Professional Documents
Culture Documents
Analytical and Numerical Modeling of Consolidation by Vertical Drain Beneath A Circular Embankment
Analytical and Numerical Modeling of Consolidation by Vertical Drain Beneath A Circular Embankment
Abstract: In the analysis of axisymmetric problems, it is often imperative that aspects of geometry, material properties, and loading
characteristics are either maintained as constants or represented by continuous functions in the circumferential direction. In the case of
radial consolidation beneath a circular embankment by vertical drains 共i.e., circular oil tanks or silos兲, the discrete system of vertical drains
can be substituted by continuous concentric rings of equivalent drain walls. An equivalent value for the coefficient of permeability of the
soil is obtained by matching the degree of consolidation of a unit cell model. A rigorous solution to the continuity equation of radial
drainage towards cylindrical drain walls is presented and verified by comparing its results with the existing unit cell model. The proposed
model is then adopted to analyze the consolidation process by vertical drains at the Skå-Edeby circular test embankment 共Area II兲. The
calculated values of settlement, lateral displacement, and excess pore-water pressure indicate good agreement with the field
measurements.
DOI: 10.1061/共ASCE兲1532-3641共2008兲8:3共199兲
CE Database subject headings: Numerical models; Finite element method; Soil consolidation; Drainage; Sand; Embankment.
Introduction
Uh = 1 − exp − 冉 冊 8Th
共1a兲
Installation of vertical drains is one of the most widely used tech-
niques for improving the engineering characteristics of soft clays. In the above equation
The main function of vertical drain application is to accelerate ch . t
soil consolidation by shortening the drainage path and activating Th = 共1b兲
radial drainage, thereby increasing the shear strength of the soil d2e
冉冊 冉 冊 冉 冊
while reducing its postconstruction settlement 共Holtz et al. 1991兲.
The dissipation of excess pore-water pressure occurs faster in the n kh kh
= ln + ln共s兲 − 0.75 + 共2lz − z2兲 共1c兲
radial direction due to the greater coefficient of soil permeability s ks qw
in the horizontal direction and the reduced drainage path. The
consolidation time can be reduced to achieve a required degree of n = 共de/dw兲 共1d兲
consolidation by selecting a suitable drain spacing and an appro-
priate installation pattern 共Jamiolkowski et al. 1983兲. s = 共ds/dw兲 共1e兲
The theoretical solution of radial drainage consolidation was
first proposed by Barron 共1948兲 based on a unit cell 共i.e., single where Uh⫽average degree of lateral consolidation; Th⫽time fac-
drain surrounding by a soil cylinder兲. Further studies on the unit tor; ch⫽coefficient of consolidation in radial direction; t⫽time;
cell consolidation were conducted by Yoshikuni and Nakanodo de⫽diameter of soil cylinder; ⫽factor accounting for smear
共1974兲 and Hansbo 共1981兲 which lead to the following well- and well resistance; dw⫽equivalent diameter of the drain; ds⫽dia-
known equation: meter of smear zone; l⫽drain length; qw⫽discharge capacity of
the drain; z⫽depth of horizontal plane under consideration; and kh
1 and ks⫽coefficients of horizontal permeability in the smear zone
Professor of Civil Engineering, Faculty of Engineering, Univ. of
Wollongong, Wollongong City, NSW 2522, Australia 共corresponding and undisturbed zone, respectively.
author兲. E-mail: indra@uow.edu.au The unit cell analysis is accurate when applied at the embank-
2
Assistant Professor, Dept. of Civil Engineering, Univ. of Baghdad, ment centerline where the lateral displacements are zero. In
Iraq; presently, Endeavour Research Fellow 共DEST兲, Civil Engineering practice, the subsoil in which hundreds of drains are installed is
Division, Faculty of Engineering, Univ. of Wollongong, Wollongong City, usually not uniform, and the process of consolidation is not al-
NSW 2522, Australia. ways a one-dimensional problem 共Indraratna et al. 1992兲. The
3
Research Associate, Civil Engineering Division, Faculty of numerical techniques including the finite-element method 共FEM兲
Engineering, Univ. of Wollongong, Wollongong City, NSW 2522, are essential for the analysis and design of multidrain systems,
Australia. apart from the effect of variability in soil properties at a site 共Hird
Note. Discussion open until November 1, 2008. Separate discussions
et al. 1992; Indraratna and Redana 2000; Chai et al. 2001;
must be submitted for individual papers. To extend the closing date by
one month, a written request must be filed with the ASCE Managing Indraratna et al. 2005兲. Various studies including Olson 共1998兲
Editor. The manuscript for this paper was submitted for review and pos- have indicated that conducting a finite-element analysis for
sible publication on July 20, 2007; approved on October 9, 2007. This multidrains is often time consuming, and that it is related to the
paper is part of the International Journal of Geomechanics, Vol. 8, No. convergence of a solution that inevitably requires large computer
3, June 1, 2008. ©ASCE, ISSN 1532-3641/2008/3-199–206/$25.00. memory and a large number of iterations. However, the complex-
Mathematical Formulation
7. Across the inner and outer boundary of each revolving prism 9 0.666975
of soil 共Fig. 3兲, no flow is allowed, and for relatively long 10 0.666917
vertical drains, only radial 共horizontal兲 flow is permitted. 11 0.666873
12 0.666840
Development of Mathematical Formulation 13 0.666815
14 0.666794
Considering the inner hollow cylindrical soil wall 共ri − S / 2 ⱕ r
15 0.666778
ⱕ ri兲, the flow rate in the radial direction from the inner imper-
meable boundary to the hollow cylindrical drain wall is expressed 16 0.666764
by Darcy’s law 17 0.666753
18 0.666744
Q kh u 19 0.666736
= A 共2兲
t ␥w r 20 0.666729
kh u where
2r共dz兲 = 共共ri − S/2兲2 − r2兲dz 共4兲
冤 冥
␥w r t i 1
2共i − 0.5兲4 ln + 共2i − 0.5兲共− 2i2 + 3i − 0.75兲+
The excess pore pressure gradient can be derived by rearranging ␣2 i − 0.5 4
冉 冊
ring =
Eq. 共4兲 as i i + 0.5 1
2共i + 0.5兲4 ln − 共2i + 0.5兲共2i2 + 3i + 0.75兲
i 4
u ␥w 共共ri − S/2兲 − r 兲
2 2
= for ri − S/2 ⱕ r ⱕ ri 共5兲 2␣2
r 2kh t r ⬇ 共9b兲
3
Integrating Eq. 共5兲 in the radial direction with the boundary con-
dition 共u = 0 at r = ri兲, the distribution of excess pore pressure u in
S = ␣de 共9c兲
Zone A 共ri − S / 2 ⱕ r ⱕ ri兲 can be expressed by
␣ = 0.887 and 0.952 for drains installed in a square pattern and an
uA =
␥w
2kh t
冉
共ri − S/2兲2 ln
r
冉冊
ri
1
− 共r2 − r2i 兲
2
冊 for ri − S/2 ⱕ r ⱕ ri
equilateral triangular pattern, respectively 共Holtz et al. 1991兲.
It is of interest to note that the value of ring converges to
2␣2 / 3 for all values of 共i ⬎ 1兲 共Table 1兲.
共6兲
Combining Eq. 共9兲 with the well-known compressibility rela-
Similarly, excess pore pressure u in Zone B 共ri ⱕ r ⱕ ri + S / 2兲 is tionship 共 / t = −mv ū / t兲 yields
determined by
␥w ū 1 2
uB =
␥w
2kh t
冉
共ri + S/2兲2 ln
r
冉冊
ri
1
− 共r2 − r2i 兲
2
冊 for ri ⱕ r ⱕ ri + S/2
ūring = −
kh
mv d ring
t 8 e
共10兲
冉 冊
It is noted that subscripts A and B are denoted for Zones A and B,
respectively, and subscript ‘ring’ is denoted for circular loading. ūring − 8Th,ring
= exp 共11兲
The average excess pore pressure 共ū兲 is determined from 1 ring
2 2
␣
kh,ring 3
冋冉冊 册
= 共13兲
kh n kh 3
ln + ln共s兲 −
s ks 4
Model Validation
Fig. 4. 共a兲 Nodes and integration points for single eight-node
Finite-element program ABAQUS 共Hibbitt et al. 2006兲 based on biquadratic displacement, bilinear pore pressure axisymmetric
Biot’s consolidation theory was used with assumed soil properties element; 共b兲 mesh discretization for single hollow cylinder of
to validate the proposed analytical solution. An elastic analysis soil-drain 共not to scale兲
was conducted with mv = 10−3 m2 / kN and simulating the condi-
tion of no lateral strain 共 = 0兲 for validating the proposed solution
关Eq. 共11兲兴. A total of 3,200 axisymmetric elements 共eight-node Application to Case History
biquadratic displacement and bilinear pore pressure兲 were used in
the finite-element model to simulate a 10 m long vertical drain In 1957, the Swedish Geotechnical Institute and the Swedish
with a spacing of 1.5 m 关Figs. 4共a and b兲兴. A ch value of Road Board designed and constructed four circular test embank-
0.32 m2 / year and de of 1.575 m 共i.e., equivalent to S = 1.5 m ments to study consolidation behavior of the soft clay and to
for triangular drain pattern兲 were used. The aspect ratio of obtain information for construction of a new airport 共Hansbo
the finite elements was kept below three. The horizontal undis- 1960兲. The site is located on an island about 25 km west of Stock-
turbed soil permeability 共kh兲 was taken as 10−10 m / s, and the ratio holm, Sweden. Sand drains were installed at three of these test
of the undisturbed permeability to the smear zone permeability embankments. This study deals with circular test embankment
共kh / ks兲 and the value of 共ds / dw兲 ratio were assumed to be 4.0 and 共Area II兲 which was constructed between June and July 1957,
2.0, respectively. The top, bottom, and outer boundaries were set with a base diameter of 35 m, a slope of 1.5H : 1V, and height of
as impermeable 关see Fig. 4共b兲兴. The vertical loading pressure 1.5 m. Gravel of unit weight of 17.9 kN/ m3 was used as the
共1 = 50 kPa兲 was applied at the top of the cell, and only vertical embankment surcharge after removing 0.25 m of the top soil.
displacement was allowed. Rigid elements were selected at the Sand drains of 0.18 m diameter were installed in a triangular
soil surface to ensure the equal strain condition. pattern at a spacing of 1.5 m. The water table was located 1.0 m
Fig. 5 shows a comparison between the analytical and numeri- below the ground surface. The settlement plates, piezometers, and
cal models before and after conversion for i = 2 and 15. It can be horizontal settlement gauges were installed to measure the verti-
seen from Eq. 共11兲 that the degree of consolidation of a particular cal displacement, pore-water pressure, and lateral displacements,
soil-drain wall does not depend on the value of i, hence, for the respectively. The locations of the instrumentations were shown
purpose of clarity, the numerical results for i = 2 and 15 are com- earlier in Fig. 1.
pared separately in Figs. 5共a and b兲. Before incorporating the
equivalent soil permeability from Eq. 共11兲, a good agreement
Site Geology and Geotechnical Properties
between the numerical and analytical results was obtained by
Eq. 共13兲. Small deviations are noted for the range 300⬍ t The site geology and geotechnical properties of soil were de-
⬍ 1,000 days with a maximum error of about 5% in the degree of scribed in detail by Hansbo 共1960兲 and have been presented in
consolidation. It can be seen after incorporating Eq. 共13兲 in the several publications 共Holtz and Broms 1972; Hansbo 2005兲. Only
numerical model, that the consolidation response agrees with a summary is given here. The geologic stratification of the site
Eq. 共1兲. In general, the above matching procedure confirms the area can be described as composed of very recent glacial and
reliability of the proposed procedure after transforming the true postglacial clay deposits. The ground surface is about 2.5 m
field condition to the equivalent axisymmetric drain rings. above the mean level of the Baltic Sea. The deepest sediments
Fig. 5. Comparison of analytical model with equivalent model before and after conversion: 共a兲 i = 2; 共b兲 i = 15
consist of glacial clays, which are about 7,500 years old while the ment was simulated to exploit the symmetry. The relevant soil
upper postglacial soils were deposited within the past 4,500 years. parameters of four subsoil layers are summarized in Table 2 based
In general, these deposits can be described as soft, normally con- on the laboratory test results provided by Hansbo 共1960兲. For the
solidated clay of moderate sensitivity. Site investigations showed topmost overconsolidated crust, the Mohr–Coulomb model was
that the subsoil conditions are relatively uniform throughout the considered appropriate. Based on Hansbo 共1997兲, it was assumed
site, consisting of a weathered crust formed by a cyclic wetting that the diameter of the smear zone 共ds兲 was 0.36 m and that both
process together with natural cementation with a total thickness of the permeability of the smear zone 共ks兲 and the vertical soil
1 m, overlying the soft clay layer extending to about 8 – 10 m permeability 共kv兲 are 0.25 times the horizontal undisturbed soil
below the surface, followed by bedrock or very dense glacial permeability 共kh兲. The well resistance is neglected in the analysis
deposits at a depth of 10– 12 m. The groundwater table is about due to sufficient drainage capacity of the drains ⬎150 m3 / year.
1 m below the ground surface. The bulk unit weight generally For the multidrain analysis, the permeability of soil for each layer
increased from 15 kN/ m3 near the ground surface to 17 kN/ m3 at under the circular embankment loading was calculated based on
the bottom of the soil profile. The compression and consolidation Eq. 共13兲. Embankment surcharge loading of 27 kPa was simu-
characteristics of the clay were determined using a conventional lated by applying linearly incremental vertical loads to the upper
oedometer test. boundary for 30 days, followed by the rest period.
As shown in Fig. 1, surface and subsurface settlement plates,
hydraulic piezometers, a concrete block, and inclinometer were
installed to monitor the embankment behavior. However, due to
mulfunction of the inclinometer, only the surface lateral displace-
ment measured by the concrete block is available 共Holtz and
Broms 1972兲.
Discussion of Results
In this section, the predictions based on the multidrain analysis
under both axisymmetric and plane strain conditions are com-
pared with the field measurements. Eq. 共13兲 was incorporated in
the axisymmetric condition, whereas the equivalent plane strain
approach proposed by Indraratna and Redana 共2000兲 was adopted
for the plane strain condition. Figs. 7共a and b兲 show a compari-
son between the predicted and recorded field settlements: 共a兲 at
the centerline of the embankment at the ground surface; 共b兲 at a
depth of 5 m; and 共c兲 at a depth of 1.5 and 7 m away from the
embankment centerline. It can be seen that the predicted settle-
ments from the axisymmetric conditions using Eq. 共13兲 agree
with the measured results as well as with those predicted by
Hansbo 共1997兲, whereas the plane strain analysis underpredicts
the field results but gives the same ultimate settlement. Undoubt-
edly, the excess pore pressure predictions from both cases at 5 m
depth and at a lateral distance of 0.75 m away from the em-
bankment centerline also agree with the field measurements
关Fig. 8共a兲兴. However, the predicted excess pore pressures from the
plane strain analysis dissipate faster than that under axisymmetric
conditions 关Fig. 8共b兲兴. Fig. 9 illustrates the surface settlement pro-
file after 452 days, which shows that plane strain condition gives
less settlement at the centerline and more heave at the embank-
ment toe.
The comparisons between the measured and predicted lateral
movements at 1.5 m from the embankment toe are shown in
Fig. 10. The predictions from the axisymmetric condition agree
well with the measured results, whereas the plane strain analysis
overpredicts the measurements 关Fig. 10共a兲兴. Due to insufficient
field data, only the predicted lateral displacements are compared
in Fig. 10共b兲. The maximum lateral displacement is observed at
Fig. 7. Settlements: 共a兲 surface at centerline; 共b兲 5 m depth at the middle of the soft clay layer at a depth of 6 m 关Fig. 10共b兲兴.
centerline; and 共c兲 1.5 m depth at 7 m away from centerline The lateral displacement profiles at the ground surface are shown
Fig. 9. Surface settlement profile after 452 days Fig. 11. Lateral displacement profiles at surface after 250 days
in Fig. 11. It is seen that the predictions from the axisymmetric radial drainage towards these equivalent walls was derived by
condition are always less than those of the plane strain. In gen- matching the degree of consolidation of Hansbo’s unit cell model
eral, for a circular embankment improved by vertical drains, the 共Hansbo 1997兲. An equivalent value for the coefficient of hori-
axisymmetric analysis with an appropriate conversion procedure zontal permeability could be obtained to analyze the multidrain
is essential to obtain accurate predictions, in terms of settlement, problem under circular loading.
excess pore pressure, and lateral displacement. A multidrain analysis based on the proposed conversion was
adopted to evaluate the performance of a selected full-scale cir-
cular embankment at Skå-Edeby, Sweden, using the finite-
Conclusions element code, ABAQUS. The effect of smear associated with the
sand drains was also considered in the analysis. Unlike unit cell
The paper presented a new technique to model consolidation by analysis, the settlements, excess pore-water pressures, and lateral
vertical drains below a circular loaded area where the system of movements elsewhere in the embankment were analyzed and
vertical drains in the field was transformed by a series of equiva- compared with the available field data by employing the proposed
lent concentric cylindrical drain walls. A rigorous solution for conversion procedure. Comparisons made with the corresponding
predictions using the proposed conversion and Hansbo’s solution
showed that the predicted results from proposed conversion are
more accurate than those from Hansbo’s solution apart from the
centerline of the embankment. The multidrain analysis under
equivalent axisymmetric condition using the proposed conversion
gives more accurate prediction, whereas the equivalent plane
strain FEM analysis tends to underpredict settlements and excess
pore pressure by approximately 20%. Moreover, lateral displace-
ment prediction under plane strain condition is almost two times
that from the axisymmetric condition. The predictions of soft soil
improved by vertical drains under circular loading are more ac-
curate when using multidrain analysis under equivalent axisym-
metric condition.
Acknowledgments
References