Bearing capacity beneath tapered blades of open dug caissons in sand

Abstract

An open-dug caisson shaft is a form of top-down construction in which a concrete shaft is sunk into the ground using the weight of the shaft and additional kentledge, if required. Excavation at the base of the caisson shaft wall allows the structure to descend through the ground. A thorough understanding of the interaction between the caisson shaft and soil is essential to maintain controlled sinking of the caisson. In this paper, the failure mechanisms developed beneath caisson blades in sand are investigated. A series of laboratory tests were carried out at the University of Oxford to explore how varying blade angles affect the performance of the bearing capacity beneath the caisson. Cutting angles of 30°, 45°, 60°, 75° and 90° (flat) were penetrated into sand under plane strain conditions; forces were monitored using a Cambridge-type load cell while soil displacements were recorded using Particle Image Velocimetry (PIV) techniques. The aim of this study is to understand how the soil failure mechanism develops and to determine the optimum cutting angle. The results of the laboratory tests can be scaled to predict the likely behaviour in the field. Results show that the bearing capacity is significantly dependent on the cutting angle; in a dense sand a steep cutting angle may be used to aid sinking of the caisson.

Full text

ABSTRACT: An open-dug caisson shaft is a form of top-down construction in which a concrete shaft is sunk into the ground
using the weight of the shaft and additional kentledge, if required. Excavation at the base of the caisson shaft wall allows the
structure to descend through the ground. A thorough understanding of the interaction between the caisson shaft and soil is essential
to maintain controlled sinking of the caisson. In this paper, the failure mechanisms developed beneath caisson blades in sand are
investigated. A series of laboratory tests were carried out at the University of Oxford to explore how varying blade angles affect
the performance of the bearing capacity beneath the caisson. Cutting angles of 30°, 45°, 60°, 75° and 90° (flat) were penetrated
into sand under plane strain conditions; forces were monitored using a Cambridge-type load cell while soil displacements were
recorded using Particle Image Velocimetry (PIV) techniques. The aim of this study is to understand how the soil failure mechanism
develops and to determine the optimum cutting angle. The results of the laboratory tests can be scaled to predict the likely
behaviour in the field. Results show that the bearing capacity is significantly dependent on the cutting angle; in a dense sand a
steep cutting angle may be used to aid sinking of the caisson.
KEY WORDS: Caisson; tapered blade; bearing capacity; laboratory testing.
1 INTRODUCTION
An open-dug caisson shaft is a form of top-down
construction. They have many functions in industry such as
launch and reception pits for tunnel boring machines and
underground storage tanks for foul and storm water attenuation.
Caissons can be sunk through various soil types including sand,
clay and rock. It can be a very safe and efficient form of
construction as the permanent structure is used to retain the soil
and water during excavation as shown in Figure 1.
Figure 1 30 m diameter reinforced concrete caisson at
Anchorsholme Park, Blackpool, constructed by Ward and
Burke Construction
In order for a caisson to sink into the ground, both the skin
friction that develops between the soil and the concrete shaft in
addition to the bearing capacity of the soil beneath the shaft
walls must be overcome. A common technique to aid the
sinking process is the use of a tapered blade, or cutting edge,
beneath the wall of the shaft. The purpose of the cutting edge is
to cut into the ground, anchor the caisson horizontally and
maintain verticality. It mobilises the failure mechanism of the
soil towards the centre of the shaft, so that the soil in this area
can be easily excavated to allow the caisson to sink. In order to
achieve controlled sinking of the shaft, a thorough
understanding of the bearing failure that develops in different
soils is therefore essential. It ensures operatives in the field
know where to excavate to induce bearing failure beneath the
shaft walls. Moreover, a reduction in the bearing capacity
beneath the caisson blades means less kentledge will be
required to get the shaft to formation level.
The bearing capacity of sloped caisson footing depends on
the angle of the tapered blade, roughness of the interface, angle
of friction of the soil, width of the footing and the unit weight
of the soil. In this paper it is assumed that the caisson diameter
is large, 2D plane strains will develop with minimal conical
action. While some work has been undertaken on sloped
conical footings [1-2], limited information is available on 2D
plane strain conditions for tapered footings.
Tomlinson [3] recommends various tapered angles
depending on the soil type; flatter cutting angles are
recommended for sand compared to clay. Nonveiller [4]
describes various potential rupture surfaces. As the caisson
penetrates into the soil, the resisting forces increase until a new
state of equilibrium is achieved.
A method for quantifying the bearing capacity factor, Nγ,
using limit-equilibrium theory was proposed to quantify the
bearing capacity of tapered footings beneath caissons [5-6].
According to this approach, the value of Nγ is less sensitive to
the angle of friction compared to the blade angle and
embedment depth of the wall. However, some important
parameters are neglected in this method, such as the interface
friction between the soil and sloped blade of the caisson.
Bearing capacity beneath tapered blades of open dug caissons in sand
Ronan Royston1, 2, Bryn M. Phillips1, 2, Brian B. Sheil1, Byron W. Byrne1
1 Department of Engineering Science, University of Oxford, UK
2 Ward and Burke Construction, Kilcolgan, Co. Galway
Email: ronan.royston@eng.ox.ac.uk, bryn.phillips@hertford.ox.ac.uk, brian.sheil@eng.ox.ac.uk, byron.byrne@eng.ox.ac.uk

The aim of this study is to explore the performance of various
tapered angles in sand through a series of laboratory tests.
Particle image velocimetry (PIV) analysis is employed to track
the failure planes developed in the soil during testing. In
addition, a number of tests were carried out to examine the
influence of overburden pressure during sinking.
2 EXPERIMENTAL EQUIPMENT
A three degree of freedom loading rig, developed at the
University of Oxford [7], was used for the laboratory testing in
this study (see Figure 2 and 3). A Cambridge-type load cell,
attached to the end of the loading arm, records vertical,
horizontal and moment reaction.
Figure 2 Three degree of freedom loading rig
The loading rig was located on top of a 600 mm x 300 mm x
95 mm testing tank which has a perspex front, which allows the
sand movements to be recorded during testing, see Figure 4. A
Nikon DS3200 camera located 700 mm from the front of the
tank was used to capture images at a frequency of 1 Hz during
testing. A downward penetration rate of 0.5 mm/sec was
adopted for all tests; rate effects in sand are not expected to be
significant. PIV analysis was carried out by processing the
images using MATLAB module GeoPIV [8].
The experiments were conducted using a dry, yellow
Leighton Buzzard DA30 silica sand, the properties of which are
summarised in Table 1.
Table 1 Properties of Leighton Buzzard DA30 sand
Property
Value
D10, D30, D50, D60, D90, (mm)
0.36, 0.45, 0.51, 0.54, 0.65
Specific Gravity, Gs
2.73
Minimum dry density, min (kN/m3)
14.5
Maximum dry density, max (kN/m3)
17.1
Critical state friction angle, ’cs (o)
32
Loose sand samples were prepared using a sand raining
procedure in conjunction with a low drop height. Dense
samples were prepared by vibrating the testing tank after sand
raining. To ensure a repeatable bulk density was achieved for
each sample, the tank was filled using the same drop height for
all tests.
Figure 3 Test setup and soil failure wedge
A schematic of a typical cross section of a cutting edge and
possible soil failure plane is provided in Figure 3, where B is
the width at embedment, A is the length of the face, β is the
tapered cutting angle, QV and QH are the vertical and horizontal
reactions, R is the resultant reaction of the forces, and X and Y
are the width and depth of the failure plane. It is worth noting
that value of β generally used in industry is 45o, derived from
on-site experience.
Aluminum pieces were created as the test pieces of varying
cutting angles, β. In order to reduce the friction between the test
pieces and the sides of the tank, 1 mm polytetrafluoroethylene
(PTFE) sheets were placed either side of the piece with
compressible foam placed in between the PTFE and the test
piece. The test piece spanned the width of the test tank to ensure
plane strain conditions.
Figure 4 Testing tank
Figure 5 Cone Penetration
3 VALIDATION OF TESTING PROCEDURES
Cone tests were performed, with an 8 mm diameter cone and
a 60° cone angle, to examine the uniformity of the sample. The
cone was penetrated 150 mm into soil samples prepared with
three different relative densities, ID, as per Equation (1), where
γ is the density of the prepared sample. The cone was attached
to the rig as shown in Figure 5 and penetrated through the sand
at a rate of 1 mm/s. From the results presented in Figure 6, the
cone penetration resistance appears consistent with depth thus
indicating a uniform sample. Moreover, it is obvious that at
higher relative densities, there is a commensurate increase in
the cone resistance.

 
 
(1)

Figure 6 Cone penetration results
A series of preliminary tests were carried out to validate the
present sand properties, sample preparation and experimental
techniques. Careful consideration of the stress level is required
in order to extrapolate model-scale laboratory testing to
expected behavior in the field. At higher stress levels the
dilatancy of soil is suppressed; relationships proposed by
Bolton [9] are used to relate the critical state angle of friction
to the peak angle of friction based on relative density and the
stress state of the sand in the laboratory testing. Equations (2)
and (3) are based on plane strain conditions for the relationship
between critical and peak angles of friction based on the
isotropic stress and density of the sample:
  
(2)

(3)
where ’max and ’cs are the peak and critical state friction
angles, respectively, p’ is the mean effective stress in the soil at
failure and ID is the relative density of the soil.
In order to validate the test results and sample preparation, a
90° piece (flat piece) was used to compare present
measurements to published literature. The relative density of
each test sample was calculated using the data presented in
Table 1 which, in turn, was used to determine the value of ’max.
Bearing failure was assumed to occur at 0.1B, neglecting soil
cohesion and influence of overburden, the bearing capacity, qrd,
can therefore be defined as follows:
 

(4)
where B is the width of the flat footing, Nγ is a bearing capacity
factor and  is the effective unit weight of the soil.
Test results for the flat pieces are shown in Figure 7.
Theoretical bearing loads are based on the approaches for
calculating Nγ by Hansen [10] and Meyerhof [11]. Test results
are consistent with theory and provide additional confidence in
the experimental set up and the Bolton method [9].
Figure 8 shows an example of the incremental displacements
in the soil obtained using PIV; the predicted failure plane
according to Rankine theory, with a friction angle of 32°, has
also been superimposed on the image. The sample in Figure 8
is for a dense sample which has a peak angle of friction of 45o
when applying Equations (2) and (3). This is not consistent with
the 32o overlaid as the dilation of the sand is suppressed at
higher densities.
The development of a triangular active wedge beneath the
footing, in addition to the passive wedges, is obvious from this
output.
Figure 7 Theoretical and measured bearing capacities
Figure 8 PIV of flat piece test
4 TESTING OF TAPERED ANGLE PIECES
A series of tests were carried out using cutting angles, β, of
30°, 45°, 60° and 75°. The test set-up for the cutting edges is
shown in Figure 3. Each test was carried out on a medium-
dense sample. The medium-dense tests have a relative density
ranging between 0.44 and 0.54.
Influence of cutting angle, β
Figure 9 plots the influence of β on the variation of R with
penetration. The resultant forces are higher for the shallower
angles; this is attributable to the much greater bearing width of
the flatter angles for the same penetration.
The influence of β on the relationship between QH and QV is
examined in Figure 10 and Figure 11. The relationships
presented in Figure 10 are remarkably linear where the steeper
cutting angles reduce the vertical reaction. The ratio of
horizontal to vertical force is shown in Figure 11 based on a
best fit line to the results in Figure 10. In general, there appears
to be a linear variation in QH/QV with β. It is worth noting that
QH is as high as 0.9QV for a 30° cutting angle which could result
in a large hoop tension force in the wall of the caisson.
ID=0.16
ID=0.48
ID=0.67
0
20
40
60
80
100
050 100 150
Reaction (N)
Penetration (mm)
Medium
Dense
0
50
100
150
200
250
300
350
400
0.00 0.20 0.40 0.60 0.80 1.00
qrd (kPa)
Relative Density, Id
Bearing Capacity -
Hansen [10]
Bearing Capacity -
Meyerhof [11]
Bearing Capacity
Test Results
Symmetry

Figure 9 Resultant force of angle pieces
Figure 10 Horizontal against vertical forces
Figure 11 Ratio of Horizontal to vertical forces
Figure 12 shows the vertical bearing capacity of the footing,
qrd,v, against the embedment width. All angled footings
illustrate a similar increase in bearing capacity with embedded
width. The theoretical vertical bearing capacity of a flat footing
is also plotted for various values of B. The bearing capacity of
the flat footing is approximately 1.4 times the bearing capacity
of the angle piece at values of B less than 50 mm. As the
bearing pressure increases, there is a change in the rate in the
increase of the bearing capacity. This could be attributable to
the stress-state of the soil as penetration progresses; additional
numerical work is being conducted to explore this aspect.
Figure 12 Effect of cutting angle on Nγ
Figure 13 Average Face Pressure against footing width
In Figure 13, the variation of the pressure at the cutting face,
R/A, is plotted against B. This framework appears to provide
improved agreement between tests and appears to be relatively
invariant to β. The slight differences in resultant pressures
could be based on the slight differences in the relative densities
and test discrepancies.
The failure mode for the different footing varies depending
on the tapered angle, β. Figure 14 (a-d) shows the incremental
total displacements of the soil; an embedment width of B=40
mm was chosen for these comparisons. At this stage of testing,
0
250
500
750
1000
020 40 60
R (N)
Penetration (mm)
β=30
β=45
β=60
β=75
0
50
100
150
200
0 200 400 600 800 1000 1200
QH(N)
QV(N)
β=30
β=45
β=60
β=75
0.0
0.2
0.4
0.6
0.8
1.0
30 45 60 75 90
QH/QV
β
0
20
40
60
80
100
120
525 45 65 85
qrd,v (kPa)
B (mm)
Flat Footing
β=30
β=45
β=60
β=75
0
20
40
60
80
100
525 45 65 85
R/A (kPa)
B (mm)
β=30
β=45
β=60
β=75

the face pressures are similar (see Figure 13). The failure
mechanism occurs towards the excavation side, as shown in
Figure 14, as the tapered angles move into the active wedge of
soil according to the Rankine theory for soil bearing failure. For
the flatter angles, as the failure stresses in the soil develop, the
soil begins to fail towards the excavation and overburden side,
similar to a traditional flat footing.
FigureFigure 15 show how the zone of influence varies with
β. The depth and width of the failure zone is much higher for
the steeper tapered angles at B=40 mm.
(a)
(b)
(c)
(d)
Figure 14 Resultant displacement at B=40mm: (a) β=75o,
(b) β=60o, (c) β=45o, (d) β=30o
(a)
(b)
Figure 15 (a) Horizontal and (b) vertical zone of influence;
B=40mm
Influence of overburden on one side
In the field, there will always be a surcharge outside the
caisson from the overburden soil as excavation and sinking of
the shaft commences. This surcharge will increase as the
caisson sinks further into the ground. A surcharge was applied
by applying a weight on the overburden side (see Figure 3).
Surcharges of 12.5kPa, 25kPa and 50kPa were considered.
The overburden has two effects on the soil and the resulting
failure mechanism. Firstly it increases the initial stress in the
soil and secondly, it will encourage failure of the soil towards
the excavation side. Figure 16 andFigure 18 shows the effects
of the overburden against the base case of a level surface.. It
can be clearly seen that the overburden increases the bearing
capacity of the soil beneath the footing.
The increase in bearing capacity is more sensitive to the
shallower tapered angles, as the increase in resultant reaction is
larger for β equal 75° compared to the β equal 45° test. The
failure mechanism of the soil is shown in Figure 17 and Figure
19. For the higher surcharge pressures, the failure mechanism
varies and is pushed towards the excavation side and the failure
plane becomes larger.
The failure mechanism occurs towards the excavation side,
as shown in Figure 14 for flatter angles (c-d), as the tapered
angles move into the active wedge of soil according to the
Rankine theory for soil failure. For the shallower tapered
angles, as the failure stresses in the soil develop, the soil begins
to fail towards the excavation and overburden side, similar to a
traditional flat footing. This can be clearly seen in Figure 19
(g), as the soil failure mechanism develops on both sides of the
75° piece.
Figure 16 Influence of surcharge on resultant force; β=45,
(a)
(b)
Figure 17 (a) 12kPa surcharge (b) 50kPa surcharge;
B=40mm, β=45, see Figure 16
0.0
2.0
4.0
6.0
30 45 60 75
X/B
β
0.0
1.0
2.0
3.0
30 45 60 75
Y/B
β
(a)
(b)
0
100
200
300
400
500
020 40 60
R (N)
Penetration (mm)
50kPa
12kPa
No Surcharge

Figure 18 Influence of surcharge on resultant force; β=75,
(a)
(b)
(c)
(d)
(e)
(f)
(g)
Figure 19 (a+b) No surcharge (c+d) 25kPa surcharge (e+f)
50kPa surcharge (g) No surcharge, B=100mm and 35mm
penetration; β=75, see Figure 18
5 CONCLUSIONS
In this paper, a suite of tests carried out on small-scale
tapered footings in medium-dense sand has been presented. The
authors have arrived at the following conclusions arising from
this study:
a) The forces that develop at the base of a caisson are
largely dependent on the tapered angle of the footing.
While steeper footings have a much reduced vertical
resistance, at the same penetration, this is offset by an
increase in the horizontal reaction. The bearing pressure
on the face of the footing appears to be invariant to the
cutting angle of the test piece and varies linearly with the
embedment width of the piece.
b) Applying an overburden pressure on one side of the
footing best models conditions in the field. The
overburden has a greater influence on flatter cutting
angles since failure wedges are developed on both sides
of the cutter unlike steep angles where the failure wedge
develops on the excavation side only. The overburden
forces failure to occur on the excavation side, thereby
increasing the bearing capacity.
c) Output from the PIV analysis of the present tests can be
used to guide excavation on site in order to induce failure
of the soil beneath the caisson.
d) Further work is being carried out at University of Oxford
to extend these results and confirm the applicability to
conditions in the field. In particular, further comparisons
with full-scale measurements will be carried out to
ensure that the proposed methods are reliable and valid.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial support
provided by Ward and Burke Construction Ltd.
REFERENCES
[1] M. J. Cassidy and G. T. Houlsby, “Vertical bearing capacity factors for
conical footings on sand,” Géotechnique, no. 52, pp. 687–692, 2002.
[2] G. T. Houlsby and C. M. Martin, “Undrained bearing capacity factors for
conical footings on clay,” Geotechnique, vol. 53, no. 5, pp. 513–520,
2003.
[3] M. J. Tomlinson, Foundation Design and Construction, 7 edition.
Harlow, England; New York: Prentice Hall, 2001.
[4] E. Nonveiller, “Open caissons for deep foundations,” J. Geotech. Eng.,
vol. 113, no. 5, pp. 424–439, 1987.
[5] N. B. Solov’ev, “Use of limiting-equilibrium theory to determine the
bearing capacity of soil beneath the blades of caissons,” Soil Mech.
Found. Eng., vol. 45, no. 2, pp. 39–45, 2008.
[6] F. Y. Yan, Y. C. Guo, and S. Q. Liu, “The Bearing Capacity Analyses of
Soil beneath the Blade of Circular Cassion,” in Advanced Materials
Research, 2011, vol. 250, pp. 1794–1797.
[7] C. M. Martin, “Physical and numerical modelling of offshore foundations
under combined loads,” University of Oxford, 1994.
[8] S. A. Stanier, J. Blaber, W. A. Take, and D. White, “Improved image-
based deformation measurement for geotechnical applications,” Can.
Geotech. J., no. ja, 2015.
[9] M. D. Bolton, “The strength and dilatancy of sands,” Geotechnique, vol.
36, no. 1, pp. 65–78, 1986.
[10] J. Hansen, “A general formula for bearing capacity, Danish Geotechnical
Institute Bulletin,” No.(11), 1961.
[11] G. G. Meyerhof, “Some recent research on the bearing capacity of
foundations,” Can. Geotech. J., vol. 1, no. 1, pp. 16–26, 1963.
(c)
(e) (d)
(f)
(a) (b)
(g)
0
400
800
1200
1600
2000
020 40 60 80 100
R (N)
B (mm)
50kPa
25kPa
No Surcharge