See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268289558

PREDICTION OF CONSOLIDATION BEHAVIOR OF SRI LANKAN PEATY CLAY

USING AN ELASTO-VISCOPLASTIC THEORY

Article

CITATIONS READS

3 285

4 authors, including:

Fusao Oka Sayuri Kimoto

Kyoto University Osaka Sangyo University
225 PUBLICATIONS   3,346 CITATIONS    128 PUBLICATIONS   1,552 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Numerical Analysis on Liquefaction Damage of Embankment Considering Air Bubble Injection as a Countermeasure View project

Proc. 11thIWBDG View project

All content following this page was uploaded by Fusao Oka on 26 August 2015.

The user has requested enhancement of the downloaded file.

PREDICTION OF CONSOLIDATION BEHAVIOR OF SRI LANKAN PEATY CLAY
USING AN ELASTO-VISCOPLASTIC THEORY

Karunawardena, W.A., Oka, F., Kimoto, S.

Graduate School of Engineering, Department of Civil and Earth Resources Engineering, Kyoto University, Kyoto,
Japan
Kulatilaka, S.A.S.
University of Moratuwa, Sri Lanka

ABSTRACT: The Consolidation behavior of Sri Lankan peaty clay is analyzed using the elasto-viscoplastic model. The model
takes into account all the main features involved in the peat consolidation process, namely: finite strain, decreasing permeability,
and the secondary compression effect which is assumed as a continuous process. In addition, the model can accommodate the
structural degradation effect on consolidation. The material parameters required for the analysis and their evaluation procedures
using the standard laboratory and field tests are explained. Initially, the model performance on predicting consolidation behavior of
peaty clay is verified by back analyzing the laboratory consolidation test data, and results indicate that the observed creep behavior
is well captured by the model. Then, the field behaviors of peaty clay due to actual construction works under different field
conditions are simulated. In this context, firstly, the model is used to simulate the field consolidation behavior of a peaty clay layer
which was subjected to one dimensional compression due to a construction of compacted earth fill. Thereafter, consolidation
behaviors of peaty clay due to the construction of two embankments on peaty clay are simulated. One embankment was
constructed without vertical drains and the other was constructed with prefabricated vertical drains in the peaty clay. A comparison
of the predicted results with the field observations has shown good agreement.
Keywords: Amorphous peat; Finite strain; Secondary creep; Structural degradation; Elasto-viscoplastic consolidation analysis.

process, it is important to understand and to

1. INTRODUCTION
quantify the consolidation response of peaty soils in
Soft compressible amorphous peat deposits can be order to employ such techniques and to predict the
found in many places around Sri Lanka, including long-term settlement of structures supported by
the capital city of Colombo and its suburbs. The these soils.
process of urbanization and increasing population
Many researches have extensively investigated the
with associated developments in the country has
consolidation behavior of peaty clays by empirical
compelled utilization of these areas for new housing
and experimental means in the field and laboratory
developments, industrial estates and embankment
conditions. An important conclusion which emerges
for roads and highways. However, constructions
from these studies is that there is a significant
over peaty deposits always create special problems
influence of secondary compression, rapid change
due to the poor engineering properties of peaty soil.
of permeability and the resultant large strain on
The excessive settlements are the major problem
consolidation analysis of peaty clay. The studies
associated with the highly compressible nature of
also reveal that the composition of natural deposits
peat soil. These settlements appear to be quickly
of peat may vary considerably between different
and may continue for a long period of time due to
sites and, in consequence, so do their mechanical
the dominant creep behavior. In addition, low shear
properties [1]. Therefore, it is necessary to identify
strength often causes stability problems in peat soils the type of peat under consideration, as a resulting
when they are subjected to external loading. consolidation behavior may be region-specific.
Therefore, in order to prevent differential There had been only very few research studies on
settlements and subsequent potential failures, it is this subject and they indicated that consolidation
necessary to improve the strength and the stiffness behavior of peaty clay found in Sri Lanka could not
properties of the natural peat deposits using be satisfactorily modeled by the conventional
appropriate ground improvement techniques. In this theory, Barden model and Bjerrun model or by the
1
model based on modified Cam clay theory above process has been occurring for a long time
(e.g.Karunawardena [2], Munasinghe [3], Kethee- and as a result, now there exists around 2500 acres
swaravenayagam [4]). of marshy land underlay with peat in and around
In this paper, the consolidation behavior of Colombo city. Investigations reveal that these
amorphous peaty clay found in Sri Lanka is marshy lands consist of peat layers having a
analyzed using the elasto-viscoplastic model thickness of 5m-12m and yield following basic
proposed by Kimoto and Oka [5]. This model can properties:
describe the secondary behavior as a continuous Table 1. Basic properties of Sri Lankan peat
process and it can also account for the effect of Peat type Amorphous peat
structural degradation on the consolidation analysis. Moisture Content 200% - 800%
An important feature of the analysis in this paper is Initial Void Ratio 2.0-8.0
that it takes into account the main features involved Organic Content 20%-50%
in the peat consolidation process, namely: Specific Gravity 1.5-2.2
secondary compression, finite strain, and changes in PH value Less than 3.0
permeability. The model also considers decreasing Liquid Limit Often less than natural
compressibility at higher stress levels as indicated moisture content (%)
in the experimental results.
In order to make a clear picture of Sri Lankan peat, Preliminary design of structures in and on peats can
this paper begins with a report on basic properties profit from empirical correlations sometimes even
and some useful empirical correlations that exist more so than when dealing with clays due to the
between engineering and index properties of Sri difficulty of obtaining high quality peat samples and
Lankan peaty clay. Then, the constitutive equations the extreme variability of peat deposits. There are
used in the model and the procedure adapted to no well established and widely known correlations
account for the above mentioned features in the between simpler index properties and more difficult
analysis are described. The description of the mechanical parameters for peat [6]. As such, some
material parameters used in the model and their observed correlations for Sri Lankan peaty clays
evaluation procedures using the standard laboratory reported by Karunawardena [7] are presented in
and field tests are explained. Then, the model Fig.1 and Fig.2.
applications to predict consolidation behavior of 5
peaty clay are described. Firstly, the results of the Peat
Compression Index

laboratory consolidation tests are simulated. 4 Cc = 0.007

Secondly, the model is used to simulate the field O.clay
Organic Clay
3
consolidation behavior and compared with
observations in the construction of a compacted 2
earth fill and two embankments on peaty clay, one
1
with prefabricated vertical drains (PVD) and Natural moisture content(%)
another without PVD. 0
0 100 200 300 400 500 600
2. PROPERTIES OF SRI LANKAN PEAT AND
SOME EMPIRICAL CORRELATIONS Fig. 1. Relationship between Cc vs. N.M.C (%)

Sri Lanka lies in the monsoon region of South-East 5

Asia. Colombo, the capital city, situated on the Peat Cc = 0.367 e 0
Compression Index

Western coast of the island, has a humid tropical 4

climate with an average annual rainfall of 2500 mm. O.Clay
Organic Clay
3
In the Colombo region, there exists flood plains of
Kelani Gaga to the North and there are some water 2
bodies such as the Kotte Lake, Bolgoda Lake etc in 1
the coastal belt. The fluctuation of water table due Initial Void Ratio
0
to excessive rainfall provides suitable condition for
0 2 4 6 8 10 12 14
the accumulation of peat deposits in the areas
mentioned, as these are poorly drained grounds. The Fig. 2. Relationship between Cc vs. e0

2
As it is indicated in Fig.1 and Fig.2 some useful swelling index, and the superimposed dot denotes
relationships could be established between the the time differentiation.
compression index and basic index properties of The viscoplastic strain rate tensor is calculated
peat found in Sri Lanka. Also, Fig.3 shows the using the following flow rule which is based on
observed Cα /C c relationship for Sri Lankan peat. Perzyna’s viscoplastic theory:
The pattern agrees well with similar results reported
∂f p
elsewhere but numerical values of the correlations ε&ijvp = γ Φ 1 ( f y ) (4)
are different. This might be due to the lower organic ∂σ ij′
matter content reported in Sri Lankan peat [7].
where Φ1 is a material function indicating the rate
0.16 sensitivity, and f y and f p are the static yield and
the potential functions, respectively, which are in
Coeff. of Secondary
consolidation (C α)

C α /Cc =0.0341
0.12 the shape of the original Cam clay type.
0.08 ⎛ σ′ ⎞
f y = η * + M * ln⎜ m ⎟=0
~
⎜ σ ′( s ) ⎟
(5)
0.04 ⎝ my ⎠

Compression Index (Cc) ~ ⎛ σ′ ⎞

0 f p = η * + M * ln⎜ m ⎟=0 (6)
⎜ σ mp ⎟
0 1 2 3 4 ⎝ ′ ⎠

Fig. 3. Relationship between Cα vs. Cc In the above equations, σ my ′(s ) denotes the mean
effective stress in the static equilibrium state, where
3. ELASTO-VISCOPLASTIC MODEL stress may be reached after an infinite period of
~
time. M * is the failure envelope slope in σ m′ − 2 J 2
The elasto-viscoplastic constitutive model proposed
by Kimoto and Oka [5] is based on Perzyna’s space ( 2J 2 is the second invariant of the deviatoric
viscoplastic theory and the Cambridge elasto-plastic stress tensor). η * is the stress parameter that
theory combined with empirical evidences. The represents the anisotropic consolidation history
important assumption taken in the derivation of the defined by:
model is that: “At the end of consolidation, the state
of the clay does not reach the static equilibrium η * = (ηij* − ηij*( 0) )(ηij* − ηij*( 0) ) (7)
state but is in non equilibrium state” [8]. In the
model, Terzaghi’s effective stress is used as in the where ηij* = S ij / σ m′ , ηij* is the pressure normalized
following form: deviatoric stress tensor and η ij*( 0) is the value for ηij*
σ ij′ = σ ij −u w δ ij (1) before the deformation occurs (the initial state).
In the model, the hardening rule is based on the
where σ ij′ is the effective stress and u w is the pore
overconsolidation (OC) boundary surface which
pressure. The total strain rate tensor consists of the delineates the OC region ( f b < 0 ) from the normally
elastic strain rate tensor ε&ije and the viscoplastic
consolidated (NC) region ( f b ≥ 0 ) and given as
strain rate tensor ε&ijvp :
follows:
ε&ij= ε&ije + ε&ijvp (2)
~ ⎛ σ′ ⎞
f b = η * + M * ln⎜⎜ m ⎟⎟ = 0 (8)
⎝ σ mb
The elastic strain rate tensor is given by a ′ ⎠
generalized Hooke type of law, namely,
1 & κ
in which σ mb′ is the hardening parameter and all the
ε&ije = S ij + σ& m′ δ ij (3) other symbols have the same meaning as previously
2G 3(1 + e0 )σ m′
defined. The overconsolidation boundary
where Sij is the deviatoric stress tensor, σ m′ is the surface, f b , the static yield function, f y , and the
mean effective stress, G is the elastic shear viscoplastic potential function, f p , for η ij* = 0 , are
modulus, e0 is the initial void ratio, κ is the schematically described in the σ ′m - Sij S ij space for
NC region in Fig.4.

3
 residual stresses. Structural parameter β controls
S ij S ij *
the rate of structural changes and z is an
M
accumulation of the second invariant of the
viscoplatic strain rate as follows:
fp = 0
t
Current stress state z = ∫ z&dt , z& = ε&ijvp ε&ijvp (13)
f b= 0 0

fy = 0 Due to this modification, the degradation of the soil

σ m′
structure makes a slower increase in the size of the
σ my
′(s ) σ mb
′ σ mp
′ σ mb
′ surface with an increase in viscoplastic strains
Fig. 4. OC boundary surface, static yield function than that predicted by the original case (Eq.10).
and potential function for normally
consolidated clay (Oka 2005) The same procedure is adopted in calculating σ my ′(s )
in the static yield function (Eq.5), to take into
Based on the experimental results of the strain-rate account the degradation effect. Therefore, static
constant triaxial tests, the material function Φ1 yield function also increases its size in much lower
mentioned in the Eq.(4) is defined as: rate than that of pure volumetric hardening case.
⎧⎪ ⎛ ~ σ m′ ⎞ ⎫⎪ 3.2. Model Features
γΦ 1 ( f y ) = Cσ m′ exp ⎨m ′⎜⎜η * + M * ln ⎟⎟ ⎬ (9)
⎪⎩ ⎝ σ mb
′ ⎠ ⎭⎪
3.2.1. Change of Permeability and Shear Modulus
where m′ and C are viscoplastic parameters related The change of permeability coefficient, k , is taken
to the coefficient of secondary consolidation and to into account using the following empirical
the initial volumetric strain rate, respectively. relationship:
k = k0 exp( e − e0 ) /C k (14)
3.1. Account of Structural Degradation Effect
In the original model [9], the change in size of the where k 0 is the initial value of k at e = e0 and C k is
σ mb
′ surface was controlled only with respect to the material parameter called the permeability
the viscoplastic volumetric strain as follows: change index.
⎛ 1 + e0 vp ⎞ The elastic shear modulus is assumed to be
σ mb
′ = σ mbi
′ exp⎜ εv ⎟ (10) proportional to σ m′ as
⎝ λ −κ ⎠

However, it has been observed that the behavior of σ m′

G = G0 (15)
natural soils is influenced by the degradation of the σ mo
′
microstructure induced by the applied loads [10].
This effect has been incorporated into the model by in which G0 is the value for G when σ m′ = σ mo
′ .
introducing viscoplastic strain softening in addition
to the strain hardening with viscoplastic volumetric 3.2.2. Account of Finite Deformation
strain. This process was explained by Kimoto and In the finite element formulations based on the
Oka [5] as follows: finite deformation theory, an updated Lagrangian
method with the objective Jaumann rate of Cauchy
⎛ 1 + e0 vp ⎞ stress for a weak form of the equilibrium equation is
σ mb
′ =σ ′ maexp⎜ εv ⎟ (11)
⎝ λ −κ ⎠ adopted. An eight node quadrilateral element with
′ + (σ mai
′ −σ ′ maf )exp(− β z )
reduced Gaussian two-point integration is used for
σ ma
′ = σ maf (12)
the displacement. Biot type two-phase mixture
where σ ma′ is assumed to decrease with an theory is applied with a velocity pore pressure
increasing viscoplastic stain, and σ mai
′ and σ ′maf formulation and the pore water pressure is defined
are the initial and the final values for at the four corner nodes of the element. The
σ ′ ma respectively. σ mai
′ corresponds to the complete form of the model and the finite element
consolidation yield stress and σ ′maf is determined formulation based on the finite deformation theory,
from the difference between the peak and the are shown elsewhere [8].

4
3.2.3. Relationship between Viscoplastic However, it has been observed that permeability
Parameter m′ and Cα C c calculated using the above relationship often
Secondary compression is often more significant in underestimates the field value. On the other hand,
peat soils than in inorganic clays, and, therefore, it in-situ horizontal permeability coefficient (kh ) can
should be considered in the peat consolidation be reliably evaluated through in situ tests. Field
analysis. Parameters m′ and the Cα C c concept permeability tests carried on Sri Lankan peaty clay
proposed by Mesri et al. [11] to predict the sec- indicated that the in-situ value for k h is in the order
ondary consolidation, were interrelated by Oka [8] of 10-7m/s. Permeability change index C k can also
as follows: be deduced from the variations in permeability and
the void ratio observed from the above described
λ −κ
α= (16) methods. Also, Mesri et al. [11] reported that for
m ′(1 + e) M *
peat deposits the value of C k = Δe Δlog k v was close
where α is the secondary compression rate given to e 0 4 ,where e0 is the initial void ratio.
by v p = α ln(t / t 0 ) + v 0p under the condition of v p = ε kkvp Viscoplastic parameter m′ is determined using the
that exists during the secondary creep and M * is the relationship mentioned in Eq.(17). It has been
failure envelope slope in σ m′ − 2J 2 space. In terms experienced that C α determined from the long-term
of the coefficient of secondary consolidation, Cα , consolidation tests, provides a more representative
Eq.(16) can be expressed as: value in the case of peaty clay.
Cc − Cs The value of coefficient of the earth pressure at rest
m′ = (17)
Cα M * used here, K 0 ,is based on the reported data by Edil
Often in peat soils C s ≈ 0.1C c and M * ≈ 1 , and the and Dhowian [12]. They pointed out, K 0 calculated
viscoplastic parameter can be approximated as from empirical correlations based on the measured
m ′ = 0.9C c / Cα . Mesri et al. [11] reported that C c / Cα
value of the angle of internal friction of the soil (e.g.
Jaky formula) rarely agrees with the measured value
for amorphous peat varies from 20~25 and that the
of K 0 for peat, and thus it has been suggested that
corresponding m′ becomes 18~23 according to the
the following values be used for engineering
above relationship.
purposes:
4. DETERMINATION OF MODEL Fibrous peat - 0.3 Amorphous peat - 0.53
PARAMETERS Compression yield stress σ mbi ′ is related to the
The material parameters required for the analysis preconsolidation pressure determined through the
are determined as follows: oedometer test. Viscoplastic parameter C ,
degradation parameter β , structural parameter
For the compression index and the swelling index:
σ maf
′ and initial shear modulus G0 are determined
λ = −de / d ln(σ v′ ) = 0.434C C , κ = − de / d ln(σ v′ ) = 0.434C s by the curve fitting of the stress-strain curve of the
undrained triaxial tests.
where CC and C S are the compression and the
recompression index determined from the
oedometer test. Stress-Strain curve
60
The initial permeability coefficient (k0 ) of the soil
Deviator Stress (kPa)

50
can be determined by direct measurement of soil
40
permeability with consolidation test under the
laboratory conditions or initial vertical 30
permeability (kv 0 ) can be calculated using the 20 Experiment
Terzaghi’s one-dimensional theory of consolidation 10 Simulation
using the following relationship: 0
k v = mv cv γ w (18) 0 5 10 15 20
Axial Strain(%)
where c v is the coefficient of consolation, mv is the Fig. 5. Simulation of stress-strain behavior
coefficient of volume compressibility, γ w is the unit
weight of water. For this purpose, the results of the undrained triaxial
tests conducted on peaty clay are numerically

5
simulated using the one element program based on elasto-viscoplastic model can predict the observed
the constitutive relations of the model. The results creep behavior of peaty soil quite well, after the
of the numerical simulation, which was performed completely dissipation of excess pore water
to evaluate the above parameters for the analysis of pressure, as well as during the dissipation.
the field consolidation behavior due to the
Table 3. Parameters used in the model validation
construction of a compacted earth fill on peaty clay,
are shown in Fig.5. Parameters used in the Parameters for the elasto-viscoplastic model
simulation are listed in Table 2. Initial void ratio e0 5.80
Initial vertical effective stress σ 22
′ ( 0) (kPa) 5.0
Table 2. Parameters used for the triaxial simulation
Coeff. of earth pressure at rest K0 0.53
Parameters Values Coeff. of permeability k 0 (m/s) 4 x 10-8
Strain Rate ε& (%/min) 0.11 Change of permeability index Ck 0.95
Initial void ratio e0 5.7, 5.4, 5.0 Initial shear modulus G 0 (kPa) 450
#
Initial elastic shear modulus G 0 (kPa) 800 Compression index λ 0.8207
Compression index λ 1.7163 Swelling index κ 0.08946
Swelling index κ 0.1151 0.98
Stress ratio at critical state M*
Stress ratio at critical state M * 0.95 Compression yield stress σ mbi ′ (σ mai
′ ) (kPa) 3.5
Compression yield stress σ mbi′ (σ mai
′ ) (kPa) 35, 50, 75 Viscoplastic parameter m′ 18
Viscoplastic parameter m′ 22 Viscoplastic parameter C (1/s) 1.0x 10-9
Viscoplastic parameter# C (1/s) 1.0x 10-10 Structural parameter σ maf
′ (kPa) 3.0
Structural parameter #
σ maf
′ (kPa) 0.7 σ mbi
′ Structural parameter β 5
Structural parameter # β 5 Parameters for Terzaghi model
#
from the Triaxial test stress–strain relationship
Coeff. of volume compressibility m v (kN/m2) 0.010
Coeff. of consolidation c v (m2/year) 5.0
5. MODEL VALIDATION
This section demonstrates the capability of the
constitutive model [5] to capture the one 0.001 0.01 0.1 1 10 100
Tme(days)
dimensional consolidation characteristics of peaty 0
clay under laboratory conditions. For this purpose, 5
nd
2 loading
Vertical Strain(%)

data on the medium scale consolidation model test rd

10 3 loading
(275mm in thickness and 203mm in diameter) th
15 4 loading
conducted on remolded peaty clay obtained from Terzaghi model
sites along the Colombo-Katunayaka Expressway 20
Elasto-viscoplastic model
project reported by Kugan et al. [13] was used. The 25
normally consolidated behavior for load increments Experiment
30
of 5-30 kPa (in steps of 5kPa at 2 days time
intervals) was predicted using the Elasto- 35
viscoplastic model and Terzaghi’s model and
Fig. 6. Vertical strain vs. time profile
compared with the measured strains. The material
parameters used in the analysis are listed in Table 3
and it should be noted that viscoplastic parameter C 6. PREDICTION OF FIELD CONSOLIDATION
and structural parameters σ maf
′ and β were assumed, BEHAVIOR DUE TO A CONSTRUCTION
while all the other parameters were based on the OF COMPACTED EARTH FILL
experimental results. As indicated in Fig.6, the
A compacted earth fill was constructed in the Fill
strain predicted by the elasto-viscoplatic model is
Area 2 because of the Madiwela Government
higher than that of Terzaghi’s model at the end of
project in Sri Lanka in the year of 2000. The
each load increment. Moreover, it shows that the
thickness of the peat layer was around 5m and
development of strain ceases in Terzaghi’s model
classified as amorphous type peat. The fill was
after the dissipation of excess pore water pressure
about 2.4 hectares and instrumentations were
whereas the development of strain progresses in
located around the centre of the fill area. Therefore,
slow rate in the elasto-viscoplastic model, as
one dimensional consolidation behavior was
observed in the experiment. This illustrates that the
6
expected in the peat layer and analysis was done degradation effect on the consolidation behavior of
with a finite element mesh consists twenty five peaty clay, a separate analysis was carried out by
elements (size 0.2m x 0.2m). The bottom boundary considering finite strain together with the
was assumed to be perfectly drained due to the degradation effect; the corresponding analysis is
existence of medium dense to dense sandy silt layer referred to as FS & SD.
underneath the peat layer. Investigation revealed
that the compacted fill behaved as an impermeable 6.1. Excess Pore Pressure predictions
layer and therefore top boundary was assumed to be The predicted and the observed excess pore water
impermeable for simulating the field conditions. pressures in the middle of the peat layer (2.5m in
The material parameters used in the analysis were depth) are presented in Fig.7. The loading curve
determined using the laboratory and field tests showing the construction history of the fill is also
carried out during the project period, following the shown in the same figure. It indicates that the field
procedure outlined previously and listed in Table 4. pore water pressure reached around 30kPa at the
completion of the fill due to a load of 46kPa. After
Table 4. Parameters used in the analysis the completion of the fill, the quantity of dissipated
Parameters Values excess pore water pressure was only about 10kPa
Initial void ratio e0 6.28 during the following year.
Initial vertical effective stress σ 22
′ ( 0) (kPa) 10.25#
Coeff. of earth pressure at rest K0 0.53 35
IS
Coeff. of permeability k 0 (m/s) 1.5 x 10-7

Excess Pore Pressure(kPa)

30 FS
Change of permeability index Ck 0.8 FS & SD
25
Initial shear modulus G 0 (kPa) 800 Field Data
Compression index λ 1.7163 20
Swelling index κ 0.1151 15
Stress ratio at Critical state M* 0.95
10
Compression yield stress σ mbi ′ (σ mai
′ ) (kPa) 12.5#
5
Viscoplastic parameter m′ 22 Time(days)
Viscoplastic parameter C (1/s) 1.0x 10-10 0
Structural parameter σ maf
′ (kPa) 0.7 σ mbi
′ 0 200 400 600 800 1000
Structural parameter β 5 Fig. 7. Excess pore water pressure-time profile
#
at the depth of 2.5m
When this behavior is compared with the predicted
The loading curve used in the analysis, which behavior, it shows that the prediction made by the
IS theory overestimates the excess pore water
shows the construction sequence of the fill, is
pressure; in contrast, the FS theory underestimates
shown in Fig.7 together with the excess pore water
the same when it is compared with the observed
pressure-time profile. The stress in the loading
field data. Only the predictions made by FS&SD
curve was calculated as the product of the fill
agree remarkably with the observed field behavior.
thickness and the total unit weight of the fill based
The above results indicate that when predicting
on the field records. Field density tests results excess pore water pressure behavior in highly
indicated that the total unit weight of the fill varied compressible material like peat, it is necessary to
form 15kN/m3 to 18 kN/m3. consider the changes in geometry (decrease in soil
In the analysis, both finite strain (FS) and layer height) in order to account for the correct
infinitesimal strain (IS) computations were made to drainage path. Moreover, it is essential to consider
explore how these two approaches describe the the effect of structural degradation along with other
observed behaviors. In the finite element analysis variables in order to simulate the stagnated pore
considering FS, the method outlined in the section pressure observed after the construction.
(3.2.2) was adopted while in IS case, a four-node
quadrilateral element with reduced Gaussian two- 6.2. Settlement predictions
point integration was used for the displacement, and A comparison of predicted and actual settlements is
the pore water pressure was defined at the center of presented in Fig.8. Field settlement data indicates
each element [5]. In order to investigate the that, the settlement advances even with the

7
existence of stagnated excess pore water pressure in dense sand. The grass mat existing on the original
field. ground surface that gave additional strength to the
subsoil was modeled as a thin elastic element. In the
analysis, the drainage was permitted only at the top
IS
0.25 boundary and excess pore water pressure was
FS
initially assumed to be zero in the subsoil. At the
FS & SD vertical boundaries the displacement was const-
Settlement(m)

0.75 Field Data rained horizontally but allowed to move vertically.

At the bottom boundary movements in both
directions were restrained. Decreasing compress-
1.25
ibility was considered in the peaty clay by varying
the compressibility parameters with the specific
Time(days) volume of the soil. The embankment loading was
1.75
0 200 400 600 800 1000
simulated by applying incremental vertical loads
according to the actual construction sequence of the
Fig. 8. Settlement-time profile beneath the Fill embankments. For this purpose, the total unit
weight of 20kN/m3 and 17.4kN/m3 were assumed
This behavior has been observed in many for the initial sand mat and the subsequent gravel
embankment constructions around the world. Many fill used in the construction respectively.
researches suggest that this behavior occurs due to
the effect of the structural degradation of the 7.1. Embankment analysis on Natural Soil
compressible layer. Kimoto and Oka [5] simulated The trial embankment at Ch.11+950 in the
this behavior numerically by considering the Colombo-Katunayaka Expressway (CKE) project
degradation effect in the consolidation process. A was constructed using the preloading with a
comparison shows that, as expected, the predicted surcharge as the ground improvement technique.
settlement assuming IS gives a higher value than The embankment has a height of 5.69m (with a
that assuming FS, and in this case, the difference is surcharge of 0.94m fill) with a crest width 13.6m
more significant due to the resultant large strain of and 1:2 side slopes. Ketheeswaravenayagam [4]
around 20% in the field. Prediction made by the IS reported that there exists at the surface a highly
theory overestimates the field settlements whereas compressible peat layer, 1m in thickness, followed
FS theory shows qualitative agreement with the by a very loose fine sand layer, up to 2m in depth.
observed data even though it underestimates the Below that, a soft silty sand/clay layer is
actual settlement quantitatively. Similar to the pore encountered. It is 4.5m in thickness and contains
water pressure predictions, FS&SD shows the better pockets of peat. The assumed geotechnical profile
qualitative and quantitative agreement with the considered in analysis is shown in Fig.9. A loading
observed field data for the parameters used in the curve that reflects the construction history of the
analysis. embankment is shown in Fig.10.

0.0 m peat
7. SIMULATION OF FIELD BEHAVIORS DUE very loose sand 8
Fill height (m)

2.0 m
TO EMBANKMENT CONSTRUCTIONS very soft silty 6
gravel fill
6.5 m clay/peat 4
In this section results of the back-analysis of two loose to medium 2 sand fill
trial embankments are presented. One embankment dense sand Time(days)
0
was constructed on natural subsoil and the other 13.5m stiff clayey silt 0 100 200 300 400
15.5m
was constructed on prefabricated vertical drain
(PVD) improved subsoil. Two dimensional elasto- Fig. 9. Subsurface profile Fig. 10. Loading curve
viscoplatic finite element analyses were performed used in analysis
to simulate the observed field behaviors in both
The material parameters required for the analysis
cases. Analyses were carried out by considering were determined based on the data reported by
finite deformation theory together with structural Ketheeswaravenayagam [4] and other reported
degradation effect. In the modeling, elasto- literature relevant to the amorphous peaty clay. The
viscoplastic behavior was assumed in peat, very determined model parameters for the peaty clay are
loose sand, soft silty sand/clay and stiff clayey silt listed in Table 5. It should be noted that soft silty
layers and elastic behavior was assumed in loose to
8
clay/peat layer indicated in the subsurface profile In addition to excessive settlements, the
was modeled in the analysis with the same embankments constructed on peaty clays are
properties of the peaty clay by considering observed subjected to large outward lateral displacements. As
large settlement. This assumption is further indicated in Fig. 12, a large lateral deformation
confirmed by the borehole data taken at Ch.11+900. occurred beneath the toe of the embankment.
Table 5. Parameters used for the peat layer C
0.0
Parameters Values
Initial void ratio e0 6.5 1 2 3 4
Initial vertical effective stress σ 22
′ ( 0) (kPa) 9.0
0.53 15.5m
Coefficient of earth pressure at rest K0
Coefficient of permeability k 0 (m/s) 8.0 x 10-7 0.0 10.0 22.0 41.0 65.0 97.0m
Change of permeability index Ck 1.0 Fig. 12. Deformed mesh after 341 days
Initial elastic shear modulus G 0 (kPa) 800
Compression index λ 1.505
Swelling index κ 0.096
Stress ratio at Critical state M* 0.95
Compression yield stress σ mbi
′ (σ mai′ ) (kPa) 9.0
Viscoplastic parameter m′ 22
Viscoplastic parameter C (1/s) 1.0x 10-9 Fig. 13. Excess pore pressure (kPa) variation during
Structural parameter σ maf
′ (kPa) 0.7 σ mbi
′ construction (after 215 days)
Structural parameter β 5
7.1.2. Verification of the Stability
7.1.1. Prediction of Deformation and Excess Pore This lateral deformation is mainly caused by the
Water Pressure shear stress induced by the embankment load, and if
Fig.11 shows the predicted and the measured the shear stress is large enough, it will cause shear
settlements just below the center of the failure within the subsoil. Therefore, the stability of
embankments. The calculated results show a good an embankment on peaty soil is very important due
agreement with observed data which indicates that to the soft consistency of peat. With regard to this,
the resultant large strain in the field is well captured the effective stress paths were analyzed, followed
by the model. The deformed geometry of the by the different soil elements under the
embankment foundation after 341 days is shown in embankments
Fig. 12 and it shows that the maximum settlement
was not predicted at the center. Similar observations 50
K0
M*
Deviatoric stress (kPa)

have been made in some wide embankments 40 2

constructed on soft soil elsewhere in the world (e.g. 3 1
Indraratna et al. [14]). However, the maximum 30
excess pore water pressure dissipation was always 20 1. under the center
predicted around the center area and calculated 4 2. 10m from center
excess pore water pressure after 250 days (during 10 3. 16m from center
4. 22m -under the toe
construction of embankment) is shown in Fig.13. 0
0 20 40 60 80 100
0.0 Mean effective stress (kPa)
0.5 Predicted Settlement
Fig. 14. Stress path under the embankment
Settlement(m)

1.0 Field Settlement

1.5
As shown in Fig. 14, the stress paths of elements
under the center area move along the K0 path,
2.0
whereas the stress paths of elements under the toe
2.5 area rise towards the critical state line indicating the
3.0 toe area is more prone to plastic instability than the
Time(days) center region under this embankment geometry.
3.5
0 100 200 300 400 This is further illustrated by the distribution of the
Fig. 11. Comparison of settlement prediction accumulated viscoplastic shear strain (ε qvp ) of the
9
embankment foundation and viscoplastic volumetric and the smear zone respectively, k v ~ the vertical
strain (ε vvp ) vs. shear strain (ε qvp ) graphs as shown hydraulic conductivity of the natural soil,
in Fig. 15 and Fig. 16 respectively. In Fig.15, it can l ~ drainage length of the PVD, and q w ~ discharge
be clearly seen that localized strain develops capacity of PVDs.
underneath the toe region of the embankment. And The trial embankment at Ch.11+250 in the CKE
Fig.16 indicates that the dε vvp dε qvp of the element project was constructed on the PVD improved
under the toe area approaches zero during some subsoil. The embankment has a height of 7.99m
stages of the embankment construction. Here, with a crest width of 13.6m and 1:2 side slopes. The
ε qvp = ∫ ( deijvp deijvp )1 / 2 ,where deijvp is the viscoplastic
subsoil condition under the embankment was
shear strain increment tensor. deduced using the data reported by Ketheeswa-
ravenayagam [4]. The PVD was installed 108 days
after the construction started. It was installed to a
depth of around 8 m from the original ground
surface in square pattern with spacing of 1.1m.
The foundation soil was modeled up to 15m depth
below the ground surface and 78m horizontally
Fig. 15. Viscoplastic shear strain variation during from the embankment center line. The finite
construction (after215 days) element mesh indicating the PVD improved zone
together with the subsurface profile relevant to this
0.5
1. under the center 1
2 analysis is shown in Fig.17. Loading curve showing
Viscoplastic Vol. Strain

0.4 2. 10m from center

3
the construction sequence of the embankment is
3. 16m from center
shown in Fig.18. Material parameters used for the
0.3 4. 22m -under the toe
subsoil were the same as listed in Table 5 except the
-7
0.2 e0 , k 0 , λ and κ where value of 5.8, 1x10 m/s, 1.315
4
and 0.087 were used respectively. The parameters
0.1 related to the PVD behavior are listed in Table 6.
Viscoplastic Shear Strain
0
Using these parameters, Eq.(19) yields an
equivalent vertical permeability of 7.87 k v .
0 0.2 0.4 0.6 0.8 1 1.2
Table 6. Drain parameters used in the analysis
Fig. 16. The variation of viscoplastic volumetric
strain vs. viscoplastic shear strain. d w (m) D e (m) d s (m kh / ks k h / k v q w (m3/yr)
0.0515 1.25 0.4 5 1 100
7.2. Embankment analysis on PVD-Improved Soil
Consolidation behavior when PVDs are installed in 7.2.1. Settlement
the peaty clay was predicted by the model using the The predicted settlements at just under the center
method proposed by Chai et al. [15]. It proposed a and the crest (12.5m away from center) of the
method to calculate the equivalent vertical hydraulic embankment are presented together with
conductivity of PVD-improved subsoil, which can representative field measurements in Fig.19. Also,
be used to evaluate the value of k ve , i.e., the mass predictions made for the settlement under the center
vertical hydraulic conductivity of PVD improved of embankment without considering the PVD effect
zone. is shown in the same figure in order to demonstrate
⎛ 2.5l 2 k h ⎞⎟ the effectiveness of the PVD installation to expedite
k ve = ⎜1 + kv (19)
⎜ μDe2 k v ⎟⎠
⎝ the field settlement. The instrumentations were
installed after the construction of sand mat and
⎛n⎞ 2l 2 k h
ln (s ) − + π
kh 3 therefore, the field measurement for the initial
μ = ln ⎜ ⎟ + (20)
⎝s⎠ ks 4 3q w period is not available. According to the field
In the formulas mentioned above, De ~ diameter of measurement, the maximum settlement occurred not
the unit cell (containing a PVD and its improvement under the center but under the crest area of the
area), n ~ De d w ( d w is the diameter of the drain), embankment. This pattern was well predicted by the
s ~ d s d w ( d s is the diameter of the smear zone), k h
model. Predicted settlement just under the center
and k s ~ hydraulic conductivities of the natural soil has also shown a good quantitative agreement with
the measured settlement.
10
 Fill height (m)
8 Loading curve
6
peaty clay 4 gravel fill
PVD-improved zone
very loose silty/clay sand 2 sand fill
15.1m
medium dense clayey silt/sand 0
Time(days)
medium
medium dense
dense to dense sand
silty/clayey sand
0 100 200 300 400
78.0m
Fig. 18. Loading curve
Fig. 17. Finite element mesh showing PVD-improved zone,
subsurface profile 25
Measured-1m depth
Model prediction for settlement under the crest of
20 Predicted -1.25m
the embankment is underestimated about 0.2m from

Excess pore water

pressure(kPa)
the measured values. However, this is acceptable 15
considering the non-homogeneous nature of peaty
soil which has not been considered in the present 10
analysis.
5
0 100 200 300 400 500
Time(days)
0.0 0
Time(days) 0 100 200 300 400 500
0.5
Fig. 20. Comparison of excess pore pressure
1.0
Settlement(m)

center 7.2.3. Lateral Displacements

1.5 measured
predicted
The predicted lateral Lateral displacement(m)
2.0 predicted(nodrain) displacements profiles 0 0.5 1 1.5 2 2.5
crest see are shown in the 0
2.5 measured Fig.21 (dotted lines). 25day
predicted However, correspond- 50day
3.0 2
ding field horizontal
Fig. 19. Comparison of settlements displacement profiles 445day Depth from original G.L(m)
4
form inclinometers are
7.2.2. Excess Pore Water Pressure not available for the 6
The excess pore water pressure predicted 1.25m comparison. The
Lateral deformation
below the embankment and the excess pore water maximum computed 8 corresponding to
measured 1m below the bottom of the embankment lateral displacements surface settlement of
are compared in Fig.20. Since the installations of are located at a depth 10 1.5m

pizometer done after the PVD, the sudden drop of ratio z / H of the with drain

excess pore water pressure indicated in the order of 0.25-0.375. 12 without drain
prediction could not be captured in the field H is the compressible
monitoring data. Also, all sharp peaks of numerical layer thickness, and in 14
predictions that accurately represent the sudden rise this case it is around Fig. 21. Prediction of lateral
of pore pressure associated with loading, do not 8m. displacement under the
embankment toe
show in the field measurements as some dates of the
field measurement do not coincide with the exact It has been reported that vertical drains not only
times at which the maximum peaks are predicted. increase the settlement rate, but also influence the
magnitude of the lateral displacement of subsoil
However, the agreement is generally good for both
[16]. In order to verify this fact, lateral displacement
generation and dissipation of excess pore water
profiles predicted with and without considering
pressures and results could be further improved by
drains corresponding to surface settlement of 1.5m
considering well résistance, different k h / k v ratio or
are presented in Fig.21. It shows that although the
using different discharge capacity for the drains. analysis without drains yielded a smaller settlement,

11
it calculated a larger maximum lateral displacement REFERENCE
than that of PVD case. 1. Berry, P.L. & Poskitt, T.J. (1972) The consolidation of
peat, Geotechnique, 22(1), 27-52.
8. CONCLUSION
2. Karunawardena, W.A., Kulatilaka, S.A.S. (2003) Field
This paper presented a detailed study on the Monitoring of a Fill on Peaty Clay and its Modeling, In
prediction of the consolidation behavior of peaty Proc. of 12th Asian Regional conf. on Soil Mechanics
& Geotechnical Eng., Singapore, 4-8 August 2003, eds.
clay found in Sri Lanka using the elasto-viscoplastic Lung et al.,159-162,World Scientific Publishing Co.
model [5]. In the model, the secondary compression
3. Munasinghe, W.G.S. (2001) Method for improvement
was taken into account by viscoplastic parameter of engineering properties of peat –A comparative study,
m′ , which is related to the C α / C c ratio of the soil Master Thesis, University of Moratuwa, Sri Lanka.
and the value of permeability was varied with the 4. Ketheeswaravenayagam, R. (2006) Finite Element
void ratio throughout the consolidation process. Modeling of Highway Embankments Over Soft Sub Soil
Material parameters were obtained using the Conditions, Master Thesis, University of Moratuwa, Sri
available laboratory and field test results and Lanka.
empirical correlations were used in the case of lack 5. Kimoto, S. and Oka, F. (2005) An Elasto-Viscoplastic
of test data. The conclusions obtained from this model for clay considering destructuralization and
consolidation analysis of unstable behavior, Soils and
study are as follows: Foundations, 45-2, 29-42.
1) The model assumption of occurrence of creep
6. Edil, T.B. and den Haan, E.J. (1994) Settlement of
(or secondary) settlement as a continuous process Peats and Organic Soils, In Proc. of the Conf. on
agrees well with the observed consolidation Vertical and Horizontal Deformations of Foundations
behavior under the laboratory conditions. and Embankments, Texas, USA,16-18 June, eds. Albert
2) The simulation of the field consolidation T. Yeung and Guy Y. Felio, Part 2, 1543-1572, ASCE
behavior of peaty clay due to the construction of 7. Karunawardena, W.A. (2000) A Study of
compacted earth fill reveals that the finite Consolidation Characteristics of Colombo Peat, In
Proc. of 1st Int. Young Geotechnical Engineering
deformation theory is more representative than the Conf., Southampton, United Kingdom
infinitesimal strain theory for the peat consolidation
8. Oka, F. (2005) Computational modeling of large
process. Also effect of structural degradation has deformations and the failure of geomaterials, Theme
significance in the prediction of both the observed lecture, In Proc. of 16th ICSMFE, Vol.1, 47- 95.
settlement and the stagnated excess pore pressure 9. Adachi, T. and Oka, F. (1982) Constitutive equations
simultaneously. for normally consolidated clay based on Elasto-Visco-
3) The embankment analysis of natural peaty plasitity, Soils and Foundations, 22-4, 57-70.
clay indicates that the model can be successfully 10. Burland, J.B. (1990) On the compressibility and shear
applied to predict both deformation and stability strength of natural clays, Geotechnique, 40(3), 329-378.
aspects of structures constructed on peaty clays. 11. Mesri, G., Stark, T.D., Ajlouni, M.A. & Chen, C.S.
4) The model incorporating equivalent vertical (1997) Secondary Compression of Peat with or without
permeability for PVD improved embankment Surcharging, J. of Geotech. Engrg., ASCE, 123(5),
411- 421.
foundation yielded acceptable results in terms of the
observed excess pore water pressure and 12. Edil, T.B. and Dhowian, A.W. (1981) At-Rest Lateral
Pressure of Peat Soils, J. of Geotech. Engrg, ASCE,
settlements. Also, the analysis indicates that install- 107(GT2), 201- 217.
ation of PVD not only accelerates the consolidation
13. Kugan, R., Puswewala, U.G.A., Kulathilaka, S.A.S.,
rate, but also influences the deformation pattern of Peiris,T.A. (2004) Consolidation Testing of Peaty Clay
the subsoil due to embankment loading. , In Proc. of Engineering Research Unit, University
Because of the general good agreement obtained in of Moratuwa ,Sri Lanka.
this paper in both laboratory and field conditions, 14. Indraratna, B., Balasubramaniam, A.S. and
the constitutive model used here can be Balachandran, S. (1992) Performance of test
recommended for a practical application of embankment constructed to failure on soft marine clay,
J. of Geotech. Engrg, ASCE, 118(1), 12- 33.
construction on Sri Lankan peaty clays.
15. Chai, J.-C,Shen,S.-L.,Miura,N.and Bergado,D.T.(2001)
ACKNOWLEDGEMENTS Simple method of modeling PVD improved subsoil, J.
of Geotech. Engrg, ASCE, 127(11) ,965- 972.
Authors are grateful to Road Development
Authority, University of Moratuwa and National 16. Chai, J.-C, Shen, S.-L., Miura, N., Sakajo, S. and
Bergado, D.T. (1995) Behavior of vertical drain
Building Research Organization of Sri Lanka for improved subsoil under embankment loading, Soils and
providing necessary data for this research work. Foundations, 35-4, 49-61.

12
View publication stats