Sei sulla pagina 1di 27

INTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS

Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877


Published online 20 October 2008 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/nag.745

Consolidation around stone columns. Influence


of column deformation

Jorge Castro and Cesar Sagaseta,


Group of Geotechnical Engineering, Department of Ground Engineering and Materials Science,
University of Cantabria, Avda. de Los Castros, s/n 39005 Santander, Spain

SUMMARY
A solution is presented for the radial consolidation around stone columns under constant surcharge load.
The solution considers the influence of vertical and radial deformation of the column, either in elastic
and elastoplastic regimes.
The solution is in terms of the average excess pore pressure in the soil. It is based on previous
solutions, initially developed for rigid column, or including only vertical deformation. For elastic column,
the solution gives the variation of strains and stresses between the undrained and final states, for which
it coincides with the existing elastic solutions.
All the results are given in closed form, and both the elastic and plastic deformations of the column
lead to an equivalent coefficient of consolidation for the radial flow, which enables the application of the
existing methods of integration of the consolidation equation.
A parametric study is presented, showing the influence of the main problem features. A design example
is used to illustrate the application to practical cases. Copyright q 2008 John Wiley & Sons, Ltd.

Received 23 August 2007; Accepted 23 August 2008

KEY WORDS: stone columns; consolidation; elastoplastic; analytical solution

INTRODUCTION

Stone columns, either by the vibro-replacement or vibro-displacement methods, are one of the most
common improvement techniques for foundation of embankments or structures on soft soils. The main
effects usually considered with respect to the untreated ground conditions are: improvement of bearing
capacity, reduction of total and differential settlements, acceleration of consolidation, improvement
of the stability of embankments and natural slopes, and reduction of liquefaction potential.

Correspondence to: Cesar Sagaseta, Group of Geotechnical Engineering, Department of Ground Engineering and
Materials Science, University of Cantabria, Avda. de Los Castros, s/n 39005 Santander, Spain.

E-mail: sagasetac@unican.es

Contract/grant sponsor: Spanish Ministries of Public Works and of Education

Copyright q 2008 John Wiley & Sons, Ltd.


852 J. CASTRO AND C. SAGASETA

Unlike other improvement techniques, stone columns are considered not to affect significantly
the properties of the surrounding ground. They act mainly as inclusions with a higher stiffness,
shear strength, and permeability than the natural soil.
The basic mechanisms by which these three properties produce the above effects are reasonably
well understood. Owing to the relatively high column/soil stiffness ratio, part of the applied
load is taken by the columns, and this means a reduction of soil settlement. The acceleration of
consolidation is mainly governed by the column permeability and spacing, in a similar way to
vertical drains. On the other hand, the shear strength of the treated soil is a weighted average of
the strengths of soil and gravel. In conventional calculations, final settlements (and sometimes,
the immediate ones) are calculated first, using several available methods for the stress distribution
between soil and columns (settlement reduction analysis). Then, the evolution of these settlements
with time is worked out independently by means of more or less simplified formulations of the
theory of consolidation.
However, there are some couplings between these mechanisms, which must be taken into
consideration in a refined analysis. The column/soil stiffness ratio varies along time due to soil
volume change during consolidation. So, the fraction of the vertical load carried by the columns
also varies. On the other hand, the columns deform and bulge under this vertical load, exerting
some horizontal pressure against the surrounding soil. This horizontal pressure, also variable with
time, has an influence on the consolidation process of the surrounding soil.
All of the above effects can be included in refined numerical models. However, closed-form
analytical solutions are useful for the identification of their relative influence, and still form the
basis of methods of calculation commonly used in practice. The aim of the present paper is to
develop a simplified and closed-form analytical solution for the consolidation around stone columns,
including their radial deformation. Soil is assumed to be elastic, but plastic strains (dilatancy) are
considered for the columns. The time-dependent stress distribution between columns and soil is
compatible with the conditions at the start and the end of the consolidation period. The presence
of zones around the column with different permeabilities, associated to smear or contamination of
gravel with fines, is not considered.
The proposed solution is an extension of the existing ones, including the above features. So, a
review is presented of these basic approaches.

REVIEW OF PREVIOUS ANALYSES

Final and undrained states. Soilcolumn stress transfer


The analysis of the stress distribution between columns and soil is based on imposing the conditions
of vertical equilibrium and compatibility of vertical deformations.
Under embankments or large uniformly loaded areas, it is convenient to consider a representative
cylindrical unit cell formed by a column and its surrounding soil (Figure 1). The boundary
conditions at the outer wall are: zero shear stress, zero radial displacement, and no water flow.
The top (surface) and bottom (bedrock) boundaries are considered as smooth (zero shear stress).
When a vertical uniform unit load pa is applied at the surface, the equilibrium condition in vertical
direction is

pa = zc ar +zs (1ar ) (1)

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 853

pa

sz sr
rc rs d rl
rc
u
zc
zs
L

STONE
SOIL kz
COLUMN
kr rl
Es s
rc
Ec c c c
d

rc
rl

Figure 1. Unit cell.

where zc and zs are the total vertical stress increments on the column and on the soil, respectively
(zc /zs is usually called stress concentration ratio). The upper bar denotes average value of the
variable along the radius. ar is the area replacement ratio (area of the column, Ac , divided by the
area of the cell, Al ), sometimes defined also in terms of diameters or radii, N =rl /rc = (1/ar )1/2 .
Regarding the compatibility of deformations, it is usually assumed that the settlement of the
ground surface (soil and columns) is uniform (equal vertical strain). This condition has proved
to be more realistic under foundation slabs or embankments than the other extreme alternative (the
so-called free vertical strain).
The formulation of this condition in terms of stresses requires the use of some constitutive
model. Linear elastic behaviour is a reasonable assumption for the soil, because the partial lateral
confinement provided by the columns produces small shear strains. However, the columns are
surrounded by a softer material, and even for moderate loads they can reach the active limit
condition. So, elastic behaviour is regarded only as a first approximation for the columns.
In the following, different elastic parameters are used: Youngs (E), shear (G), bulk (K ) and
oedometric (E m ) moduli, Lames constant (), Poissons ratio (), in order to get simple and clear
expressions. Obviously only two of them are independent.
The simplest approach is to assume full lateral confinement in the columns and in the soil. Then,
the final vertical strains at the end of consolidation are related to the vertical effective stresses
through the respective oedometric (constrained) moduli:
zs,f zc,f
zs,f = zc,f = = (2)
E ms E mc
In the following, the upper bars are removed when the variable is uniform, as is the case for the
vertical stresses in the soil at the end of consolidation, and at any time in the column.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
854 J. CASTRO AND C. SAGASETA

Equations (1) and (2) can be solved for the final stresses on the columns and soil:
pa
zs,f =   (3)
E mc
1+ar 1
E ms

The effect of the columns is usually expressed as an improvement factor, n, defined as the quotient
between the final settlement without and with columns. The direct proportionality between strains
and effective stresses (2) means that the improvement factor is
pa
 
sz0 z0 E E mc
n= = =  ms = 1+ar 1 (4)
sz z zs,f E ms
E ms

Equation (2) means that the final stress concentration ratio is equal to the constrained modular
ratio. However, this is in contradiction with experience: the modular ratio is usually in the range
1050, while the stress concentration ratio measured in actual cases is much lower, in the range
310 [1]. Under the lateral confinement assumption, this range of stress concentration factors can
be obtained only using either extremely low modular ratio (E c /E s 5), or unrealistic values of
the soil Poissons ratio (s 0.45) [2].
The reason for this discrepancy is commonly attributed to the influence of the column lateral
deformation. Several methods have been developed to include, at least partially, this effect.
The semi-empirical method by Priebe [3, 4] is probably the most successful and popular. The
column is considered as rigid-plastic, with infinite modulus of elasticity, yield limit at the active
state and plastic deformation at constant volume (zero dilatancy). Some assumptions of semi-
empirical nature are made along the analysis [5]: the vertical strain in the soil is calculated with
the condition of zero horizontal strain, and then the radial strain is calculated from the cylindrical
cavity expansion solution with zero vertical strain; the variation of stresses with depth is also
introduced as an empirical correction. The solution is given as an improvement factor of the same
general form of (4) but with a different stiffness ratio.
Van Impe and De Beer [6] present a solution including soil and columns lateral deformation. In
order to get a closed-form solution, the problem is simplified to a plane strain case, substituting
the columns by equivalent infinite trenches. The elasticperfect plastic behaviour of the column
material is modelled as a constant volume deformation.
On the other hand, the elastic solution by Balaam and Booker [7, 8] is more rigorous, with
simultaneous consideration of the horizontal and vertical components of the deformation. The
vertical soil stress, written in the form (3) is
pa
zs,f =   (5)
E mc 2c F
1+ar 1
E ms +2s Far /(1ar )

where
(1ar )(c s )
F=
2[ar (s + G s c G c )+c + G c + G s ]

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 855

The improvement factor is


 
E mc 2(c s )F
n = 1+ar 1 (6)
E ms

Comparing Equations (3) and (5), it is clearly seen that the lateral deformations always reduce
the stress concentration ratio by subtracting a term to the oedometric modulus of the column and
adding another term to that of the soil. However, the stress concentration factor (SCF) is still
higher than observed, due to the elastic behaviour assumed for the column. It must also be noticed
that, comparing Equations (4) and (6), the improvement factor is also reduced, but with a different
modification of the column and soil moduli.
Taking the appropriate elastic parameters, the analysis is applicable not only at the end of
consolidation, but also at the undrained loading situation. Of particular interest is the vertical strain,
which gives rise to a non-zero immediate settlement:
pa
zs,u = (7)
H (13ar )(G c G s )

Also, the excess pore pressure in the soil, which is equal to the total octahedral normal stress:

u u = oct,s,u = H zs,u (8)

Pulko and Majes [9] consider the column material as elasticperfect plastic, with the Mohr
Coulomb yield criterion and non-associated plastic flow with constant dilatancy. A closed-form
solution is also obtained, with the stresses varying with depth. The stress concentration is lower than
in the elastic case. However, the solution is only applied to the final state, without consideration of
the proper loading history (undrained loading and consolidation process). As it will be shown later,
this has a significant influence on the final solution, particularly on the depth reached by plastic
state in the column.

Consolidation analyses
Stone columns accelerate the consolidation rate providing a shorter drainage path, in a similar way
as prefabricated vertical drains. Hence, the classical solutions of Barron [10] and Hansbo [11],
available for radial flow around PVD, were taken and applied also for stone columns.
The general mass conservation equation for 3-D consolidation was originally formulated by
Biot [12] based on the following assumptions:

(a) The soil is fully saturated.


(b) The water and soil particles are incompressible, i.e. the change in volume is only due to the
leak of water.
(c) Darcys law is obeyed.
(d) Elastic soil.
(e) Small strains.

For the consolidation around drains and stone columns it is also convenient to work with the
described unit cell, using cylindrical coordinates (r, z, ) (see Figure 1). Then the general equation,

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
856 J. CASTRO AND C. SAGASETA

in the case of anisotropic permeability, is:


 2  2
kr * u(r, z, t) 1 *u(r, z, t) k z * u(r, z, t) *vol,s (r, z, t)
+ + = (9)
w *r 2 r *r w *z 2 *t

The right-hand member can be transformed in terms of the excess pore pressure with the aid of
the following additional simplifying assumptions:
(f) The soil deforms under lateral confined conditions, with zero horizontal strain at any point.
This is based on the boundary conditions of zero radial displacement at the outer wall, and
also at the soildrain interface (for stone columns, it is only a first approximation). This
means a significant simplification, because the volumetric strain, vol , coincides with the
vertical strain, z , which in turn by hypothesis (d) is related to the soil vertical effective
stress, zs :

zs
vol,s = zs = (10)
E ms
(g) The soil vertical strain, zs , and then also, zs , are uniform in any horizontal plane (equal
vertical strain). Then, although u and zs vary with r , their difference is constant (Figure 2),
and they can be substituted by their respective average values along the radius:

zs (r, t) = zs (r, t)u(r, t) = zs (t)u(t) (11)

Then
*vol,s *zs 1 *zs 1 *(zs u) 1 *(zs u)
= = = = (12)
*t *t E ms *t E ms *t E ms *t

z (average) 'z
u, z

z' (average)
u

u (average)

Soil Outer wall


Drain / column of cell

rc rl
r

Figure 2. Radial variation of vertical stresses and excess pore pressure in Barrons solution.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 857

(h) The average total vertical stress in the soil is constant with time and equal to the applied
load, pa . This is clear with vertical drains, but for stone columns it must be considered only
as a crude first step:
zs = pa , zc = 0 (13)
Then, from Equation (12):
*vol,s 1 *( pa u) 1 *u
= = (14)
*t E ms *t E ms *t
For the case of pure radial consolidation (no vertical flow):
 2 
* u 1 *u *u
cvr + = (15)
*r 2 r *r *t
where cvr is the radial coefficient of consolidation of the soil:
kr
cvr = E ms (16)
w
BarronHansbo solution of Equation (15) is
     
u r 1 r 2
u= ln ar 1 (17)
f (ar ) rc 2 rc

with the average:


u = u u e8Tr / f (ar ) (18)
where Tr is a time factor (Tr = cvr t/4rl2 ), u u is the undrained excess pore pressure (u u = pa ),
and
ln ar 3ar
f (ar ) = (19)
2(1ar ) 4
This solution has also provided the basis for the equivalence conditions from the actual 3-D
configuration to plane strain approximations [13] of PVD under embankments.
In the case of combined vertical and radial flows, they can be uncoupled as demonstrated by
Carrillo [14] for homogeneous soil. Recently, new fully coupled solutions have been developed
[15]. This allows including the effect of the smear zone and the well resistance around drains in
a more refined way because the uncoupling is not strictly valid under these conditions.

Influence of column vertical stiffness


As stated above, the stone columns carry a substantial part of the applied load. Lane [16] had
already pointed out that Barrons solution ignored the effect of the different stiffnesses between the
sand drains and the surrounding soil on the consolidation rate. This problem was later forgotten
due to the reduced diameter and stiffness of prefabricated drains. However, it emerged again for
the application to stone columns, with a column/soil modulus ratio typically of 1050, and area
ratio ar of 530%.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
858 J. CASTRO AND C. SAGASETA

The simplest way of considering this feature is to abandon partially the assumption (h) of the
above list, and to take an average total vertical stress on the soil, zs , constant with time, but different
from the applied load. It is usually calculated at the end of consolidation. For consistency with
assumption (f), it must be calculated with lateral confinement (Equation (3)). However, in practice
any of the methods described in the preceding section can be used. Then, the Barron/Hansbo
solution to Equation (9) is applied directly with u u = zs .
The obvious result of this approach is that the settlements are reduced with respect to the untreated
case in direct relation to the ratio zs / pa , but the relative consolidation time rate is unchanged.

Time-dependent vertical stiffness ratio


The column/soil stiffness ratio varies considerably during consolidation. Initially, soil deforms
in undrained condition, and hence, with a relatively high stiffness (zero settlement if lateral
confinement is assumed). With consolidation, soil apparent modulus gradually reduces towards its
drained value.
This means that at the undrained stage, the load transfer to the columns is less important, and it
increases as consolidation proceeds. Inversely, the soil is subjected to a higher load at the beginning,
implying some degree of pre-loading with respect to the final soil stress. This produces a faster
consolidation compared with the case of constant load.
This has been considered by Han and Ye [2]. It implies the full abandon of assumption (h).
Equation (9) holds, and the evaluation of vol (14) must be modified to include the variation of
total stress with time.
This can be done, imposing equilibrium (1) and compatibility condition (2), extended to any time:
zs zs u zc
zc = zs = = = (20)
E ms E ms E mc
Solving Equations (1) and (20) for the total soil stress, a generalization of Equation (3) is obtained,
in terms of the excess pore pressure:
E mc
pa + ar u(t)
E
zs (t) = ms  (21)
E mc
1+ar 1
E ms
Substitution into (12) gives
*vol,s 1 *u
= a (22)
*t E ms + E mc
r *t
1ar
Comparing this expression with Equation (14), it comes out that the consolidation equation (15)
keeps the same form as before, but with a modified coefficient of consolidation
 
E mc ar
ze
cvr = cvr 1+ (23)
E ms 1ar
where the superscript ze refers to the influence of elastic column vertical deformation. It is always
ze >c , so this implies a faster consolidation than for constant load.
cvr vr

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 859

Figure 3. Time-dependent stress transfer in Han and Ye [2] solution.

Figure 3 shows the results obtained in an application example [2]. The stress concentration
ratio increases with time from zero to the final value (3). The example chosen has a constrained
modular ratio E mc /E ms = 5, lower than the usual range in real soils.

PROPOSED SOLUTION

All the above solutions keep the assumption of lateral confinement during consolidation, ignoring
the horizontal interaction between column and soil. The main objective of the proposed solution
is to include the radial deformation of columns in the radial consolidation analysis, removing the
assumption (f) of the above list, and imposing the equilibrium of radial stresses at the soilcolumn
interface. Nevertheless, the solution is kept as simple as possible.
In a first step, elastic behaviour is assumed for the soil and the column. Consolidation takes
place starting from the elastic solution by Balaam and Booker [7] for undrained loading. Then,
plastic deformation of the column is considered, both in the starting undrained condition and during
consolidation until reaching the final state.

Horizontal deformation. Basic assumptions


The severe condition of zero horizontal strain is substituted by a state of radial deformation of the
column and of the soil at each horizontal plane. For the calculation of the soil volumetric strain,
Equation (9) must be replaced by
1 
vol,s =  (24)
K s oct,s
The determination of the octahedral effective stress in the soil implies the solution of a fully coupled
problem. However, a reasonably accurate simple solution can be obtained by evaluating the soil
volumetric strain taking a constant pore pressure, equal to its average value along the radius, u.
This is the main assumption of the present approach, and it enables to decouple the problem. It is
a natural development of the Barron/Hansbo approach, which also uses the average pore pressure

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
860 J. CASTRO AND C. SAGASETA

Figure 4. Equilibrium and compatibility conditions between soil and column in the elastic case.

to evaluate the right-hand member (time derivative) of the consolidation equation (9). It can be
justified by the fact that the horizontal strains are by far much smaller than the vertical ones, and
they can be treated in this approximate way. This means in any case a substantial improvement
with respect to the previous solutions, which assume them to be zero.

Elastic solution
Figure 4 shows a horizontal slice of the elementary cell. The column is a vertical solid cylinder
subjected to a vertical uniform pressure zc and a radial pressure r c at its lateral wall. The soil is
a cylinder with a central cylindrical cavity, subjected to a vertical uniform effective pressure zs ,
a radial pressure r s (= r c ) at the cavity wall (soil/column interface), and an internal excess pore
pressure u. These five pressures determine the stresses and strains at any point of the soil and the
column.
The conditions of vertical (1) and radial equilibrium and vertical and radial compatibility of
deformation at the soilcolumn interface must be imposed. These four equations allow to express
the above four vertical or radial pressures in terms of the pore pressure u and the applied vertical
pressure pa only. Given the linear character of the problem, all the stresses and strains components
at any radial distance, r , have the general form
(, ) = e u +e pa (25)
where the functions ( ,  ) depend on the radial coordinate, r , problem geometry and material
e e

properties. Superscript e refers to the elastic nature of the solution. Only the pore pressure u is
a function of time.
This linearity allows for definition of a unique degree of consolidation, valid for any component
of stress or strain
(, )(, )u
U= (26)
(, )f (, )u
In particular, for the average excess pore pressure, the final value is zero, so
u u u u u
U= = 1 = 1 (27)
u u uu oct,s,u

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 861

The solution is detailed in Appendix A. In particular, the volumetric strain in the soil is constant
with the radius, and is given by

vol,s = evs u +evs pa (28)

and its time derivative:


*vol,s *u
= evs (29)
*t *t
A useful expression for evs can be found if Equation (28) is applied to the undrained (u) and final
(f) states:
 
vol,s,f vol,s,u 1 oct,s,f oct,s,u
evs = = (30)
u f u u Ks u f u u
The final pore pressure is obviously zero (u f = 0). In the undrained stage, the volumetric strain is
also zero (vol,s,u = 0). Then

vol,s,f 1 oct,s,f
evs = = (31)
uu K s oct,s,u

Substituting (31) in (29):



vol,s,f 1 oct,s,f
evs = = (32)
uu K s oct,s,u

Comparing this expression with Equation (14), the consolidation equation is again the same as in
the basic case, but with a new modified coefficient of consolidation:
K s oct,s,u 1+s oct,s,u
zr e
cvr = cvr  = cvr (33)
E ms oct,s,f 3(1s ) oct,s,f

where the superscript zre refers to the influence of vertical and radial elastic deformation of the
column. The value of cvr zr e can be calculated from Equation (33) once the stresses and strains are

known in the undrained and final stages. For this purpose, the solution by Balaam and Booker
[7, 8] is consistent with all the assumptions made and can be used. The final result is

[ar (c +2G c )+(1ar )(s +2G s )][H (c s )](1ar )(c s )2
zr e
cvr = cvr (34)
(s +2G s )[H (13ar )(G c G s )]

with H = (1/ar )(c + G c + G s )(G c G s ).


The quotient (cvr
zr e /c ) depends only on Poissons ratios of the soil and of the column ( ,  ),
vr s c
the modular ratio (E c /E s ), and the area replacement ratio (ar ). With the resulting value of cvr
zr e ,

the general equation of consolidation can be integrated by any of the usual methods. The result
is the variation of the average excess pore pressure u with time. Then, Equation (27) defines the
degree of consolidation, U . Equation (26) is then used to interpolate any stress or strain component
between the undrained and final states.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
862 J. CASTRO AND C. SAGASETA

Plastic deformation of the column


Plastic strains in the column can be adequately modelled with MohrCoulomb yielding criterion
and a non-associated flow rule for the plastic strains, with a constant dilatancy angle (c  = c ):

r c 1sin c
= = kac (35)
zc 1+sin c
p
zc 1sin c
p = = k c (36)
2r c 1+sin c

In the following derivation, the increments of elastic strains in the column during plastic deformation
are neglected. Hence plastic components, when they exist, are equal to the total strains, and
superscripts p can be suppressed.
In the elastic analysis presented in the preceding section, the problem was formulated in terms of
stress increments produced by the applied load, which in the elementary cell considered (Figure 1),
were constant with depth. Hence, the radial consolidation process was the same in any horizontal
slice, and the vertical coordinate z could be ignored in the derivation. In the elasticplastic analysis,
the effective stresses in the yield condition (35) must include also the previous stresses existing
before the load application. As these stresses can in general vary with depth, the analysis must
be carried out in terms of z. Being an elasticplastic analysis, the final state is highly dependent
on the loading path followed. Therefore, the analysis must be performed along the time, keeping
track of the development of stresses and strains at each depth during the process.

Second step
First step
rc/zc=kac rc/zc=kac
/zc=kac
rc
I: initial
k0c U: undrained loading k0c F k0c
Y: yield
F: final, drained
F
Y

Y
Vertical stress, zc

Vertical stress, zc
Vertical stress, zc

U
F
F I
1.0 1.0 1.0

U U

kpc kpc kpc


U
I I
I

O O O
(a) Radial stress, rc (b) Radial stress, rc (c) Radial stress, rc

Figure 5. Stress paths in the column in typical conditions: (a) elastic case;
(b) elasticplastic case; and (c) two-step load.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 863

Figure 5 illustrates the column stress paths in the (r c , zc ) plot in a typical case. If the effects
of the columns installation are neglected, the initial stresses (point I) are

zs,i = s z

zc,i = c z
(37)
r s,i = r c,i = k0s zs,i

ui = 0

All the column is in elastic state, because k0s >kac always, and in most cases k0s >k0c . Immediately
after the load application, the soil undrained stiffness is higher than the drained stiffness of the
column, and so it takes more vertical load than the column. Therefore, the column is well confined
by the surrounding soil, with a vertical stress increment even lower than the radial one (point U).
When the soil starts to consolidate, part of the vertical load is progressively transferred from the
soil to the column, and at the same time the lateral stress in the columnsoil interface decreases.
The column can remain elastic until the end of consolidation (case a), point (F), or reach the plastic
state at some time (case b), point (Y).
As the initial stresses (point I) increase with z and their increments (path IUF) are independent
of depth, the worst condition happens at the surface. Yielding starts at the surface and it progresses
downwards with time. Once yielded, the vertical stiffness of the column decreases, and also its
radial deformability increases due to dilatancy. This affects the transfer of vertical load from the soil.
Then, the vertical load on the column varies with depth. This means that for vertical equilibrium
some shear stresses must appear at the columnsoil interface. However, this is considered as a
second order effect, and it is ignored in the analysis. In any case, below the limit of the plastic
zone, stresses are constant with depth, and the preceding elastic solution is valid.
Then, at any depth, z, the time of yielding can be obtained from the yield condition (35) applied
to the elastic solution at the elasticplastic interface:

r c,y k0 s z +r c,y 1sin c


=  = = kac (38)
zc,y c z +zc,y 1+sin c

where the stresses increments (denoted as ) are due to the load application and the suffix y
refers to the moment of yielding.
Each stress increment can be expressed according to (26):

y = u +Uye (ef u ) (39)

where Uye is the elastic degree of consolidation at the moment of column yielding at the depth
considered, defined as the interpolation factor between the undrained stress increments and the final
ones if the column would behave elastically until the end of consolidation, ef . Owing to plastic
deformation, the true final stresses will differ from ef , and so Uye is not the true interpolation
factor. In fact, it will not be the same for all the components. Only for the pore pressure the final
value is obviously unchanged with respect to the elastic case (u f = 0), so

u y = (1Uye )u u (40)

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
864 J. CASTRO AND C. SAGASETA

Hence, Uye is the true degree of consolidation in terms of average excess pore pressures at the time
of yielding. From (38) and (39)
(k0 s kac c )z +r c,u kac zc,u
Uye = (41)
kac ezc,f rec,f +r c,u kac zc,u
In Equation (41), only the first term in the numerator depends on z. The rest are undrained or final
elastic stress increments, all of them proportional to the applied load pa . Hence, Uye depends on
the depth z and on the load pa only through the factor (k0 s kac c )z/ pa . Obviously, its value at
the surface (z = 0) is independent of the magnitude of the applied load.
Equation (41) defines the time for yielding at any depth z. Alternatively, it can be viewed as
defining the depth z c reached by the plastic zone for any given time or elastic consolidation ratio U e .
The determination of stresses and strains during the plastic phase is detailed also in Appendix A.
Table II shows the increments of stresses, strains, and displacements from the moment of yielding.
The elastic strains in the plastic phase have been neglected. It must be noticed that the plastic
deformation affects only the column, the soil is always elastic.
Once the undrained and final situations are defined also for the plastic phase, the consolidation
process for the plastic increment can be studied using any of the available solutions of the radial
consolidation equation, in the same way as it was done for the elastic column. The modified
coefficient of consolidation is now
zr p 1+s uy
cvr = cvr (42)
3(1s ) oct,s,f oct,s,y


where u y is defined by Equations (40) and (41), and the final and at yield octahedral stress
increments can be taken from Table II with u = 0 and u = u y , respectively. The final result is
 
ar s J
(1ar )+ +
zr p (s +2G s ) kc kac
cvr = cvr    (43)
ar ar
1+ 1ar +
(1ar )kc kac
with J = s + G s +ar (s + G s )/(1ar )kc .
zr p
The quotient (cvr /cvr ) depends only on Poissons ratio of the soil (s ), the plastic parameters
zr p
of the column ( c , c ), and the area replacement ratio (ar ). With the resulting value of cvr , the
solution follows the same procedure as for the elastic phase.
Unlike the elastic case, the final state reached after consolidation does not coincide with the
elasticplastic analysis of the case of fully drained loading (infinite soil permeability) [9], because
the stress paths followed are very different. As shown in Figure 5, all the undrained loading takes
place within the elastic range, and this affects to the magnitude of the radial column displacements,
which in turn govern the soil/column stress transfer.
All the above is for the case of an instant load, pa , applied instantaneously to the cell initially
subjected only to its self-weight. In the more general case of multi-step loading, the procedure can
be applied to each loading step, taking as initial stresses the final ones after the previous load step.
Even if the column is in the plastic range, the undrained loading brings it back to elastic range,
due to the increase of lateral stresses (see Figure 5(c)).
This general multi-stage load is not considered here. The following paragraphs are restricted to
the case of a single step load.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 865

Integration for the whole column


As the moment of yielding Uye is different for each depth, the surface settlement at any time must
be obtained by integration of the vertical strain along the column length, L.
There are some options: (i) analytical integration; (ii) numerical integration; (iii) use of the
situation at a representative point for the whole column (the mean depth z = L/2 gives the best fit
to the analytical integration).
The settlement-time law is by far the most important result from the analysis, because surface
settlement is the variable most often monitored in actual cases. For this reason, the result of
the analytical integration is given here, using the BarronHansbo closed-form solution for the
consolidation at any depth, z. Integrating the consolidation rate for the whole column, the following
expressions are obtained for the settlement of the column head, s, in the three different phases:
Phase A (z c 0), the whole column is elastic:
sz = L[z,u +(z,f z,u )(1e(8Tr / f (ar ))cvr
zr e /c
vr )] (44)
Phase B (0<z c <L), the upper part of the column (zz c ) is plastic and the rest (zz c ) is
elastic:
sz = (L z c )[z,u +(ez,f z,u )(1e(8Tr / f (ar ))cvr
zr e /c
vr )]+ z c [z,u +(ez,f z,u )P(z c )]
p
z zr p
(8Tr / f (ar ))cvr /cvr
+
y u u {z c [1 P(z c )]+ Q(z c )e } (45)
u
Phase C (Lz c ), the whole column is plastic:
p
z zr p
(8Tr / f (ar ))cvr /cvr
sz = L[z,u +(ez,f z,u )P(L)]+ y u u {L[1 P(L)]+ Q(L)e } (46)
u
where
z 2ep

u  e 
ef u f
z 2ep ef
P(z) = 2e , Q(z) = eu
f u (2ep )
ef u f

k0 s kac c


zr p
kac zc r c cvr
= ,
= , ep = zr e
kac kac cvr
and where ( u , z,u ) and ( ef , ez,f ) refer, respectively, to the undrained state and to the final
p
one in the assumption of elastic behaviour, as defined from Equation (39). The term z /u y
can be taken from the first row of Table II for u = 0.

PARAMETRIC STUDY AND COMPARISON WITH OTHER PREVIOUS SOLUTIONS

Consolidation
As it has been shown, the consolidation process around a deformable stone column is governed by
the same differential equation as for a prefabricated drain (negligible stiffness), but using equivalent
coefficients of consolidation to cover different assumptions for the column deformation.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
866 J. CASTRO AND C. SAGASETA

Figure 6 shows the equivalent values of cvr for elastic column (Equation (34)). The equivalent
coefficient of consolidation is always greater than the basic one, indicating a faster consolidation in
any case. In the figure, the proposed solution is compared with the assumption of zero horizontal
strain [2]. The lateral confinement leads to much higher values of cvr . This effect was also
observed for the analysis of the final stress transfer (Equation (4)), where abnormally high stress
concentration ratios were also obtained. The introduction of the column radial deformation reduces
drastically this effect. In a typical case, with ar = 0.20 and E c /E s = 20, the correction factor to
cvr with lateral confinement (Equation (23)) would be 6.0, and it reduces to 3.55 when lateral
deformation is considered.
If area replacement ratio increases approaching 1 (i.e. almost all the soil is replaced with gravel),
the equivalent cvr goes to infinite for lateral confinement, but when radial strain is considered it
decreases and it tends back to the basic value. This apparently slower consolidation is balanced
by a very high immediate settlement, which was not considered assuming lateral confinement.
The influence of the column plastic behaviour (Equation (43)) is shown in Figure 7. The plastic
deformation makes the column softer, thus reducing its capacity to carry load, and leading to
equivalent values of cvr even lower (0.80.9 times) than the basic one. This means a considerable
reduction with respect to the elastic regime, where the coefficient of consolidation is typically 23
times the basic one.
Figure 8 shows a case of elastic column analysed by Balaam and Booker [7] using a numerical
model for coupled radial consolidation. It is compared with the present solution, and also with the
assumption of lateral confinement [2], and the basic solution with constant vertical load (Barron).
In the original paper, the results of the numerical analysis were given in terms of consolidation
settlements, subtracting the immediate settlements to allow for comparison with Barrons solution.
Unfortunately, the eliminated immediate component was not given in the paper, so the numerical
results shown in Figure 8 have been obtained by adding the immediate settlement calculated from
the analytical solution. As can be seen, consolidation under constant load (Barron solution) is
much slower than the numerical analysis. On the contrary, the assumption of variable load with

10
Present solution
9 No radial deformation [2]

8 (c=s=0.3)

7
Ec/Es=40
cvr /cvr , cvr /cvr

6
ze

5
zre

Ec/Es=20

Ec/Es=10
3

Ec/Es=5
2

Ec/Es=1
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Area replacement ratio (a r )

Figure 6. Equivalent coefficient of consolidation. Elastic column. Influence of modular ratio.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 867

1.00

c=20

0.95

cvr /cvr
zrp 0.90

c=0
0.85
c=10 c=50
c=40
c=30

(s= 0.3)
0.80
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Area replacement ratio (ar)

Figure 7. Equivalent coefficient of consolidation. Plastic column. Influence


of the friction and dilation angles.

0
Ec/Es=10
as=0.25
c=s=0.3

20
Elastic settlement (%)

40

60

80 Balaam and Booker [7] modified


Present solution
No radial deformation (Han and Ye [2])
Constant load (Barron, [10])

100
0.0001 0.001 0.01 0.1

Time factor Tr

Figure 8. Comparison of elastic solutions. Case analysed by Balaam and Booker [7].

lateral confinement [2] results in a too fast process, particularly for consolidation rates above 50%
(immediate settlements are not considered).
The present solution shows a better agreement with the numerical results. The differences are
greater for degrees of consolidation below 30%. However, this happens also with all the other
approaches. Balaam and Booker [7] attribute this fact to the assumptions that are inherent to the
Barrons solution, regarding the absence of radial displacements, which are particularly important
in this part of the process. In any case, even in this region the results of the present method mean
a substantial improvement with respect to the previous ones.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
868 J. CASTRO AND C. SAGASETA

Column yielding

0.2

Degree of consolidation, U = 1 - u / uu
pa=50 kPa z=0 m
0.4
ar=0.11
Ec/Es=40
Es= 1000 kPa z=5 m
0.6 c=s=0.33
c=40
c=0
3 z=10 m
's=10 kN/m
0.8 3
'c=10 KN/m
k0s=0.6
Elastic

1.0
0.0001 0.001 0.01 0.1 1
(a) Time Factor (Tr )

Column yielding

0.005

Elastic
Vertical strain, z

0.010

z=10 m
0.015

z=5 m
0.020

z=0
0.025
0.0001 0.001 0.01 0.1 1
(b) Time factor, T r

Figure 9. Solution for a typical case. Elasticplastic column: (a) pore pressure and (b) vertical strain.

A general case, involving plastic strains in the whole column, is presented in Figure 9. The degree
of consolidation in pore pressures for different depths is shown in Figure 9(a). There is a basic
unique elastic curve, valid for all depths. The degree of consolidation at the moment of column
yielding is 38% at the surface, 55% at mid-depth (z = 5 m), and 72% at the base (z = 10 m). The
strains are shown in Figure 9(b). The differences in the moment of yielding produce differences in
the final strains, which are constant in the elastic region, but greater in the upper part of the column.
Owing to linearity with z of Equation (41), the vertical strain at any time varies linearly with z,
so the settlement of the plastic part of the column is given defined by the strain at its mid-depth.

Stress concentration factor


The influence of the horizontal deformation and plastic behaviour of the column on the distribution
of stresses between soil and column is pointed out in Figure 10. With lateral confinement [2],

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 869

45

Ec /Es=40 No radial deformation


40 c =s =0.33 (Han and Ye, 2001)
ar=0.25
35 z=0
=40

30

SCF (zc /zs)


25 Present solution
elastic column

20

15

10 Present solution
elastic-plastic column
(=20)
5
(=0)
yielding point
0
0.0001 0.001 0.01 0.1 1

Time Factor (Tr )

Figure 10. Stress concentration factor. Influence of radial deformation and plastic strains.

3.0
Ec/Es=40
c=s=0.33
2.5 ar=0.25
zc a
Averaged vertical stresses, /p and /p

z=0 Column
=40
=20
2.0
zs a

1.5
yielding point

1.0

Soil (Total)
0.5

Soil (Effective)
0
0.00001 0.0001 0.001 0.01 0.1 1

Time Factor (Tr )

Figure 11. Time development of soil and column stresses.

the SCF (SCF = zc /zs ) starts from zero and reaches a final value equal to the confined modular
ratio (E mc /E ms = 40). This is not realistic, as commented above. The consideration of radial
deformations, keeping elastic behaviour, reduces this final value to 25, showing a non-zero initial
value, due to the presence of immediate settlement. Plastic strains in the column reduce further the
final value of SCF to about 5, with a small influence of the dilation angle of the column material.
The evolution of total and effective stresses in soil and column is detailed in Figure 11 for the
case of c = 20 .

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
870 J. CASTRO AND C. SAGASETA

6
Priebe (uncorrected)
=0
c
=10
c
=20
5 c

Improvement factor, n
4

3 c=50

c=30

2
c=40

1
0 0.1 0.2 0.3 0.4 0.5

Area replacement ratio, ar

Figure 12. Improvement factor, n. Comparison with Priebes method (uncorrected).

Settlement reduction
The settlement reduction factor, or its inverse, the improvement factor n, are used in practice to
evaluate the efficiency of the stone columns. In Figure 12 the final values of n obtained with the
present solution are compared with the Priebe [3, 4] basic method. This means that the column
elastic strains are ignored, and no account is made for the dilatancy angle. So, for comparison, the
present solution is applied using an infinite Youngs module for the column. The agreement is good,
and the influences of the area replacement ratio and of the column friction angle are adequately
reproduced in Priebes method. The present solution allows for an explicit consideration of the
influence of the depth and of the gravel dilatancy, which is clearly pointed out in the figure.

DESIGN EXAMPLE

An example has been selected to show how to use the solutions presented in this paper. The case
is an embankment applying an uniform vertical unit load of 100 kPa, on a soft clay layer 10 m
thick, reinforced with stone columns with a diameter of 0.7 m and a spacing of 2 m in triangular
pattern.
The column material properties are: submerged unit weight, c = 10 kN/m3 ; Youngs modulus,
E c = 30 MPa; Poissons ratio, c = 0.33; friction angle, c = 40 ; and dilatancy angle, c = 10 . The
soil relevant parameters are: submerged unit weight, s = 10 kN/m3 ; Youngs modulus, E s = 2 MPa;
Poissons ratio, s = 0.33; coefficient of consolidation, cvr = 107 m2 /s; and at-rest lateral pressure
coefficient, k0s = 0.6.
The area replacement ratio is ar = 0.11, and the modular ratio, E c /E s = 15.
For the application of the present method, the procedure is as follows:
(a) The equivalent coefficients of consolidation for elastic and plastic conditions can be calcu-
lated from Equations (34) and (43).

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 871

(b) The elastic stresses and strains at the undrained (u) and final (fe) situations can be calculated
from Balaam and Booker solutions [7, 8], or from Table I with u = oct,s and u = 0,
respectively.
(c) The Barron/Hansbo solution (18) is applied between these undrained and final elastic states
with the elastic equivalent coefficient of consolidation, cvrzr e , and the average excess pore-

pressure u is obtained as a function of time.


(d) All the stress and strain components are obtained from pa and u using Table I.
(e) The initiation of plastic condition at the surface is evaluated as the corresponding degree
of consolidation, Uye using Equation (41). If (Uye >1), the case is elastic and calculation is
finished. If (Uye <1), the case is elasticplastic.
(f) If it is an elasticplastic case, the yielding moment is kept as the initial point for the plastic
phase, with an initial excess pore pressure u y . From this point, the Barron solution (18) is
zr p
used again, with the plastic equivalent coefficient of consolidation, cvr . The average excess
pore pressure u is obtained again as a function of time.
(g) All the stress and strain increments are obtained from u using Table II, and accumulated to
their respective values at the moment of yielding.

The results are summarized in Table III, using four methods: in the last two columns, the present
solution, in the assumption of elastic columns (Table I) and elasticplastic columns (Table II). The
first two columns include as a reference the analyses without lateral deformation of the columns,
with the conventional assumption of constant column/soil stiffness ratio (Barron/Hansbo solution
with constant load calculated for the final situation), and with variable stiffness ratio, with the
method proposed by Han and Ye [2].
All the results given are for stresses and displacements at the top surface (z = 0). The settlement,
sz , is calculated using the integrated expressions (44)(46).
For this case, the column reaches the limit active state (kac = 0.22) at the top surface for a
degree of consolidation of 53%, in 12 days. The plastic area progresses downwards, reaching the
column mid-depth for U = 69% in 19 days and the bottom for U = 85% in 31 days (these results
are not shown in Table III).
The surface final settlement without soil improvement is 337.5 mm. Assuming lateral confine-
ment, it is reduced to 132.9 mm (improvement factor of 2.54). However, it increases to 160.8 mm
due to the elastic radial deformation of the columns (n = 2.10), and to 220.0 mm (n = 1.53) by the
effect of the column plastic deformation. As a comparison, Priebes method gives a basic settle-
ment improvement factor of 1.59, increasing to 2.02 with the empirical corrections for overburden
pressure and column deformability (keeping k0s = 1.0).
The radial deformations of the column are obviously zero with the lateral confined approach.
With the present solution, they are apparently small if compared with the settlements, but the unit
strains are of the same order of magnitude. It can be seen that for the undrained situation, the
columns shrink laterally by a small amount, due to the condition of no volume change in the soil,
but this tendency is reversed as consolidation proceeds.
The dissipation of excess pore pressure follows the trends discussed in the paper: with lateral
confinement, the variation of the stiffness ratio means a very significant speed up of the process
(the degree of consolidation at 40 days increases from 71 to 97%). The radial deformation reduces
this trend, and the values at 40 days are 91% for elastic column and 78% for plastic deformation.
The SCF is also shown in the table. For the lateral confinement approach, it increases from zero
at the undrained situation to 15 (equal to the modular ratio) at the end of consolidation. This value

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
872

Copyright q
Table I. General elastic solution. Stresses and strains for any load pa and excess pore pressure u.
Column (c) Soil (s)
pa [1+ar (2F1)]u

2008 John Wiley & Sons, Ltd.


Vertical strain, z (c +2G c )ar +(s +2G s )(1ar )2ar (c s )F
   
Radial displacement, sr u Far 1(r/rl )2 u
r F  z c s 1ar (r/rl )2 r  z c s
    
u 2Far Gs u
Radial effective stress, r c z 2(c + G c )F z c s s z + 1ar
s + G s + (r/r 2  z c s
l)
    
u 2Far Gs u
Hoop effective stress,  c z 2(c + G c )F z c s s z + 1ar
s + G s (r/r 2  z c s
l)
   
u 2Far u
Vertical effective stress, z (c +2G c )z 2c F z c s (s +2G s )z + 1a
r
s  z c  s
        
u 2Far 2Far u
Octahedral effective stress, oct c + 23 G c (12F)z +2F c + 23 G c c s s + 23 G s 1+ 1a
r
 z 1a r c  s
J. CASTRO AND C. SAGASETA

 
Volumetric strain, vol u 2Far 2Far u
(12F)z +2F c s 1+ 1a z 1a c s
r r

F = (1ar )(c s )/2[ar (s c + G s G c )+ c + G c + G s ].

DOI: 10.1002/nag
Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
CONSOLIDATION AROUND STONE COLUMNS 873

Table II. Plastic increments of strains and stresses from the moment of yielding at constant load pa , for
excess pore pressure increment u = u u y .

Column (c) Soil (s)


 
p 1ar + kaac
r u
Vertical strain, z
(s +2G s )(1ar )+ akr s + akracJ
c
p p
z ar 1(r/rl )2 z
Radial displacement, sr r 2k r 1a
c r (r/rl )2 2kc

s + G s (rc /r ) +ar (G s +s )
2
Radial effective stress, r
p p
J z + u (1ar )kc z

s G s (rc /r ) ar (G s +s )
2
Hoop effective stress, 
p p
J z + u (1ar )kc z
 
Vertical effective stress, z
p ar s p
kac (J z + u) s +2G s + (1a z
1
r )kc
  
(2+1/kac )
Octahedral effective stress, oct
p p
3 (J z + u) s + 23 G s 1+ (1aar )k z
r c
   
p p
Volumetric strain, vol 1 k1 z 1+ (1aar )k z
c r c

J = s + G s +ar (G s + s )/(1ar )kc .

is reduced to 10.22 by the consideration of the column radial deformation, and to 3.87 for plastic
deformation. Priebes method gives SCF = 6.2.
With respect to the ratio of radial to vertical stresses in the column, the methods with lateral
confinement give no valid result, because radial stresses in the soil and column are different,
implying a discontinuity at the column wall, due to differences in the values of k0 and vertical
stresses between both materials. This is shown in the last rows in Table III. The differences are very
significant, and radial stresses in the column are between 2 and 10 times the radial soil stresses.
The consideration of radial deformation eliminates this discontinuity. For elastic column, the stress
ratio starts at a value of 1.31 (well above k0 ), and decrease to 0.07 at the end. Consideration of
plastic deformation limits this value between the active and passive bounds. In this case, the active
condition is reached, as commented above.

CONCLUSIONS

A solution has been presented for the radial consolidation around stone columns under constant
surcharge load. The solution considers the influence of vertical and radial deformation of the
column, either in elastic and elastoplastic regimes, and in this case, with a non-associated behaviour
(constant dilatancy angle).
The solution is in terms of the average excess pore pressure in the soil. For elastic column,
the solution gives the variation of strains and stresses between the undrained and final states, for
which it coincides with the elastic solution by Balaam and Booker [7, 8].
The radial column deformation has a significant influence on the distribution of stresses between
the soil and the column. If only vertical deformation is considered, the column/soil stress ratio
is abnormally high, equal to the respective confined modular ratio. The column radial elastic

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
874 J. CASTRO AND C. SAGASETA

Table III. Results for the design example. Stresses and displacements at the top surface (z = 0).
Vertical and radial
Only vertical deformation deformation (present solution)
Constant stiffness ratio Variable stiffness Elastic Elasticplastic
Variable Time (days) (Barron/Hansbo) ratio [2] column column
sz (mm) 0 0 0 3.5 3.5
40 93.7 128.8 147.1 159.4
132.9 132.9 160.8 220.0
n 2.54 2.54 2.10 1.53
sr c (mm) 0 0 0 0.5 0.5
40 0 0 1.5 2.7
0 0 1.7 4.5
u (kPa) (U %) 0 112.5 112.5 102.4 102.4
40 33.2 (71%) 3.5 (97%) 8.9 (91%) 22.6 (78%)
0 0 0 0
zc /zs 0 0/112.5 0/112.5 76.3/103.0 76.3/103.0
(kPa) (SCF) (0.00) (0.00) (0.74) (0.74)
40 416.3/94.1 572.2/45.1 467.6/54.1 296.9/75.4
(4.42) (12.68) (8.65) (3.94)
590.6/39.4 590.6/39.4 504.9/49.4 293.3/75.8
(15.00) (15.00) (10.22) (3.87)
r c /zc 0 (0112.5)/0 (0112.5)/0 99.8/76.3 99.8/76.3
(kPa) (k ratio) (1.31) (1.31)
40 (148.769.8)/416.3 (204.428.2)/572.2 39.9/467.6 64.5/296.9
(0.360.17) (0.360.05) (0.09) (0.22)
(211.023.6)/590.6 (211.023.6)/590.6 34.2/504.9 63.8/293.3
(0.360.04) (0.360.04) (0.07) (0.22)
Radial stress discontinuity at the columnsoil interface. First and second numbers shown in brackets correspond
to the total radial stresses in the column and in the soil, respectively.

expansion reduces this final value to about one half of the modular ratio, which is still too high.
Plastic strains in the column reduce further the final SCF to values in the range found in real cases,
with a small influence of the dilation angle of the column material.
All the results are given in the closed form, and both the elastic and plastic deformations of the
column lead to an equivalent coefficient of consolidation for the radial flow, which enables the
application of the existing methods of integration of the consolidation equation.
Application to typical cases shows good agreement with existing semiempirical methods of
analysis, and the role of the governing parameters is identified.

NOTATION

ar area replacement ratio: ar = Ac /Al


cv coefficient of consolidation

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 875

k coefficient of permeability
k0 coefficient of at rest lateral pressure
ka , kp coefficient of active and passive pressure
n improvement factor (ratio of settlement without and with columns)
pa applied vertical unit load
rl ,rc radius of the unit cell, of the column
s displacements
sz0 settlement without columns
A cross-section
E Youngs modulus
Em oedometric (constrained) modulus: E m = [E(1)]/[(1+)(12)]
G shear modulus: G = E/[2(1+)]
K bulk (volumetric) modulus: K = E/[3(12)] = +2G/3
N column spacing ratio: N =rl /rc
friction angle
 dilatancy angle
 Lames constant:  = 2G/(12) = E m 2G
 Poissons ratio
Subscripts/superscripts
c, s, l column, soil, elementary cell
e, p elastic, plastic
i, u, f, y initial (previous), undrained, final, at yielding
r, z,  cylindrical coordinates
(upper bar) average value along the radius

APPENDIX A: STRESSES AND STRAINS FOR THE UNIT CELL WITH EXCESS PORE
PRESSURE IN THE SOIL

Elastic solution
The procedure followed by Balaam and Booker [7, 8] for undrained and final states is generalized
for the case of an excess pore pressure, u, as superposition of two states:
(A) a vertical uniform compressive strain, z A , with no pore pressure and with both the soil
and the column being horizontally confined. This was also the solution used by Han and
Ye [2], and all the strains and stresses can be put in terms of z A in a straightforward way.
However, it implies a discontinuity in the horizontal stress at the columnsoil interface:

r cA = c z A in the column
r cA  = r sA (A1)
r sA = s z A in the soil

(B) different effective radial pressures (r cB , r sB ) applied on the column and on the soil at
their interface in order to cancel the unbalanced radial stresses in state A. The soil and
column are now vertically confined, with zero vertical strain (plane strain).

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
876 J. CASTRO AND C. SAGASETA

In state B, the column and the soil are, respectively, a cylinder and a thick tube, in plane strain
and axial symmetry. Both problems have also a simple analytical solution. All the stress and strain
components can be expressed as functions of the applied radial pressures (r cB , r sB ) and the
excess pore pressure u in the soil. The radial displacements at the column wall are
sr c,c 1
= r cB in the column
rc 2(c + G c )
(A2)
sr c,s 1ar
=  in the soil
rc 2[G s +ar (s + G s )] r sB
Then, the conditions of compatibility (sr c,c = sr c,s ) and equilibrium (r c = r s +u) at the soil
column interface provide
c z A +r cB = s z A +r sB +u
1 1ar (A3)
r cB = 
2(c + G c ) 2[G s +ar (s + G s )] r sB
These two equations can be solved for the radial stresses (r cB , r sB ), in terms of z A and u.
The last step is to express vertical equilibrium between the applied load, pa , and the vertical
stresses on the top of the cell. This allows for elimination of z A , and to express all the stresses
and strains as functions of pa and u.
The final complete results are summarized in Table I. The solutions of Balaam and Booker
[7, 8] for the fully drained and undrained conditions can be recovered by introducing the value of
the excess pore pressure u as zero or equal to the octahedral total stress, respectively.

Plastic solution
The only difference of this case with respect to the elastic one lies on the column behaviour, because
the soil is always elastic. Only the column deforms plastically. If elastic strains are neglected in the
plastic zone, the solution is even easier, because the problem is uncoupled: the stresses (radial and
vertical) in the column are linked by the MohrCoulomb failure condition (35), regardless of the
strains. Given the absence of cohesion, the failure condition can be established in terms of the
stress increments. On the other hand, the vertical and radial strains are related by the flow rule
(36), with no influence of the stresses. As a result, the plastic phase can be formulated in terms
of stress and strain increments from the moment of yielding, independently of their values in the
elastic phase.
The solution process is as follows:
(a) For any given value of the incremental pore pressure, u = u u y , if the resulting incremental
vertical strain is z (elastic for the soil, plastic for the column), the radial strain increment
in the column, r c , is given by the flow rule (36).
(b) These radial and vertical strains are then imposed to the soil (this means that the compatibility
condition at the soilcolumn interface is automatically fulfilled), and all the soil stresses
and strains can be calculated in terms of z and r c as an elastic thick tube with known
vertical and horizontal strains. Particularly, the incremental vertical and radial effective
stresses (zs , r s ).
(c) From the radial equilibrium condition at the column wall, the radial stress increment r c
in the column is determined.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag
CONSOLIDATION AROUND STONE COLUMNS 877

(d) The vertical stress increment zc is then defined by the yield condition (35).
(e) Finally, the vertical equilibrium condition at the top surface (constant load) means
ar zc +(1ar )zs +(1ar )u = 0 (A4)
(f) Equation (A4) can be solved for the incremental vertical strain z , which can now be
introduced in step 1, to get all the variables as functions of u.
The resulting increments of stresses and strains are shown in Table II.

ACKNOWLEDGEMENTS
The work presented is part of a research project on Improvement of soft soils with stone columns for
foundation of embankments. Analysis of the process and design criteria, for the Spanish Ministry of
Public Works (Ref.: 03-A634). The first author has received also a Grant from the Spanish Ministry of
Education.

REFERENCES
1. Barksdale RD, Bachus RC. Design and construction of stone columns. Report FHWA/RD-83/026, National
Technical Information Service, Springfield, Virginia, 1983.
2. Han J, Ye SL. A simplified solution for the consolidation rate of stone column reinforced foundations. Journal
of Geotechnical and Geoenvironmental Engineering 2001; 127(7):597603.
3. Priebe HJ. Abschatzung des Scherwiderstandes eines durch Stopfverdichtung verbesserten Baugrundes. Die
Bautechnik 1978; 55(9):281284.
4. Priebe HJ. Design of vibro replacement. Ground Engineering 1995; 28(10):3137.
5. Dhouib A, Wehr J, Soyez B, Priebe HJ. Methode de Priebe: origine, developpement et applications. Symposium
International sur lAmelioration des Sols en Place, Paris, Presses de lENPC et LCPC, 2004; 131146.
6. Van Impe W, De Beer E. Improvement of settlement behaviour of soft layers by means of stone columns. 8th
European Conference on Soil Mechanics and Foundation Engineering, Helsinki, vol. 1, 1983; 309312.
7. Balaam NP, Booker JR. Analysis of rigid rafts supported by granular piles. International Journal for Numerical
and Analytical Methods in Geomechanics 1981; 5:379403.
8. Balaam NP, Booker JR. Effect of stone column yield on settlement of rigid foundations in stabilized clay.
International Journal for Numerical and Analytical Methods in Geomechanics 1985; 9:331351.
9. Pulko B, Majes B. Simple and accurate prediction of settlements of stone column reinforced soil. 16th International
Conference on Soil Mechanics and Geotechnical Engineering, Osaka, Japan, vol. 3, 2005; 14011404.
10. Barron RA. Consolidation of fine-grained soils by drain wells. Transactions of ASCE 1948; 113:718742.
11. Hansbo S. Consolidation of fine-grained soils by prefabricated drains. 10th International Conference on Soil
Mechanics and Foundation Engineering, Stockholm, vol. 3, 1981; 677682.
12. Biot M. General theory of three-dimensional consolidation. Journal of Applied Physics 1941; 12:155164.
13. Hird CC, Pyrah IC, Russell D. Finite element modelling of vertical drains beneath embankments on soft ground.
Geotechnique 1992; 42(3):499511.
14. Carrillo N. Simple two- and three-dimensional cases in the theory of consolidation of soils. Journal of Mathematical
Physics 1942; 21(1):15.
15. Leo CJ. Equal strain consolidation by vertical drains. Journal of Geotechnical and Geoenvironmental Engineering
2004; 130(3):316327.
16. Lane KS. Consolidation of fine-grained soils by drain wellsdiscussion. Transactions of ASCE 1948; 113:
743747.

Copyright q 2008 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2009; 33:851877
DOI: 10.1002/nag

Potrebbero piacerti anche