Design Recommendation for Stone Column

Reinforced Soft Clay Deposit

Sudip Basack
Bengal Engineering & Science University, Howrah, West Bengal, India.
Buddhima Indraratna & Cholachat Rujikiatkamjorn
Centre for Geotechnical & Railway Engineering
University of Wollongong, New South Wales, Australia.

ABSTRACT
Reducing long-term settlement of infrastructure and providing cost-effective foundations with sufficient load-bearing
capacities are national priorities for infrastructure development in most countries. Among various methods of soft soil
improvement, installing stone columns is one of the well established and effective techniques practised worldwide. The
stone columns not only act as reinforcement to the surrounding soil, but also speed up the time-dependant dissipation
of excess pore water pressure due to surcharge loading by shortening the drainage path. A novel numerical model has
been developed and validated by the authors to analyse the response of stone column reinforced soft soil under
embankment loading, adopting the free strain approach and considering arching, clogging and smear effects. Using the
model, a design methodology associated with a series of charts and curves for various clogging and smear zone
parameters has been suggested by the authors. Utilizing them, a typical design example for stone column reinforcement
in a soft clay deposit has been presented.

RÉSUMÉ
Dans la majorité des pays, le développement des infrastructures constitue une priorité nationale pour construire sur des
foundations à coût raisonnable, offrant une portance suffisante et une réduction de tassement à long terme. L‟exécution
des colonnes ballastées représente l‟une des techniques d‟amélioration efficace des sols mous largement pratiquée à
l‟échelle internationale. En plus du renforcement du sol environnant, les colonnes ballastées accélèrent la dissipation
des surpressions interstitielles, générées par le chargement, par la réduction de la distance de drainage. Une nouvelle
methode numérique, mise au point et validée par les auteurs, a été déveoppée pour l‟analyse du comportement d‟un
sol mou renforcé par colonnes ballastées sous l‟action d‟un remblai. Cette méthode se base sur une approche à
deformation libre et tient compte des effets de voûte, de l‟obturation et du remaniement. Avec ce modèle, les auteurs
proposent une méthode de dimensionnement illustrée par des abaques et de courbes avec variation des paramètres de
l‟obturation et de remaniement. Un exemple type de renforcement par colonnes ballastées d‟une argile molle est
étudiée à l‟aide de ces outils.

1 INTRODUCTION Various analytical and numerical solutions have

already been developed by many researchers for
It is imperative to apply adequate ground improvement understanding the load transfer mechanism of soft soil
technique to the existing soft soils before any construction reinforced with stone column, e.g., Han et al.(2000 &
in order to prevent unacceptable excessive and 2002), Ambily et al. (2007), Malarvizhi et al. (2008), Wang
differential settlement and increase the bearing capacity (2009), Lo et al. (2010) and Murgasen et al. (2010). All
of the foundations (Indraratna et al. 1992). these solutions are based upon unit cell analysis
Amongst various methods of soft soil improvement, assuming „equal strain‟ hypothesis. However, such an
reinforcing the ground by installation of stone column is assumption is applicable whenever the surcharge load
one of the well established and effective techniques applied on the ground surface possesses adequate
followed worldwide (Wang, 2009). As reported by Guetif flexural rigidity initiating uniform ground settlement,
et al. (2007), there are basically three components which thereby resulting in an unequal distribution of stress
contribute to the degree of improvement achieved by induced on the soil surface. In case of embankment, the
installation of stone columns in soft clay deposits, which flexible nature of applied surcharge loading is most likely
are: (i) The stone column, possessing greater strength to induce an equal distribution of surface load resulting in
and stiffness in comparison to the surrounding soft soil, an uneven ground settlement, i.e., the „free strain‟
acts as a reinforcement increasing the overall load (Barron, 1948).
bearing capacity of the ground and decreasing the In case of a fill embankment, the behaviour of the
settlement, (ii) During installation of stone column, soil-stone column system is time-dependent. Initially,
densification of soil in the vicinity of the interface takes most of the imposed total stress is borne by the pore
place, and (iii) The stone column speeds up the time water. Due to dissipation of the resulting excess pore
dependant dissipation of excess pore water pressure due pressure, progressive settlement of the soft clay occurs
to surcharge loading by shortening the drainage, thereby and arching takes place resulting in an uneven
accelerating the improvement the strength and stiffness distribution of vertical stress on the ground surface. This
of the surrounding soil within the zone of influence.
phenomenon is duly supported by other researchers (Low
et al., 1994, Abusharar et al., 2009 and Deb, 2010). H
As reported by Han et al. (2002), a smear zone is
developed in the soil adjacent to the soil-column interface
due to installation. Also, because of the migration of clay Impervious rigid boundary (a) Stone column
particles from soil into the pores of the column, a clogged Unit cell idealization q = Q + γe He
zone may be formed within the column in the vicinity of
the soil-column interface (Adalier et al., 2004). r
Based on „free strain‟ hypothesis and considering the
arching, smear and the clogging effects, a numerical
model (finite difference method) have already been H
developed and validated by the authors (Indraratna et al.,
2010). Using the model, a design methodology Unit cell
associated with appropriate design curves has been
proposed and a typical design example has been
illustrated in this paper. Un (b) Z
it
cel (
2 NUMERICAL ANALYSIS l Undisturbed
b zone
The details of the numerical model developed have been )
described elsewhere (Indraratna et al., 2010). A brief Smear zone
illustration of the analysis carried out is given herein.
Clogged
The analysis was done based on the assumption that rs column zone
the compressive strains of the soil column occur only in
the vertical direction, and ignoring the elastic settlements re Unclogged
which are insignificant compared to the consolidation column zone
settlement. The soil was assumed to be fully saturated
with incompressible water, the flow of water through the
soil to be purely horizontal (radial towards the column) (c)
following Darcy‟s law and no flow of water to take place
through the cylindrical boundary and the impervious base Fig.1. (a) A typical stone column reinforced soft clay
of the unit cell. The coefficients of permeability (kh) and deposit supporting an embankment. (b) Unit cell
compressibility (mv) of the soil was assumed to remain idealization. (c) Cross section of the unit cell.
constant during the process of consolidation.
The idealised problem is depicted in Figures 1(a) and Radial Stone column Ground surface Vertical
(b). The soft clay layer of thickness H has been assumed divisions:
to overlay on an impervious rigid boundary, and is divisions: 1 2 3 . . . . . . . . . i . . . . . . . . . n-1 n1
improved by a group of stone columns having a radius rc δz
each, extended to the bottom of the clay layer. The 2
average load intensity on the ground surface ( q ), was δz
expressed as a sum of the uniform surcharge load
intensity Q on the embankment fill and the self weight of 3
δr δr .
the embankment (γeHe). The radius of influence of the
.
unit cell re could be calculated following the method of δz .
Wang (2009). The cross section of the entire zone of the .
unit cell was divided into four distinct zones (Fig.1c), viz., j
the unclogged column zone, clogged column zone, smear .
zone adjacent to the column and the outer undisturbed .
soil zone. As shown in Fig.2, the soil mass within the unit .
cell had been divided both radially as well as vertically .
into (n-1) number of equal divisions; n being a positive n-1
integer greater than unity, such that each of such
divisions may be expressed respectively as: δr
Impervious rigid boundary n
= (re – rc)/(n-1) and δz = H/(n-1). The total time interval of re - rc
computation tt is divided into (n-1) number of equal
Figure 2. Soil element discritization within the unit cell.
divisions, i.e., δt = tt /(n-1). In this paper, these separators
are denoted as „nodes‟. The soil elements are
Q Sand
understandably ring-shaped. The primary objective of the
blanket analysis was to compute the excess pore water pressures
and the effective stresses developed at each separator at
He Embankment the corresponding time, and thereby compute the other
Embank time-dependant variables such as the degree of
ment consolidation and settlement.
 The smear zone parameters, viz. rs and ks may be consolidation of the soil = kh / (mv γw ), γw being the unit
reasonably estimated following the recommendations of weight of water. Since both rc and re are variables, the
Han et al. (2002), Walker et al. (2006) and Wang (2009), constant parameter H is used for normalizing the time.
where, rs and ks are the radius and the horizontal The computation was carried out by means of a user-
permeability of soil in the smear zone respectively. The friendly computer software written in FORTRAN 90
clogging effect was quantified by the non-dimensional language. The relevant flowchart is shown in Fig.3.
parameters α and αk. Because of the different column to
soil stiffness ratio, the soil arching beneath the Start
embankment occurs. Following the analysis of Abusharar
et al. (2009), the arching effect was analysed to compute Input soil data, column parameters,
the vertical stress distribution q(r) on the ground surface imposed load and time
which was found to be parabolic and a unique function of
N (=re/rc), ns, rc, r, q and Kp (passive earth pressure Compute excess pore water pressures
developed in the soil for all values of i and k
coefficient of the embankment). The term ns, referred as
the stress concentration ratio, was defined as the ratio of
stiffness between soft soil and stone columns and a Compute effective stresses
function of height of embankment, properties of the soils induced in the soil for all values of r, z and t
in the embankment and that of the soft clay deposit. The
values of ns can be reasonably estimated following the
recommendations Castro et al. (2009). Compute nodal displacement
The nodal excess pore pressures were computed of soil at all nodal points
based on the radial consolidation theory of Barron (1948).
Using appropriate boundary conditions, the following
matrix equation was evaluated: Compute undrained shear strength
A {u} {b} ……(1) for all values of r, z and t
where, A coefficient matrix; {u} unknown vector
for excess pore water pressure at nodes; {b} augment Compute improvement factor and
vector. settlement factor for all t
Solving the Equation (1), the nodal excess pore
pressures, and hence the degree of consolidation, were
computed. The nodal displacements within the soil mass Print output results
of the unit cell at time t was given as:
t z
Stop
mv
u (r , z )
dzdt ……(2)
0 0
t Figure 3. Flowchart for the computer program.
The nodal and average ground settlements were
computed by carrying out numerical integration.
The effective stress developed in the soil mass at any 3 VALIDATION
point (r, z, t) in the space-time coordinate may be Comparison of the average degree of consolidation by
expressed by (Khan et al., 2010): radial drainage only has been made with the existing
r , z, t z q(r ) u (r , t ) ……(3) solutions of Barron (1948), Hansbo (1981), Han et al.
where, γ’ is the effective unit weight of the soil mass. (2000 & 2002) and Wang (2009). The variations of
During consolidation, the undrained strength and average degree of consolidation with time factor are
stiffness of the soil increase progressively. Guetif et al. presented in Fig.4. It was observed that the results
(2007) carried out extensive finite element analysis to obtained using the present model are in acceptable
investigate the improved soft clay characteristics due to agreement with the other solutions and close to those of
stone column installation. The undrained cohesion at any Han et al. (2002).
point in the space-time coordinate system was computed Redana (1999) analysed the performance of two test
following the analysis of Umezaki et al. (1993). The embankments T1 and T2 constructed at a naval dockyard
increased soil strength was quantified by a non- of Pom Prachul, Thailand. For improvement of soft
dimensional improvement factor β which was defined as ground, sandwitch drains were installed in a square
the minimum value of the ratio of the post-consolidation
to initial undrained cohesion at ground surface. Similarly,
the increase in stiffness of the soft soil was expressed as
a settlement factor ξ defined as the ratio of the average
ground settlements of the reinforced to unreinforced soils
at 90% consolidation.
For design purpose, a modified time factor T’90 has
been introduced herein, which is expressed as:
2
T’90= cvr t90 / H ……(4)
where, t90 is the time required for the average degree of
consolidation to attain 90% and cvr = coefficient of radial
 (a)

α
α
N=2, kh/ks=10, rs/rc=1.1,
H/rc=20, ns=3, α=1 Period (days) Thick- Unit
Fill Rest ness weight
(m) (kN/m3)
T1 T2 T1 T2
0-12 0-6 12-20 6-10 0-0.35 16.2
20- 10- 38-110 23-98 0.35-1.1 18
Figure 4. Comparison of computed rates of consolidation 38 23
using different methods. 110- 98- 130- 121- 1.1-2.35 19
130 121 onward onward

grid pattern and filled with compacted dry sand. The field
test results were compared with those obtained by using
the model and that of Han et al. (2002). The
computational parameters adopted are same as
thosefound by Redana (1999) and presented in Table1.
The variations of average ground settlement and excess
pore water pressure with time are shown in Figures 5(a) &
(b) respectively. As observed, although the time-
settlement variation obtained by the present model is in (b)
reasonably good agreement with the field data, the
solutions of Han et al. (2002) slightly over-predicts the
values. The assumption of free strain and the parabolic
distribution of vertical stress on ground surface in the
present analysis have significantly affected the results. As
regards to the time variation of excess pore pressure, the
present solution yields promising results as compared to Figure 5. Comparison of numerical results with field test
the field data for t > 100 days, where as the solution of data of Redana (1999) for: (a) settlement versus time.
Han et al. (2002) slightly under-predicts the values. For t (b) excess pore pressure versus time.
< 100 days, both the solutions were observed to give
relatively lower values compared to the field test results.
Solutions are also obtained considering clogging effect (α of T’90 with N for different values of H/rc and ns are
= 0.5, αk = 1) which predicts the settlement and the presented in Fig.6, assuming reasonable smear and
excess pore water pressure even closer to the field clogging parameters. Similarly, the variation of the
values. stiffness factor ξ with N for different values of H/rc and ns
are presented in Fig.7. Lastly, the variation of the
4 DESIGN RECOMMENDATION improvement factor β with normalized imposed stress
q/cu0 were plotted (Fig.8) for different values of N and ns ,
Indraratna (2009) reported ground improvement at the cu0 being the initial undrained cohesion at ground surface.
Ballina Bypass for construction Pacific Highway linking It has been observed that the variation of β with H/rc is
between Sydney and Brisbane. This site had a floodplain insignificant, and therefore not considered.
consisting of highly compressible and saturated marine For other values of smear and clogging parameters,
clay deposits. A soft silty layer of clay approximately 10 m similar design curves can be prepared using the model
thick was underlain by moderately stiff, silty layer clay and the relevant computer software developed.
located 10-30m deep, which was in turn underlain by firm
clay. The groundwater level was almost at the ground 4.1 Design Example
surface. The relevant field data are utilized herein to
carry out a hypothetical case study followed by a design In this section, how the solutions developed and the
illustration. The parameters used for computation in this above-mentioned design curves are used in actual design
case study are given in Table 1. Using these values, a set are described. The following are the design requirements:
of typical design curves are developed. The variation (i) The consolidation is expected to be completed 90 %
by 1 year after the completion of the embankment
construction.
 (ii) The height of embankment above the virgin ground
level should be 4.3 m. ks/kh =0.1; rs/rc = 1.15;
(iii) The desired improvement factor should not be below α = 0.5; αk = 1;
2.75.
(iv) From the serviceability criteria, the average

T’90 (10-
settlement should not exceed 500 mm at 90%

)
consolidation.

1
Although the value of ns actually depends upon the
relative stiffness of the column and the soil which should
be estimated reasonably from the laboratory tests and the
method of installation, it is hereby assumed as 6 (for
illustration).
From the given data, T’90 = 0.00263. Using Fig.6, the
value of N and H/rc can be estimated as 4 and 40
respectively, from which the column parameters may be N = re/rc
chosen as: rc = 0.5 m and re = 2 m, although the column
radius might be chosen depending upon the installation Figure 6. Variation of T’90 with N.
technique. The improvement factor obtained from the
Fig.8 for t = 1 year is 2.926 which is well above the
allowable limiting value of 2.75. Using Terzaghi‟s one
dimensional consolidation theory, assuming the vertical
permeability to be same as the horizontal permeability, ks/kh = 0.1; rs/rc = ns:
the average surface settlement of the untreated soft 1.15; α=
ground at the end of 90% consolidation at the design 2
0.5; αk = 1;
embankment loading has been estimated as 4320 mm. 6
The allowable settlement factor is therefore calculated as
500/4320 = 0.116, as against the actual value of 0.11 10
estimated from the Fig.7.

Table 1: Input parameters for field study.

Material Parameter Values

Redana Indraratna
(1999) (2009)
-10 -9
kh 2 x 10 1 x 10 m/s
m/s
mv 6.13 x 10
-7
3 x 10
-6
Figure 7. Variation of ξ with N.
2 2
Soil m /N m /N
$
ks / kh 0.1* 0.333
$
rs / rc 1.15* 2.5
H 17 m 20 m N: ns :
He As reported 4.3 m Time = t90
3
γe by Redana 20 kN/m
Embankment (1999)
Improvement factor, β

Q 0
[Given in
Fig.5(b)]
* *
Kp 3 3
6; 2
rc 0.025m 2: 2
Stone column re 0.8475m To be
/Vertical drain (T1); designed
1.413m
(T2)
ns 4.72 As given in ks/kh = 0.1; rs/rc = 1.15;
α As given in Figures 6-8 α = 0.5; αk = 1;
αk Fig.5
* $ Assumed values ($ as per Indraratna, 2009)
Normalized imposed stress, q / cu0

Figure 8. Variation of β with normalized

imposed load intensity.
5 CONCLUSIONS Han, J. and Ye, S. L. (2002), “A Theoretical Solution for
Consolidation Rates for Stone Column Reinforced
Considering the characteristics of stone column Foundations Accounting for Smear and Well
reinforced soft clay, a numerical solution based on unit Resistance Effects”, Int. J. Geomechanics, ASCE,
cell theory was developed by the authors for computing 2(2), 135-151.
the rate of consolidation, stress distribution, settlement Hansbo, S. (1981), “Consolidation of Fine Grained Soil by
and degree of post-consolidation ground improvement th
Prefabricated Drains”, Proc., 10 Int. Con. Soil Mech.
achieved. The free strain hypothesis is adopted for And Found. Engrg., vol.3, Balkma, Roterdam, The
analysis which appears to be more realistic for Netherlands, 677-682.
embankment loading when the arching effect and Indraratna, B., Balasubramaniam, A. and Balachandran,
clogging are taken into account. The comparison of the S. (1992). “Performance of Test Embankment
numerical results with the available solutions and field Constructed to Failure on Soft Clay.” Journal of
data indicates acceptable agreement which justifies the Geotechnical Engineering, ASCE, 118 (1), 12-33.
validity of the model. Based on the methodology, design Indraratna, B. (2009), “Recent Advances in the
recommendations followed by a set of design curves are Application of Vertical Drains and Vacuum Preloading
developed. Considering a hypothetical case study on a in Soft Clay Stabilization”, 2009 E H Davis Memorial
recent ground improvement project at Ballina bypass, Lecture, Brisbane, Australia.
Australia, a design example is illustrated. It is expected Indraratna, B., Basack, S. and Rujikiatkamjorn, C. (2010),
that the work reported herein will go a long way in design “A Numerical Solution for Stone Column Improved
of stone column reinforcement in soft clay deposit. Soft Soil considering Arching, Clogging and Smear
Effects”, J. Geotech. Geoenviron. Engrg., ASCE.
6 ACKNOWLEDGEMENT [communicated].
The authors gratefully acknowledge the financial support Khan, A. P., Madhav, M.R. and Reddy E. S. (2010),
received from the Department of Education, Environment “Consolidation of thick clay layer by radial flow ---
and Workplace Relations (DEEWR), Australian nonlinear Theory”, Geomechanics and Engineering,
Government through Austraining International to carry out vol.2, no.2, pp.: 157-160.
the research. Special thanks are given to Professor Lo, S. R., Zhang, R. and Mak, J. (2010), “Geosynthetic-
Mounir Bouassida for helping the authors in French encased Stone Columns in Soft Clay: A Numerical
translation of the abstract. Study”, Geotex. and Geom., 28 (2010), 292–302.
Low, B.K., Tang, S.K. and Choa, V. (1994), “Arching in
REFERENCES Piled Embankments”, J. Geotech. Engg., ASCE,
120(11), 1917–1938.
Abusharar, S.W., Zheng, J. J., Chen, B.B. and Yin, J.H. Malarvizhi, S. N. and Ilamparuthi, K. (2008), “Numerical
(2009), “A Simplified Method for Analysis of a Piled Analysis of Encapsulated Stone Columns”, Proc., 12th
Embankment Reinforced with Geosynthetics”, Intl. Conf. of IACMAG, Goa, India, 3719-3726.
Geotextiles and Geomembranes, 27, 39–52. Murugesan, S. and Rajagopal, K. (2010), “Studies on the
Adalier, K. And Elgamal, A. (2004), “Mitigation of Behavior of Single and Group of Geosynthetic
Liquefaction and Associated Ground Deformations by Encased Stone Columns”, J. Geotech. Geoenviron.
Stone Columns”, Engineering Geology, 72(3-4), 275- Engrg., ASCE,136(10), 129-139.
291. Redana, I.W. (1999), “Effectiveness of Vertical Drains in
Ambily, A. P. and Gandhi S. R. (2007), “Behavior of Stone Soft Clay with Special Reference to Smear Effect”,
Columns Based on Experimental and FEM Analysis”, PhD Thesis, University of Wollongong, Australia.
J. Geotech. Geoenviron. Engrg., ASCE, 133 (4), 405- Umezaki, T., Ochiai, H. and Hayashi, S. (1993),
415 “Undrained Shear Strength of Clay during
Barron, B. A. (1948), “Consolidation of Fine Grained Soil Consolidation”, Proceedings of the Eleventh
by Drain Wells”, Trans. of Am. Soc. Civ. Engrg., 113, Southeast Asian Geotechnical Conference,
712-748. Singapore, 269-274.
Castro, J. and Sagaseta, C. (2009), “Consolidation Walker, R. and Indraratna, B. (2006), “Vertical Drain
around Stone Columns: Influence of Column Consolidation with Parabolic Distribution of
Deformation”, Int. J. Num. An. Methods Geomech., 33, Permeability in Smear Zone”, J. Geotech. Geoenviron.
851-877. Engrg., ASCE, 132 (7), 937-941.
Deb, K. (2010), “A Mathematical Model to Study the Soil Wang, G. (2009). “Consolidation of Soft Soil Foundations
Arching Effect in Stone Column-Supported Reinforced by Stone Columns under Time Dependant
Embankment Resting on Soft Foundation Soil”, App. Loading.” J. Geotech. Geoenviron. Engrg., ASCE,
Math. Modelling, 34, 3871–3883. 135(12), 1922-1931.
Guetif, T., Bouassida, M. and Debats, J.M. (2007).
“Improved Soft Clay Characteristics due to Stone
Column Installation” Computers and Geotechnics,
34(2007), 104-111.
Han, J. and Ye, S. L. (2000), “Simplified Method for
Consolidation Rate of Stone Column Reinforced
Foundations”, J. Geotech. Geoenviron. Engrg., ASCE,
127(7), 597-603.