Professional Documents
Culture Documents
Kamarudin Ahmad
Associated professor, Department of Civil Engineering, University Technology
Malaysia; 81310 Johor; Malaysia; e-mail: Kamarudin@utm.my
Nazri Ali
Senior lecturer, Department of Civil Engineering, University Technology
Malaysia; 81310 Johor; Malaysia; e-mail: nazriali@utm.my
ABSTRACT
Cement columns are widely used to improve the load settlement characteristics of soft soils below
shallow foundations. This paper investigates the performance of cement columns in soft peat. A
series of laboratory tests were conducted on undisturbed peat for finding deformation properties.
Finite element analyses were used to evaluate settlement of soft peat stabilised with floating and
end- bearing cement columns using PLAXIS 3D foundation software. A mat foundation under
vertical stress over treated peat was considered as a three-dimensional problem. For finite element
simulation, embedded pile model was considered for the cement columns while soft soil creep
model presented the behaviour of peat. In the case of floating columns settlement results from the
software were compared to Broms analytical methods. The results indicated that the settlements
were produced by the software are close to those obtained by analytical methods. On the other and
the analytical methods predict slighter settlement than those by the software.
KEYWORDS:
INTRODUCTION
Peat soils are well known for their high compressibility and low shear strength. Construction of
any structures on these soft soils requires the consideration of stability and settlement and causes
serious problems. In the last decade Geotechnical engineers have developed several alternatives to
solve these problems, including use of vertical drain, preloading, Geosynthetics, concrete pile, stone
column and deep mixing columns. Deep mixing method is a method to stabilize the soft soil by
adding dry or wet binders to the soil in order to increase the stability and to prevent large or
unsymmetrical settlements and post construction settlement of embankments and shallow foundations
(EuroSoilStab, 2002). Furthermore this method has been used to prevent sliding failure, control
seepage, prevent shear deformation, improve the dynamic performance of soft ground and treat
- 2243 -
2244
contaminated soil (Boussida and Porbaha, 2004). Nowadays this method is proven to be more
economical and requires minimum time (Hebib and Farrell, 2003). Since peat has high water content,
the required water for cement reaction comes from peat. Therefore dry mixing method is more
effective for peat compared to wet mixing methods (Alwi, 2007).
According to Ahnburg et al (2002) originally lime was the only binder to stabilize soft soil but
the cement was replaced due to higher strength since mid-1980s. In field conditions, cement is
continuously injected using a pumping system and mixed with the soft soil by special tools attached
to the rotary device (Porbaha, 1998). The cement hydrates and reacts with soft soil to form a solid
material stiffer than ambient soft soil to control settlement of super structures and to increase overall
stability. The amount of settlement reduction depends on the area replacement ratio and the stress
concentration ratio, which is function of column stiffness compared to unimproved soil (Alwi, 2007).
The diameter of these columns ranges from .5 to .75m, the spacing is generally 1 to 1.5m center to
center and the length varies from 10 to 30m in field constructions (Larsson, 2003; Topolnicki, 2004;
Nozu 2005).
Cement columns play an important role in foundation of building and lightweight structures. In
Japan cement columns with a 1m diameter have been used to tolerate five story building. The
columns were installed in square or rectangle pattern. The UCS (unconfined compression strength) of
stabilized soil was in a range of 2 to 4 mPa with cement content of 200 to 300 (kg/m3). The final
settlement has been general 10 to 30 mm (Hibino, 1996). In China these columns with .5m diameter
and a 22% area ratio, have been used to support up to 12 to 15 story building. The bearing capacity of
the columns was 520 to 650 kPa at the cement content of 20% (Yuewen, 1996). Soyez et al (1983)
have reported that lime columns decreased maximum settlement of embankment up to 50% compared
to unstabilised soil and settlement reduction was small when the column spacing increased. Holm et
al (1983) stabilised very soft clay with undrained shear strength of 6 to 9 kPa and compression
modulus of 60 to 175 kPa using lime columns. For an applied load of 50 kPa the settlement were
attenuated by 50% after 2.6 years. The diameter of columns was .5m with 1.4m spacing.
The thickness of mat foundation can be reduced by these columns since the mat will be supported
at a large number of points by the columns. The span length and bending moments will be smaller
with the mat rested on high capacity rigid piles (Broms, 2004). Although many investigations have
been done regarding numerical and experimental aspects of cement columns (e.g. Bergado et al.,
1994; Black, 2007; Broms, 2004; Fang, 2006; Karstunen, 1999; Ahnburg, 2002; Chai et al., 2010),
there is a need for further studies, particularly comparison of analytical methods with finite element
modelling. Therefore the main objective of this study is to address the aspect of initial settlement
reduction through modelling soft peat soil stabilization via cement columns with FEM program Plaxis
3D foundation V2. In addition an attempt has been made to compare the initial settlements obtained
by FE analysis with analytical methods.
ANALYTICAL SETTLEMENT
The quantitative assessment of the deformation behaviour of soft soil stabilized with cement
column is important for Geotechnical analysis. Several theoretical approaches have been presented to
predict the settlement of soft soil by using column stabilization methods. EuroSoilStab, (2002)
mentioned that the following factors can influence on stabilized soil with column element inclusion:
2245
Load distribution within soft soil and cement columns has an important impact on settlement
analysis. In a method proposed by Broms and Boman (1979), two theories were considered. Based on
first theory when the load carried by column exceeds its yield strength, the stress in the column is
equal to yield strain and the rest of applied load will be transferred to unimproved soft soil, while a
second theory implies that applied load will be shared into the soft soil and column in proportion to
the stiffness of materials. The maximum settlement of group columns is usually calculated based on
two assumptions:
1. The axial deformations of columns are the same as deformations of surrounding unstabilised
soil.
2. Reinforced soil acts like composite materials, similar to over consolidated clay.
q2
q1
q=q1+q2
q2
q1
q=q1+q2
2
1
Stress in column
Ecol
Creep load
2246
According to Figure 2, the load deformation curve is assumed to be linear up to long term
strength of columns. The slope of the curve shows Young moduli of columns. During the loading
stages on soft soil and columns, two settlements are compared. Settlement of the column ( ) and
soft soil (S ). Equations 1 to 3 depict the calculation of settlements:
h q
.
a E
(1)
h
q
.
1a M
(2)
S =
where:
S : Settlement of column, m
h: Thickness of layer, m
A
a:
A
A : Total area of columns
A : Total area of improved soil
q Load on columns, kPa
E : Young's modulus of column, kPa
S :
where:
a. E
h. q
+ (1 a)M
(3)
and
2. If S < S then column can't stand any more load and then the settlement S
is equal to the calculated settlement S in unstabilised soil.
which occurs
FEM MODELLING
Finite element (FE) methods have been used in Geotechnical engineering to appraise complex
problems. 3D modelling can simulate field conditions properly however analysis is time consuming
2247
and complicated. The finite element software of PLAXIS 3D foundation version 1.6, developed by
the Delf technical university, is selected in this paper. The proposed modelling considers cement
columns as embedded pile. This model has been developed in PLAXIS 3D foundation in which the
pile is assumed as a slender beam element that can be located in soil in any arbitrary orientation .The
interactions of the pile with surrounding soil are simulated by the skin tractions and the tip forces.
Sliding between the pile nodes and soil nodes are provided by skin tractions whilst the relative
movement at the pile tip determines the tip forces (Engin, 2007).
The analysis is carried out using soft soil creep (SSC) model considering the elastoplastic
behaviour of peat. According to Plaxis manual this model is appropriate for near-normally
consolidated clay, clayey silt and peat. Some basic characteristics of the SSC model consist: stress
dependent stiffness, the distinction between primary loading and unloading-reloading, time dependent
compression and failure behaviour according to Mohr-Coulomb criterion.
A drained analysis is assumed for soft soil. Although the problem is symmetric in geometry and
loading, full 3D model is considered in the analysis. Figure 3 shows the modelling of reinforced soil
by a group of cement columns. The initial vertical stress due to gravity load has been considered in
the analysis. A fictitious case but with realistic dimension and materials is simulated. The grid of
columns considered is squared with spacing of 1.5m and the cement columns diameter is equal to 1m.
The area ratio, which is the proportion of columns are to the whole area, is equal to 34.9%. Cement
columns usually are extended to hard layer, but occasionally because of thick soft layers, floating
columns are also installed. The peat layer is 10 m thick, over a rigid stratum. In this study five cases
are considered by varying the cement column length to the diameter of the column ( l d ) from 2,3,4,5
(floating columns) and 10 (end bearing- columns). It is assumed that mat foundation is rigid and both
cement columns and soft soil undergo the same amount of settlement, furthermore static load
simulates structure's load which is applied via a mat foundation using the plate element. As mentioned
interactions of cement columns with soft soil, is simulated by skin traction and tip forces that are
assumed in figure 3.In this study the skin resistance is assumed to be equal the undrained shear
strength c , and point bearing resistance corresponds to 9c . Test analysis is performed under four
stages. On the first stage the excavation is simulated by removing 0.8 m of the soil in preparation.
Cement columns installation is in second stage while construction of mat foundation is defined in the
third stage. Finally vertical stress starts from 20 kPa and increases in interval of 10 kPa continuously
and separately until 100 kPa.
Proper choices of material properties are vital in order to simulate initial settlement of peat and
cement columns. The peat used in the numerical research was obtained from Pontian, west of Johor,
Malaysia. Undisturbed and disturbed samples of peat were collected. First the peat soil was sundried
and passed through specific sieves to obtain proper engineering properties for FE analysis. Initial
results showed that the peat is classified as fibrous and the natural water content, organic content,
liquid limits, specific gravity and bulk densities ranged from 350-550%, 80-95%, 220-550%, 1.291.42 and .9-1.03 Mgm respectively. Undrained shear strength of the peat was measured by the
unconfined compression strength (UCS) test. Although the consolidation settlement didnt take into
consideration but compression indexes were measured in the conventional Oedometer test. Table 1
depicts different properties of unstabilised peat. EuroSoilStab (2002) summarized properties of
cement columns in organic soils in design guide soft soil stabilisation. In this project full scale tests
were executed in several countries. Overall results indicated useful correlations between cement
columns and peat soil properties as follows. In cemented peat columns the unit weight is put equal to
that of unstabilised peat and typical value of undrained shear strength is put equal to 100-150 kPa
2248
(100 kPa in this study), while the characteristic value of Youngs modulus is put equal to 50-100 c ,
But the proposed value for peat is 50 c . In this study UCS test on undisturbed peat with constant
strain control showed the average value 10 and 350 kPa for undrained shear strength and compression
modulus respectively. The Poissons ratio of the soil and cement columns were assumed based on
typical values. Table 2 and 3 show the properties of cement columns and mat foundation, used in this
study.
Peat
m
(kN m ) k( day)
11
.86
C (kPa)
()
10
.87
.174
.043
.15
(kN m )
11
E(kPa)
d(m)
5000
T
N
( m)
31.4
F (N)
I (m )
70.68
.2
.049
(kN m )
24
E(kPa)
h(m)
.8
.2
310
2249
10
20
Stress (kpa)
30
40
50
Stress (kpa)
60
70
0
100
20
30
40
50
60
70
80
90 100
100
300
400
500
600
700
Continuous loading
800
Seprate loading
900
Displacement(mm)
200
Displacement (mm)
10
200
300
400
500
l/d=2
l/d=3
l/d=4
l/d=5
600
l/d=10
700
2250
Stress (kpa)
10 20 30 40 50 60 70 80 90 100
Stress (kpa)
0
10 20 30 40 50 60 70 80 90 100
100
100
Penetration of columns
Displacement (mm)
200
300
400
l/d=2
500
l/d=3
600
l/d=4
700
l/d=5
800
900
200
300
l/d=2
400
l/d=3
500
l/d=4
600
l/d=5
700
20 kPa
0.7
30 kPa
0.6
SRR
0.5
40 kPa
0.4
50 kPa
0.3
60 kPa
0.2
0.1
70 kPa
0
0 1 2 3 4 5 6 7 8 9 10
1
0.9
0.8
0.7
0.6
SRR 0.5
0.4
0.3
0.2
0.1
0
l/d=2
l/d=3
l/d=4
l/d=5
l/d=10
10
20
30
40
50
60
70
80
l/d
Stress (kpa)
225
51
Settlement comparison FE
EM & ana
alytical
The
T compariso
ons of finite element
e
resultts with analyttical methods are made in this section. IIn
this study
y the effect of applied streess to the soiil below cem
ment columns is ignored. T
Two analytical
methods based
b
on Bro
oms assumptiions, are com
mpared with F
FE results. Inn first methodd it is assumeed
that appliied stress is carried by so
oft soil and cement
c
colum
mns in the pproportion of stiffness (SC
CEquation 1). In second
d method 100% of applieed load is carrried by cem
ment columns due to higheer
stiffness (C-Equation
3).Two comp
(
parisons are made
m
in this section. In ccomparison bbetween FE &
analyticall methods ressults, based on
o figure 10 there
t
is signiificant deviattion between the FE resullts
and both analytical
a
meethods in casee l/d=2 . This is due to the fact that the thickness of soft soil below
cement co
olumns is high
h and the effeect of applied load in this aarea is ignored. In floatingg columns therre
are a few differences between
b
analy
ytical method
ds and FE anaalysis and thee results are ccompatible buut
both meth
hods show lesss settlement compared
c
to FE
F results.
In co
omparison bettween both analytical
a
metthods, it is obbvious that inn all floating columns therre
are small differences between the an
nalytical meth
hod results annd the settlem
ment value of the first theorry
(SC) has been
b
always lower
l
than thee second theo
ory (C).
(
Stress (kpa)
0
10
20
2
30
40
50
60
70
80
90
100
1110
0
l/d=2 (FEM)
100
l/d=2 (Analytical-SC)
200
l/d=2 (Analytical-C)
l/d=3 (FEM)
Displacement (mm)
300
l/d=3 (Analytical-SC)
400
l/d=3 (Analytical-C)
l/d=4 (FEM)
500
l/d=4 (Analytical-SC)
600
700
800
l/d=4 (Analytical-C)
l/d=5 (FEM)
l/d=5 (Analytical-SC)
l/d=5 (Analytical-C)
900
Figu
ure 10: Comp
parison betw
ween FEM annd analyticaal theories
2252
CONCLUSION
A series of numerical analysis were carried out to evaluate the settlement of mat foundation rested
on soft peat reinforced by a group of cement columns. Triaxial and unconfined compression tests
performed on the peat showed the average value of 10 kPa for undrained shear strength while
Oedometer test indicated the average compression index of 3. The FE analysis utilized soft soil creep
model (SSC) for peat where cement columns were simulated as embedded pile. Static loads were
applied at the mat foundation continuously and separately from 20 kPa until 100 kPa. It was found
that that the maximum stress the peat can resist without failure was about 70 kPa. In floating columns,
as the length of the column increases the vertical displacement of mat decreases due to more skin
interactions. The performance of end bearing columns was superior to floating columns as the load
was transferred to the hard ground in the case of end-bearing columns. The results showed that the
average settlement reduction in floating and end-bearing columns were 40 and 60% respectively.
Comparison of analytical theories in settlement calculations yielded a close agreement between both
analytical methods. In the case of floating columns the results showed that the settlements were
obtained by the software are close to those by analytical methods. It was found that the analytical
methods anticipate less settlement than those by the FE analysis.
REFERENCES
1. hnberg H, Johansson SE, Retelius A, Ljungkrantz C, Holmqvist L, och Holm G (1995)
Cement and lime for deep stabilization of soil, In: Physico-chemical study with
stabilizing effect, pp. 14-89.
2. Alwi A. (2007) Ground Improvement on Malaysian Peat Soils Using Stabilized PeatColumn Techniques. PhD. Thesis, University Malaya, Kuala Lumpur.
3. Bergado, D.T., Anderson L.R., Miura, N., and Balasubramaniam, A.S. (1994)
Lime/cement deep mixing method. Improvement Techniques of Soft Ground in
Subsiding and Lowland Environments, Rotterdam, A.A.Balkelma, 99 130.
4. Black, J.A. (2007) The settlement performance of a footing supported on soft clay
reinforced with vibrated stone columns, PhD Thesis. Queens University, Belfast.
5. Boussida, M. and Porbaha, (2004a) Ultimate bearing capacity of soft clays reinforced by
a group of columns application to a deep mixing technique, Soils and Foundations, Vol.
44, No. 3, 91 101.
6. Broms, B.B. (1984a) Stabilisation of Soil with Lime Columns, Design Handbook, Third
Edition, Lime Column AB, 51 pp.
7. Broms, B.B. (2004) Lime and lime/cement columns Ground Improvement, 2nd edition,
252-330, Spon Press.
8. Chai J-C, Miura N, Kirekawa T, Hino T (2010) Settlement prediction for soft ground
improved by columns, Ground improvement. Proc Inst Civil Eng UK 163(2),109-119
9. Engin, H.K. (2007) Report on tension pile testing using embedded piles, Plaxis internal
report. Delft (The Netherlands)
2253
10. EuroSoilStab (2002) Development of Design and Construction Methods to Stabilize Soft
Organic Soils, Design Guide Soft Soil Stabilization, Industrial & Materials Technologies
Programme (Brite- EuRam III), European Commission, CT97-0351, Project No. BE 963177, pp. 15-60.
11. Fang, Z. (2006) Physical and numerical modelling of the soft soil ground improved by
deep cement mixing method, PhD Thesis, The Hong Kong Polytechnic University.
12. Hebib S, Farrell FS (2003) Some experience on the stabilization of Irish
peats,Canadian Geotechnical J. 40: 107-120.
13. Hibino, S., (1996) Monitoring of Subsidence of Building on Ground Improved by Deep
Mixing Method, Proc. 2nd Int. Conf. On Ground Improvement Geosystems, 14-17 May,
1996, IS-Tokyo96, Vol. 1, 595-601.
14. Holm, G., Trnk, R., Ekstrm, A. and Torstensson, B.A., (1983) Lime Columns under
Embankments, A Full Scale Test. Improvement of Ground.Proc.8th EuropeanConf. on
Soil Mech. a. Found. Engng, Helsinki, Finland, Vol. 2, pp 909-912.
15. Karstunen, M. (1999) Alternative ways of modeling embankments on deep stabilized
soil, Proc. of the Int. Conf. on Dry Mix Methods for Deep Soil Stabilization, 221 228.
16. Larsson, S. (2003) Mixing processes for ground improvement by deep mixing, PhD
thesis. Royal Institute of Technology, Stockholm, Sweden.
17. Nozu, M. (2005) Regional report, Asia. Proc. of International Conference on Deep
Mixing Best Practice and Recent Advance, R3 R17.
18. Porbaha, A. (1998) State of the art in deep mixing technology, part I. basic concepts and
overview. Ground Improvement, Vol. 2, No. 2, 81 92.
19. Porbaha, A., Shibuya, S., and Kishida, T. (2000) State of the art in deep mixing
technology: part III Geomaterial characterization. Ground Improvement, 3, 91110
20. Soyez, B. and Delfaut, A., (1983) Loading Tests on a Clayey Hydraulic Fill Stabilised by
Lime Treated Soil Columns. Proc. European Conf. on Soil Mech. a. Found. Engng,
Improvement of Soil, Vol.2, pp 951-954.
21. Topolnicki, M. (2004) In situ soil mixing. Ground Improvement. New York, Spon
Press, 331-428.
22. Yang SD, Yagihashi JN, Yoshizawa SS (1998) Dry jet mixing for stabilization of very
soft soils and organic soils, Geotechnical Special Publication No. 81, ASCE, Reston, VA
23. Yuewen, Z., (1996) Deep-Cement Mixing Piles Stabilizing the Saturated Loess, Proc.
2nd Int. Conf. on Ground Improvement Geosystems, 14-17, May, 1996, IS Tokyo 96, Vol.
1, 573-576.
2012 ejge