Vous êtes sur la page 1sur 15

Color profile: Disabled

Composite Default screen

349

Consolidation modelling of soils under the test


embankment at Chek Lap Kok International
Airport in Hong Kong using a simplified finite
element method
Guofu Zhu, Jian-Hua Yin, and James Graham

Abstract: This paper models consolidation of the foundation soils under a test embankment at the new Chek Lap Kok
International Airport in Hong Kong. The modelling used a simplified finite element method and material parameters
derived from results in the original site investigation report. Various features that need to be considered in applying the
simplified method are illustrated through this case study. Good predictions of settlement results are obtained. Relatively
large discrepancies in pore-water pressure predictions suggest that the nonlinear nature of hydraulic conductivity needs
to be taken into account when large compressions are likely to occur. Geological conditions are shown to be a key factor in successful modelling of consolidation behaviour.
Key words: consolidation, pore-water pressure, case modelling, finite element method, vertical drains, settlement.
Rsum : Cet article modlise la consolidation des sols de fondation sous un remblai dessai au nouvel aroport international Chek Lap Kok de Hong Kong. La modlisation a utilis une mthode simplifie dlments finis et des paramtres du matriau drivs des rsultats du rapport de linvestigation originale du site. Diverses caractristiques qui
doivent tre considres dans lapplication de la mthode simplifie sont illustres dans cette tude de cas. De bonnes
prdictions des rsultats de tassement ont t obtenues. Des divergences relativement importantes dans les prdictions
des pressions interstitielles portent penser que la nature non linaire de la conductivit hydraulique doit tre prise en
compte lorsque des compressions importantes peuvent vraisemblablement se produire. On montre que les conditions
gologiques sont un facteur cl pour modliser avec succs le comportement en consolidation.
Mots cls : consolidation, pression interstitielle, modlisation de cas, mthode dlments finis, drains verticaux, tassement.
[Traduit par la Rdaction]

Zhu et al.

363

Introduction
Vertical drains are often installed in soft-soil engineering
projects where subsoils consist of fine-grained soils with low
hydraulic conductivity. The intention of the drains is to
shorten the drainage path and hence speed up the consolidation process.
Following derivation of the differential equation by
Rendulic (1935) for one-dimensional (1D) radial dissipation
of excess pore-water pressure, Carrillo (1942) showed that
two-dimensional (2D) flow problems can be uncoupled. As a
result, solutions to vertical and radial consolidation problems
can be combined to give solutions to the entire 2D problem.
Received January 18, 2000. Accepted October 13, 2000.
Published on the NRC Research Press Web site on
April 9, 2001.
G. Zhu and J.-H. Yin. Department of Civil and Structural
Engineering, Hong Kong Polytechnic University, Hung Hom,
Kowloon, Hong Kong, China.
J. Graham.1 Department of Civil and Geological
Engineering, The University of Manitoba, Winnipeg, MB
R3T 5V6, Canada.
1

Corresponding author (e-mail: jgraham@cc.umanitoba.ca).

Can. Geotech. J. 38: 349363 (2001)

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:08 AM

Probably the best known study of this topic was by Barron


(1948). He assumed two types of vertical strains that might
occur in a uniform clay layer: (i) free vertical strain resulting from a uniform distribution of surface load, and
(ii) equal vertical strain resulting from imposing the same
vertical deformation on the entire surface of the clay. Later,
Horne (1964) presented a formal solution to the layered consolidation problem with vertical drains. Yoshikuni and
Nakanodo (1974) gave a rigorous solution taking well resistance into consideration. Olson (1977) obtained an approximate solution for the case of vertical drainage under ramp
loading using the equal strain assumption. Zhu and Yin
(2001) used the free-strain assumption to develop a mathematical solution for consolidation analysis of soil with vertical and horizontal drainage subject to ramped loading.
Simplified solutions were also obtained by other researchers, for example Hansbo (1981), Zeng and Xie (1989), and
Xie et al. (1994). These closed-form solutions cannot conveniently be extended to account for layered systems, timedependent loading, well resistance, variable coefficients of
consolidation, and inelastic stressstrain behaviour.
To overcome these difficulties, some researchers (Hart et
al. 1958; Olson et al. 1974; Atkinson and Elered 1981;
Onoue 1988; Lo 1991) resorted to numerical solutions using

DOI: 10.1139/cgj-38-2-349

2001 NRC Canada

Color profile: Disabled


Composite Default screen

350

finite differences. The finite element method has also been


used to investigate the consolidation behaviour of soils with
vertical drains. Runesson et al. (1985) studied the efficiency
of partially penetrating vertical drains based on an assumption of linear free strains. Bergado (1993) analyzed the effects of smear zone for Bangkok clay using a linear model.
General two- and three-dimensional (2D and 3D) procedures
(Siriwardane and Desai 1981; Selvadurai 1996; Lewis and
Schrefler 1998; Cheng et al. 1998) are well developed, and
various methods of matching the effect of drains under
axisymmetric and plane strain conditions are also available
(Hird et al. 1995; Indraratna and Redana 1997). Some special problems have also been investigated, for example, the
sensitivity of consolidation behaviour to mesh discretization
and the resulting damage of the poroelastic medium
(Mahyari and Selvadurai 1998).
However, application of 2D and 3D finite element procedures for the design of vertical drains may not be practical
because of (i) difficulty in the determination of various model
parameters for 2D and 3D conditions, (ii) the large amount
of computation, and (iii) the frequent occurrence of numerical instability and convergence problems for nonlinear cases.
Recently, Zhu and Yin (2000a) used a general 1D soil
model to develop a simplified finite element (FE) procedure
for 2D consolidation analysis of soils with vertical drains.
Using a 1D soil model has the advantage that the soil parameters needed for the analysis can be easily found using conventional oedometer tests. By comparing results from the
simplified method with those from a fully coupled 3D finite
element analysis, the authors showed that the simplified FE
procedure is efficient and numerically stable.
This paper uses the simplified FE method of Zhu and Yin
(2000a) to produce true predictions of the consolidation of
foundation soils under a previously constructed test embankment at the new Chek Lap Kok International Airport in
Hong Kong. True prediction is here meant to be as much
like class A prediction as possible, although this is in fact an
after-the-fact simulation. All the parameters adopted in the
modelling were suggested in the original site investigation
report except for the coefficients of hydraulic conductivity
for the upper alluvial crust. These were estimated from the
original report.
Lessons from applying the new simplified method lead to
useful conclusions that are described later in the paper. The
following section presents a brief summary of the basic
equations and provides an understanding of how the FE consolidation model was used in this study. More details can be
found in Zhu and Yin (2000a).

Basic equations
As in Barron (1948), the consolidation problem of soils
with vertical drains is simplified to an axisymmetric one, as
shown in Fig. 1. The solution assumes (i) the soil is fully
saturated, (ii) water and soil particles are incompressible,
(iii) Darcys law is valid, (iv) strains are small, and (v) all
compressive strains within the soil mass occur in the vertical
direction. Assumption (v) can be justified as follows. In
most practical applications, vertical drains are installed in a
regular pattern at close spacing in soils where the layers are
approximately horizontal and the surface area is extensive.

Can. Geotech. J. Vol. 38, 2001


Fig. 1. Geometry of the simplified model for consolidation of
soils with vertical drains. D, depth of clay layer; H, depth of
drain; r, radial coordinate; rd, equivalent radius of vertical drains;
re, equivalent radius of influence of vertical drains; z, vertical coordinate.

The thickness D of the soil layer that contains the vertical


drains (length H) is normally much less than the dimensions
in plan. Average strains (and deformations) in soil with vertical drains occur almost exclusively in the vertical direction. Engineers are normally concerned only with the
average settlement (in plan) of soil layers with vertical
drains and give much less attention to differential settlements in the localized area surrounding a vertical drain. The
stressstrain behaviour of the soil can therefore be simplified on average to be 1D. Numerical results from the simplified method and from a fully coupled 3D finite element
analysis demonstrate that the simplification is reasonable
(Zhu and Yin 2000a).
The governing equations for finite element consolidation
modelling in this paper are given in the following sections.
The continuity equation
The continuity equation for axisymmetric consolidation
problems can be written
[1]

q =

q r q r q z v z
+
+
=
=
r
r
z
t
t

where q = (qr , qz)T; qr and qz are the radial and vertical flow
rates, respectively; v and z are the volume and vertical
strains, respectively (positive for compression); r is the radial coordinate; z is the vertical coordinate; and t is time.
The constitutive equations
Zhu and Yin (1999) wrote the general 1D constitutive
model in the form
[2]

z f ( z)
=
+ g ( z, z )
t
t

where z is the vertical effective stress.


For the nonlinear elastic model used in part of this study,
[3]

f = ln

z
V

g = 0
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:09 AM

Color profile: Disabled


Composite Default screen

Zhu et al.

351

where /V is a constant in which is the compression index


(ln-scale) and V is the specific volume.
Most of the analysis described later used the 1D elastic
viscoplastic (EVP) model suggested by Yin and Graham
(1989, 1994):

[4]

f = ln
z

exp z
g =
Vt 0

z


0

where /V is a constant related to elastic compression, where


is the recompression index (ln-scale); /V is a constant related to a reference time line (approximately the normally
consolidated compression line); 0 is a constant in units of
stress locating the position of the reference time line; and
/V and t0 (in units of time) are two constants related to
creep of the soil.
Darcys law
Darcys law for axisymmetric problems can be expressed
as

[5]

qr
r
= K
u
qz

and

r 0

K = w
kz

0
w

where u is the excess pore-water pressure, K is the permeability constant, kr is the radial coefficient of permeability, kz
is the coefficient of permeability in a vertical direction, and
w is the unit weight of water.
Vertical total stress
For simplicity, the vertical total stress is calculated by assuming that shearing stresses on every cylindrical surface
are zero.
Equations [1], [2], and [5] are the governing equations for
the consolidation problem. Although the governing equations are simplified greatly by assuming z = v, these equations are still coupled in the solution. Several iterations are
required to obtain the related solutions for vertical strains
and pore-water pressures (Zhu and Yin 2000a).

Project background
In the 1970s, it was proposed to build a replacement airport for Hong Kong by levelling the islands of Chek Lap
Kok and Lam Chau and reclaiming 600 ha of land from the
sea. At the site, the seawater was up to 10 m deep and the
tidal range was 2 m. Reclamation would involve placing approximately 80 000 000 m3 of fill to a thickness of up to
20 m.
Site investigations (RMP ENCON Ltd. 1982a; Koutsoftas
et al. 1987) for the area revealed that the entire seabed is
covered by soft to very soft, dark grey, plastic, marine clay
(upper marine clay) with pockets of shells. The thickness of
the clay varies considerably over the site, from as little as
1.0 m (or less) to over 15 m, but is generally in the range of
68 m. Examination of undisturbed samples of the upper

marine clay showed no lamination or layering. The clay is


underlain at some locations by loose to medium-dense marine sand up to 3 m thick. Underlying the upper marine clay
(and the irregular sand) is an alluvial stratum (upper alluvial
crust) consisting of interbedded layers of mottled, oxidized,
and discoloured very stiff clay and dense sand. The thickness of this deposit varies in an erratic manner over the site
but rarely exceeds 7.5 m. The third stratum is a light grey to
dark grey medium stiff to stiff clay (lower marine clay)
interbedded with medium-dense sand lenses and occasional
layers of very stiff mottled reddish and brown clay. Geologically, this is believed to have been deposited during a
period of high sea levels. Underlying the lower marine clay
is an alluvial deposit (lower alluvium) consisting primarily
of very dense, coarse to fine sands, grading into a layer of
gravel and cobbles. Occasional clay pockets are encountered
within this deposit and occasionally below the gravel. The
thickness of the lower alluvial deposit ranges from 0 to
10 m. Below the lower alluvial deposit (or the lower marine
deposit where the lower alluvium is absent) is a layer of
completely decomposed granite. The compressibility of this
layer is very small. Experiments that examined the dissipation of excess pore-water pressures demonstrate that it can
be viewed as a free-draining layer.
These soil conditions presented obvious geotechnical difficulties for the development of the new airport. The option
of removing and replacing the soft marine clay would be
very expensive, involving excavation of approximately
37 000 000 m3 of the material and transporting it for disposal about 20 km from the site. The alternative of leaving
the soft clay in place would result in settlements of up to
4 m. The settlements could be expected to extend over many
years because of the high compressibility and low permeability of the deposits. Settlement tolerance after completion
of the airport was limited, because settlements would also
cause unacceptable differential settlements. Reclamation by
simply placing fill over the soft clay might result in the development of mud waves and lead to serious construction
problems that could jeopardize the project.
To assess the feasibility of using vertical drains with fill
placement techniques that would reduce or prevent mudwave formation, an instrumented test embankment (RMP
ENCON Ltd. 1982b; Cheung and Ko 1986; Koutsoftas et al.
1987) was constructed between 1981 and 1983 on the west
shore of Chek Lap Kok Island (Fig. 2). The main test area
was a 100 m 100 m square in plan and was divided into
four quadrants. Alidrains were installed in the northwestern
and northeastern quadrants at 1.5 and 3 m triangular spacing, respectively. The Alidrains were prefabricated band-shaped
vertical drains with width b = 100 mm and thickness t =
7 mm. The bottoms of the drains were positioned at 21 m
PD (Hong Kong principal datum). Displacement sand drains
500 mm in diameter were installed in the southwestern
quadrant at 3 m triangular spacing. The southeastern quadrant was used as a control area, with no additional drains.
The fill and foundation soils were heavily instrumented
(RMP ENCON Ltd. 1982b; Handfelt et al. 1987) to monitor
their performance during and after construction. The instrumentation consisted of pneumatic and hydraulic
piezometers, settlement plates and pipes, subsurface settlement anchors, and other probes.
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:09 AM

Color profile: Disabled


Composite Default screen

352
Fig. 2. Plan of the test embankment at the new Chek Lap Kok
International Airport.

The test fill provided nearly 1D loading conditions. This


permits the formulations presented by Zhu and Yin (2000a)
to be used for this application. The following sections analyze
the consolidation behaviour of foundation soils in the northwestern quadrant of the test fill. The analysis uses soil parameters that were selected from data provided by the original
program of site investigation and laboratory testing. Results
of the analysis are then compared with measured values.

Index and consolidation test results


RMP ENCON Ltd. (1982a) and Koutsoftas et al. (1987)
reported an extensive program to determine physical and engineering properties of the major strata within the limits of
the site. The laboratory program included oedometer tests,
K0-consolidated undrained triaxial compression tests, K0consolidated undrained direct simple shear tests, unconsolidated undrained triaxial compression tests, isotropically consolidated undrained triaxial compression tests, and index
tests. Only the oedometer and index test results are used in
the following analysis of consolidation and settlement.
The oedometer tests were performed using conventional
incremental loading procedures. Generally, small load increments were used in the recompression region and until the
soil had been stressed above its maximum past (preconsolidation) pressure. Most of the tests included unloadreload cycles in the normal (first-time) compression range.
A load increment ratio of 1.0 was generally used, with each
increment being applied for approximately the time required
to achieve 90% consolidation plus 1 h. For some tests, the
loads were left on for 24 h to obtain sufficient data to define
the coefficient of secondary compression more clearly.
A number of special consolidation tests were also performed to evaluate the effects of surcharge on the coefficient
of secondary compression in the upper and lower marine
clays (RMP ENCON Ltd. 1982a; Koutsoftas et al. 1987).
These special tests required four identical test specimens to
be prepared from the same sample tube. The four specimens

Can. Geotech. J. Vol. 38, 2001


Fig. 3. Definition of compressibility parameters.

were loaded using conventional incremental loading procedures to the same value of maximum vertical stress, selected
to be well into the normally consolidated region. At the end
of primary consolidation under the final load increment,
three of the four specimens were unloaded to simulate the
effect of removing various amounts of surcharge. The fourth
specimen served as the control test and was allowed to consolidate further at the applied stress level. The test specimens were monitored for a period of 3 days to collect
secondary compression data. The results allowed evaluation
of the effects of surcharge on material behaviour.
The following section reproduces test results of the seabed
strata identified by the site investigation (RMP ENCON Ltd.
1982a; Koutsoftas et al. 1987). Figure 3 defines the various
compressibility parameters obtained from the consolidation
tests, and Fig. 4 plots results of Atterberg limit tests on a
Casagrande plasticity chart. Essentially all the data fall into
groups parallel to and slightly above the A line of the plasticity chart. This is typical of inorganic marine clays.
Figures 57 show index data and maximum past pressures
versus depth below the mudline for the three cohesive strata
at the site. Natural moisture contents for the upper marine
clay (Fig. 5) are typically at or above the liquid limit, indicating a very soft and (or) sensitive material. The
overconsolidation ratios of the upper marine clay are in the
range 1.52.0. The very stiff upper alluvial crust is overconsolidated (Fig. 6), with maximum past pressures ranging
from 200 to 600 kPa, and usually above 300 kPa. Maximum
past pressures for the lower marine clay (Fig. 7) typically
range from 200 to 400 kPa.
Figure 8 summarizes compression indices (abbreviated to
CI) versus natural water content wn. There is a general trend
in Fig. 8 for the value of CI to increase to about wn = 70%
with increasing water content for all soil types. The soft upper marine clay has the highest values of CI, ranging from
0.3 to 0.5. Figure 9 presents the recompression index (RI)
versus the compression index for the upper and lower marine
clays. The recompression indices for the upper marine clay
are on average 0.07 times the compression indices, and for
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:10 AM

Color profile: Disabled


Composite Default screen

Zhu et al.
Fig. 4. Summary of Atterberg limits data for upper marine clay,
upper alluvial crust, and lower marine clay.

353

posed granite placed above +2 m PD up to +10.8 m PD


could be chosen as 1.9 Mg/m3. Figure 11 shows a simplified
version of the incremental vertical loading (total stress at the
top of the soft marine clay) calculated using these values.
The figure is for the loading of the northwestern quadrant
and has been used for the consolidation analysis in the following sections.

Soil profile and soil parameters


The general geology and sequences of stratification were
described in RMP ENCON Ltd. (1982a) and Koutsoftas el
al. (1987). As indicated earlier (Figs. 57), the subsoils at
the site can generally be classified as consisting of four layers, namely upper marine clay, upper alluvial crust, lower
marine clay, and lower alluvium (Fig. 12). A number of rising- and falling-head permeability tests were performed in
the field during the original site investigation to measure hydraulic conductivity coefficients in the upper and lower marine clay.
For design of the reclamation works for the airport, RMP
ENCON Ltd. (1982a) suggested the soil parameters outlined
in the following paragraphs (see also Table 1). All the parameters (kv, kr, /V, /V, /V, unit weights, and maximum
past pressure) were suggested in the site investigation report
except the coefficients of hydraulic conductivity for the upper alluvial crust. These have been estimated by the authors
from data given in the original site investigation report.

the lower marine clay the recompression indices are on average 0.14 times the compression indices. Coefficients of secondary consolidation C are plotted in Fig. 10. The values
of C increase with increasing natural water content.

Construction of the main test area


To construct the main test embankment, a 2 m thick layer
of hydraulic sand fill was first pumped over the main test
area. Second, the centre of the embankment was filled to elevations above sea level. Third, the vertical drains and instrumentation were installed from the newly formed ground.
Fourth, the central portion was raised to elevation 6.4 m PD
during the period 13 June to 2 July 1982. This completed the
initial construction of the test fill. After approximately
8 months of settlement, the northwestern quadrant was finally raised to 10.8 m PD from 28 December 1982 to 21
January 1983 in the second stage of testing (RMP ENCON
Ltd. 1982b; Cheung and Ko 1986; Koutsoftas et al. 1987).
The 10.8 m PD elevation approximated the highest anticipated loading from the reclamation.
The density t of the fill is very important for accurately
assessing the incremental vertical stress. After careful examination of the test data and other relevant material, Cheung
and Ko (1986) suggested that the saturated unit weight of
the hydraulic fill material and the bulk density of the decom-

Upper marine clay


For the upper marine clay (with plasticity index Ip ranging
from 40 to 65%; Fig. 5), a density value of 1.45 Mg/m3 appears suitable. The soil has low compressibility in the
overconsolidated range, but is highly compressible when the
maximum past pressure (yield stress) is exceeded.
Recompression indices (Fig. 3) range from 0.02 to 0.03. In
the normal consolidation range, the soil is highly compressible, with CI ranging from 0.30 to 0.50. Representative values of 0.025 for RI and 0.40 for CI are considered
appropriate for this soil for design purposes. The upper marine clay exhibits very low secondary compression at
stresses below the maximum past pressure. However, in the
normal consolidation range coefficients of secondary consolidation are typically above C = 1.5% per logarithm cycle
of time. A value of 1.75% per logarithm cycle of time has
been recommended for design. Coefficients of vertical hydraulic conductivity calculated from the consolidation tests
are in the range of 2 109 to 5 109 m/s, with a mean
value of 2.2 109 m/s. The horizontal coefficients of hydraulic conductivity from variable head field permeability
tests range from 3 109 to 5 109 m/s. A value of 4
109 m/s appears to be a suitable average.
Upper alluvial crust
The upper alluvial crust is a medium-plasticity clay with
plasticity index in the range 2035%. A bulk unit density
value of 1.95 Mg/m3 was recommended for design.
Recompression indices RI range from 0.015 to 0.035. In the
normal consolidation region, the compressibility is comparatively small, with CI ranging from 0.10 to 0.20. Representative design values of RI = 0.025 and CI = 0.15 were
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:10 AM

Color profile: Disabled


Composite Default screen

354

Can. Geotech. J. Vol. 38, 2001

Fig. 5. Index results and maximum past pressure profile of upper marine clay. wn, natural water content; wl, liquid limit; wp, plastic
limit.

Fig. 6. Index results and maximum past pressure profile of upper alluvial crust and stiff lenses within lower marine clay.

selected. Coefficients of secondary compression C in the


normal consolidation range are about 0.8%, and this value
appears to be a suitable average. No permeability data were

available for this stratum. The vertical coefficient of hydraulic conductivity was estimated as 6 109 m/s from the
site investigation report (RMP ENCON Ltd. 1982a), and
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:11 AM

Color profile: Disabled


Composite Default screen

Zhu et al.

355

Fig. 7. Index results and maximum past pressure profile of lower marine clay.

Fig. 8. Compression index versus natural water content for upper marine clay, upper alluvial crust, and lower marine clay.

this value was used in the analysis. We selected a value of


12 109 m/s for the horizontal hydraulic conductivity.
Lower marine clay
The lower marine clay is of medium plasticity (25 Ip
40%). A saturated unit density of 1.85 Mg/m3 was recom-

mended for design. In the recompression stress range, compressibility is low, with RI ranging from 0.02 to 0.05. In the
normally consolidated range, the soil is quite compressible,
with CI ranging from 0.20 to 0.35. Representative values of
RI = 0.035 and CI = 0.25 were selected. Coefficients of secondary compression C in the normal consolidation range
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:12 AM

Color profile: Disabled


Composite Default screen

356

Can. Geotech. J. Vol. 38, 2001

Fig. 9. Recompression index versus compression index for upper and lower marine clay.

Fig. 10. Coefficient of secondary consolidation versus natural water content for upper marine clay, upper alluvial crust, and lower marine clay (normal consolidation region).

vary from about 1.7 to 3.0%, and 1.7% has been selected for
the analysis. Coefficients of vertical hydraulic conductivity
from laboratory consolidation tests are typically 2 1010 to
3 1010 m/s, with a mean value of 2.5 1010 m/s. The in
situ horizontal hydraulic conductivities from variable-head
permeability tests are 4 1010 to 8 1010 m/s. A value of
6.2 1010 m/s was selected as a suitable average.
Lower alluvium
In this is very dense layer, the bulk density was chosen as
2.01 Mg/m2. The preconsolidation (yield) pressure of this
layer was larger than the final stresses imposed by the fill.
Therefore, a small recompression index of 0.01 was adopted

for the analysis. The vertical hydraulic conductivity was


chosen as 1 108 m/s, and the horizontal hydraulic conductivity was 2 108 m/s.
Location of instrumentation
Figure 12 shows the elevations of pneumatic piezometers
(PP) and Sondex anchors in the northwestern quadrant.
These instruments were placed at the centre of a triangular
grid (in plan) of the vertical drains (RMP ENCON Ltd.
1982b; Handfelt et al. 1987). The construction drawings
generally gave only approximate locations for the Sondex
rings.
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:13 AM

Color profile: Disabled


Composite Default screen

Zhu et al.

357

Fig. 11. Simplified fill loading.

Fig. 12. Soil profile and instrumentation setup.

Table 1. Modelling parameters.


Parameter

Upper marine clay

Upper alluvium

Lower marine clay

Lower alluvium

Soil model

Yin and Graham


(1994) 1D EVP
model
2.2109
4.0109
1.086102
0.174
7.6103
1
1.45

Yin and Graham


(1994) 1D EVP
model
6.0109
1.2108
1.086102
0.065
3.47103
1
1.95

Yin and Graham


(1994) 1D EVP
model
2.51010
6.21010
1.52102
1.086101
7.38103
1
1.85

Nonlinear
elastic

Vertical coefficient of permeability, kz (m/s)


Radial coefficient of permeability, kr (m/s)
Elastic compression constant, /V
Reference time line constant, /V
Creep constant, /V
Creep constant, t0 (days)
Bulk density, t (Mg/m3)

Soil model and parameters


Analysis of the settlements used the 1D EVP soil model
in eq. [4] (Yin and Graham 1989, 1994) for the top three
layers. The simpler nonlinear elastic model in eq. [3] was
adopted for the lower alluvium. The parameters for these
models are the values originally suggested for the design of
the reclamation site and outlined in preceding paragraphs.
These are listed in Table 1. The parameters kr and kz in Table 1 are the horizontal and vertical hydraulic conductivities,
respectively, and are taken as constants in the analysis. The
parameter t is the bulk density. Although these values are
typical for the whole site, no borehole was located in the immediate location of the test embankment.
Initial and boundary conditions
As boundary conditions, the top surface of the upper marine clay and the bottom surface of the lower alluvium were
treated as free drainage boundaries. The initial stresses and
the modelling of the maximum past pressures used in the
calculations are plotted in Fig. 13. Initial strains were calculated using the methods suggested by Zhu and Yin (2000b).

Vertical drain characteristics and smear


zone
As mentioned earlier, the Alidrains (width b = 100 mm
and thickness t = 7 mm) were arranged in a triangular pat-

1.0108
2.0108

4.343103

2.01

tern with drain spacing of 1.5 m and their tips at 21 m PD.


The equivalent radius of vertical drains, rd, can be determined in several ways (for example, Hansbo 1979; Atkinson
and Elered 1981; Long and Alvaro 1994). It seems that the
formula rd = (b + t)/4 + t/10 suggested by Long and Alvaro
(1994) agrees well with experimental values most closely
and has been adopted in this analysis. On this basis, rd is
27.45 mm. For triangular installation patterns, the equivalent
radius re of influence of the vertical drain is 0.525 times the
drain spacing (Fig. 14). That is, re = 0.525 drain spacing =
0.7875 m (Barron 1948).
Installation of vertical drains creates a region of disturbed
soil, called the smear zone, with outer radius rs around the
drain. Installation procedures that use a mandrel of radius rm
cause the most severe disturbance. Outward displacement of
the soil distorts the adjacent ground. The zone of soil near
the drain is remolded and dragged first downwards and then
upwards as the mandrel is pushed into and then pulled from
the ground. In soft soils where the technique is most useful,
the overall effect is to produce a disturbed soil zone of reduced permeability, reduced preconsolidation pressure, and
increased compressibility (Johnson 1970). The analysis assumed rs to be five times the equivalent radius of the vertical
drain, that is, 137 mm.
A diamond-shaped mandrel with external dimensions of
75 mm and 166 mm was used to install the Alidrains. The
equivalent radius of the mandrel is r m = 63 mm. These
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:13 AM

Color profile: Disabled


Composite Default screen

358

Can. Geotech. J. Vol. 38, 2001

Fig. 13. In situ vertical effective stress and maximum past pressure.

Fig. 14. Geometry of the simplified model for consolidation of soils with a vertical drain. rs, outer radius of the smear zone.

values of rs and rm correspond to a value of rs/rm = 2.2,


within the range proposed by Mesri and Lo (1991).
Results of fully coupled finite element analysis by Zhu
and Yin (2000a) show that vertical effective stresses within
five times the equivalent radius of the vertical drain are
much higher than in other parts of the domain. These higher
stresses will reduce the permeability of the soil near vertical
drains regardless of the installation method. In addition, remolding due to drain installation may reduce the permeability in the smear zone. However, the reduction and size of the
smear zone are still not exactly known. In the following

analysis, the horizontal and vertical hydraulic conductivities


in the smear zone are assumed equal to the vertical hydraulic conductivity of undisturbed soil (Broms 1987). Since the
equivalent cross section of the Alidrain is small and the
drain length is up to 17.1 m, the effect of internal resistance
to water flow in the drain needs to be considered. In the
analysis, the hydraulic conductivity coefficient of the vertical drain is assumed equal to 1.2 m/day (this is essentially
the hydraulic conductivity of clean sand, and since it is
much larger than the hydraulic conductivity of the various
clay layers, the solution is not sensitive to this assumption).
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:14 AM

Color profile: Disabled


Composite Default screen

Zhu et al.

359

Fig. 15. Comparison between measured settlement (points) and computed settlement (lines) using design values.

The deformation behaviour of the Alidrain material allows it


to adopt the recompression stressstrain relationships of the
surrounding soil.

Finite element analysis results and


discussions
The computed settlements at the elevations of the Sondex
settlement gauge anchors are plotted in Fig. 15. Measured
values were reported earlier by Cheung and Ko (1986) and
are also shown in the figure. In the early stages of loading,
the measured values are larger than the computed results.
This may have been caused by shear straining and lateral
movement of the soils from under the fill, particularly in the
very soft upper marine clay. After the final loading stage, the
computed settlements are larger than the measured results.

Taking into consideration the effort here to produce true predictions in the sense defined earlier, and the conservatism in
the design parameters, the results are quite good. A subsequent laboratory program of oedometer tests (Cheung and
Ko 1986) on 17 undisturbed samples of the upper marine
clay close to the location of the test embankment showed a
maximum compression index of 0.333. This is smaller than
the value of 0.40 suggested by RMP ENCON Ltd. (1982a)
from the overall investigation and used in the analysis.
Using a lower value of 0.32 instead of 0.40 for the upper
marine clay in the finite element analysis produced better
agreement (Fig. 16) between computed settlements and measured results. The predicted settlements in the upper layer
are still larger than the measured settlements.
Computed and measured pore-water pressures are shown
in Fig. 17. The computed values were again obtained using
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:14 AM

Color profile: Disabled


Composite Default screen

360

Can. Geotech. J. Vol. 38, 2001

Fig. 16. Comparison between measured settlement (points) and computed settlement (lines) using test embankment site experiment data.

the soil parameters in Table 1 that came from the original


testing for design of the reclamation works. Figure 17 shows
that, although the trends have been modelled well, the computed pore-water pressures (shown as lines) during the initial loading stages are higher than measured results (shown
as symbols). After the final loading stage, however, the
computed pore-water pressures are lower than measured values. In other words, the rates and durations of the pore-water
pressure dissipation and the settlements have not been well
modelled. Using vertical total stress changes determined
from Boussinesq elastic stress distributions produces results
that are almost the same as those from the new modelling.
The simulation is from the start of construction (contract day
136), whereas field measurements only began after the ini-

tial loading stage (after contract day 201). Therefore, there


are no records available for the initial loading as simplified
in the loading curve. When plotted in the figures, there are
only two large increases in measured pore-water pressure,
whereas the predictions suggest there will be three increases.
One reason for the lack of agreement may be nonlinear
changes of hydraulic conductivity with increasing effective
stress and the resulting decreases in void ratio. Hydraulic
conductivities are larger at the beginning of loading and become smaller as consolidation proceeds. The soils in this
study undergo relatively large deformations, and the effects
of nonlinear hydraulic conductivity may be significant. Although the program used for the analysis can take some
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:15 AM

Color profile: Disabled


Composite Default screen

Zhu et al.

361

Fig. 17. Comparison between measured pore pressure (points) and computed pore pressure (lines) using design values.

2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:20 AM

Color profile: Disabled


Composite Default screen

362

account of nonlinear hydraulic conductivity, suitable information was not available from the laboratory program. The
measured pore-water pressures at PP60 near the top of the
upper marine clay during early loading dissipate more
slowly than the predicted values. This suggests that the upper boundary (the mudline) may not be a free drainage
boundary or that some destructuring was taking place. Computed pore-water pressures at PP43 agree well with measured values. Relatively large differences were noted,
however, between computed and measured pore-water pressures at PP42, PP41, PP40, PP35, and PP38.
The hydraulic conductivity at PP35 (located in the lower
marine clay) is examined in more detail in the following
paragraph. The measured pore-water pressure increased
from 15.0 kPa at contract day 213 to 63.4 kPa at contract
day 232. This corresponds to a vertical total stress increment
of 99.8 kPa in the second loading step. The measured porewater pressure increased again from 28.4 kPa at contract day
411 to 107.14 kPa at contract day 437, corresponding to a
vertical total stress increment of 104.5 kPa in the third loading step. From the test arrangement, deformations can be
considered as 1D. To estimate a lower limit of hydraulic
conductivity of the lower marine clay using the measured
value at PP35, it is assumed that the vertical drain is completely free draining. In the second loading step, 51.5% of
the increased excess pore-water pressure dissipated; and in
the third loading step, 24.7% of the increased excess porewater pressure dissipated. The analytical solution developed
recently by Zhu and Yin (2001) suggests that this corresponds to time factors of T = 1.93 for the second loading
step and T = 0.77 for the third loading step. The respective
coefficients of radial consolidation are 0.0815 and
0.0213 m/day for the two loading steps. Substituting the
recompression index RI = 0.035 and the initial effective
stress (105.5 kPa), the radial coefficient of permeability will
be 1.33 109 (m/s) for the second loading step and 3.48
1010 (m/s) for the third loading step. The coefficient of permeability in the second loading step is much larger than that
used in the calculations. It can also be seen that the coefficient of permeability in the third loading step is about 30%
of the value in the second loading step.
Examination of the borehole log obtained during the 1984
testing program described by Cheung and Ko (1986) in the
northwestern quadrant of the test fill indicates a layer of medium-dense, dark grey, fine to medium sand from 21.4 to
21.8 m PD. These elevations place it in the range of the
lower marine clay in the soil profile (Fig. 12) used for the
calculations. Also, a borehole log for the southwestern quadrant exposes a medium-dense, greyish brown, fine to medium sand layer from 11.0 to 11.9 m PD. These elevations
are in the upper alluvium crust. It is likely therefore that the
hydraulic conductivities and drainage boundaries at the site
may be rather different from those determined from the original investigation and used in the modelling.
Although it appears that settlement magnitudes can be
predicted with some success, attempts to compare predicted
and measured excess pore-water pressures under embankment projects have generally been less successful. Excess
pore-water pressures vary rapidly in the horizontal direction
around sand drains and wicks and can produce changes in
hydraulic conductivity. As a result, relatively small variation

Can. Geotech. J. Vol. 38, 2001

in the positions of the wicks or piezometers can lead to large


potential errors (Olson 1998).
The compression of the lower three layers in Fig. 12 is
relatively small compared with that of the upper marine clay.
Thus, the relatively large discrepancies in the modelling of
pore-water pressure in the lower three layers are of little importance for the prediction of total settlements of the reclaimed land.
Validation of the simplified finite element procedure
through modelling of a case study has proved very helpful in
developing confidence in the ability of the model. Case history projects of this nature are invaluable in providing factual results for developing models, validating the
assumptions, and learning how to use them successfully in
practical applications.

Conclusions
This paper presents a simulation of settlements of a test
embankment at the new Chek Lap Kok International Airport
in Hong Kong. The modelling was done using the finite element method suggested in Zhu and Yin (2000a). The predictions were done using, for the most part, results from the
original laboratory testing and site investigation report. Good
results were obtained for predictions of settlement magnitudes. However, relatively large discrepancies were encountered in modelling pore-water pressures. This has been
related to nonlinear characteristics of hydraulic conductivity
that should be taken into account when the soil experiences
large compression. Localized coarser layers in the geological
sequence are also key factors that influence the possibility of
successful modelling of consolidation behaviour.

Acknowledgements
Financial support (Grant No. H-ZJ73) from the Hong
Kong Polytechnic University, a research grant (Grant No.
PolyU 63/96E) from the Research Grants Council of UGC
of the Hong Kong SAR Government of China, and support
from the Natural Sciences and Engineering Research
Council of Canada are gratefully acknowledged. We also appreciate permission from the Civil Engineering Office, Civil
Engineering Service Department of the of Hong Kong SAR
Government, to use information about materials and performance at the test embankment site at the Chek Lap Kok International Airport. The authors acknowledge thoughtful and
helpful comments from the reviewers.

References
Atkinson, M.S., and Elered, P.J.L. 1981. Consolidation of soil using vertical drains. Gotechnique, 31(1): 3343.
Barron, R.A. 1948. Consolidation of fine-grained soils by drain
wells. Transactions of the American Society of Civil Engineers,
113(2346): 718742.
Bergado, D.T. 1993. Prediction of vertical-band-drain performance
by the finite-element method. Geotextiles and Geomembranes,
12(6): 567586.
Broms, B.B. 1987. Soil improvement methods in Southeast Asia
for soft soils. In Proceedings of the 8th Asian Regional Conference on Soil Mechanics and Foundation Engineering, Kyoto, Japan, Vol. 2, pp. 2964.
2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:21 AM

Color profile: Disabled


Composite Default screen

Zhu et al.
Carrillo, N. 1942. Simple two- and three-dimensional cases in the
theory of consolidation of soils. Journal of Mathematics and
Physics, 21: 15.
Cheng, A.H.-D., Detournay, E., and Abousleiman, Y. 1998.
Poroelasticity. International Journal of Solids and Structures,
Maurice A. Biot Memorial Issue, 35: 45135031.
Cheung, R.K.H., and Ko, S.W.K. 1986. Replacement airport at
Chek Lap Kok civil engineering design studies. Study Report
No. 2B continued monitoring of the test embankment,
Vol. 1, main report; Vol. 2 appendices D, E, and F; Vol 3 appendices H, I, and J. Unpublished Report, Civil Engineering Department, Hong Kong, SAR.
Handfelt, L.D., Koutsoftas, D.C., and Foott, R. 1987. Instrumentation for test fill in Hong Kong. Journal of Geotechnical Engineering, ASCE, 113(GT2): 127146.
Hansbo, S. 1979. Consolidation of clay by band-shaped prefabricated drains. Ground Engineering, 12(5): 1625.
Hansbo, S. 1981. Consolidation of fine-grained soils by prefabricated drains. In Proceedings of the 10th International Conference on Soil Mechanics and Foundation Engineering,
Stockholm, Sweden, Vol. 3, pp. 667682.
Hart, E.G., Kondner, R.L., and Boyer, W.C. 1958. Analysis for partially penetrating sand drains. Journal of the Soil Mechanics and
Foundations Division, ASCE, 84(1812): 115.
Hird, C.C., Pyrah, I.C., Russell, D., and Cinicioglu, F. 1995.
Modelling the effect of vertical drains in two-dimensional finite
element analyses of embankments on soft ground. Canadian
Geotechnical Journal, 32: 795807.
Horne, M.R. 1964. The consolidation of a stratified soil with vertical and horizontal drainage. International Journal of Mechanical
Sciences, 6: 187197.
Indraratna, B., and Redana, I.W. 1997. Plane-strain modelling of
smear effects associated with vertical drains. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE,
123(5): 474478.
Johnson, S.J. 1970. Foundation precompression with vertical sand
drains. Journal of the Soils Mechanics and Foundations Division, ASCE, 96(SM1): 145175.
Koutsoftas, D.C., Foott, R., and Handfelt, L.D. 1987. Geotechnical
investigations offshore Hong Kong. Journal of Geotechnical Engineering, ASCE, 113(GT2): 87105.
Lewis, R.W., and Schrefler, B.A. 1998. The finite element method
in the static and dynamic deformation and consolidation of porous media. Wiley, Chichester.
Lo, D.O.K. 1991. Soil improvement by vertical drains. Ph.D. thesis, University of Illinois at UrbanaChampaign, UrbanaChampaign, Ill.
Long, R.P., and Covo, A. 1994. Equivalent diameter of vertical
drains with an oblong cross section. Journal of Geotechnical Engineering, ASCE, 120(9): 16251629.
Mahyari, A.T., and Selvadurai, A.P.S. 1998. Enhanced consolidation in brittle geomaterials susceptible to damage. Mechanics of
Cohesive Frictional Materials, 3: 291303.
Mesri, G., and Lo, D.O.K. 1991. Field performance of prefabricated vertical drains. In Proceedings of GEO-COAST 91, Yokohama, pp. 231236.
Olson, R.E. 1977. Consolidation under time-dependent loading.
Journal of the Geotechnical Engineering Division, ASCE,
103(GT1): 5560.
Olson, R.E. 1998. Settlement of embankments on soft clays. Journal of Geotechnical and Geoenvironmental Engineering, ASCE,
124(4): 278288.

363
Olson, R.E., Daniel, D.E., and Liu, T.K. 1974. Finite difference
analyses for sand drains problems. In Proceedings of the Conference on Analysis and Design in Geotechnical Engineering,
912 June, University of Texas, Austin, Tex., Vol. 1, pp. 85
110.
Onoue, A. 1988. Consolidation of multilayered anisotropic soils by
vertical drains with well resistance. Soils and Foundations,
28(3): 7590.
Rendulic, L. 1935. Der hydrodynamische spannungsausgleich in
zentral entwasserten tonzylindern. Wasserwirtschaft und technik,
2: 250253.
RMP ENCON Ltd. 1982a. Replacement airport at Chek Lap Kok
civil engineering design studies. Study Report No. 1 site investigation, Vol. 1, main report. Unpublished Report, Civil Engineering Department, Hong Kong, SAR.
RMP ENCON Ltd. 1982b. Replacement airport at Chek Lap Kok
civil engineering design studies. Study Report No, 2 test embankment, Vol. 1, main report. Unpublished Report, Civil Engineering Department, Hong Kong, SAR.
Runesson, K., Hansbo, S., and Wiberg, N.E. 1985. The efficiency
of partially penetrating vertical drains. Gotechnique, 35(4):
511516.
Selvadurai, A.P.S. 1996. Mechanics of poroelastic media. ASME
Special Publication, Kluwer Academic Publishers, Dordrecht,
The Netherlands.
Siriwardane, H.J., and Desai, C.S. 1981. Two numerical schemes
for nonlinear consolidation. International Journal of Numerical
Methods in Engineering, 17: 405426.
Xie, K.H., Lee, P.K.K., and Cheung, Y.K. 1994. Consolidation of a
two-layer system with ideal drains. In Proceedings of the 8th
International Conference on Computer Methods and Advances
in Geomechanics, Morgantown, W.Va., Vol. 1, pp. 789794.
Yin, J.-H., and Graham, J. 1989. Viscouselasticplastic modelling
of one-dimensional time-dependent behaviour of clays. Canadian Geotechnical Journal, 26: 199209.
Yin, J.-H., and Graham, J. 1994. Equivalent times and onedimensional elastic viscoplastic modelling of time-dependent
stressstrain behaviour of clays. Canadian Geotechnical Journal,
31: 4252.
Yoshikuni, H., and Nakanodo, H. 1974. Consolidation of soils by
vertical drain wells with finite permeability. Soils and Foundations, 14(2): 3546.
Zeng, G.X., and Xie, K.H. 1989. New development of the vertical
drain theories. In Proceedings of the 12th International Conference on Soil Mechanics and Foundation Engineering, Rotterdam, The Netherlands, Vol. 2, pp. 14351438.
Zhu, G.F., and Yin, J.-H. 1999. Finite element analysis of consolidation of layered clay soils using an elastic visco-plastic model.
International Journal for Numerical and Analytical Methods in
Geomechanics, 23: 355374.
Zhu, G.F., and Yin, J.-H. 2000a. Finite element analysis of consolidation of soils with vertical drain. International Journal for Numerical and Analytical Methods in Geomechanics, 24: 337366.
Zhu, G.F., and Yin, J.-H. 2000b. Elastic visco-plastic consolidation
modelling of Berthierville test embankment. International Journal for Numerical and Analytical Methods in Geomechanics, 24:
491508.
Zhu, G.F., and Yin, J.-H. 2001. Consolidation of soil with vertical
and horizontal drainage under ramp load. Gotechnique. In
press.

2001 NRC Canada

I:\cgj\Cgj38\Cgj02\T00-103.vp
Wednesday, April 04, 2001 8:16:21 AM

Vous aimerez peut-être aussi