Você está na página 1de 9

Small-Scale Testing of Laterally Loaded Bucket Foundations in Dense Sand

Lars Bo Ibsen, lbi@civil.aau.dk; Aligi Foglia, afo@civil.aau.dk


Aalborg University, Department of Civil Engineering. Sohngaardsholmsvej 57, 9000 Aalborg, Denmark.
Abstract
Monopod bucket foundations promise to become a reliable and cost-effective option for offshore wind turbine. In this paper, six small-scale tests of a steel bucket foundation subjected to
quasi-static lateral load, are presented. When conducting small-scale experiments on soil, scale
effects can considerably affect the tests outcome. To overcome this issue, a novel testing system based on the application of suction between the soil surface and a membrane, is employed.
By means of the suction the effective stress of the soil is increased. The tests are conducted
at stress levels of 0 kPa, 10 kPa, 15 kPa, 20 kPa, and 30 kPa respectively. The test results
successfully prove reliability and consistency of the method. The comparison between the tests
conducted at stress level of 0 kPa, and the tests with stress level increased, shows remarkable differences. In particular, the moment rotation curves of tests with pressure applied, show
more similar trend to large-scale tests. The relationship between scaled overturning moment,
M/( d4 ), and rotation, is well represented by a power law. The exponent of the power law is
consistent among all tests carried out with stress level increased. Besides, attention is given to
the instantaneous centre of rotation distribution. Further cyclic loading studies may be carried
out adopting this testing system to investigate the accumulated rotation of bucket and monopile
foundations.

1 Introduction

paper will be possibly adopted to conduct research on long term cyclic loading behaviour of
pile and bucket foundations. It is well known that
Monopod bucket foundations are likely to bewhen conducting small-scale tests, the scale efcome a reliable and cost-effective option for
fect greatly threatens the experiment outcome.
offshore wind turbine foundations.
As reported in LeBlanc (2009) such foundations can
be installed without using heavy vessels, and
thereby, considerably lowering the cost of installation. Bucket foundations have been studied over the last decade especially at Aalborg
University, Oxford University and University of
Western Australia. While in the early years
the ultimate capacity of the structure was addressed, more recently attention focused on lateral cyclic loading (Zhu et al., 2010). Wind
turbine foundations are characterized by strong
horizontal forces caused by wind and waves,
and a rather low vertical load. An offshore wind
turbine mounted on a bucket foundation support
is sketched in Figure 1 where h is the height of
the wave loading, d the skirt length and D the
the diameter of the bucket. Waves and winds
are naturally cyclic, consequently, a wind turbine is exposed to millions of cycles over its lifetime. For this reason investigating the wind turbine foundations behaviour under cyclic loading,
shall be considered as necessary, for a correct
design framework. The new method of static Figure 1: Wind turbine mounted on a monotesting of bucket foundations presented in this pod bucket foundation. After LeBlanc (2009).
1

Lars Bo Ibsen, Aligi Foglia


When a soil medium is object of study, the smallscale effective stress condition is far off from
that of the real case. This, inevitably leads to different soil parameters between large and smallscale. The friction angle is profoundly affected
by the effective stress only for very low stress
states. Furthermore, it presents rather uncertain values at those stresses. When the effective stress increases, the friction angle variation and fluctuation become less evident, with
an overall benefit to the scale effect. In a bid to
reduce the scale effect a pressure tank or a centrifuge can be utilized. Differently, the tests presented in this paper were performed by creating
a negative pressure between the sand surface
and a membrane. As a result of that the effective stress of the soil was increased.

Figure 2: 300 mm diameter, 300 mm skirt


bucket foundation.

Table 1: Test program.


Test No.

Throughout this paper, results obtained by conducting monotonic tests on a bucket foundation
in dense sand at different soil effective stress
are presented and analysed. The data elaboration is mainly aimed at investigating the following features:

1
2
3
4
5
6

Overburden Pressure
[kPa]
0
0
15
30
20
10

Consistency of the testing method.


Effect of the different effective stress of the 2.1 Sand Box
soil on the dimensionless moment-rotation
The test rig consists of a sand box and a loadcurve.
ing frame mounted on it, cf. Figure 3 and Fig Ultimate moment of the bucket foundation ure 5. The sand box is made of steel with an
under different effective stress of the soil. inner width of 1600 mm x 1600 mm, and an inner depth of 1150 mm. The box is equipped
Displacements evaluation of the bucket with a drainage system on its bottom. The
drainage system consists of perforated pipes,
under different overburden pressure.
100 mm of draining material (gravel) and a sheet
These points are considered significant for fur- of geotextile. The perforated pipes are equally
ther cyclic loading studies.
distributed in order to provide a homogeneous
drainage condition to the entire area. The geo2 Laboratory Tests
textile sheet needs to keep the drainage maThe test program consisted of 6 monotonic hor- terial clear of sand grains, which could potenizontal loading tests of the bucket foundation at tially occlude the drainage system. The water
five different effective stress level. All experi- level inside the box is regulated by means of a
ments took place at the Geotechnical Engineer- valve on the inlet pipe, and controlled through a
ing Laboratory at Aalborg University. Accord- piezometer directly connected to the bottom of
ing to the scaling relationship proposed in Foglia the box. The water is supplied by a water tank
(2011), the tests features were chosen in order placed higher than the soil surface. On the horto model the behaviour of a real bucket foun- izontal part of the loading frame are mounted
dation with d = 16 m, D = 16 m and h = 32 the electric motor for the installation and the hym. The arm of the horizontal load , h, refers to draulic jack for the cone penetration tests. On
as the waves action. The foundation adopted to the vertical part, at a height of 600 mm from the
carry out all tests was a steel bucket of diame- soil surface, the electric motor for the horizontal
ter, D = 300 mm, with a skirt of d = 300 mm, load is placed.
cf. Figure 2. The arm of the horizontal force h 2.2 Preparation of the Soil
was for every test 600 mm from the soil surface.
Tests were performed at different soil effective Through the drainage system the water level
pressure. The test program is listed in Table 1. was raised on purpose of loosening up the sand.
2

Small-Scale Testing of Laterally Loaded Bucket Foundations in Dense Sand

Figure 3: Test rig.

Figure 4: Bucket foundation during loading


phase without overburden pressure.

Figure 5: Test rig, dimensions in mm.


A wooden frame was placed on the sand layer
to retain the water due to the level raise. Once
the soil had reached the complete saturation,
the compaction could commence. In order to
carry out a uniform compaction, a wooden plate
with 64 holes was placed on the wooden frame.

Thereby, the entire area was equally divided.


The compaction was performed with a rod vibrator by penetrating every hole in a systematic manner. The compaction procedure was
repeated two times before the first test and one
time before every subsequent test. At soil vibra-

Lars Bo Ibsen, Aligi Foglia


tion concluded the alignment operation could be
carried out and the water level was set to the final soil surface. 1-2 cm of water was poured
from above to ensure the complete saturation of
the sand.

0
50
100

Depth, mm

150

2.3 Parameters of the Soil


The medium used was for all tests was Aalborg University Sand No 0. The sand properties are listed in Table 2. For each test, after
soil compaction, three cone penetration tests
(CPTs) were conducted to verify the homogeneity of the soil and to calculate the soil parameters. Each CPT series was performed in the
middle of the box and 400 mm right and left from
the middle. The penetration velocity of the CPT
was set to 5 mm/s. Typical results from the CPT
are shown in Figure 6, where cone penetration
resistance-depth curves for the three CPTs performed prior to test no. 2 are compared. From
these curves it can be noted an overall homogeneous compactness of the sand. A mean
value comparison of the CPTs performed prior
to the six tests, is illustrated in Figure 7. From
Figure 7 it is seen that fairly similar values are
reached prior to each test. In order to obtain
the density index, and thereof, the other parameters of the soil from solely CPT data some calculations were employed. The whole method
was based on the relationship between density index and secant friction angle proposed by
Schmertmann (1978). Such relationship was
characterized for Aalborg University sand No. 0
on the base of a series of triaxial tests by Ibsen et al. (2009). The soil parameters, determined by conducting CPTs and successive calculation, are presented in Table 3 where tr and
tr are friction angle and dilation angle, respectively. Note that the subscripts tr underlines the
triaxial test-based derivation of the parameters.
Tests in normal overburden pressure condition
(tests no. 1 and 2) were conducted in saturated
sand. When applying the increase in effective
stress (tests no. 3, 4, 5 and 6) the soil was
brought to dry condition. For all tests the effective unit weight of the soil was calculated and
used for interpreting results.

200
250
300
350
400
450

200

400
600
Tip Resistance, N

800

1000

Figure 6: Series of CPTs conducted prior to


Test 2.

0
50
100

Depth, mm

150
200
250
300
350
400
450

100

200

300

400

500

600

700

800

900

Tip Resistance, N

Figure 7: Mean result from CPTs performed


prior to each test.

Table 3: Sand parameters for the six tests.

Test 1
Test 2
.
Test 3
Test 4
Test 5
Test 6

tr
[ ]
52.31
52.28
51.92
51.87
51.63
51.25

tr
[ ]
17.72
17.67
17.21
16.13
16.85
16.37

ID
[-]
0.77
0.76
0.74
0.72
0.72
0.70

[kN/m3 ]
10.64
10.54
10.34
10.14
10.14
9.96

Table 2: Index properties of Aalborg University


sand No. 0. (Ibsen, 1999).
Property
d50 = 50% - quantile
Cu = d60 /d10
Specific grain density ds
Maximum void ratio emax
Minimum void ratio emin

Value
0.14
1.78
2.64
0.86
0.55

2.4 Overburden Pressure

Unit
[mm]
[-]
[g/cm3 ]
[-]
[-]

To increase the effective stress of the soil suction was applied between the soil surface and
a membrane. The suction system consisted of
a membrane with four suction spots, each of
them connected to a vacuum pump through a
4

Small-Scale Testing of Laterally Loaded Bucket Foundations in Dense Sand


suction hose. The membrane was placed beneath the soil surface and fixed to the sand box
edges by means of a steel frame and clamps.
Between membrane and steel frame a washer
was inserted in order to avoid leaks of air. Furthermore, between sand and membrane a filter
was put to prevent any sucking of sand grains.
An additional spot was placed on the membrane
and connected to the data acquisition device
by means of a pressure transducer of the type
HPM P6A. The pressure was regulated with a
control valve and kept steady during each test.
The four vacuum pumps applied the suction
through a sealed glassed cylinder to collect the
water that unavoidably came out from the box
together with air. As a result of that, for all tests
with suction applied, i.e. with increased effective
stress, the soil was not in saturated condition. A
picture of the suction system is illustrated in Figure 8.

The load cell, type HBM U2A 100 kg, was


mounted on the vertical bar, and connected in
series to the horizontal loading device with a
steel cable. When the effective stress state was
increased a load cell of the type HBM U2A 10
kN was utilized. For tests no. 1 and no. 2
two force transducers, type HBM U2A 10 kN,
were installed on the caisson to measure the
vertical forces, created in turn, by applying the
horizontal load. Due to the rather high overturning moment acting on the bucket when applying the overburden pressure, the latter transducers were removed in all the other tests. The
LVDTs were placed on two perpendicular plates
mounted on the caisson. These plates formed
a rigid body with the bucket. Two LVDTs, type
HBM W20TK, were vertical, and one type HBM
WA/50, was horizontal. The horizontal and vertical displacement of the caisson were deduced
by computing the measurements of the LVDTs
on the perpendicular plates. The displacement
transducer, was placed at the same height of
the horizontal loading device in order to measure the displacement of the force application
point. A photograph of the equipment is shown
in Figure 4. A PC-based data acquisition, HBM
SPIDER 8, was used to transfer data from the
measurement devices to the computer. Vertical displacement w, horizontal displacement u,
rotation , and loads applied to the foundation
were assessed according to the sign conventions proposed by Butterfield et al. (1997), which
is shown in Figure 9.

2.5 Measuring System


The bucket foundation was instrumented with
one force transducer (or load cell), three linear
variable differential transducers (LVDTs), and
one displacement transducer.

2.6 Testing Procedure


By using an electric motor the bucket was driven
in the middle of the sand box, as illustrated in
Figure 10. The installation velocity was kept extremely low (approximately 80 mm/h) in order
to avoid pore pressure building up. A load cell,
type HBM U2B 50 kN, was mounted between
the bucket and the installation rig to measure
the installation force. The installation procedure
was a delicate operation, the correct horizontal
alignment of the bucket was constantly kept under control by means of a spirit level. Once the
bucket was installed, a vertical steel bar was
bolted on the bucket lid and a steel cable was
horizontally connected between the bar and an
electric motor placed on the loading frame, cf.
Figure 5. The loading phase consisted of pulling
the bucket foundation with the electric motor at
a constant velocity of approximately 0.01 mm/s.
No vertical load exceeding the foundation self
weight was added. The tests with suction applied were subjected to an unloading-reloading

Figure 8: Suction equipment.

Figure 9: Sign convention for loads and displacements. After Butterfield et al. (1997).
5

Lars Bo Ibsen, Aligi Foglia


phase.

3000
0 kPa
0 kPa
10 kPa
15 kPa
20 kPa
30 kPa

2500

Moment, Nm

2000

1500

1000

500

0.5

1.5

2
2.5
Rotation,

3.5

4.5

Figure 11: Raw rotation-moment data for all


tests.

3.1 Force-Displacement Curves

moment and rotation. From Figure 11, it can be


observed that the moment-rotation gradient diminishes significantly after 2 of rotation for test
no. 3, 4, 5, and 6, and after 1 of rotation for
Test no. 1. In tests no. 3, 4, and 5, a series
of sudden load-loss immediately followed by a
load-recover, can be noted starting from 1.5 of
rotation. This phenomenon might be attributed
to the failure mode of the foundation which involves more and more soil during its development. Indeed when this phase ceases the load
increases at a lower rate than before. Such phenomena does not occur when the overburden
pressure is not increased and when the suction
is 10 kPa. Figure 11 illustrates also how, the applied moment, evidently depends on the overburden pressure. For instance, the bucket foundation rotation equal to 1 is obtained with a moment of approximately 130 Nm for test no. 1 and
2. With respect to the latter value, the moment
increases with a factor of approximately 6.4, 8.2,
9.7, and 12.7 for test no. 6, 3, 5, and 4 respectively. The residual plastic deformation seems
to be independent of the overburden pressure
value. After the unloading phase the plastic
deformation falls within the range 80-90 % of
the deformation reached at the end of the loading phase. Experiments conducted by Larsen
(2008), with normal overburden pressure conditions, revealed the same residual plastic deformation.

The raw data of the moment, as collected with


rotation, for all tests, are presented in Figure 11.
In the figure legend, the overburden pressure at
which each test was carried out, is stated. In every test the soil was brought to failure. As it was
expected, when the effective stress is increased
the failure occurs at higher value of both applied

In Figure 12 raw data of the bucket foundation


rotation against the dimensionless overturning
moment, M/( d4 ), are plotted on double logarithmic scale. The records of tests with increased effective stress show a distinct similar
slope which contrasts with that of test no. 1 and
2.

Figure 10: Installation rig.


3 Results
Part of the following results will be presented
by employing the dimensionless moment M/(
d4 ) and the dimensionless horizontal force
H/( d3 ). The choice of these normalized
values relies on the results found in Foglia
(2011) regarding the dimensionless relationships moment-rotation, and horizontal loaddisplacement. Hereby, attention is mainly given
to the relationship between dimensionless moment and rotation, which is more relevant for
bucket foundation fatigue design.
Test no. 1 and test no. 2 were performed to have
a double check of the case without overburden
pressure increased. Indeed, they give basically
the same response in terms of loading as well
as in terms of deformation. Although test no. 3
was too early interrupted, its result is considered
significant and it is not dismissed from the data
analysis.

Small-Scale Testing of Laterally Loaded Bucket Foundations in Dense Sand

Dimensionless moment M/( d )

10

10

10

10

10

Table 4: Power law coefficients for every test.

Test 1
Test 2
.
Test 6
Test 3
Test 4
Test 5

0 kPa
0 kPa
10 kPa
15 kPa
20 kPa
30 kPa

10
Rotation, rad

10

10

[-]
0.24
0.30
0.49
0.49
0.49
0.48

Table 5: Ultimate moment capacity.

Figure 12: Raw dimensionless momentrotation data for all the tests on a logarithmic
plane.

Test no.

.
10
9
4

C
[-]
4.24
3.79
51.06
64.39
76.14
93.39

10

Dimensionless moment, M/( d )

Suction
[kPa]
0
0
10
15
20
30

8
7

0 kPa
10 kPa
15 kPa
20 kPa
30 kPa

1
2
6
3
4
5

Ultimate Moment
Capacity
[Nm]
213
234
1008
1444
1674
2401

6
5

The ultimate moment capacity of the bucket


foundation, under different overburden pressure
3
induced, is taken as the moment measured at 3
2
of rotation. For test no. 3 a spline interpolating
1
function is adopted to obtain the moment at 3
0
0
0.002
0.004
0.006
0.008
0.01
of bucket rotation. Test no. 1 and test no. 2 do
Rotation, rad
not reach 3 of rotation, their moment ceases
to increase after 1 of rotation. Therefore, the
Figure 13: Dimensionless moment-rotation
ultimate moment capacity of these tests was
curves, each of them fitted with a power law.
taken as the maximum moment reached during the tests. Note that the ultimate moment
Focusing now on the first 0.5 of rotation, i.e. for test no.1 and no.2 does not precisely coinsignificant for fatigue limit states, a power law by cide. The two tests were conducted at different
velocities, and the strain rate dependency of the
the analytical form


sand caused the discrepancy. Table 5 lists the
M
C
(1) ultimate moment for all tests.
=
d4
3.2 Displacements Analysis
can be adopted to represent the records (Foglia
, 2011). In Figure 13 the rotation as a function of By computing the LVDTs measurements, the inthe dimensionless moment is given only for the stantaneous centre of rotation position during
first 0.5 of rotation. In order to avoid data su- the tests could be evaluated. In Figure 14 the
perposition test no. 2 is not displayed in Figure istantaneous centre of rotation for test no. 1 are
13. Although the power law shows slightly lower illustrated. A typical instantaneous centres of
values of M/( d4 ) for rotation < 0.05 , it rotation distribution for tests with effective stress
is considered well representative of the normal- increased is illustrated in Figure 15, where the
ized moment-rotation trends. In Table 4 a com- loading phase of test no. 5 is considered. In
parison of the power law parameters is given. both figures the arrow above the graph, refers
The exponent of the tests with suction applied is to as the loading direction. The majority of the
consistent and differs from the exponent of tests instantaneous centres of rotation was found beno. 1 and no. 2. This suggests that might be low the middle of the skirt, toward the direction
of the loading. In spite this general similarity, a
a constant of the soil in real-scale.
4

Lars Bo Ibsen, Aligi Foglia


Loading Direction

0.4

15

Horizontal displacement u, mm

Normalized vertical position w/d

0.2
0
0.2
0.4
0.6

10

0 kPa
0 kPa
10 kPa
15 kPa
20 kPa
30 kPa

0.8
1
0.5

0.4

0.3

0.2 0.1
0
0.1
0.2
0.3
Normalized horizontal position u/D

0.4

0.5

Rotation,

Figure 14: Instantaneous centres of rotation


for test no. 1.

Figure 16: Horizontal displacement against rotation.

Loading Direction

0.4

300

250

0
200
Moment, kNm

Normalized vertical position w/d

0.2

0.2
0.4

150

100

0.6
50

0.8
0

1
0.5

0.4

0.3

0.2 0.1
0
0.1
0.2
0.3
Normalized horizontal position u/D

0.4

0.01

0.5

0.02

0.03
Rotation, rad

0.04

0.05

0.06

Figure 17: Raw rotation-moment data for the


large-scale test.

Figure 15: Instantaneous centres of rotation


for test no. 5.

The foundation adopted was a 2 m diameter, 2


m skirt length and 12 mm thickness steel bucket.
In Larsen (2008) the testing procedures such as
installation and loading phase are described. As
shown in Figure 17 the moment-rotation curve
of the large scale test appears to follow the trend
of small-scale tests with overburden pressure.
In fact the soil failure of the large scale test occurs after 2 of rotation. This reveal that the
testing method is capable of better representing
the actual soil-structure interaction. Although,
scaling laws are now needed to scale up the
physical model results to the prototype-scale.

little difference between the two plots can be


noted. The centres of rotation for test no. 1
are concentrated lower, and more toward the
loading direction. This indicates that, the failure mode of the two cases is, to some extend,
different. This is plausible, and can be attributed
to the different stress states distribution in case
of overburden pressure. More support to the latter observation is given by Figure 16 where the
horizontal displacement, u, is plotted as a function of the rotation, , for each test. Every record
exhibits a linear trend. A gap between tests conducted in normal stress conditions, and those at
increased effective stress, is evident.

5 Conclusions

Bucket foundations may become a costeffective option for offshore wind turbine supWithin a research led by Aalborg University port. Over the last decade small-scale experiand MBD-Offshore-power A/S seven large- ments and field trials have proven the reliability
scale tests of bucket foundation were conducted of this foundation. The loads acting on an offat the Frederikshavn test cite facility.
shore structure are mainly cyclic and the current
4 Comparison with large-scale tests

Small-Scale Testing of Laterally Loaded Bucket Foundations in Dense Sand


aim of researchers is to address the cyclic loading response of such structures. This study investigates the capability, for a new small-scale
testing system, of representing the real-scale
behaviour. Such new system may be used for
a long-term testing program about bucket foundation under lateral cyclic loading.

mation for future cyclic loading researches was


obtained.
References
Butterfield R, Houlsby GT, Gottardi G
(1997). Standardised sign conventions
and notation for generally loaded foundations.Geotechnique 47(4), 1051-1054.

In a bid to minimize the scale effect, which is unavoidably related to small-scale physical mod- Foglia A (2011). Behaviour of Bucket Foundaels, a new testing method is set out in this pation under Horizontal Load in Dense Sand,
per. The novel method consists of applying sucMSc Thesis, Department of Civil Engineering,
tion between a membrane and the soil surface
Aalborg University, Denmark.
in order to increase the soil effective stress and
Larsen KA (2008). Static Behaviour of Bucket
to reduce thereby the scale effect. Six monoFoundations, PhD Thesis, Department of Civil
tonic tests were conducted at different overburEngineering, Aalborg University, Denmark.
den pressure applied. The results clearly shown
how the moment applied on the bucket depends LeBlanc C (2009). Design of Offshore Wind Turon the stress state of the soil.
bine Support Structures: Selected Topics in
the Field of Geotechnical Engineering, PhD
The consistency of the method was achieved by
Thesis, Department of Civil Engineering, Aalshowing that, stiffer responses of the soil correborg University, Denmark.
spond to higher overburden pressure applied by
means of the suction. The plastic response of Ibsen LB (1999). The mechanism controlling
the soil was not affected by the different overstatic liquefaction and cyclic strength of sand,
burden pressure applied. The overturning moProceedings of the International Workshop on
ment capacity was calculated for every test.
Physics and Mechanics of Soil Liquefaction,
This will be useful to calibrate the load magA. A. Balkema.
nitude of cyclic loading tests. Due to the disIbsen LB, Larsen KA, Nielsen SA (2005). Intinct boundary conditions, tests in normal connovative Foundation Solutions, The Novel
ditions showed a slightly different failure mode
Bucket Foundation. Copenhagen Offshore
than tests with overburden pressure increased.
Wind Conference and Exhibition, 26-28 OcThis conclusion was reached by plotting rotatober 2005, Denmark.
tion against horizontal displacement curves and
by analysing the instantaneous centres of rota- Ibsen LB, Hanson M, Hjort T, Thaarup M (2009).
tion locations of different tests. Further numeriMC-ParameterCalibration for Baskarp Sand
cal simulations should take heed of the different
No. 15, DCE Technical Report No.62 .Departfailure mode presented by the tests with effecment of Civil Engineering, Aalborg University.
tive stress increased.
Schmertmann, JH (1978). Guidelines for cone
The method capability of representing realpenetration test, performance and design, US
scale behaviour, was proved by making a comFederal Highway Administration, Washington
parison between large and small-scale test. The
DC, Report, FHWA-TS-78-209, 145.
moment-rotation curve of the large-scale test
Zhu B, Houlsby GT, Byrne BW (2010). Cyclic
was found to some extend similar to that of the
moment loading of suction caisson in sand,
new small-scale experiment.
Proceedings of the international symposium
The overall method was found to be consistent
on Frontiers in offshore geotechnics II, Perth
and gave the expected results. Interesting inforAustralia, November 2010.

Você também pode gostar