Professional Documents
Culture Documents
47-70, 1997
0 1997 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
PII: SO266-352X(96)00013-4 0266-352X/97 $17.00+ .OO
ELSEVIER
ABSTRACT
The eficiency of end bearing stone columns and end bearing lime columns in
reducing the settlement of a foundation system is examined in this paper. The
foundation system is assumed to consist of a large number of regularly spaced
stone columns of equal length installed in a weak soil layer and supporting a
rigid mat. The analysis examines the influence of such factors as the column
spacing, the weak soil properties, properties of the granular medium used in
constructing the column, the in situ stresses caused by the installation technique,
the depth of the bedrock relative to the tip of the columns and the magnitude
of the load carried by the supported raft foundation (expressed as a untformly
distributed load, UDL). This paper is concluded by presenting a number of
charts that may be used in the analysis and design of such a foundation
scheme. 0 1997 Elsevier Science Ltd. All rights reserved.
INTRODUCTION
For low-rise buildings, lightly loaded foundations, earth structures and storage
tanks that can tolerate appreciable movement, stone columns provide an
economical method of support in compressible and cohesive soils [1,2]. The
problem of the settlement of such a foundation system has been tackled before,
notably by Poulos and Davis [3] who used the elastic theory and by Priebe [4].
A stone column is essentially a vertical cylindrical “hole” dug in the soft
soil layer and filled with compacted stone fragments and gravel. The columns,
which have afinished cross sectional diameter in the range of 75-150 cm, are
usually extended to bedrock or a hard layer, but occasionally floating columns
are also installed. This paper deals with end bearing stone columns only. A
48 H. B. Poorooshasb and G. G. Meyerhof
rigid raft
oft clay
stone column
bedrock or
hard layer
(a) (b)
2b
t l
UDL
raft
P
strain of the column and V is the volumetric strain of the granular medium
having a negative value for a dilating (expanding) material.
Other factors which influence the performance of the raft-stone column-
soil system include the initial stresses caused by the installation (compaction
of the material in the column) and the distance between the tip of the piles to
the hard layer. Short columns installed in a deep layer of soft soil deposit will
not be effective in reducing the settlement of the foundation system. It is the
objective of this paper to evaluate the influence of the various factors
involved and thus arrive at a practical design procedure.
Lime columns are made by mixing 7-10% of lime in situ with soft clay to
make columns of up to 800 mm diameter with a maximum height of
15 meters. Lime columns are much stiffer than stone columns and are likely
to behave linearly during the loading process: a fact that will facilitate the
analysis considerably as will be seen later.
This paper is written in the following sequence. First the equations gov-
erning the discussions are stated without proof leaving the small amount of
mathematics involved to the two Appendices provided at the end of the
paper. Next the various factors which are likely to influence the performance
of the system are stated and their effects evaluated. Based on these evalua-
tions, a set of design charts are prepared and presented. Finally the results of
some field tests are briefly discussed.
Throughout this paper geometric linearity will be assumed. Thus the small
strain theory is applicable. The weak soil is assumed to possess material
linearity with a Young’s modulus Es and a Poisson’s ratio v. The stress
deformation characteristics of a stone column are represented by the two
relations;
where o1 is the maximum principal stress (the load intensity carried by the
column), a3 is the minor principal stress, ~1 is the axial strain (the vertical
strain experienced by the column and the surrounding soils), .s3 is the lateral
strain and V is the volumetric strain of the stone column respectively. A
formal derivation of Eqns (1) and (2) may be found in the Appendix.
50 H. B. Poorooshasb and G. G. Meyerhoj
The minor principal stress a3 appearing in Eqn (1) is the sum of two
components. First, a component which is developed at the column-soil
interface due to the action of the surface loads and the tendency of the stone
column to expand laterally plus a residual stress component due to the self
weight of the soil and those caused by the installation activities (in particular
compaction of the gravel in the hole). Thus;
a, = a3 + ores (3)
(1 -
p=1_2v2_v
v)&
[
E’-G-_E3
2v a2
1 (4)
(1 + v)a* + (1 - v)b2
o3 = &p - EsE3 (1 _ V2)(b2 _ a2) (5)
The above set of six equations may be solved simultaneously to obtain the
value of the six unknowns E], Ed, cl, 03, a3 and p in terms of the given value
of UDL. Of these six unknowns, cl is of prime interest since the settlement of
the foundation system, 6, is equal to LEE.
Since Eqns (1) and (2) are non-linear a process of trial and error must be
used to solve for the unknowns. This approach is usually a lengthy process
which may require many iterations and which may not converge to a solu-
tion at all. A more intelligent approach is to assume values for the unknowns
appearing in the non-linear equations (thus removing them from the set) and
Behavior of stone columns 51
then solving for the other set of unknowns which are encountered in the
linear equations. The process is somewhat similar to what is generally called
the inverse process and is immensely easier to deal with. Thus, assuming a
value of Ed, Eqn (2) may be used to evaluate c3. Equation (4) can now be
solved to obtain the value of p. With the value of p just calculated cr3 from
Eqn (5) can be obtained and then, with the aid of Eqn (3) and o1 of Eqn (I),
substitution for the values of p and cl in Eqn (6) renders the solution. That
is, the relationship between UDL, the load carried by the column/soil system
and the settlement that this UDL produces is established. Obviously the
above formulations apply only to the cases where the stone columns are end
bearing.
In the case of lime columns the material in the columns acts as a linear
material having a Young’s modulus EC and a Poisson’s ratio of v,. Thus it
may be shown, as in the Appendix, that;
UDL b2 - a2
-=,4{1+B~,}~ + {Ec + 2v,[AC(l + Bv,) + Dvc]}; (7)
@IL)
where 6 is the settlement of the system, L is the height of the column and the
constants A,B,C and D have the values;
A= cl-‘) E.
1 - 29 -v S’
C=L.
1 -v’
D = (1 + v>a2
+ (1 - VP2 Es
(1 - v2)(b2 - a2)
n = (UDWG)
(8)
W)
52 H. B. Poorooshasb and G. G. Meyerhof
This last equation, in which n is the settlement ratio, A, is known as the area
ratio since it represents the area of the cross section of the column to the area
of the cross section of the influence cell (i.e. A, = a2/b2), is due to Priebe [4].
EVALUATIONS
foundation settlement h. cm
foundation settlement h, cm
I I I I I 1 1
0 IO 20 3c
foundation settlement b, cm
Fig. 2. Typical results for very compact, compact and less compact stone fills, E, = 1000 kPa.
v = 0.2, spacing = 2 diameters.
Behavior of stone columns 53
For a well compacted stone fill (@= 44”) the performance ratio is 0.31.
Thus if the settlement of a foundation system without stone columns is, say,
20 cm, inclusion of the columns would reduce its settlement to 0.31 x20, or
about 6 cm. For a less compact stone fill ($= 38”) the settlement ratio is
about 0.5. Under the circumstances just stated the settlement would be
54 H. B. Poorooshasb and G. G. Meyerhof
Physical dimensions
Column spacing
Column spacing is the factor which is likely to influence the behavior of the
system most strongly. To demonstrate this consider the data presented in
Fig. 3. The gravel in this case was assumed to be well compacted (@=44”)
and all other quantities are kept the same as shown in Fig. 2 except that here,
the spacing (= 2b) was increased to 6 times the diameter of a column. The
following conclusions may be drawn. The system is very inefficient, note that
the maximum UDL that can be sustained is only 40 kPa. At higher loads the
columns develop shear bands and fail. Also note that the value of the per-
formance ratio is very high, being equal to 0.85. Thus, in the example given
foundation settlement h, cm
1
30
foundation settlement b, cm
z --5
B _ (b) E = loo0 kPa, v=.2, +=44.2
G- spacing= 6 diameter
z _
I I I I I I
0 IO 20 30
foundation settlement h. cm
Fig. 3. Performance of the system when stone columns are installed at 6 diameter spacing.
Behavior of stone columns 55
TABLE 1
Performance ratio (PR) for various spacings and compaction, Es = 1000 kPa, v = 0.2
6,cm 0 L=5m,2a=l m
I 6
.d
0 L=10 m, 2a=I m
3 + L=15 m, 2a=1 m
(b) i
E
< I I I I I I
kl I
10 20 30
foundation settlement 6, cm
E
Possibility
of shear band
formation
0.8
.6 00 L=S
L=lOm,
m, 2a=.S
2a=1 mm
5
foundation settlement 6, cm
b
s=-
a
Behavior of stone columns 57
Poisson’s ratio v
Like ES, the influence of v on the performance ratio is quite small. For
example, assuming a column spacing of 2 and a very compact gravel
(4 = 449, the variation in this ratio (i.e. the PR value) is from 0.32 to 0.31 for
a Poisson’s ratio ranging from 0 to 0.5. The corresponding values for less
compact stones (4=41”) are 0.414.38 and for the uncompacted column
(4= 38”) 0.524.45., as in Fig. 5. In the range of recommended values for soft
clays, 0.15-0.25 [8], the variation is quite small and negligible even for an
uncompact column. It is concluded, therefore, that based on the assumptions
made in this paper, a soft soil’s mechanical properties do not appear to play
a major role in the overall performance of the foundation system. Recall that
performance is measured by the PR value only.
The two relevant mechanical properties of the stone fills are the maximum
angle of friction $J and a measure of the stiffness of the stone fragments after
placement and compaction in the holes. It is customary to specify the stiff-
ness of any granular material by its so called “initial tangent modulus”
which is the slope of a tangent to the stress-strain curve drawn at the origin
of the plot (i.e. where &1= 0). A more objective measure is the failure strain,
defined as the strain ,sl required to reach the peak stress (failure) point in the
stress-strain curve. In the present paper the failure strain, which will be
denoted by Er, is used. Note that in the stress-strain curves shown in Fig. 2a
the failure strains are 4% for @=44”, 5% for $=41” and 7% for $= 38”.
These curves (shown in Fig. 2a) will be referred to, in this paper, as the
standard curves and the columns filled with these as the standard columns.
I I I 1 I
.I .2 .3 .4 .5
Poisson’s ratio. v
Failure strain ef
Table 3 shows the effect of the failure strain on the computed PR values. In
this table the middle column in each category represents the PR associated
with the standard column. The numbers to the left of the middle columns are
those associated with a stiffer column (the failure strain, Er, for the material
used in their construction is chosen to be equal to one half of the standard
failure strain) while those to the right belong to a “less stiff column” for
which &r= 2&r (standard). The column stiffness does have an influence on the
PR value (see Fig. 6) although the influence does not appear to be as pro-
found as expected. Nevertheless its influence must be taken into account
when preparing the design charts.
Before leaving this section it must be pointed out that in the preparation of
Table 3 the total volumetric strain experienced by the “non-standard” sam-
ples has been assumed to remain the same as those of the standard samples.
Thus, for example, for the standard sample with a $=44” (Er=4%) the total
volumetric strain is, from Fig. 2b, about 10%. This value has been retained
in both the case of Er= 2% and Er= 8%.
Before the columns are installed the horizontal component of stress in the
ground is given by the equation Kay ‘z where z is the depth below grade and
K0 is the coefficient of the at rest earth pressure for the soft clay. Installation
of the columns increases this pressure to a higher value. If the soft soil were a
non-yielding material then the value of the initial ambient stress would be
Key’z where K0 would now be equal to the at rest earth pressure coefficient
TABLE 3
Effect of failure strain on the performance ratio
&f=2a &f=4 Ef”5 &f= 2.5 &f=s &f= 10 &f= 3.5 &f=l &f= 14
1.5 0.1 0.15 0.21 0.15 0.20 0.27 0.24 0.27 0.34
2 0.23 0.31 0.41 0.31 0.39 0.51 0.45 0.48 0.59
3 0.46 0.56 0.67 0.56 0.64 0.75 0.70 0.72 0.81
4 0.62 0.71 0.80 0.71 0.78 0.85 0.81 0.83 0.90
“Value of .q in %.
Behavior of stone columns 59
I c
Figure 6, a- Performance
I Ratio vs.
Spacing for Stone Columns, Standard I
= 38’, q=14%
,=41° , q=lO%
= 44‘-‘, Ed=8%
I I
9 = 38O, ~,=3.5%
$I=410 , q=2.5%
0 = ‘MO, Er=2%
Fig. 6. Performance ratio vs spacing for (a) standard stone columns; (b) stone columns less
stiff than standard; (c) floating stone columns stiffer than standard.
60 H. B. Poorooshasb and G. G. Meyerhof
of the stones. However, the soft soil would yield and during the act of com-
paction the lateral stress would certainly rise above this level. Once the
compaction is stopped relaxation causes the stresses to reduce to a lower
level, the magnitude of which cannot be decided with any accuracy. In spite
of this, and since the aim of the study at this stage is a sensitivity test, it was
decided to run an additional test by assuming a K0 value 25% in excess of the
K0 of the stone fillings. That is, Ke = 1.25( 1-sin@). The results are shown in
Table 4. A comparison of the PR values from this table and those presented
in the second column of Table 1 shows that the initial stresses are not likely
to be an issue. In this context it must be emphasized that the present analysis
is unable to account for the initial shear stresses produced at the column-soft
soil interface. The magnitudes of the inter-boundary shear stresses are prob-
ably small however and may be ignored.
This completes the study of the performance of end bearing stone columns.
The main conclusion to be drawn is a simple one and may be represented by
the equation;
PR = PR(s,4, q) (11)
DESIGN CHARTS
The design charts produced here are based on derived in the last section. For
a proposed s, 4 and ar they provide the value of PR which is, in turn, given
by the equation;
E’ (12)
HUDL
In Eqn 12 6 is the settlement of the foundation system, His the height of the
soft soil layer, UDL is the magnitude of the uniformly distributed load sup-
ported by the mat plus the self weight of the mat per unit area and E’ is the
modified Young’s modulus for the soft soil corrected for the Poisson’s ratio effect.
Three charts (see Fig. 6) are produced here assuming a value of v= 0 as
this provides the most conservative estimate of PR. The case where s = 1
represents a situation where the soft soil is completely dug out and replaced
by stones! Spacings larger than 4 diameters are not considered efficient.
TABLE 4
Performance ratio for & 25% in excess of K0
Spacing, s 1.5 2 3 4 6 10
-7
-6
Eq.
10 20 30 40
Area ratio Ar(=A,/A), %
Fig. 7. Settlement ratio vs area ratio for stone columns and lime columns. Comparison of
analytical results with field data.
64 H. B. Poorooshasb and G. G. Meyerhof
Also shown in the figure are certain field results from tests conducted on
lime columns. Here, the Priebe equation, which as pointed out earlier, is a
special case of the more general equation derived in this study, Eqn (7), agrees
well with the field test results. The Priebe equation is represented by the line
marked (d) for which the following data has been used; E, = 15000 kPa,
E, = 650 kPa. Using the same data and upper bound values for the Poisson’s
ratio of the soil (v,=O.25) and the lime column (v,=O.5), Eqn (7) is repre-
sented by the line marked (e) in the figure. Note that two of the test results
fall within the zone bordered by the Priebe equation and Eqn (7) which may
be viewed as the lower and the upper bounds for the settlement ratio,
respectively.
CONCLUSIONS
Based on the studies presented in this paper it may be concluded that the
factors that most severely effect the performance of a stone column founda-
tion scheme are the spacing (or the area ratio) and the degree of compaction
of the material in the columns which, in turn, control their strength, stiffness
and dilatation properties.
REFERENCES
1. Broms, B. B. and Bowman, P., Stabilization of Soil with Lime Columns. Design
handbook, Department of Soil and Rock Mechanics. Royal Institute of Tech-
nology, Stockholm, 1977.
2. Mitchell, J. K., Fundamentals of Soil Behavior. J. Wiley and Sons, Inc., New
York, NY, 1991.
3. Poulos, H. G. and Davis, E. H., Pile Foundation Analysis and Design. J. Wiley
and Sons, Inc., New York, NY, 1980.
4. Priebe, H., Abschatzung des setzungsverhaltens eines durch stopfverdichtung ver-
besserten baugrundees Die Bautechnik, 1976,54, 16&162.
5. Balaam, N. P. and Booker, J. R., Analysis of rigid rafts supported by granular
piles. Int. J. for Num. and Anal. Methods in Geomech., 1981, 5, 379403.
6. Poorooshasb, H. B. and Madhav, M. R., Application of rigid plastic dilatancy
model for prediction of granular pile settlement. Proceedings of the 5th ICON-
MIG, Nagoya, Japan, 1985, pp. 1805-1808.
7. Schweiger, H. F. and Pande, G. N., Numerical analysis of stone column sup-
ported foundations . Computers and Geotechnics, 1986, 2(6), 347-372.
8. Das, B. M., Principles of Geotechnical Engineering. PWS-KENT Publishing
Company, Boston, USA, 1990.
9. Castelli, R. J., Sarkar, S. K. and Munkfakh, G. A., Ground treatment in the
design and construction of a wharf structure. Proceedings of the International
Conference on Advances in Piling and Ground Treatment for Foundations.
Institution of Civil Engineers, London, 1983, pp. 275-28 1.
Behavior of stone columns 65
10. Greenwood, D. A. and Kirsch, K., Special ground treatment by vibratory and
dynamic methods. Proceedings of the International Conference on Advances in
Piling and Ground Treatment for Foundations. Institution of Civil Engineers,
London, 1983, pp. 1745.
11. Hughes, J. M. 0. and Withers, A. J., Reinforcing of soft cohesive soils with
stone columns. Ground Engineering, 1974, 7, 4249.
12. Watt, A. J., deBoer, J. J. and Greenwood, D. A., Loading tests on structures
founded on soft cohesive soils strengthened by compacted granular columns.
Proceedings of the 3rd Asian Conference on Soil Mechanics and Foundation
Engineering, Haifa, 1967, Vol. 1, pp. 248-251.
13. Munkfakh, G. A., Sarkar, S. K. and Castelli, R. J., Performance of a test
embankment founded on stone columns. Proceedings of the International Con-
ference on Advances in Piling and Ground Treatment for Foundations. Insti-
tution of Civil Engineers, London, 1983, pp. 259-265.
14. Meyerhof, G. G., Closing address. Proceedings of the International Conference
on Advances in Piling and Ground Treatment for Foundations. Institution of
Civil Engineers, London, 1983, pp. 293-297.
15. Bredenberg, H. and Broms, B. B., Lime columns as foundation for buildings.
Proceedings of the International Conference on Advances in Piling and Ground
Treatment for Foundations. Institute of Civil Engineers, London, 1983, pp.
133-138.
APPENDIX
The increment of the principal major strain d&i has two components; the
elastic and the plastic component. The elastic component is usually small and
may be ignored. The plastic component is derived from a potential function
CJJand with a magnitude depending on the amount of yielding, represented by
the function5 Thus
(Al.l)
Equation (A1.2) may now be integrated to show .sl =E~(v). The inverse of
this relation, viz.;
or &3= am. And since V= &t(q) + 2~~(n) = V(q) = V [I] = V(E~) in view of
Eqn (A1.3). Thus Eqn (2) is established.
(0
(ii)
Fig. A.1. Thick cylinder under internal pressure p and constrained from movement at its
outer radius.
Behavior of stone columns 67
exterior wall. A rigid frictionless piston forces the upper surface of the
cylinder downwards and in doing so causes the hole to reduce its diameter.
This is analogous to the action of the rigid mat supported by the stone
columns. The case shown in (ii) represents the column-soil interaction. It is
the well known thick cylinder problem. The stone column, because of its
tendency to deform laterally, exerts certain stresses on the interior wall of the
cylinder and causes the hole to expand. The solution to the problem (i.e.
Eqns (4) and (5) of the main text) are obtained by making the displacement
of the two cases compatible.
Case (i) will be treated first. In view of the axisymmetry of the problem and
with reference to the coordinate system shown in the figure the only non-zero
components of the stress tensor are err, O@ and oZZ= p. It will be assumed
that p is constant, a fact that will be demonstrated later on. The corre-
sponding strain components are Ebb,&06and E,, =E]. The use of p and &1
are in conformity with the notation employed in the main text. To ensure
equilibrium and compatibility of strains the following relations must hold;
da,,
-==l(q(j-orr) (A2.1)
dr r
(A2.2)
where
and
%9 = ; [%x3
- V(Gr+ P)] (A2.4)
1
E v(1 + v)
CT --
rr - 1 - Err + VEee + (A2.5)
y2
[
E ’
(A2.6a)
and
(A2.6b)
68 H. B. Poorooshasb and G. G. Meyerhof
Substituting the values of err and 0 88-~,, from Eqns (A2.5) and (A2.6a) in
Eqn (A2.1) and noting that E~Q= -u/r, F,,.= -du/dr (note the geomechanics
sign convention) results in;
(A2.7)
B
u=Ar+- (A2.8)
r
which must be solved subject to the condition that at r =b, u=O and that at
r = a, B,, = 0. The first condition immediately yields;
(A2.9)
Substituting for the values of A and B in Eqn (A2.8) results in the equation;
U v(l + v) b2 - a2
- r=a = - (A2.11)
r E ’ (b2 + a=) - v(b2 - a=)
Substituting for see + E,, in Eqn (A2.6b) and using the results in conjunc-
tion with the constitutive law F,, = ,sl = l/E Ip-v (CT,,+ am)] yields;
Note that for constant Ed, p is a constant (and does not depend on r). This
justifies the claims regarding the constancy of p made earlier. Also note that
if a=0 then;
Behavior of stone columns 69
l-v
Et + E’=
1 -v-2$
The results expressed by Eqns (A2.11) and (A2.12) are for case (i) of
Fig. A. 1. To complete the solution of the problem the results from case (ii) in
Fig. A.l, must be obtained and superimposed on these results. Thus the final
results obtained are;
_U 1 (b2 - a2)(1 + v)
(b2 + a2) - v(b2 - a2)
(A2.13)
r !.=*= -E*
1
2v 1 -v-2v2 (1 + v)u2
&zz = El = l_v 2v (A2.14)
+ (b2 + a2) - v(b2 - a2) h
For reasons of convenience Eqns (4) and (5) of the main text may be written
as;
a3 = Cp - DEB
where A,B,C and D are as defined in the main text (following Eqn (7)). Also,
the axial strain is given by the equation;
81 = i(q
c
- 2v,q) (A3.2)
Finally the equilibrium of the raft in the vertical direction requires that;
b2 - a2
UDL=pi--to,; (A3.3)
Noting that &3= -VIES, the first equation of the set (A3.1) reduces to;
70 H. B. Poorooshasb and G. G. Meyerhof
Substituting for this value of p in the second equation, the same set
provides the value of cr3;
Substituting for p and CT] from Eqns (A3.4) and (A3.5) in Eqn (A3.3)
results in Eqn (7) of the main text.