ÞK 2
B2 ¼ z = vertical coordinate;
F a1 n2 εvi = vertical strain; and
η = decreasing ratio of the vacuum pressure along the drain.
8K 3 h22
B3 ¼ ;
F a2 n2
References
8h22 ðn2  1ÞK 4 Atkinson, M. S., and Eldred, P. J. L. (1981). “Consolidation of soil using
B4 ¼
F a2 n2 vertical drains.” Géotechnique, 31(1), 33–43.
Bergado, D. T., Balasubramaniam, A. S., Fannin, R. J., and Holta,
8K 3 h22 R. D.(2002). “Prefabricated vertical drains (PVDs) in soft Bangkok
B5 ¼ clay: A case study of the new Bangkok International Airport project.”
n2 F a
Can. Geotech. J., 39(2), 304–315.
Bo, M. W. (2004). “Discharge capacity of prefabricated vertical drain
8h22 ðn2  1ÞK 4
B6 ¼ and their field measurements.” Geotext. Geomembr., 22(1–2), 37–48.
F a2 n2 Chu, J., and Yan, S. W. (2005). “Estimation of degree of consolidation for
vacuum preloading projects.” Int. J. Geomech., 5(2), 158–165.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Conte, E., and Troncone, A. (2009). “Radical consolidation with vertical
ðΘ=C þ B1 þ B2 Þ þ ðΘ=C þ B1 þ B2 Þ2  4ðΘ=CÞB2 drains and general time-dependent loading.” Can. Geotech. J., 46(1),
a1 ¼ 25–36.
2
Durbin, F. (1974). “Numerical inversion of Laplace transforms: An efficient
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi improvement to Dubner and Abate’s method.” Comput J. (UK), 17(4),
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 371–376.
ðΘ=C þ B1 þ B2 Þ  ðΘ=C þ B1 þ B2 Þ2  4ðΘ=CÞB2 Fox, P. J., Di, N. M., and Quigley, D. W. (2003). “Piecewise-linear model
a2 ¼
2 for large strain radial consolidation.” J. Geotech. Geoenviron. Eng.,
129(10), 940–950.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ghandeharioon, A., Indraratna, B., and Rujikiatkamjorn, C. (2010).
ðΘ þ B3 þ B4 Þ þ ðΘ þ B3 þ B4 Þ2  4ΘB4 “Analysis of soil disturbance associated with mandrel-driven prefabri-
b1 ¼ cated vertical drains using an elliptical cavity expansion theory.” Int. J.
2 Geomech., 10(2), 53–64.
Hansbo, S., Jamiolkowski, M., and Kok, L. (1981). “Consolidation by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi vertical drains.” Géotechnique, 31(1), 45–66.
ðΘ þ B3 þ B4 Þ  ðΘ þ B3 þ B4 Þ2  4ΘB4 Holtz, R. D., Jamiolkowski, M., Lancellotta, R., and Pedroni, S. (1991).
b2 ¼ Prefabricated vertical drains: Design and performance, Butterworth-
2
Heinemann, London.
Hird, C. C., Pyrah, I. C., and Russell, D. (1992). “Finite element modelling
SK 3 h22 of vertical drains beneath embankments on soft ground.” Géotechnique,
Θ¼
n2 42(3), 499–511.

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012 / 41


Indraratna, B., Bamunawita, C., and Khabbaz, H. (2004). “Numerical Olson, R. E. (1977). “Consolidation under time-dependent loading.”
modeling of vacuum preloading and field applications.” Can. Geotech. J. Geotech. Engrg. Div., 103(1), 55–60.
J., 41(6), 1098–1110. Rujikiatkamjorn, C., Indraratna, B., and Chu, J. (2008). “2D and 3D
Indraratna, B., and Redana, I. W. (1998). “Laboratory determination of numerical modeling of combined surcharge and vacuum preloading
smear zone due to vertical drain installation.” J. Geotech. Geoenviron. with vertical drains.” Int. J. Geomech., 8(2), 144–156.
Eng., 124(2), 180–184. Runesson, K., Hansbo, S., and Wiberg, N. E. (1985). “The efficiency of
Indraratna, B., and Redana, I. W. (2000). “Numerical modeling of vertical partially penetrating vertical drains.” Geotechnique, 35(4), 511–516.
drains with smear and well resistance installed in soft clay.” Can. Sathananthan, I., Indraratna, B., and Rujikiatkamjorn, C. (2008). “Evalu-
Geotech. J., 37(1), 132–145. ation of smear zone extent surrounding mandrel driven vertical drains
Indraratna, B., Rujikiatkamjorn, C., and Sathananthan, I. (2005a). “Analyti- using the cavity expansion theory.” Int. J. Geomech., 8(6), 355–365.
cal and numerical solutions for a single vertical drain including the Shang, J. Q., Tang, M., and Miao, Z. (1998). “Vacuum preloading consoli-
effects of vacuum preloading.” Can. Geotech. J., 42(4), 994–1014. dation of reclaimed land: a case study.” Can. Geotech. J., 35(5),
Indraratna, B., Sathananthan, I., Rujikiatkamjiorn, C., and 740–749.
Balasubramaniam, A. S. (2005b). “Analytical and numerical modeling Tang, X. W., and Onitsuka, K. (2000). “Consolidation by vertical drains
of soft soil stabilized by prefabricated vertical drains incorporating under time-dependent loading.” Int. J. Numer. Anal. Meth. Geomech.,
vacuum preloading.” Int. J. Geomech., 5(2), 114–124. 24(9), 739–751.
Kim, S. S., Kim, Y. Y., Han, S. J., Kim, K. N., and Kang, B. Y. (2009). Walker, R., and Indraratna, B. (2006). “Vertical drain consolidation
“Field application and numerical analysis of suction vertical drain with parabolic distribution of permeability in smear zone.” J. Geotech.
method.” Proc., 17th Int. Conf. on Soil Mechanics and Geotechnical Geoenviron. Eng., 132(7), 937–941.
Engineering, IOS Press, Amsterdam, Netherlands, 2151–2154 Wijeyakulasuriya, V., Hobbs, G., and Brandon, A. (1999). “Some experi-
Leo, C. J. (2004). “Equal strain consolidation by vertical drains.” ences with performance monitoring of embankments on soft clays.”
J. Geotech. Geoenviron. Eng., 130(3), 316–327. Proc., 8th Australia New Zealand Conf. on Geomechnics, Vol. 2,
Mesri, G., Lo, D. O. K., and Feng, T. W. (1994). “Settlement of embank- Australian Geomechanics Society, Hobart, Australia, 783–788.
ments on soft clays.” . Proc., Settlement 94, A. T. Yeung and G. Y. Felio, Yan, S. W., and Chu, J. (2003). “Soil improvement for a road using the
eds., ASCE, New York, 8–65. vacuum preloading method.” Ground Improv., 7(4), 165–172.
Mohamedelhassan, E., and Shang, J. Q. (2002). “Vacuum and surcharge Zhu, G. F., and Yin, J. H. (2000). “Finite element consolidation analysis
combined one-dimensional consolidation of clay soils.” Can. Geotech. of soils with vertical drain.” Int. J. Numer. Anal. Methods Geomech.,
J., 39(5), 1126–1138. 24(4), 337–366.

42 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2012


Copyright of International Journal of Geomechanics is the property of American Society of Civil Engineers and
its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.

You might also like