Recommendations

for the design,

calculation, construction
and quality control of stone
columns under buildings
and sensitive structures
Forward
This report is a newand revised version of the recommendations published in
the RFGNo. 111 in 2005. It takes into account both the experimental results and
observations of the French Union Syndicale Géotechnique (USG) in the interest of
better understanding soil-structure interactions. This version also includes chap­
ters on design based on laboratory and cone penetration tests (CPT). It presents
the guidelines to be followed by all building contractors in France (according to
Article R111-40 of the French Building and Residential Construction Rules) and
oversight entities dealing with the design, calculation, installation and quality
control of stone columns and their interaction with buildings and settlement-sen­
sitive structures.
The following participated in drafting this document:
P. Aguado (Apave)
P. Berthelot (Bureau Veritas)
L. Carpinteiro (Socotec)
F. Durand (Fugro Geotechnique)
M. Glandy (Soletanche Bachy Pieux)
P. Liausu (Menard)
B. Pezot (Menard)
C. Poilpre (consultant)
S. Lambert (Keller Fondations Spéciales)
J.-P. Volcke (Franki Fondation)
G. Billoet (Keller Fondations Spéciales) and J.R. Gauthey (Spie Fondations)
also participated in discussions.
This document has been approved by the Union Syndicale Géotechnique (USG,
the French geotechnical union association) and the Comité Français de Mécanique
des Sols (CFMS, the French societyfor soil mechanics, ISSMGEmember society).

51
REVUE FRANÇAISE DE GÉOTECHNIQUE
N° 144-145
3eet 4etrimestres 2013
 2
Introduction Fields of application
(1) The columns referred to in these guidelines are
vertical piles made of cohesionless granular materials 2 .1
that are driven into the soil and then compacted by
repetitive action. By type of structure
Comment 1: These columns do not contain anypar­ (1) The most common uses of stone column treat­
ticular type of binder at any level. In this document, the ment involve structures with concrete slabs or raft
term ""stone column"is limited to the definition above. foundations with surface loads sensitive to settlement:
(2) The columns can be constructed in regular or ■ Hangars and warehouses.
variable grids in rows or groups, or even in isolation. ■ Industrial and commercial buildings.
(3) Their design takes into account the type of struc­ ■ Silos and tanks of all kinds.
ture to be built, the kind of load, tolerances, and requi­
rements regarding absolute and differential settlement, ■ Watertight hydraulic structures (tanks, sewage
as well as the nature and characteristics of the soil to treatment plants, etc.)
be treated. (2) By extension, they can be used under other types
(4) The purpose of stone columns is to give the of structures, too, as long as the residual deformations
soil under the structure to be built new general and/ of the treated soil and underlying layers are consistent
or local characteristics so that the structure's various with the structure in question and with associated
infrastructure elements (isolated or strip footing, raft technical measures:
foundations, concrete slabs, embankments, etc.) will ■ On-shore civil engineering works (roads,
demonstrate predictable and justifiable behavior embankments, bridges, retaining walls), and maritime
consistent with the regulations and tolerances that structures (seabed, lake and riverbed reinforcement).
apply to the structure of the building and its intended ■ Under shallowbuilding foundations.
use.
(5) Treating soil with stone columns involves a com­ (3) They can also be used for stable heterogeneous
bination of the following actions, of which one or more embankments where systematic treatment with an
may be intended: appropriate and regular grid can help improve and/or
homogenize the soil characteristics to make it suitable
■ Improving bearing capacity. for surface construction.
■ Reducing settlement.
■ Improving uniformity of geotechnical characte­ 2.2
ristics.
■ Increasing the consolidation rate by creating Use in seismic zones
drainage elements.
■ Increasing equivalent ground mass characteris­ (1) It is also possible to use stone columns in seismic
tics (horizontal shear strength, angle of internal friction zones where they may reduce the risk of soil liquefac­
and deformation parameters). tion and increase shear resistance.
(6) A stone column is a soil improvement measure. (2) Refer to the guidelines entitled "Procédés d'amé­
It is neither a foundation element, nor a deep founda­ lioration et de renforcement de sols sous sollicitationssis­
tion element. miques" published by the French Association of Earth­
(7) The foundation of a structure built on soil treated quake Engineering (AFPS).
by stone columns is always superficial (isolated or strip
footing, raft foundations, concrete slabs, etc.). It can 2 .3
also be part of the "foundation" of an embankment.
(8) The objective is to control the behavior of these Limitations of use
structures, and in particular to reduce the settlement.
Comment 1: Depending on the existing grid density (1) The lateral earth pressure exerted by the sur­
and geotechnical conditions, this coefficient is usually rounding soil is a determining factor in column
between 1.5and 4 under distributed loads. construction, behavior and capacity.
(9) The need for soil treatment should be analyzed (2) Therefore:
by the project's geotechnical engineer, who should ■ Stone columns must not be used in soil that exhi­
provide and approve the calculation hypotheses. bits a potential risk of a loss in volume and/or mecha­
Comment 2: A reminder: the stone column geo­ nical soil characteristics, especially waste dumps and
technical study should examine not only the soil to be peat deposits. More generally, stone columns should
treated, but any soil that could potentially settle under not be used in soils with a loss on ignition of more than
the treated layer. 5%as specified by standard XP 94-047.
Comment 3: The contractor must involve the geo­ Comment 1: There is also standard NF P 94-055
technical engineer in the design and the execution of which specifies a chemical method for determining the
the project in compliance with standard NF P 94-500 organic material weight content of asoil.
concerning geotechnical procedures in France. ■ Unless particular measures have been provided

52
REVUE FRANÇAISE DE GÉOTECHNIQUE
N° 144-145
3eet 4etrimestres 2013
for, such as preloading or consolidation, or a specific
compatibility study, treatment with stone columns in
highly compressible soils (silt, mud and soft clay) with
a thickness greater than 50 cm and weak characteris­ Comment 1: Los Angeles test, standard DIN EN
tics (Cu<20 kPa or qc<300 kPa) is not possible. 1097-2; Micro De-val test, standard NFEN 1097-1.
■ In any case, essential design verifications must (4) The particle size test depends mainly on the
be carried out to calculate potential stone column fai­ backfill material. Bottom-feed vibrators are more sen­
lure and deformation as described in Chapter 5 of this sitive and an unsuitable particle size can plug the pipe.
document. Specific stresses related to the site must (5) The following indicative values can be used as a
also be taken into account. reference:
■ Bottom-feed vibrators with a material transfer
3 pipe on the side: the most commonly used particle size
is 8/40.
Procedure ■ Other methods: the most commonly-used par­
ticle size is 20/75.
.13 (6) The standard for particle homogeneity: less than
5%particles smaller than 80 m.
Methods and materials
(1) The two following methods meet the definition 4
given in the forward (see glossary for a detailed des­
cription): Construction provisions
■ Vibro-compaction.
■ Vibro-stone columns: |
4.1
- vibro-displacement by a "dry" method where
compressed air jetting is used, or Diameter of stone columns
- vibro-replacement by a "wet" method where (1) The diameter of the stone columns depends on:
water jetting is used. ■ the device used and its suitability to the soil at
Comment 1: Any other method requires its own the site,
technical specifications. ■ the layers of soil penetrated and their characte­
(2) Which device, wet or dry method, characteris­ ristics,
tics and the implementation method to use depends
largely on the nature and the level of soil saturation, ■ the total amount of energy used (power output,
the desired objectives, and the characteristics of the vertical force and running time).
gravel fill materials. It is up to the contracting company (2) The diameter of the column may vary over its
to decide on the basis of its expertise and experience length, depending on the resistance of the different
which process is best adapted to each situation. layers penetrated.
Comment 1: The diameter of most dry method
.m columns is between 50and 80 cm.
(3) All things equal, the diameter of wet method
Pre-drilling columns is usually greater than dry method columns
due to the soil extraction caused by water jetting.
(1) Soil displacement through compacted layers or
obstacles can be facilitated by pre-drilling, with or with­
out earth excavation. 4.2
(2) Excavated or altered volumes should be filled
and compacted with the column material. Interface between the treated soil
and the building structure
3.3 (1) If the foundation element does not have enough
inertia to distribute load bearing in a homogeneous
Gravel backfill way on the initial column grid, it is necessary to add
(1) The gravel backfill must be of high quality, and a load transfer platform between the foundation ele­
its particle size must be as homogeneous as possible. ments and the treated soil. The purpose of this plat­
form is to improve the load bearing distribution.
(2) In most cases natural, pea gravel or crushed (2) If the stone columns are being used for their
stone gravel is used. draining properties, a drainage layer (with outlet)
Comment 1: Except in the case of specific docu­ should be added at the top of the columns.
mentation showing an absence of short and long-term Comment 1: In practice, isolated and strip footings
problems (expansion, pollution, physical-chemical reac­ and raft foundations do not require aload transferplat­
tions, etc.), recycled materials may not be used. form, as opposed to concrete slabs (reinforced or not)
(3) The minimum characteristics for gravel backfill which do.
are: (3) With few exceptions, in general, soil treatment
■ Los Angeles test loss <35%. with stone columns does not in and of itself improve

53
■ Micro De-val test loss <30%. the behavior of surface soil between columns where
■ Combined Los Angeles + Micro D-val test loss there is no increase in the values for either the EV2
<60%. modulus or the Westergaard coefficient.

REVUE FRANÇAISE DE GÉOTECHNIQUE

N° 144-145
3eet 4etrimestres 2013
 Installation of footings and raft foundations Configuration of stone columns
(1) Footings and raft foundations must be construc­
ted in the same way as on untreated soil. They can be
cast directly into the trench or into a formwork. Structures with uniform distributed loads
(2) When the footing base is less than 50 cm below
the work platform level, the head of the column must (1) In the case of structures with uniform distribu­
be re-compacted to ensure the characteristics specified ted loads, the columns are laid out in a regular grid,
in the calculations. usually square or triangular.
Comment 1: This re-compacting can be done in (2) The grid spacing depends on the geotechnical
conjunction with platformre-compacting. characteristics and applied loads.
Comment 1: Under certain geotechnical and/or
load conditions, part of the structure may not be loca­
ted on treated ground as long as the absolute and dif­
Installation of a load transfer platform ferential settlements between areas treated or loaded
differently remain consistent with the structure and its
(1) The minimum thickness for a gravel load trans­ usage.
fer platformto distribute the load is 40 cm. Comment 2: Except in special cases, it is not neces­
(2) In the case of concrete slabs, the top layer of gra­ sary to extend the column layout beyond the concerned
vel should at least have the characteristics of subgrade structure.
as defined by DTU13.3 and Guide GTR92.
Comment 1: For the purposes of this document, the
followingshould be noted for the subgrade layer:
■ Class Fmaterials are not permitted (according to Isolated and strip footing
standard NFP 11-300). (1) The footing dimensions will depend on both the
■ The minimum thickness is 25 cm. allowable stress and allowable settlement after soil
■ The modulus of elasticity EV2is greater than treatment.
50MPa. (2) The stone column grid spacing is based on the
(3) It is up to the soil treatment designer to define least favorable criterion.
the thickness and the minimum characteristics of the Comment 1: For the same structure, it is entirely
platform according to the nature of the treated soil. possible that some footings may be placed on improved
Comment 1: Bear in mind that the criteria applied soil, while others with lighter loads may be placed on
to gravel layer calculations are its punching resistance untreated ground. This situation is allowed in conditions
(ifnecessary), its thickness and the modulus of elasticity. where the uniformity of settlement can be ensured for
(4) The concrete slab designer must verify that the footings subject to different load levels (e.g. secondary
load transfer platform's thickness and characteristics frame structures that support siding for industrial han­
are sufficient to resist the pressure exerted by the gars or warehouses).
concrete slab, including the punching resistance, and (3) For individual columns or a single row of
that the values remain above the minimum concrete columns, in the absence of appropriate verification,
slab requirements (standard NF P 11-213 ref. DTU13.3 the edge of the footing in relation to the theoretical
on concrete slabs). position of the bare outer part of the columns must be
(5) The load transfer platform can be partially or at least equal to the installation tolerance specified in
entirely installed before the stone columns and there­ §4.9.
fore can be used as a work platform. Comment 1: Verification involves making sure that
(6) However, any re-grading, final compacting, even if the column is not fully under the footing (pos­
and re-treatment of subgrade, as well as any gravel sibly before installation and mandatorily after verifica­
additions to increase thickness should be performed tion if there is any column deviation), that the column's
after the installation of the stone columns so that the deformation and maximum stress results remain allo­
characteristics in compliance with the project remain wable.
consistent.

54 FIG. 1

REVUE FRANÇAISE DE GÉOTECHNIQUE

N° 144-145
3eet 4etrimestres 2013
Load transfer platform.
 4.8

Stone columns under an embankment

supporting a foundation
(1) It is possible to treat all types of soil in the scope
of application with stone columns, and then build a
substitute or raised embankment on the improved soil
using fine, quality-controlled materials and demons­
trating the utmost care during installation and strict
attention to compaction, for example according to the
1980 recommendations by COPRECLCPC.
FIG. 2 Case of columns lying outside of footings.

(4) For columns that are not aligned, it is recom­

mended to adopt the same configuration (footing edge
equals at least the installation tolerance). Otherwise, FIG. 3 Installation of stone columns under an
the column settlement and stress must be verified if embankment supporting a foundation.
the columns do not conformto the theoretical layout.
Comment 1: Appendix I includes some standard
configurations. (2) The configuration of the columns takes into
Comment 2: Under strip footings, placing columns account the superposition of stresses generated by
in alternate rows limits the incidence of column devia­ the embankment itself and also the various structural
elements built into the fill thickness. This configura­
tion. tion requires justification in particular if the height of
the embankment is less than half of the reference grid
4 .6 measurement.
Maximum reference grids 4 .9
(1) For soil to be considered treated by stone
columns, and whatever the intended objective, the Tolerances
maximumreference grid should be 9 m2and the subs­
titution rate should be greater than 3%.
(2) For strip footings with a single row of columns
and no load transfer platform, the maximum grid spa­ Columns in grid networks
cing without specificjustification is 2.5 m. (1) Due to their own characteristics, stone columns
can provide new characteristics to soil that are equiva­
lent to soil's basic lattice structure located at its center.
(2) In this case, the concept of column installation
Minimum reference grids deviation does not apply, though it is applicable for
localized foundation elements.
(1) The minimumreference grid should be 2.25 m2.
(2) For strip footings and groups of two to five
columns, the spacing between the columns axes can­
not be less than 1.5 times the diameter of the stone Columns under footings
column, or 1.20m.
Comment 1: A high substitution rate can lead to a (1) The isolated and strip footings installed on trea­
risk of raising the platform. ted soil must be considered foundation elements.
Comment 2: To address specific problems (if part (2) The tolerance for stone columns is therefore
of structure is too small, for example), it is possible to ± 20 cm.
reduce the spacing in (1) and (2) by conducting a feasi­ Comment 1: This value can be difficult to assess
bilitystudy. due to the irregular shape of the columns when they are
Comment 3: For columns executed with the wet vibrated.
method, a process rarely used for earthen sites due to (3) Above this value, the company must prove that
the soil extraction required, it is possible to use larger
diameter stone columns (1.00 to 1.20 m) more closely
spaced than those installed using the dry method.
the footing behavior is compatible with the building
structure. If there is a gravel layer under the footings,
its characteristics are taken into account.
55
REVUE FRANÇAISE DE GÉOTECHNIQUE
N° 144-145
3eet 4etrimestres 2013
 Comment 1: The constructions built on the columns • The overall allowable bearing capacity for the
are designed to remain unchanged as long as installa­ improved soil after treatment and justification of the
tion is completed within the tolerances. different types of foundations regarding failure, and
(4) The choice of whether to opt for lower tole­ • The absolute settlement of the various structural
rances is left up to the company, but it must ensure aspects for each construction and justification of the
strict compliance and the tolerance may not be the differential settlement for the structure, or between the
same in both directions. structure and the concrete slab according to the allo­
wable tolerances for each construction and any rele­
vant regulations in effect.
4.10
(3) When there are other objectives set, the fol­
Reduction factor for the volumes lowing additional information must also be provided:
of material used • In the case of soil liquefaction, it must be shown
that the columns will minimize this risk.
(1) There is a difference between the theoreti­ • If the columns are intended to act as vertical
cal volumes of the columns designed, and the actual drains to accelerate consolidation, the consolidation
volume of the material supplied and thus used onsite. time is calculated.
This difference is due to losses during installation • In the case of slope stabilization, the calculation
(including variations from expected dimensions resul­ compares the factor of safetyto the circular failure rate.
ting fromlateral pressure) and fromcompaction of the
material in the column.
(2) Usuallythe following ratios ofthe real volume to 5.3
the theoretical volume are accepted:
• Top feed vibrated columns using the wet method: Hypotheses
1.3 to 1.5. (1) The building construction constraints (maximum
• Bottomfeed columns using the dry method: 1.2. bearing capacity and allowable deformation) must be
specified in the documents specific to the project.
5 (2) The soil behavior hypotheses are the geotech­
nical characteristics identified from soil tests: strati­
Design graphy, soil composition, hydrology, rheological and
mechanical properties (Young's modulus, Poisson's
ratio, lateral earth pressure, compression index Ccvoid
5 .1 ratio, etc.) for all soil layers.
Comment 1: When the data result from corre­
Prerequisite information needed lations, it is best to obtain written consent from the
for calculations geotechnical engineer attesting to the validity of the
hypotheses under consideration.
(1) The behavior of soil reinforced with stone (3) The hypotheses concerning the stone columns,
columns cannot be analyzed without extensive especially in terms of their modulus of deformation,
knowledge of: diameter and length depend in large part on the mate­
• The nature and characteristics of the soil to be rial used and its performance.
treated and the ground underneath it: soil boring tests, (4) Common values for non-geometrical parameters
results of onsite, mechanical or laboratory tests, etc. used in the calculations for the backfill materials des­
• The nature, characteristics and construction cribed in §3.3 and correctly employed are as follows:
methods for the actual building foundation itself.
• Regular and unusual stress levels.
• The nature and behavior of the structure to be TABLE I Common column characteristics.
built.
(2) The calculation rules specified in this docu­ Young'smodulus: averageover columnvolume Ecol=60MPa
ment do not take into account any soil improvement Inter-granular pea gravel 'c=38degrees
between columns. internal angle crushedstonegravel  'c=40degrees
Comment 1: Installing columns can improve the Poisson's ratio col=1/3
characteristics ofsome types ofsoil. It is possible to take Unitweight ofsaturatedmaterial inplace col=21kN/m3
the "improved" characteristics into account as long as
appropriate tests are conducted to make sure they can
be attained.
Comment 1: Higher or lower values are possible,
but must be explained by appropriate tests.
Criteria for determining column size 5.4
(1) Determining the size of stone columns is inex­ Maximum allowable stress for stone columns
tricably linked to the characteristics of the soil to be
treated and the implementation process. (1) To calculate the maximum allowable stress, first

56
(2) In general, the justifications that need to be pro­ determine the vertical stress-rupture point qr for an
vided inthe calculations are based on these two criteria: isolated column based on the characteristics of the

REVUE FRANÇAISE DE GÉOTECHNIQUE

N° 144-145
3eet 4etrimestres 2013
column and of the soil after treatment, and according
to the possible modes of failure illustrated below(Fig. 4
a, b and c; see Soyez; 1985):
• Bulging failure (a criterion often used to deter­ • corresponding to an average over alength of2 SC
mine size) centered on z.
• General shear failure (failure is rare, occurring
when columns are short).
• Punching effect shear failure ("floating columns"). General shear failure
(1) General shear failure can be studied when the
characteristics of the column closely resemble those
of the soil. This occurs only very rarely and the corres­
ponding calculation does not appear in this document
(see Soyez, 1985).

Punching effect shear failure

a) column bulging failure
b) general shear failure (1) The vertical shear within the column is most
c) punching effect failure (floating column) intense at the top and decreases as it moves down (see
Soyez, 1985).
FIG. 4 Failure modes. (2) In a homogenous, cohesive undrained environ­
ment Cu, the vertical shear failure compared to the
punch effect shear failure at the base of the column is
9 Cu. Therefore, the maximum stress at the head qrpis
determined according to the following formula:
5.4.1 . qrp=9 . Cu+Lc·(2 . Cu/Rc- gc)where:
Bulging failure • c: column unit weight of column material,
• I'c: column length,
(1) By analogy with the triaxial conditions, the effec­ • Rc: average radius of the column.
tive lateral expansion failure stress qreis given accor­
ding to the depth and radial stress  rby: Comment 1: Minimum value that balances out the
• qre=  r. tan2( /4 + 'c/2) Greenwood (1970) ground resistance:
(2) The radial stress value results fromthe geotech­ • ULSload: Lc Rc[(ULS  0ULS/Cu)- 9]/2
nical context and is estimated based on laboratory tests • SLS load: Lc Rc[(SLS  0SLS/CJ - 9]/2
(triaxial tests) or in situ tests (pressuremeter, vane tests • 0is the column head stress
and cone penetration tests) as part of the project's geo­ • with  uls=1.5 and SLS=2
technical study. (3) For multilayer soil, the formula becomes:
Comment 1: For example, for the pressuremeter,
the radial stress is estimated as follows for each soil • qrp=9 . Cup+Lc(2 . Cum/Rc- c)where:
layer: • Cup is the soil cohesion at the column base,
•  r#Ple* • Cumis the average cohesion along the length
If in addition,  'cis around 38 degrees, then: of the column.
• qre#4.p,/ (4) The cohesion values (Cu, Cupand Cum)result from
Where ple*is the equivalent net maximum pressure the geotechnical context and are estimated based on
calculated according to the following formula: laboratory tests (triaxial tests) or in situ tests (pressure­
meter, vane tests and cone penetration tests) as part of
• ple*= min/(ple*[z]) over the length of the column at the project's geotechnical study.
each layer Comment 1: For example, for the pressuremeter
• with ple'[z] =geometric average ofp* values over a the following can be used:
length of 2  SCcentered on z, but limited to 1.5times the • Cu#p1*/5.5 when pl*<.3MPa, Cuin MPa
lowest pl*value for alength of2  SC. • Cu#p1*V10+.025 when p1* 0.3MPa, Cuin MPa
Comment 2: For example, for CPTs, the radial stress Comment 2: For example, for CPTs the following
is estimated as follows for each soil layer:
can be used:
• r#qcc/3 • Cu# (qc- p0)/15where pois the total vertical stress
If in addition,  'cis approx. 38 degrees, then: for the level in question
•qre # 4/3 . qce Comment 3: Under the footings, the columns are
Where qceis the equivalent cone resistance calcula­ always considered non-floating according to the fol­
ted according to the following formula: lowing criterion.
• qce =min(qce[z]) over the length of the column at A non-floating column is defined as one that stops
each layer
• with qc[z] calculated according to the following
formula:
within a horizon characterized by Cup greater than
or equal to 150 kPa (or approximately p1* .8 MPa or
qc 2.5MPa) or such that 9 . Cup>qr.
57
REVUE FRANÇAISE DE GÉOTECHNIQUE
N° 144-145
3eet 4etrimestres 2013
 Comment 4: In all cases, it is necessary to calculate where qsol is the maximum stress transferred to the soil
and take into account the settlement of the soil under­ between the columns.
neath the columns. (3) Other methods are also available. It is necessary
to use the orders of magnitude obtained under special
conditions as follows.
Stresses within the columns Comment 1: These calculation methods should
make it possible to verify the bearing capacity and set­
tlement criteria described in §5.5.1, 5.5.2 and 5.6.
Failure stress
(1) Vertical failure stress qr within the column is Homogenization method for concrete slabs
equal to: and raft foundations subject to uniform, infinite loads
• qr =min(qre; qrp; 1.6 MPa)
(1) After constructing the columns, the settle­
ment for each layer i at the center of the building is
5 .5 4 .2 expressed as follows:
SLS stress
(1) At the service limit state (SLS), the allowable
vertical stress qaSLSwithin the column is obtained
by applying a safety factor of 2 to the vertical failure
stress qr:
*qaSLS=qr/ 2 =min(qre/2; qrp/2; 0.8 MPa) and the value of the stress within the column at the
layer i (Sci) can be expressed as:

ULS stress
(1) The maximum stress for calculation qaULSwithin
the column is obtained by applying a safety factor of where:
1.5 on the vertical failure stress qr: • ai :the replacement ratio for layer i
* qaULS = qr/ l ·5 =min(qre/1.5; qrp/1.5; 1.06 MPa) • Ecol : Young's modulus for the column
Comment 1: This means applying a coefficient of • Esj : Young's modulus for layer i
1.33to the allowable stress at the ULS:
*qaULS=1.33 . qaSLS • vsi : Poisson's ratio for layer i
•  t : average vertical stress exerted by the building
• hi:thickness of layer i
5. Comment 1: If there are results from laboratory
Evaluating stress and settlement oedometer tests (oedometric modulus for soil layer i:
Esoli), the above formulas become:
at the service limit state (SLS)
(1) The methods described below (§ 5.5.1 and 5.5.2)
are usually onlyvalid if:
• The additional load on the soil between the
columns (calculated according to the aforementioned
methods) remains inferior to the allowable stress for and the stress within the column at the layer i (Sci) can
the untreated soil. be expressed as:
• The column bases rest on a more compact ground
layer.
Comment 1: A layer is considered more compact
when it is characterized by Cupgreater than or equal to Comment 2: If there are results frompressuremeter
150kPa (approximatelyp1* 0.8MPa or gc 2.5 MPa) or tests (modulus EM,coefficient a), in keeping with Inter­
such that 9.Cup>qr. national Society for Soil Mechanics and Geotechnical
(2) They are based on the soil elasticity hypotheses Engineering recommendations, the ratio EM/ is com­
and on the conservation of flat, horizontal sections pared to the oedometric modulus.
and depend on the stiffness of each element (soil and In the traditional hypothesis with aPoisson's ratio of
columns).
Comment 1:Aspart ofpressuremeterandconepene­ 1:3, the Young's modulus for soil Esis then equal to 2/3
tration test methods, to remain within the elastic range of the oedometric modulus.
under the foundation element (raft foundation, concrete The above formulas become:
slab, footing, etc.) it must be verified that in the SLS:

58 • qsol <kp.p/2 +q'0forpressuremeter tests

•qsol<kc.qc/2 + q'0 for CPT

REVUE FRANÇAISE DE GÉOTECHNIQUE

| N° 144-145
| 3eet 4etrimestres 2013
and the stress within the column at the layer i (Sci)can be Comment 1: In all cases, it is necessary to calcu­
expressed by: late and take into account the settlement beneath the
columns.

Comment 3: If there are results from electric cone Bearing capacity In the Ultimate Limit State (ULS)
penetration tests (CPT), the following correlations can
be made for the most common and normally consolida­ (1) For a given footing (with area Ss = B . L) sup­
ted soil types: ported on n columns (with area Scol), with a surface
• Essolol = cC . qc avec cC common values appear in the overload qULS, the following must be verified:
table below ·(n.Scol ·qaULs+(Ss- n ·Scol . q'/ 2}>qULSSs
• with the oedometric modulus Esolfor the soil Comment 1: If there is a need for deformation at
the ULS, the method described for SLS can be used as
an initial approach.
TABLE II Values for  c.
5.6

c 4to6 4to5 2to3 Stresses other than vertical centered loads

Note: For other soil types and/or over-consolidated soils, other (1) Given a footing with dimensions B . Lsubject to
correlationscanbemade.
vertical loads and overturning moments, the elements
at the center are the vertical load Q (Q= q . B . L) and
bending moments Mxand My; Lis parallel to the x axis.
The above formulas become: (2) Assuming that there are columns along more
than one parallel axis (at the SLS and ULS):
• Along the x axis from Mx/Q>B/6
• Along the y axis from My/Q>L/6.
and the stress within the column at the layer i Sci)can be (3) The methods described in § 5.8 deal with the
expressed by: question in different ways. Below is a simplified
method.

5 .6 .1
Simplified method
Comment 4: Since pressuremeter and CPT corre­
lations can be difficult to make for certain types of soil (1) The value of Mxis increased by a factor  xequal
(saturated soft clay, unsaturated soil, etc.), oedometric to:
moduli measured in alaboratory may be preferable. • 1.25if there are columns along several axes paral­
(4) The stresses should remain belowthe maximumlel to the x axis
allowable values specified in 5.4.4.2: • and 1.5 ifthe opposite is true (all columns are alig­
*  ci < q a i
ned along the x axis).
The total settlement ( Wi, increased if necessary (2) The value of M is increased by a factor  xequal
by the settlement of layers under the bases of the to:
columns) should remain below the values set by the • 1.25 ifthere are columns along several axes paral­
operating conditions. lel to the y axis
Comment 1: If the load varies from one area to ano­ • and 1.5 ifthe opposite is true (all columns are alig­
ther and/or if the stratigraphyis not uniform, it is neces­ ned along the y axis).
sary to also check if the differential settlement values are (3) The eccentricity values ey= x. Mx/Q and ex= y
acceptable. . M y/Q should verify:
Comment 2: If the column bases do not rest on a ex < dxand ey<dy.
more compact layer, a specific explanation is offered.
(4) Therefore, we can refer to an equivalent rectan­
gular layout (Meyerhof) q':
q' = q . B . L/Sr
Columns under footings with vertical with Sr= (B- 2 . ey).(L- 2 . ex)
and centered load bearing (5) It is then possible to apply the simplified methods
describedin§5.5.2.1tothe compressedpart ofthefooting.
(6) For a column to be taken into account in the cal­
SLS loads culation, it is necessary to verify that it is located in
the part of the soil that is entirely compressed. The
(1) The steps for calculation are the following for a increase in y of 1.25 and/or 1.5 makes it possible to
given footing (whose surface Ss=B . L) supported on n avoid this verification by determining the dimensions
columns (with area Scol), bearing surface load qSLS. in advance.
59
REVUE FRANÇAISE DE GÉOTECHNIQUE
N° 144-145
3eet 4etrimestres 2013
 TABLEIII Phases ofcalculationunderfootings.
Step
0 First checkthe condition: Comment 1:Thisvalueq'umustbeprovidedorapproved
{n.Scol.qa+(Ss- n·Scol).q'u/3}>qSLS.Ss bythegeotechnical engineer.
withqamaximumallowablestress withinthe colum(see §5.2)
andq'usoil failurestress before improvement under acentered
load.n=number ofcolumns, Ss=surfaceofthefooting.
1 Calculatethe settlementwswithout treatment accordingtothe Basedonapressuremeter test:
rules ineffect. Thusthefollowingcanbe determined: ws=qsls (B .As/Ec+Ad.B/Ed)
ks=qsLS/Ws withAs= .c/9
Ad=1.2(d/0,6)a/9
andwhereEcandEdarethe equivalent pressuremeter
moduli correspondingrespectivelytothe areas
ofspherical anddeviatoricstress.
BasedonaCPT:
ws=C.qSLS.B/Esol
withEsol=c. qcwherecsee §5.5.1
andC=0.5for isolatedfootings and1.1for stripfootings.
2 Thesettlement equationfor acolumnwcolwithstress at thehead,
qcol
W ol='H·qiscothe
wchere l.Hlevel
/Ecolat whichthe settlement iscalculated
' isaratiothat shows that there isadistributionofstress from
the columntothesoil.
Comment 1:Inpractice, H=min(2.5B;Lc)isusedsinceover85%
ofsoilsettlement occursbetween 0and2.5B
Comment 2:As aninitial approach, use'=1 (nodistribution);
whenthedistributioncanbecalculated, thereis' =qmoy/qcol
whereqm oy istheaverageofstressesinthecolumn.
Forahomogenoussoil over2.5B, 'mini=0.67
3 Thecolumnstiffness canbe deducedandexpressedas:
kcol=qcol/wcol=Ecol/(' .H)
4 Thestiffness oftheentire "footing+column" structureat the
height inquestion:
k=(ks-.(Ss-n.Scol)+(n.kcolScol)}/(B.L)
5 Thefollowingcanthenbe deduced:
the settlement aftertreatment at height H
WSH
thefinal =qsettlem
SLS/k ent aftertreatment
wsf=wsH/0.85the stress under thefooting: qsol=wsf.ks
the stresswithinthe column: qcol=wsf.kcol
6 Thencheckthat thevalues are still withintherange
ofpseudo-elasticvalidity:
qsol<limit ofsoil'selasticbehavior qsol<kp . ple/2+q'o
andqcol<limit ofcolumn's elasticbehavior uptoqaSLS qsol<kc.qc/ 2+q'o
andthat thefinal settlement, includingthat ofunderlyingsoil
layers isacceptable.
Otherwise, start overfromstep 1changingthe number of
columns, their diameter and/or footingsize.

TABLEIV Eccentricity thresholds (applicable only

when: Mx =0 and My  0 or Mx  0 and
My=0).
dx
quasi-permanent SLS <L/6 <B/6
rare SLS <L/4 <B/4
ULS <L/2.2 <B/2.2

FIG. 5 Orientationofthe bending moments.

(7) Checkthat inthe rare SLS, 75%ofthe surface

ofthefootingremains compressed, that inthe quasi­ Other methods

60
permanent SLS, 100%of the surface remains com­
pressed, andtheninthe ULS, about half ofthe num­ (1) Other methods (without increasingMxandMy)
berofcolumnsremainsundertheentirelycompressed arepossible. It isthennecessarytomakesurethat the
surfaceSr. referencevalueforthesoil stressq'refafterthecalcula-
REVUEFRANÇAISEDEGÉOTECHNIQUE
N° 144-145
3eet4etrimestres2013
 • n r · S col . q aSLS + ( S r - n r · Scol) . q ' 3/ > q ' sLS · S r
with nrnumber of columns under the compresses sur­
face S in the SLS
• n r. S col · q aULS + (S r - n r . So l) . q 'u/2 > q 'ULS . S r
with nrnumber of columns under the compressed sur­
face Srin the ULS
• with qaSLS: maximum allowable stress in the
column at the SLS
• with qaULS: maximum allowable stress in the
column at the ULS
• q'u: soil failure stress before improvement
under a centered load
• Check that in the rare SLS, 75% of the surface
of the footing remains compressed, that in the quasi­
permanent SLS, 100% of the surface remains com­
pressed, and then in the ULS, about half of the num­
Taking into account columns in compressed ber of columns remains under the entirely compressed
areas. surface Sr.

tion of the distribution of the stress applied under the 5.7

footing remains within the valid range of settlement Special provisions for embankments
formulas.
q'ref = (3 . Rmax + qmin)/ 4 ' o u q 'ref = Q /(B - 2 . e y)(L - 2 .
(uniform load)
(1) The calculations for stone columns underneath
embankments are carried out as for columns under
raft foundations and concrete slabs subject to uniform
loads.
(2) These structures may require stability verifica­
tion in terms of sliding and soil creep.
(3) In this context, the following need to be distin­
guished:
• The initial model corresponding to the beginning
of the load bearing without any load transfer
• The gradual load transfer model (drainage phase)
• The final model after the full load transfer (end of
drainage).
(4) Regarding this issue, Priebe (1978) suggests
solutions for homogenization that, according to Mit-
chel (1981), can be applied over the long-term (final
load transfer model) making it possible to obtain an
equivalent, simplified model.
(5) This simple approach does not allow for consi­
deration of the load transfer mechanismand can unde­
restimate the safety factor when the incorporation rate
FIG. 7 Calculation of q'ref. increases.

m
Comment 1: As part of pressuremeter and CPT Other evaluation methods
methods, the followingmust be verified:
• The stresses in the soil must remain within allo­ (1) In all cases, step 0 described in § 5.5.2.1 must be
wable limits: applied.
• q 're f < k p . P el, / 2 2+ q 'o
• q' réf<kc qce/2 +q' (see step 6 on the table in
§5.52.1) Numerical finite element approach
• The stresses within the columns must remain wit­
hin allowable limits: (1) The finite element method consists of defining
a geometric model (soil + columns) in numerical sub­
• qcol <qaSLSin the SLS sets based on conditions of displacement and stress in
• qcol < qauLsin the ULS order to build an overall rigidity matrix.
• The overall bearing capacity is verified as follows:
61
(2) Then for each given lawof behavior and for each
• q' =q .B . L/[(B-2. ey)(L-2. ex)] load situation studied, the displacement, deformation,

REVUE FRANÇAISE DE GÉOTECHNIQUE

N° 144-145
3eet 4etrimestres 2013
 and stress are calculated precisely according to the (3) For concrete slabs, the expansion of the model
defined model. should take into account the position of the joints and
(3) This method offers the advantage of including limiting conditions. The verifications must cover the
all of the parameters of the chosen model (displace­ entire vertical load in both distributed and special
ment deformation, stress, iso-values for parameters situations.
and areas of prior plastification). (4) Determining the dimensions of the structure
(4)Nevertheless, this approach requires special atten­ should be based on a model that can take the soil-
tion to the modelization of the columns and limit condi­ structure interaction into account.
tions. The results depend a great deal on the rheological
model and its geo-mechanical parameters (inparticular,
the Young's modulus, Poisson's ratio, Cc, Cs, etc.) that
must be validated bythe geotechnical engineer. Other methods
(1) Among other methods, the following can be
cited:
Homogenization methods • The "additional moments" envelope method
involves defining corrective terms associated with the
(1) Homogenization methods offer the advantage of presence of the columns. These are added as an enve­
simplifying the "combined soil + columns" context to lope to the concrete slab stresses (bending moment)
obtain a basic model where the treated environment calculated according to Appendix C of standard NF P
is reduced to an equivalent homogenous environment. 11 213-1for the central part, panel edges and angles on
(2) The simplified homogenization technique was equivalent homogenized soil. There are two of these
first used by Priebe (1978) for stability calculations for corrective terms:
embankments on top of soft soil reinforced with stone - One that represents the influence of the columns
columns. on a continuous concrete slab.
Comment 1: The method described in § 5.5.1 is a - Another that represents the interaction between
homogenization method. the columns and the joints.
• Calculations based on the finite element method.

Possible effects on structural calculations

Footing or raft foundation with or without
a load transfer platform
Concrete slabs on a load transfer platform (1) To calculate the reinforcement for strip footings
or a raft foundation, the difference in stiffness between
(1) Several methods for calculating stresses on the the soil and the columns and the different load situa­
structures are possible. tions are taken into account.
Comment 1: For concrete slabs, the "soil onsite + Comment 1: The minimum reinforcement speci­
column" can also be assimilated with a homogenous fied in French standards for isolated footings is usually
soil (for which the modulus should be defined) in the enough if the construction provisions in Appendix I are
following cases: followed.
• Distributed load less than or equal to 30 kPa + Comment 2: It is important to be especially careful
load transfer platform of a thickness greater than or with structures for which no minimal reinforcement is
equal to 40 cm with a minimum EV2of 50 MPa and a specified in the standards.
maximumgrid of 9 m2. Comment 3: For footings with a load transfer plat­
• Distributed load between 30 kPa to 50 kPa + load form, if the cones of influence meet, the "soil onsite +
transfer platform of a thickness greater than or equal column" can be assimilated with a homogenous soil
to 60 cmwith a minimum EV2of 80 MPa and a maxi­ which canlead to using atraditional reinforced concrete
mumgrid of 7.5 m2. calculation.

Simplified method based

on the subgrade reaction modulus Quality control and acceptance

—
(1) The model provides a distribution of surface
springs equivalent to the "load transfer platform +
reinforced soil" combination. In practice, this distri­ Quality control during installation
bution can be described with two types of values: the
values applicable vertically and around the column set
by a slope of 2V/1H from the top of the column, and
those applicable to the additional surface. Calibration tests

62 (2) The structure is assimilated with a beamor plate

on elastic supports for which the values for each area
are subtracted fromthe preceding calculation.

REVUE FRANÇAISE DE GÉOTECHNIQUE

N° 144-145
3eet 4etrimestres 2013
(1) At the start of any stone column construction
project, the contracting company must carry out cali-
bration tests to validate the choice of material and Comment 1: These tests must be carried out by an
verify the compliance of soil reactions with expected expert approved by the contractor.
behavior (depths attained, consumption, possible swel­ (2) The results are entered together in the project's
ling, effects fromvibrations, etc.). quality control report.
(2) If the preliminary soil study shows remarkable (3) In addition, the installation tolerance (see § 4.9)
heterogeneity in the depths, nature or characteristics is verified for one column per 80 for concrete slabs
of the layers to be treated, calibrations should be car­ and raft foundations and by random sampling for all
ried out for each of the different areas in question. columns under footings.
(3) These tests are preferably carried out in the vici­ Comment 1: The columns installed outside the tole­
nity of the soil sampling or borings for the geogra­ rances are identified on ablueprint for verification.
phical study.
Comment 1: The calibration tests are specific infor­
mation tests.
Diameter verification
(1) Columns outside the building footprint can be
Informational tests side-stripped at different layers deep enough for dia­
meter verification.
(1) The stone column installation parameters are (2) For feasibility reasons, this excavation is usually
provided for one column per 50 (with a minimum of done from the top of the column to a minimum depth
3 per structure): depth, energy consumed and overall of 1meter beneath the platform.
volume of incorporated material.
Comment 1: These tests can be in the form of elec­
tronic data as long as all of the parameters are entered.
Checking continuity
(1) Static cone penetration tests (standard NF P
Attachments 94.113) are suited to this verification, though dynamic
CPTs (NFP94.114) can also be carried out.
(1) This reference document specifies the charac­
teristics for each column: date of installation, depth,
energy consumed, incorporated volume, etc.
(2) It also specifies the material and process to be Compaction verification
used.
Comment 1: These attachments can be in the form (1) This verification is carried out with a static CPT
of electronic data for all columns with the depth, energy (standard NF P 94.113).
consumed and volume incorporated. (2) This test must be done down to 1 m below the
tip of the column except in the case of refusal on the
underlying layer.
6.2 (3) To be consistent with the characteristics descri­
bed in § 5.3, the minimal characteristic for all points
Acceptance tests along the column axis from 1meter belowthe column
(1) Unless otherwise specified by the contractor, base must be equal to:
these tests are to be carried out by the project construc­ qcm= 10MPa
tion company and include (details in the table below): Comment 1: If the values are lower, specificjustifi­
• Checking the column diameter. cations must be provided.
• Checking continuity in case of a data entry error. Comment 2: A reminder of the definition of the
• Compaction verification. leveled tip resistance (qcm)
• Load test.
TABLE V Test frequency

recorded unrecorded recorded unrecorded

Checking diameter 1 per set of 50 columns up to 100, beyond that at least 3
Checking continuity 1/50 1/20 only if an anomaly 1/50
is detected
Compaction verification 1/80 under concrete slab or raft foundation + 1/20 under ground mass
with a minimum of 5
Load test* 1test up to 800 m and 1 per section beyond 800 m 1test up to 2,000 m and 400 columns, and at least
one more beyond 2,000 m
*For construction sites with less than 1,000 m of stone columns installed with the dry method (800 m by wet method), a load test may
not be performed, but in this case the allowable stress must be reduced by a factor of 1.5.
63
REVUE FRANÇAISE DE GÉOTECHNIQUE
N° 144-145
3eet 4etrimestres 2013
 (7) The test report includes:
• The loading/unloading plan with the measures for
comparison by time and load amount.
• ais equal to 0.5m • Achart showing settlement/time by increment.
• Dis the depth at which the characteristic is consi­ • A diagram showing the settlement after each
dered increment by load amount.
• q c(z) is the measurement obtained and leveled to • A diagram showing the creep slope by load
1.3qcm amount.
Comment 3: Complications can arise to make these (8) The test is considered conclusive if these two
tests difficult to carry out: criteria are met:
• Blocking due to large pieces of column material • The critical load ("creep") is not attained.
• Deviation of the drill pipe string, which can slip out • The settlement at the end of the service load incre­
of the column ment (Qn) is less than that estimated in the calculation
report and is compatible with the tolerances set by the
If this occurs, the contractor must provide the structure or building section.
recorded data for the column in question and suggest a
new quality control plan.
Comment 4: The other usual in situ geotechnical Glossary
tests, in particular pressuremeter tests (standard NF P
94.110) at every meter and type A dynamic CPTs (accor­
ding to standard NFP94.114) are more difficult to carry Column reference grid: The foundation area divi­
out in ballast-type materials. ded by the number of columns underneath it.
If the stability of the pressuremeter borehole walls
Replacement ratio: The ratio of the column sec­
• tion to the grid. It can vary according to the depth.
cannot be guaranteed, the pressuremeter test can
be carried out with a split spoon sampler (see NF Vibro-compaction stone column: This type of
P 94.110-1). column is formed by driving a pipe temporarily bloc­
ked at its base down to the desired depth. Granular
• To be consistent with the characteristics described material is then introduced through the pipe in small
in § 5.3, the minimal characteristic for all points quantities and gradually compacted by pounding
along the column axis starting from 1 m below the downwards. The pipe is then progressively pulled
top of the column taking into account the most out so that the volume incorporated - always greater
recent correlation must be respectively equal to: than the size of the pipe - matches the diameter in the
• p1= 1.2 MPa design calculations. The diameter of the pipe depends
• qd= 10 MPa on the soil and column characteristics. The driving
weight is adapted to the pipe. The amount of energy is
• If the values are lower, specific justifications controlled by varying the drop height.
must be provided.
Dry method vibro-stone column: This type of
column is formed by inserting a radial vibratory probe
to the desired depth. Granular material is then reintro­
duced into the soil and repeatedly compacted with the
Load test probe. The energy required and retraction speed are
(1) This load test is done at 1.5 times the SLS load adapted to the diameter and the degree of compaction
for column QNon one column at the site. specified by the design calculations. The drive system
for the vibratory probe can be electric or hydraulic.
(2) The load test requires installing a footing on the Finally, filling is done from the bottom through a pipe
top of the column, preferably leveled off under the load attached either on the side or on top of the vibratory
transfer platform. The surface of the footing should be probe.
less than 2.5 times the planned column section. Wet method vibro-stone column: Same as for
(3) This is a stress controlled test carried out with dry method stone columns, except that penetration is
compression. It involves measuring howfar the top of facilitated by water jetting (fresh or sea water).
a column (with two or three values as a basis for com­
parison) settles under a vertical load. Bottom-feed dry method vibro-stone column:
Stone column where the granular material is injected
(4) The load is applied gradually in six increments: fromthe bottomof the vibratory probe with a constant
• Qn/4, Qn/2, 3 Qn/4, Qn, 5 QN/4 and 3 QN/2. application of air pressure on the ballast, except when
(5) At each increment, the displacement is measu­ the supply pipe is being filled.
red at the following intervals: Deviation: Distance between the theoretical
• 1, 2, 4, 8, 15, 30, 45 and 60 minutes, or until the column installation location according to the design
comparison values stabilize. plans and the center of the column where it is actually
installed.
Comment 1: The deformation is considered to be
"stable"when its variation does not exceed .02mm per
minute.
Comment 2: The first increments can therefore be

64
30minutes long.
(6) The column is unloaded in four increments held
for 5minutes.

REVUE FRANÇAISE DE GÉOTECHNIQUE

N° 144-145
3eet 4etrimestres 2013
Appendix i - Standard configurations for stone columns under foundation
footings
Comment 1: Illustrations are provided for columns with a diameter of 80 cm. All dimensions are minimum mea­
surements and are in centimeters.
Comment 2:
• Without a specific feasibilitystudy, dmin Max (1.20m; 1.5 where dminis the distance from axis to axis.
• Without any special explanation, the footing overhangin relation to the theoretical outer bare part of the column
must be at least equal to the installation tolerance (e =20 cm).

It is advisable to Install the columns

in staggered rows order to avoid asymetry
in case of deviation.

65
REVUE FRANÇAISE DE GÉOTECHNIQUE
N° 144-145
3eet 4etrimestres 2013