2.

MECHANICAL IMPROVEMENT OF SOILS

BELOW GROUND SURFACE

D. A. GREENWOOD, PhD, BSc, Manager, Vibroflotation Department, Cementation Ground Engineering Ltd

Mechanical improvement of soils below ground surface may be 2. Following the first contract application at
achieved by deep compaction of cohesionless ground, or by Nuremburg in 1936, development of the vibroflot was
modification of cohesive soils by their partial replacement with pursued separately in Germany and the U S A , leading,
stronger materials introduced by mechanical means. The Paper
in the 1940s, to t w o machines of similar capabilities
relates only to means of achieving these ends by the respective
but of differing mechanical detail. Until recently
techniques of vibro-compaction and vibro-replacement, using a
large poker vibrator. The essential properties of this machine are these two basic designs, with only minor modifications,
described together with its mode of use, with or without jetting have been the most widely used in the western world,
fluids. The properties of treated ground which depend on the although little k n o w n post-war variants were pro­
machine, technique and soil type are indicated with examples. duced in Russia and Japan.
The philosophy of design of treatment is discussed for structures
on natural soils of cohesionless, cohesive and mixed stratified Machine characteristics
types, and on heterogeneous fills. Examples are quoted of settle­
ments of full-scale structures and test loads, and of other per­ 3. The main features of the vibrator are shown in
formance data. Figs 1 and 2.
4. The simple concept of compaction of loosely
Notation packed granular soils by vibration demands that the
y bulk density machine produces and maintains maximum amplitude
c undrained cohesive strength and lateral reaction at the tip for a given power, when
c' drained cohesive strength it is working under restraint of the surrounding ground.
<£' angle of internal friction in drained triaxial test
m moisture content With regard to the damping effect of the soil, the larger
z depth below surface the power and amplitude of the essentially horizontal
K p coefficient of passive earth pressure vibrations the larger is the radius of effect. There­
q applied stress fore a relatively l o w frequency is used; the original
P p passive earth pressure
German and American types work at 50 and 30 c/s
Introduction respectively for a similar input of the order of 75 kW.
In terms of practical effectiveness, the differences are
Vibroflotation or vibro-replacement techniques for
small.
strengthening weak natural soils and fills at depth are
now well established. Numerous examples have been 5. Recent British development has been to increase
1 7
p u b l i s h e d " of applications to soils ranging from soft the power and amplitude of vibration substantially
organic and alluvial clays to gravels and coarse rubble while reducing the machine diameter.
fills; the types of structure supported vary from single
to twenty storeys, and from traditional brick construc­ Factors limiting application
tion to industrial plant and oil tanks. The method is 6. Application is limited by the machine's ability
used primarily for limiting settlements resulting from to improve a given soil and, considering the degree of
vertical structural loads, and also for related applica­ improvement, depends o n whether the proposed struc­
tions such as embankment stabilization. S o m e un­ ture can accommodate settlements o f the treated
usual applications include the sinking of ground ground under load.
anchors and caissons and the reduction of soil perme­ 7. Densification of loose cohesionless deposits
1
ability ; however, these uses tend to arise from specific depends upon their ability t o compact under gravita­
j o b circumstances. tional force, when vibration destroys intergranular

II

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 PAPER TWO

friction. This is inhibited by permeability as the soil soils is determined by machine characteristics enabling
must expel pore fluids in order t o compact. In it to penetrate. This is primarily a design parameter,
cohesionless soils'of l o w permeability, compaction is dependent o n the cross-sectional area o f the machine
inefficient or uneconomic because of the time required and lateral reaction at the tip which is related t o power,
t o achieve it. There is n o clearly definable lower but the initial density of the soil also has some in­
limiting grading o f soil beyond which compaction is fluence. The uniformity of grading of coarse ma­
impracticable, possibly because testing methods are terials also affects penetrability markedly. While
unable to highlight the significant factors with suffi­ uniform cobbles or house bricks can be penetrated to
cient precision. However, as a general guide, it is useful depths, coarser soils cannot generally be pene­
suggested that soils of permeability less than 10 to trated unless they have a wider range o f grading in­
1 jim/s cannot be compacted without resort t o an im­ cluding sand sizes.
ported coarse-grained backfill for the bore formed by 9. Genuine soil cohesion also inhibits vibratory
the machine. The coarse backfill reduces attenuation compation and, together with low permeability, causes
of the vibrations and facilitates drainage. clayey soils to be virtually unaffected between treat­
8. The limit of application to coarse cohesionless ment points. Such soils can be strengthened only by
packing the bores with a coarse-grained backfill to
form a dense and relatively strong and rigid granular
column at each treatment centre. Some local c o m ­
paction may result from physical displacement during
penetration of the machine and packing of the back-

Uifting pulley

Manifold section
Water hoses
Hydraulic hoses

-Follower sections

-Top jets

.Vibration isolator

-Vibrator section

-Hydraulic motor

Water passage
to nose cone
A K
-Fins to prevent twist

-Nose cone

Fig. I. Vibroflot suspended from crawler crane Fig. 2. Essential features of vibroflot

12

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 GREENWOOD

fill, but the radial influence outside the column is 18. Thus resort to compressed air should be made
negligible. only in those soils of such low permeability that the
10. The backfill material is usually of 2 0 - 7 0 m m rate of water accumulation in the bore is negligible
size, frequently of uniform grading within this range. during the time taken to form a column. Somewhat
11. The introduction of the coarse backfill en­ paradoxically, cohesive soils of strength less than about
2
hances the compaction of surrounding ground re­ 2 0 - 2 5 k N / m cannot be treated by dry techniques.
sponsive to vibration, and thus mixed or stratified 19. The action of the vibrator in soils of low
ground of cohesive and non-cohesive types can be cohesion, especially in the presence of groundwater,
improved substantially. can form a clay slurry which, if incorporated within
12. Subject to the technique of construction, and below the column, precludes intergranular con­
granular columns can almost always be formed in soft tact and a dense state of packing of the backfill which
2
clays of strength greater than 7-5 k N / m . The bear­ are essential to its strength and rigidity.
ing capacity of this treated ground is dependent upon 20. In soft clays water circulation from the tip of
the degree of support provided by the natural clay to the machine up the bore allows penetration largely by
the columns, as well as on column diameter, consti­ jetting action within a stabilized bore. The water
tuents and state of compaction. In soft cohesive transports disturbed clay from the hole and cleans the
soils, therefore, the controlling factor on the use of descending backfill in the annular space around the
the process is usually the tolerance of the structure to machine. Thus a well defined column is formed of
settlements. maximum diameter dependent on, and in inverse pro­
portion to, the stability of soil exposed in the bore.
21. Field piezometric observations 0-3 m outside
Jetting fluids as aids to treatment a column in soft clay have indicated pore pressure
13. It is almost always necessary to circulate either fluctuations during construction with water circulation
water or compressed air through the machine into the 2
from - 4 0 to 4-30 k N / m . N o comparable figures
bore to achieve effective treatment. The choice of are available for columns formed without water.
jetting fluid is determined by technical considerations 22. Some fine-grained waterlogged cohesionless
relating to the action of the machine in the ground, soils can be successfully treated by carrying batches of
and it is quite possible that sub-standard or totally stone down with the machine during a dry penetration
inadequate results follow on neglect of these con­ phase. With this technique it is possible to form a
siderations. column, although the consumption of material tends
14. In sandy soils a considerable degree of c o m ­ to be high and the process is slow. Fill materials such
paction is achieved during penetration of the machine, as pulverized fuel ash can sometimes be treated by this
this being more significant with machines of large dia­ method.
meter. Hence, a downward water jet from the tip is 23. In the Author's experience most practical
used to induce a locally quick condition enabling the situations demand the wet technique. The dry process
machine to sink. During withdrawal of the machine can normally be used only in fine grained soils or
seepage from the uncased bore stabilizes the hole and mixed fill materials above the water table, or in coarse
saturates the soil to provide conditions in which cohesionless ground.
8
vibration can achieve maximum density.
15. In highly permeable coarse-grained cobbles or
rubble, the advantages of water circulation are neg­
Treatment of cohesionless soils
ligible and treatment is undertaken without a jetting Determination of treatment
fluid. 24. Loose natural sands frequently exist with
16. When working in soils of low permeability to widely varying density over the area of a construction
form stone columns with coarse imported backfill, site. Similarly cohesionless fills are likely to have a
water circulation should always be used where the bore density variation dependent o n the method of filling.
is unstable or where natural groundwater stands within Such variations can lead to substantial differential
the depth of the bore. settlements. Thus the major objective of treatment
17. When the machine penetrates without jetting of cohesionless soils is to produce a zone of ground
water, it tends to displace the soil laterally, resulting beneath surface spread foundations of uniform and
in a degree of local compaction or a heave at surface. high density.
The bore is almost the same diameter as that of the 25. Spread footings usually take the form of in­
machine which must be withdrawn completely to dependent pads, strips or rafts* whose settlement is
facilitate backfilling. The suction developed thereby governed primarily by the state of compaction of the
is conducive to collapse of soft cohesive soils or fine­ ground immediately beneath the footings, and which
grained silty sands, especially below groundwater. in natural situations is likely to be least dense. Treat­
Circulation of compressed air through the machine to ment therefore aims to provide a pier or raft of c o m ­
generate a back pressure in the hole improves stability, pacted ground specifically beneath each footing.
but below water level the rising air bubbles increase 26. Treatment may extend to reach compact
the risk of collapsing the hole. ground at shallow depths, but is otherwise normally

2 13

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 PAPER TWO

determined by the depth of the pressure bulb predicted to 1 - 8 m for a dune sand. Respective cone resistances
by elastic theory, i.e. usually about twice the minimum near the centres of compactions were of the order of
plan dimensions* of the footing. Where rafts or 2
8-6 M N / m and 16-5 M N / m , compared with 3-5 2

groups of closely spaced small footings stress deeper 2 2

M N / m and 6-9 M N / m for untreated ground.
zones, it is found in practice that a depth of treatment 31. Studies of the relationship between compaction
between 6 and 8 m below founding level is adequate. spacing and relative density measured directly have
This is sufficient to develop arching in the treated zone 9
been made by D ' A p p o l o n i a and Basore and B o i t a n o 10

to bridge loose pockets in untreated ground below. for fine uniform sands of effective size 0 0 8 m m and
27. Greater depth may be necessary where c o m ­ respective uniformity coefficients between 2 and 4, the
paction is required t o prevent liquefaction or settle­ latter being marginally more silty. The relative
ment of loose soils subjected to earthquake or artificial density ranged from 8 5 - 9 3 % at a spacing of 1-8 m
vibrations, t o ensure that dynamic stress levels t o about 7 5 % at 2-4 m. These results are con­
achieved during treatment everywhere exceed those sistent with measurements made in de-watered test
anticipated under working conditions. pits up to 7 m deep in a similar fine sand at Dungeness
28. Spacing of compaction centres to provide Nuclear Power Station; here the relative density
overlapping zones to cover a wide area of ground is averaged 9 1 % for tests made on ground treated at
determined by settlement tolerances for the loads t o 1-8 m centres. The untreated ground at these sites
be applied and depends on the relative characteristics averaged 6 0 - 7 5 % relative density.
o f the machine and ground. 32. These results imply that for typical unit loads
29. Most methods of estimating settlement of 2
in the range 100-430 k N / m compaction spacing of
footings o n granular soils are related directly or in­ between 2-5 m and 1-8 m is appropriate. All these
directly t o relative density. Accurate measurement field observations relate to the American/British type
of relative density is difficult and expensive especially of machine.
in waterlogged conditions; hence the c o m m o n use in
site investigation practice of penetration tests which Settlement of compacted ground
can be related approximately to relative density or 33. The use of post-treatment penetration tests for
settlement. design purposes tends to lead to conservative estimates
30. Penetration resistance of the D u t c h cone at of settlements. It would be instructive to study
various distances from compaction centres has been settlement records for full-scale structures on com­
2
determined from large numbers of tests by W e b b . pacted ground but few have been published, presum­
Soils ranging through clayey, silty and windblown ably because post-treatment testing has indicated
sands were examined on three sites. The maximum satisfactory results which were subsequently proved
radius of influence ranged from 1 - 3 m for a clayey sand during the working life of the structures. Grimes and
3
Cantlay measured total settlements of 20-25 m m one
Unit l o a d : k N / m 2
year after completion of a twenty-storey structure
founded on a 13 m wide raft resting on 13 m of loose
sands over silty sands and clays. Treatment was to a
depth of 7-5 m. In this instance calculated total
settlements were respectively 95 m m and 210 m m with
2
and without treatment for unit loadings of 193 k N / m .

2
Unit load:kN/m

I ,1.3m
3^square
»2 J bases

Location of V*4 13m sq.base

compactions I-5m sq. base

Founding depth 0 6 m Location of Treated

Treatment depth 4 - O m compaction Untreated
Founding depth 0-6m
40-* Treatment depth 3-3m
Fig. 3. Port Talbot: loading test results on treated and untreated Fig. 4. Dunham Park: loading test results on treated and un­
ground treated ground

14

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 GREENWOOD

34. T h e results of direct loading tests o n dummy Clay Silt Sand Gravel
concrete footings are presented in Figs 3 - 5 for soils \/f]
of grading shown in Fig. 6. All the footings were o f
y
reinforced concrete 1-3—1-5 m square as indicated and 1. Port Tal bot
founded 0-6-0-9 m below ground surface. Load was
T 2. Ardeer
/ 3. D i nham Park
applied by jacking against kentledge, the typical set-up /
being shown in Fig. 7.
35. The untreated ground at Port Talbot and
D u n h a m Park had similar characteristics: standard J
penetration tests gave an almost constant 15 blows
per 0-3 m within and for 3 m below the treated depth.
J
—. .—, ' —-~| ' •—i • '
36. In each case the improvement resulting from 0-0010002 0-006001 0 02 006 01 0-2 06 I 2 6 20
treatment is substantial. Comparison of results on
these t w o sites indicates the effects of additional c o m ­
Fig. 6. Particle size distribution: Port Talbot, Ardeer, Dunham
paction centres and imported coarse backfill at Port Park
Talbot, leading t o a high bearing capacity.
37. Comparison of the tests at D u n h a m Park,
Ardeer and Ocker Hill suggests that the initial state of
compaction of the surrounding soil influences the
result obtained. T h e one inch diameter continuous
cone penetration tests (Fig. 8) show that the natural
ground at D u n h a m was in a moderately compact state,
whereas that at Ardeer was considerably looser. A t
Ocker Hill the test was conducted o n a clinker ash fill
well graded from silt sizes up t o 75 m m loosely tipped
immediately before testing, and presumably in a still
looser state. The measured settlements of the test
bases for similar loads reflect the initial state. This is
recognized in practice (together with structural con­
siderations) in that a closer compaction spacing is
used beneath isolated footings than beneath large rafts.

2
Unit toad-kN/m

Fig. 7. Typical load test set-up

Blows per 0-3m

Compaction centres SPT

no Imported backfill
Fig. 8. Records of dynamic one inch cone penetration tests at
Fig. 5. Ardetr and Ocker Hill: loading test results Ardeer and Dunham Park

15

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 PAPER TWO

Free draining blanket

Passive earth
pressure JP
P

(b) Stone columns under widespread load

Firm clay

(a) Stone column under strip footing Fig. 9. Stone columns under strip footings and widespread loads

Treatment of cohesive soils straining pressure, the maximum bearing capacity of

a column will rise when the ratio of the principal
38. For treatment of cohesive soils, stone columns
stresses is also a maximum depending upon the angle
are placed directly beneath the loaded footings and
of friction of the column material. Assigning typical
generally extend to underlying granular layers or firmer 3 2
values of density, 1900 k g / m and cohesion 24 k N / m
clays, or they are used in ground the strength of which
for the clay and assuming an angle of internal friction
increases with depth. Thus they tend to act initially
of 35° for a column of diameter 0-9 m, its bearing
as somewhat flexible piles according to the relative
capacity will be of the order of 240 k N .
rigidity of the compacted column and surrounding
43. For a strip foundation the restraining pressure
soil. Dissipation of pore pressure generated by load­
will be influenced by stress from the footing and will
ing is quickly effected by drainage through the
not be equal all round the column. However, the
columns so that settlement is rapid and foundation
order of ultimate bearing capacity will not be
soil strength improves.
materially different.
39. A s yet there are n o exact mathematical
44. Lateral support of the column is developed
methods of estimating bearing capacity and settlement
quickly as the method of construction packs the back­
of treated cohesive ground. However, the following
fill tightly against the surrounding soil.
primitive approach gives conservative results.
45. Clearly, the ultimate bearing capacity of a
40. When load is applied from a spread footing it column is dependent upon its diameter. With a wet
tends to concentrate o n the column as the stronger technique of construction, backfill consumption is
element of the composite foundation soil. The 3
usually 0-65-0-85 m per lineal metre from which a
column dilates and applies lateral stress to the sur­ diameter of 0-9-1-04 m may be inferred. The cor­
rounding clay which is resisted by passive pressure. responding figures for backfill consumption for
There is thus a triaxial stress system within the column. columns constructed without water are 0-28-0-4 m 3

Conventional theory of passive pressures implies an per lineal metre giving diameters of 0-6-0-7 m.
increase of pressure with depth at the shallow depths Using the soil and column properties already assumed,
concerned (shown in Fig. 9(a)). Beneath a rigid con­ a dry column would carry approximately 140 k N .
crete footing there will be a zone of the column, in­ When a factor of safety is allowed, this compares well
dicated by dashed lines, which will not be subject to a with the result predicted by the empirical method of
significant deformation. Under a typical strip foot­ 4
Thorburn and Mac Vicar which suggests a 100 k N
ing o n a clay of essentially uniform strength, the design load for the same circumstances.
column will be weakest at the base of this zone, where
46. Under widespread loads, lateral passive re­
the lateral support is least about 1-75-2 m below
straint for the columns away from the edge of the
ground level.
loaded area is much higher due to the equal all-round
41. This hypothesis is supported by model tests influence of the applied load (Fig. 9(b)). Generally
11
reported by W i l l i a m s simulating columns partially the contribution of applied load to restraining pressure
penetrating to various depths through a uniform clay. is much larger than that due to soil strength and
The model columns failed by bulging near the top density, and the total carrying capacity of the column
below a loaded circular plate of twice the column dia­ increases until the limiting factor becomes either local
meter. Irrespective of length, the load at failure did shear in the clay due to contact stress with individual
not differ materially. backfill particles or the resistance to penetration of the
42. Under triaxial stress with equal all-round re­ column as an end bearing pile.

16

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 GREENWOOD

47. Which of these modes of failure occurs first is layers exist, it is unnecessary to found the column as
probably determined by the relative strengths of soils an end bearing pile.
around and below the columns. 51. F r o m Fig. 9(b) it is clear that columns near
48. The piezometric measurements shown in the edge of widespread loads will not have the same
Fig. 10 were taken during construction of a 9 m high bearing capacity as those under the centre. Under
embankment on a site where columns were arranged embankments where loads diminish almost in pro­
on a 2-4 m triangular pattern. They show the transfer portion to column capacity this is not critical, but it is
of load to the toes of the columns as for conventional important if loads terminate abruptly at the edge of the
piles. Since the magnitudes of pore pressures in loaded areas as indicated by the dashed line. Columns
treated and untreated ground were similar, drainage near the edge must be spaced more closely than in the
effects at the time of measurement can be discounted. centre and should not carry loads greater than those
Settlements of treated and untreated zones were not for isolated footings.
materially different, attributable to toe loading of the
52. Observations of lateral soil displacement with
columns causing local over-stressing of the clays at
borehole inclinometers in flexible casings installed in
that level, or transfer of stress t o consolidating layers
uniform soft clays around the periphery of oil tanks
below. This behaviour would be anticipated beneath
suggest that, if peripheral columns are overloaded, a
a wide load.
progressive lateral displacement of relatively large
49. It should be noted that the dangerous pres­ magnitude can occur within 3 m of the surface. The
sures at shallow depths were transferred to levels below columns are not able to provide significant resistance
the peat layer and ultimately drainage would continue to deflexions induced by lateral forces, but the pre­
to improve overall stability. However, a slip in the sence of relatively high strength frictional material of
adjacent untreated zone ruptured the piezometric the columns helps to improve the overall shearing of
tubes and a long-term record was impossible. resistance of soft clays to rotational movements.
50. In stronger clays it is probable that arching in
53. Estimation of settlement within the treated
the column transfers stress laterally to upper levels of
depth must be by empirical means, as precise mathe­
the clay with little stress transmitted to the toes of the
matical solutions ensuring compatibility of column
columns, which become virtually redundant. H o w ­
and clay deflexions have not been derived. Fig. 11
ever, the determination of stress distribution is not
shows curves for estimating consolidation settlements
practicable with any accuracy and curtailment of
of soft clay strengthened by stone columns under wide­
columns could be dangerous. In cases where a signi­
spread loads. Such curves should be used with cau­
ficant thickness of strong clay crust overlies weaker
tion within the range indicated.
clay layers, or where alternating strong and weak
54. A n important difference between stone col­
umns and conventional piles is the facility of the former
Excess p i e z o m e t r i c head o f w a t e r :m to provide for accelerated dissipation of excess pore
1,5 30 4-5 60 7.5 pressures in the foundation soils due t o load. In
G.L practice most building loads are necessarily applied
slowly, except perhaps in the case of water retaining
structures. Hence the initial factor of safety used in
30
design can be very small as the increase of clay
strength following rapid drainage quickly increases
6-0 the factor of safety.
5 5. The importance of using a coarse backfill rather
than sand to lend rigidity is illustrated by a large-scale
9-0

I2-0 Spacing of stone columns: m

20 225 25 275 30 3-25
i
Clay strength
15. OA 40kN/m 2

2
1
^ 50-1 ^ /20kN/m

p r y method " V *
P. IOCHclay strength^
2
40kN/m
2LOH 5 U n t r e a t e d zone Curves neglect immediate settlement and shear
displacements
T r e a t e d zone Columns assumed resting on firm clay, sand or
harder ground

Fig. 10. Pore pressures in soft clays strengthened by stone Fig. 11, Settlement diagram for stone columns in uniform soft
columns clay

17

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 PAPER TWO

field test at Bremerhaven for motorway slip road em­ settlement plotted against time from the settlement
bankments o n a difficult site. The ground conditions plates indicated in Fig. 12(a) which exclude peripheral
and properties are shown in Fig. 12, which also shows zones in each area. For the stone column, sand
the distribution of treatment. Before treatment a column and untreated areas, average settlements and
sand blanket was placed over the surface of highly range of settlement after 15 months were respectively
compressible peat to provide a working surface and t o 468 ± 31 m m , 645 ± 75 mm, and 765 ± 13 mm. The
ensure that the columns would be formed integrally calculated settlement for untreated ground was
with a free draining layer. 665 mm.
56. A n identical technique using water circulation 58. Thus even with the relatively wide spacing of
was employed for the construction of all columns. In 2-3 m the stone columns had reduced the settlement
one group of columns a sand backfill of grain size by almost half. The reduction with sand columns
0 - 3 m m was used, but the remainder were formed of was only 15% relative to untreated ground. A closer
gravel of size 3 0 - 7 0 m m . Average column diameters spacing would have produced less settlement.
inferred from backfill consumptions were in each case 59. The correspondence of the estimated and mea­
about 1-2 m. sured results for untreated ground is g o o d having
57. Following treatment, settlement plates were regard to secondary consolidation in the peat.
established as indicated in Fig. 12(a) and an earth fill 60. In this instance the settlement against time
load was placed over the site, including an adjacent records show only marginal improvement in the rate
untreated area. Fig. 12(c) shows records of average of drainage due to the presence of the columns, but
this is probably due to the nature of the peat and the
1 2
proximity of the treated area. W e b b has noted
-20m- that sand drains installed by vibroflotation techniques
are more efficient than those installed by mandrels.
j23m|
71
! o « o * o
Treatment of alternating clays and sands
I I
61. Practical situations do not usually involve
Igi-h ©•©•©«
wholly cohesionless or cohesive ground. Frequently
jM O•O•O € I clays and sands are interlayered. The scale of these
j ©—© O layered or lensed structures relative to the foundation
sizes and depths influences the approach when con­
COMPACTION CENTRES
sidering vibroflotation. Fig. 13(a) shows a substantial
© Stone backfill • Settlement plate
u s e d f o r m e a f t r e s u I t
O Sand backfill < >
o Settlement plate.

E
Sand blanket O o
Earth fill

Peat and ctayt

O
Fine uniform sandMB
In
i :
•::"-"Vv ::;V:.;: :
<o
C-5-IOkN/m 2
m . 400-4407o •^C#^\ Stiff clay
<p»l3-20° m «0004m /kN 2
or sand
v
Stone columns

Ca) Compacted sand raft over soft clay

(c) 1968 | 1969
I Jun Jul Aug Sep Oct Nov Dec I Jan Sep

BOOH
(b) Stone columns stiffening clay lenses

Fig. 12. Bremerhaven test Fig. 13. Treatment of mixed clays and sands

18

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 GREENWOOD

thickness of sand overlying a similar depth of soft years after occupation of the school, levels of the
cohesive soil. Providing the sand is thick enough to three-storey block were recorded, and again after a
disperse locally intense stresses beneath individual further three months. Total settlement of this block
footings, general compaction to a uniformly high was 40 m m with a differential of 18 m m ; also as far as
density will be sufficient to control differential settle­ can be detected settlement is sensibly complete.
ments within the sand layer which acts as a raft over
the compressible stratum. In such cases elastic Treatment of fills
theory predicts a concentration of stress in the upper 69. Fills frequently contain a wider range of soil
strong layer according to the ratio of moduli of particle sizes the distribution of which is much more
elasticity of the two layers. For compacted sand over irregular than in natural ground; the state of compac­
soft clay this ratio may be of the order of 100; rafting tion is also more variable. Otherwise they respond
is thus very effective. to vibration as natural soils and can be strengthened
62. If total settlement must also be reduced, treat­ by granular columns.
ment can be continued through the compressible layer 70. The successful treatment of a city refuse and
with stone columns. general dump for silo foundations loaded t o about
63. Figure 13(b) illustrates a case where the scale 2
215 k N / m has been described by L u c e . The tip had 5

of the compressible layers is much less relative to foot­ been filled over many years and contained a high pro­
ing size. If the length L to diameter D ratio of the portion of ash and granular material. Treatment
stone column is less than unity in the compressible reduced settlements to approximately 50 m m c o m ­
lenses, the columns form a relatively rigid structure pared with 150 m m for structures on untreated fill.
and settlements due to the presence of the lenses can 71. Similar tips have been treated elsewhere, but
almost be neglected. A n example of this in practice the character of modern tips with less granular content
is given in Figs 14 and 15. and a high proportion of organic and plastic materials
64. A one- to three-storey school structure was to usually precludes effective stabilization.
be founded on pad and strip footings designed for a 72. Treatment of demolition sites resulting from
2
maximum unit loading of 215 k N / m . The three- city centre re-developments is n o w almost routine.
storey block was founded on pads 1-3 m square and These sites are usually filled to depths of about 3 m
3 m apart with internal strips approximately 0-9 m over firm strata on which the original buildings were
wide, all founded about 1 m deep. founded. They contain basements loosely filled with
65. The high water table, and the loose sand would brick and rubble debris, between which clay over­
have made excavation of the peat difficult and ex­ burden from original excavations has been dumped to
pensive. Conventional piling was unattractive as the make up to present ground level.
length of the driven piles was difficult to pre-determine 73. This variety of material is normally strength­
and bored piling would be problematical in the water­ ened by granular columns to make up for compaction
logged silty sand.
66. Vibroflotation using coarse imported backfill
was used, extending to depths of only 4-6 m. Blows perO-3m
20 40 60 SO IOO 120
67. The results of one inch diameter dynamic cone _JL_
penetration tests before and after treatment are shown
in Fig. 15. The presence of at least two soft layers
can be detected from the records of pre-treatment and o-6-
post-treatment tests between columns. The resis­
tance of the columns is high. 12
68. Estimated settlements without treatment range
from 50 m m to 95 mm, according to the thickness of
peat lenses and initial density of the sand. Three

' 2-4-
IOO

30-

42-

48 I inch did. dynamic cone

penetration tests
OOI O l I
Fig. 14. Blackpool, Palatine School: ground conditions Fig. 15. Blackpool, Palatine School: penetration tests

19

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 PAPER TWO

ft i
KH
Fig. / 6 . Manchester, Hulme V: exposed columns in demolition
debris
and displacement of the fill during penetration of the
vibrator and to cater for contained clay zones, as it is
not practicable to define the boundaries of various
types of fill with sufficient precision to vary the tech­
nique. Columns revealed by excavation of treated
Total load: kN
debris at the Hulme V development, Manchester, are
O 200 , 400 6QO 8OO , IOOO . I2QO shown in Fig. 16.
74. Figure 17 shows results of loading tests o n
dummy footings at the same site where long crescents
of six-storey factory built flats are being erected for a
2
working load of 1 6 0 k N / m on cross-wall footings.
The results are comparable with those for tests on
cohesionless soils and show differential settlements
related to the ground type rather than to compaction
grouping.
75. Table 1 shows plate bearing test results o n the
same site compared with those for a number of other
similar fills, refuse and general fill, and a sand fill.
a
152 Sm Except where noted, a test load of 375 k N / m was
e"
Bos.es in Founding depth 0-6m applied at 0-6 m below ground level.
1 to4^ w Depth of treatment
2 0 to3-Om 76. The test load is approximately twice normal
^ f- according to location working load for housing sites, and even under this
Location of columns
overload the total and differential settlements are well
Fig. 17. Manchester, Hulme V: loading test records within normal building tolerances.

20

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 GREENWOOD

Table I. Results of 0-6 m dia. plate bearing tests on fills

Site Mean Number Range, Ground Remarks

settlement, of mm
mm tests
Manchester Hulme V
1-5 m square compaction
On cols 2-5 22 60 1 Clay, brick, rubble fill, pattern beneath 3 m
Between cols 3-7 3 1-2 | demolition debris wide strips

Rochdale
On cols 1-8 3 0-5 Silty sand, ash, demoli- 2-1 m compaction spacing
Between cols (wet technique) 4-5 2 11 > tion debris, brick, beneath narrow strip
On cols 1-8 1 rubble fill foundations
Between cols (dry technique) 7-9 1
Leeds, Balm Road
On cols 110 2 160
Between cols 420 1 2
Untreated 840 1 Test load 305 kN/m , 2-1 m
> Brick rubble, ashes, compaction spacing be­
Leeds, Belinda Street sandy clay fill neath narrow strip foun­
On cols 1-6 1 dations
Between cols 4-6 1
Untreated 120 1

Birmingham, Hobmoor Road

On cols 5-5 4 60 Clayey sand and gravel 2-1 m compaction spacing
fill, refuse brick rubble beneath narrow strip
foundations

Runcorn
On cols 60 7 8-6 Uniform medium sand
Untreated 11-5 1 (old fill) uniformity co­
efficient, D olD = 2
6 10

Liverpool 2
Between cols 4-3 7 3-5 Brick rubble, ashes, de­ Test load 310 k N / m
molition debris, soil

Conclusions ments in the vibrator, but field measurements of the

77. Examples quoted in the Paper and elsewhere performance of full-scale structures o n treated and
illustrate the effectiveness of vibroflotation in achiev­ untreated ground are urgently needed to enable further
2
ing bearing capacities of 200-500 k N / m o n granular development of the technique.
materials with only a few millimetres of settlement.
The process has proved reliable over 35 years. Acknowledgements
78. In the past decade granular columns t o 82. The Author wishes to thank the City Architects
strengthen cohesive ground have also proved their of Manchester, Leeds, Birmingham a n d Liverpool,
worth. However, the effect o f initial soil strength the Borough Architects of Blackpool and Rochdale,
on that of treated ground is more marked in clays the Central Electricity Generating Board, the British
than in sands, and the relationship between original Steel Corporation, Imperial Chemical Industries Ltd,
ground and the type of structure needs consideration D u p o n t Co. ( U K ) Ltd, Terrasond Grundbau G m b H
early in design studies. and Cementation Ground Engineering Ltd, for their
79. Where technically acceptable, these techniques co-operation and permission t o publish information
offer considerable economy over conventional piling contained in the Paper.
arising mainly from the limited depth of treatment and
use of conventional spread footings. References
80. Acceptance of the concept of designed settle­ 1. DOSCHER H . D . Vibroflotation. / . Ind. nat. Soc. Soil
ments and the uniformity of ground reaction after Mechs, 1968, 7 (July) No. 3, 365-385.
treatment have led t o economies arising from the 2. WEBB D . L. Effects of vibroflotation on clayey sands.
application of the technique t o a wide range of mixed J. Soil Mech. Fdn Engng Proc. Am. Soc. civ. Engrs,
soils and fills hitherto considered unsuitable for 1969, 95 (Nov.) SM6, 1365-1378.
3. GRIMES A. S. and CANTLAY W . G. A twenty-storey
foundations. office block in Nigeria, founded on loose sand.
81. Further economy will follow from improve­ Struct. Engr, 1965, 43 (2) 45-57.

21

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.
 PAPER TWO

4. THORBURN S. and MACVICAR R . S. L. Soil-stabiliza­ tion of the fundamental factors controlling loose
tion employing surface and depth vibrators. Struct. packing of sands. Proc. 2nd Int. Conf. Soil Mech.,
Engr, 1968, 46 (10X309-316. Rotterdam, 1948, 6, 80.
5. LUCE A. F. The strengthening in depth of weak soils 9. D'APPOLONIA E. Loose sands—their compaction by
by vibration at Tolka Quay. / . Ir. Instn civ. Engrs> vibroflotation. Symp. dynamic Test. Soils, American
1968 (Jan.) 3-12. Society for Testing and Materials, 1953, STP 156.
10. BASORE C. E. and BOITANO J. D . Sand densification
6. WATT A. J. et al. Loading tests on structures founded by piles and vibroflotation. / . Soil Mech. Fdn Engng
on soft cohesive soils, strengthened by compacted Proc. Am. Soc. civ. Engrs, 1969, 95 (Nov.) SM6,
granular columns. Proc. 3rd Asian Reg. Conf. Soil 1303-1323.
Mechs, Haifa, 1967,1, 248-251.
11. WILLIAMS J. D . G . Small scale tests on granular piles
7. APPENDINO M. and JAMIOLKOWSKY M. Experiences of in soft clay. BSc thesis, University of Wales, June
the uses in Italy of the vibroflotation process for com­ 1969.
paction of soils. 9th Geotech. Symp. Italian Geo-
f 12. WEBB D . L. Foundation problems of the Natal coast
technical Association, Genoa, 1968 (Oct.). area. Natal Conv., South Africa Institution of Civil
8. KOLBUSZEWSKI J. J. Discussion on General investiga­ Engineers, 1968.

22

Downloaded by [ La Trobe University] on [27/10/16]. Copyright © ICE Publishing, all rights reserved.