Geotechnical Engineering Proceedings of the Institution of Civil Engineers

Volume 166 Issue GE2 Geotechnical Engineering 166 April 2013 Issue GE2
Pages 159–169 http://dx.doi.org/10.1680/geng.11.00015
Dynamics of offshore wind turbines Paper 1100015
supported on two foundations Received 11/02/2011 Accepted 24/08/2011
Published online 09/08/2012
Bhattacharya, Cox, Lombardi and Muir Wood Keywords: dynamics/foundations/offshore engineering

ICE Publishing: All rights reserved

Dynamics of offshore wind turbines

supported on two foundations
j
1 Subhamoy Bhattacharya PhD(Cantab) j
3 Domenico Lombardi MSc
Senior Lecturer in Dynamics, University of Bristol, Bristol, UK Research Postgraduate Student, University of Bristol, Bristol, UK
j
2 James A. Cox MEng j
4 David Muir Wood MA, PhD(Cantab), FICE, FREng
Research Postgraduate Student, University of Bristol, Bristol, UK Emeritus Professor of Civil Engineering, University of
Bristol, Bristol, UK

j
1 j
2 j
3 j
4

Offshore wind turbines are currently considered as a reliable source of renewable energy in the UK. These structures,
owing to their slender nature, are dynamically sensitive at low frequencies, the first modal frequency of the system
(less than 1 Hz) being very close to that of the excitation frequencies. The majority of operational offshore wind
turbines situated in UK waters are founded on monopiles in water depths up to 30 m. For future development
rounds where water depths are up to 70 m, alternative foundation arrangements are needed. To date there have
been no long-term observations of the performance of these relatively novel structures. Monitoring of a limited
number of offshore wind turbines has indicated a departure of the system dynamics from the design requirements.
This paper summarises the results from a series of 1:100 scale tests of a V120 Vestas turbine supported on two types
of foundation: monopiles and tetrapod suction caissons. The test bed used consisted of kaolin clay and sand. Up to
1.25 million loading cycles were applied to the scaled model, and the dynamic properties of the system were
monitored. The results provide an insight into the long-term performance. Some interesting dynamic soil–structure
interaction issues are identified and discussed.

Notation 1. Introduction
D pile diameter With the current demand for clean ‘green energy’ there are very
Dsp median sediment grain diameter strong economic incentives to construct offshore wind farms. The
E Young’s modulus of pile rate of expansion of offshore wind energy is soon predicted to
EI pile bending stiffness outstrip even levels seen during the heyday of the offshore oil
emax maximum void ratio and gas industry, and as a result substantial effort has been put
emin minimum void ratio into research and development. The UK offshore wind sector has
ff forcing frequency been steadily expanding under the direction of the government.
fn system frequency So far, planning has been conducted in so-called ‘rounds’. Each
G shear modulus of soil at reference depth of these rounds has allowed applications to be submitted for a
Gs specific gravity of solids number of wind farms in the coastal waters surrounding the UK;
Kh horizontal permeability of soil applications for the third round have recently been completed.
K1 lateral foundation stiffness
L length of tower (a) Round 1. The first round of offshore development saw the
M mass of rotor and blades submission of 12 proposals for offshore wind schemes, with
N number of cycles output expected to be in the region of 1.4 GW. For these
P total lateral load on wind turbine schemes the water depths are never greater than 20 m, and no
tw thickness of pile site is further than 13 km from the shore. All sites, with one
y lateral pile head deflection exception, make use of monopile foundations.
yc point of application of total load (b) Round 2. The second round contains some 14 sites, plus their
y yield stress of pile material associated expansions (recent additions). Output is expected
łcrit critical angle of friction to be 7.3 GW. Round 2 sites are situated in more adverse
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
159
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

water conditions with a maximum depth of 37 m, with the scale model wind turbine supported on two types of
furthest site being 40 km from shore. As with the round 1 foundation: (a) monopile; (b) tetrapod suction caisson. The
turbines, the water is sufficiently shallow that most of the tests were carried out in BLADE (Bristol Laboratory for
foundations have been designed as monopiles. Advanced Dynamics Engineering) where dynamic soil–
(c) Round 3. The third round represents a large expansion in structure interaction characteristics were explored.
Britain’s renewable energy sector, with a total output
estimated to be in the region of 21 GW. Water depths are 2. Dynamic considerations for designing
expected to range up to 70 m, and some sites are located up offshore wind turbines
to 210 km offshore. For such developments in deep water Offshore wind turbines are particularly sensitive to dynamic
alternative foundation arrangements are required, as monopile loading conditions because of the combination of the slender
structures start to become uneconomical, owing to the size of structural nature of the turbine and the wide range of cyclic loads
pile required. As part of the Carbon Trust’s Offshore Wind to which the turbine is subjected. Figure 1 shows a summary of
Accelerator various alternative arrangements have been the typical forcing frequencies applied to a Vestas V120 4.5 MW
proposed, such as four-legged jackets, tripods, gravity wind turbine system: the 1P frequency signifies the rotational
foundations, floating platforms and suction caissons. As most frequency of the turbine, and the 3P frequency signifies the
round 3 sites are still in the design stages, the foundation blade-passing frequency. The 3P frequency results from shadow-
arrangements are unknown; however, it is expected that ing effects of the blade on the tower caused by a drop in the
monopiles will no longer be sufficient. upstream wind velocity in the vicinity of the tower as each of
the three blades passes in front of the tower. An example of the
1.1 Aim and scope of the paper dynamic wind loading is also shown in Figure 1, utilising the
With each successive round of offshore wind farm expansion, the Frøya wind spectrum (Andersen and Løvseth, 2006). In addition,
trend has been for developments to be in regions further offshore, the dynamic wave loads are modelled using the Pierson and
in areas where the water depth is greater. This presents some unique Moskowitz (1964) spectrum, which is considered appropriate for
engineering challenges, as the dynamic behaviour of these struc- UK North Sea locations. In order for turbines to remain
tures will be heavily dependent upon foundation type selection and unconditionally stable, and avoid unplanned resonance effects,
soil conditions at the site. The aim of the paper is therefore to they have to be designed in such a way that the magnitude of the
dynamic load applied to them can be minimised.
(a) highlight the design issues related to dynamic soil–structure
interaction of offshore wind turbines To do this, one of three design methods can be used to place the first
(b) summarise the preliminary test results obtained from a 1:100 modal (natural) frequency of the system (structure–foundation

Wind Wave (Pierson and

(Frøya model) Moskowitz, 1964)
(peak frequency ⫽ 0·1 Hz)
Power per unit frequency

1P 3P

0 0·05 0·10 0·15 0·20 0·25 0·30 0·35 0·40 0·45 0·50 0·55 0·60 0·65 0·70 0·75 0·80 0·85 0·90 0·95 1·00
Frequency: Hz

Soft–soft Soft–stiff Stiff–stiff

Figure 1. Forcing frequencies plotted against power spectral

density for Vestas V120 4.5 MW wind turbine

Delivered by ICEVirtualLibrary.com to:

IP: 131.227.131.74
160
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

combination) in either the ‘soft–soft’, ‘soft–stiff’ or ‘stiff–stiff’ the dynamic soil–structure interactions need to be quantified; at
regions, as shown in Figure 1. Typically, most turbines supported on present these issues are not well understood.
monopiles are designed to be ‘soft–stiff’, as the support structure
(foundation and transition piece) can have a greater flexibility, 3. Long-term performance of monopile-
reducing the costs associated with the additional material required. supported wind turbines
The frequencies of particular interest are those in the region from In order to understand the effects of long-term cyclic loading on
0.01 Hz to 1 Hz, as this is the region within which the first modal a monopile-supported wind turbine, a series of 1:100 scale model
frequency of a wind turbine will lie. As discussed, within this tests were conducted. Scaling laws were derived so that the test
critical band various exciting forces exist, which makes the system results can be plotted using non-dimensional groups, which can
dynamically sensitive. Table 1 summarises the first natural frequen- later be scaled up to a full-size prototype system. These scaling
cies of five offshore wind turbines, either operational or currently laws are listed in Table 2; details of their derivation can be found
being built. If the values of the natural frequency of the wind turbine in Bhattacharya et al. (2011). Despite the scientific rigour, the
system are compared with Figure 1, it becomes very apparent that experimental investigation had some limitations: only a homo-
these structures are potentially very dynamically sensitive. geneous isotropic medium was considered at this stage; the soil
stiffness was low, and was based on maintaining a representative
Monitoring of a limited number of installed wind turbines has strain field alone; the dynamic load case considered only mean
indicated a departure of the overall system dynamics from the operational conditions; and damping was not considered in detail.
design assumptions. For example, at Lely (Netherlands), the
natural frequency following 6 months of operation was 0.63 Hz, Figure 2(a) shows a schematic diagram of the wind turbine
compared with a design frequency of 0.41 Hz (Kühn, 2000; supported on a monopile, and Table 3 details the wind turbine
Zaaijer, 2006). From Figure 1, it is clear that any shift of the model parameters. To provide an initial reference point, the fixed
natural frequency of the system may coincide with the forcing base frequency of the turbine was assessed; this was accom-
frequencies, causing unplanned resonance effects and leading to plished by clamping the bottom of the tower to a rigid bench, and
rapid wear and tear of the onboard machinery and enhanced conducting a snap-back test. The fixed base frequency was
deflections. In extreme cases, this may lead to failure. So far, no measured to be around 10.27 Hz. To test whether the natural
explanation has been offered for such a change in the structural frequency changed over time, the turbine system was carefully
natural frequency; however, theories relating to cyclic strain installed in the soil test bed (detailed below), and the dynamic
accumulation (sandy soil) and pore water accumulation (clay) properties of the model were then assessed by a snap-back test.
have been proposed. As turbines increase in size and power, so The system was then subjected to a period of representative
do the associated loads. In order to maintain serviceability limits, dynamic loading: the load was applied using an actuator (replicat-

Project and reference Soil profile First natural Remarks

frequency of
system: Hz

North Hoyle: Irish sea – Upper seabed layers comprise variations 0.345 Frequency prediction based on Adhikari and
Liverpool Bay of sand and clay layers. Below is weak Bhattacharya (2012). Details of wind farm
(Vestas V80 – 2.0 MW) rock (mudstone or sandstone). can be found in Carter (2007)
Lely (A2) – inland sea, Pile passes through soft layer to stiffer 0.634 Frequency measured and reported in Zaaijer
Ijsselmeer (the sandy layer. (2006)
Netherlands)
(Nedwind 500 kW/41)
Irene Vorrink – inland Pile passes through soft layer to stiffer 0.546 Frequency measured and reported in Zaaijer
sea, Ijsselmeer (the sandy layer. (2006)
Netherlands)
(Nordtank NTK600/43)
Kentish Flat Soil at site is layered with seabed sand 0.38 Frequency prediction based on Adhikari and
(Vestas V90 – 3.0 MW) underlain by soft to firm clay on top of Bhattacharya (2012)
London Clay formation.
Frederikshavn Suction caisson founded in a fine sand 0.30 Frequency extracted from Houlsby et al.
(Vestas V90 – 3.0 MW) bed. (2005)

Table 1. First natural frequency of some operational

monopile-supported wind turbines

Delivered by ICEVirtualLibrary.com to:

IP: 131.227.131.74
161
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

Name of non- Physical meaning Remarks

dimensional group

P Average strain field in the soil around the pile and cyclic stress ratio Similar strain field, which will
GD2 (CSR) in the shear zone (i.e. ratio of shear stress to vertical effective control the degradation of soil
stress at a particular depth) stiffness
kh Rate of application of lateral loading to the model Modelling consolidation and the
ff D dissipation of pore water pressure
ff Relative spacing of forcing frequencies and natural frequencies System dynamics
fn
Py Bending strain in pile Non-linearity in material of pile
ED2 t w
Py Stress level in pile Fatigue limit state
 y D2 t w

Table 2. Scaling laws for studying monopile-supported wind

turbines

ing 3P loading) and a motor turning the turbine blades (replicat- The following summarises the main conclusions.
ing 1P loading). This loading regime was applied for a particular
time interval (or a certain number of cycles); the natural (a) Wind turbines founded in sands (both dry and saturated)
frequency was then re-measured by conducting a snap-back test. exhibited near-field soil stiffening, resulting in an increase in
During this snap-back test, the actuator was disconnected from frequency, possibly as a result of densification. The results in
the tower, and the tower was given a small impulse displacement; sand are very similar to the observations carried out by
the resulting acceleration decay of the system was recorded. Leblanc et al. (2010) on a cyclically loaded pile, where it was
observed that, as a result of cyclic loading, the stiffness of the
The tests were carried out in dry sand, saturated sand, and soft monopile increased. In comparison, when founded in clay
overconsolidated kaolin clay. Details of the sample preparation, (Figure 5), the foundation degraded, causing a reduction in
set-up and testing procedure can be found in Bhattacharya et al. the frequency with the number of cycles. This frequency drop
(2011). A sensitivity study was carried out with three different appears to be a function of the soil strain level (P/GD2 ) and
forcing frequencies (3P loading denoted by ff in Table 2) of 2 Hz, the number of loading cycles.
20 Hz and 125 Hz, applied by the electro-dynamic actuator. The (b) Under certain conditions, cyclic loading may have a negative
dynamic influence of these frequencies is illustrated in Figure 3. effect on monopile-supported wind turbines founded in
The 1P frequency of 0.66 Hz representing the rotor frequency clayey soil, as the foundation stiffness and system natural
was applied by an electric motor powered by a DC supply. frequency will exhibit a continued decrease over time. The
conditions under which such degradation will occur relate to
Figure 4 shows data from a typical snap-back test obtained from functions of: (i) strain level in the soil imposed by the
the monopile-supported wind turbine founded in sand. The test monopile; and (ii) the ff /fn parameter, the relative value of the
results are plotted in the frequency domain using the Welch (1967) system frequency in comparison with the forcing frequency.
method. The system has a dominant frequency of about 3.3 Hz; (c) From the test results obtained, it can be seen that the greater
the foundation provides significant flexibility to the wind turbine the strain level in the soil (the (P/GD2 ) term in Table 2), the
system, which had a fixed base frequency of 10.27 Hz. The ff /fn greater is the degradation observed. As a result, in practice,
ratio was varied between 0.6 and 44 in the series of model tests. large-diameter piles present a better solution than small-
diameter piles. For example, for a particular ground condition
In a series of additional tests, the stiffness of the foundation alone (fixed shear modulus G) and a particular turbine at a site
under cyclic loading was also measured, utilising the set-up where the total lateral load (P) is constant, the larger the
shown in Figure 2(b) with the superstructure not attached to the monopile diameter, the lower is the strain level, and
foundation. The pile was subjected to 5000 cycles of loading, and consequently the lower is the degradation.
then a pushover force was applied to obtain a load–displacement (d ) Using the pile-head displacement test results (presented in
relationship. The tests were repeated after 10 000 cycles, and the Figure 6), a prediction of the frequency change of the overall
new stiffness was compared with the initial static stiffness. These system was made. For the purposes of numerical analysis, the
tests were carried out to verify whether or not the change in foundation can be replaced by three springs, as shown in
foundation stiffness alone would be adequate to predict the Figure 7: KL (lateral spring); KR (rotational spring) and KV
change in natural frequency of the system. (vertical spring) (Adhikari and Bhattacharya, 2011, 2012).
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
162
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

This foundation spring analogy model is currently being used

in industry-standard software such as Bladed (2008). Table 4
shows the prediction of the system frequency based on the
Accelerometer results obtained from the pile-head displacement tests
(considering the monopile alone), and compares this with the
Force sensor observed values of the system frequency under similar cycles
of loading. From the data presented in Table 4 it can be seen
that the mathematical model based on the foundation stiffness
degradation alone overpredicts the measured frequency
change of the system, indicating that the foundation-only
degradation model cannot fully capture the dynamic soil–
structure interaction effects. This may be due to the lack of
Actuator feedback mechanisms between superstructure and the
foundations, such as damping. Further to this, it may be
inferred that the dynamic response of the system is non-
1000 mm linear. Further study is required to characterise these aspects.

600 mm 4. Long-term performance of wind turbines

supported by tetrapod suction caissons
Suction caissons have been widely used in the oil and gas
industry for both jack-up and permanent platforms, such as the
SPT/Heerema Centrica and Draupner E platforms (Ibsen and
Brincker, 2004). Suction caissons have the significant advantage
Soft clay over conventional foundations of being quick to install and easy
to remove. A suction caisson is effectively an upturned bucket
embedded in the seabed, initially by its self-weight and subse-
500 mm
quently by an induced negative pressure inside the caisson,
drawing it into the seabed. Bye et al. (1995) first proposed the
use of suction caissons as foundations for offshore wind turbines.
So far, only a single wind turbine has been constructed on a
single suction caisson; this was sited at Fredrickshaven, and is
600 mm the property of Aalborg University. At present, information is
(a) limited, as it is the focus of ongoing research (Houlsby et al.,
Accelerometer
2005).

Force sensor Using the derived scaling laws, a 1:100 scale model of an
Actuator
offshore wind turbine supported by a tetrapod suction caisson
was created. For the purposes of design it was assumed that a
Vestas V120 4.5 MW turbine would be modelled (representing
the larger type of turbine that would be used in round 3 of
offshore wind expansion). The model caissons were 7.4 cm in
diameter and 5.5 cm deep, and were spaced 40 cm apart in
Soft clay 500 mm
orthogonal directions; this design was based upon design recom-
mendations by Byrne and Houlsby (2003) (Figures 8(a), 8(b) and
9). In order to test the long-term dynamic response of the suction
caissons, the model was founded in a representative sand matrix
with properties as described in Table 5. However, the installation
of the foundation was not studied.

600 mm For this series of tests two types of sand were used, as follows.
(b)
(a) Leighton Buzzard fraction E (LED) sand (uniform graded
Figure 2. (a) Experimental set-up of wind turbine supported on fine silica sand having D50 of 0.14 mm) poured to 28%
monopile; (b) set-up for studying foundation stiffness degradation relative density. The average shear modulus of the soil at the
mid-depth of the foundation is estimated to be 4.3 MPa.
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
163
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

Parameters Values Remarks

Bending stiffness of tower, EI: N mm2 2.125 3 109 Bending rigidity of tower
Mass of rotor and blades, M: kg 1.348 Motor weighs 1.008 kg and blades
0.34 kg
Length of tower, L: m 1.0 Aluminium tube 38 mm in diameter and
1.6 mm thick
Mass per unit length of tower: kg/m 0.576 Uniform tube
Frequency of system fixed at base of tower: Hz 10.27 Measured value. Predicted value based
on a single-degree-of-freedom model
(Figure 7) is 10.66 Hz
Pile length: m 0.5 Data not directly relevant in this paper;
provided for completeness
EI (pile): N mm2 3.18 3 108
Soils used: (a) Leighton Buzzard fraction E sand; For kaolin clay, shear modulus Details of the sands are given in Table 5
(b) Kaolin clay; (c) mixture of Leighton Buzzard G was measured as 6 MPa using
fraction E and Leighton Buzzard fraction B sand bender elements

Table 3. Details of model wind turbine

⬃1 ⬃7 ⬃40 (b) A mixture of Leighton Buzzard fraction B and Leighton

Buzzard fraction E (LBED) sand was employed to model a
relatively stiff soil configuration. The mean diameter ratio
Dynamic response factor

between the two fractions of sands (D50-Fraction-B :

D50-Fraction-E  6) was considered beneficial for an increase of
packing when the two types of particles were mixed together.
The fine particles were expected to fill the voids efficiently in
the large-particle matrix. The mass composition
corresponding to the theoretical maximum packing density of
1 the mixture was calculated according to a classic model of
packing (German, 1989), resulting in
XFractionB :XFractionE ¼ 85:15. The sand mix had an approximate
relative density of 2%, and the average shear modulus at the
Forcing frequency/natural frequency
mid-depth of the foundation was estimated to be about
Figure 3. Magnitude of dynamic influence 6.6 MPa.

The shear modulus of the two sand specimens was calculated

using the method proposed by Hardin and Drnevich (1972).
These calculated values matched within 10% of the shear
0·06 modulus measurements carried out on this specimen in the Blade
Power spectral amplitude

0·05 Monopile foundation laboratories during the Reluis project; see Dihoru et al. (2010).
supported on dry sand The test bed was prepared by standard air-pluviation of the sand,
0·04
0·03
after which the model was carefully installed under a steady
vertical load. The model was then subjected to the testing regime
0·02
described previously.
0·01
0 Figure 10 shows a typical frequency response of the wind turbine
0 1 2 3 4 5 6 7 8
Frequency: Hz system supported on a tetrapod foundation (Figure 8(b)). The plot
reveals a distinct difference when compared with the similar
Figure 4. Typical test result from snap-back test on monopile- response from a monopile (Figure 4), in that two closely spaced
supported wind turbine model, plotted in frequency domain response peaks were observed. The two closely spaced natural
frequencies correspond to modes of tower vibration in orthogonal
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
164
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

15

10 P/GD2 ⫽ 5·0 ⫻ 10⫺4

5
P/GD2 ⫽ 5·0 ⫻ 10⫺4
% change in natural frequency

⫺5 P/GD2 ⫽ 2·0 ⫻ 10⫺4

⫺10
Dry sand
⫺15
Saturated sand
⫺20
Soft clay (low strain)
⫺25 Soft clay (high strain)
⫺30

⫺35 P/GD2 ⫽ 3·4 ⫻ 10⫺3

⫺40
0 5 10 15 20 25 30 35 40 45
Number of cycles (⫻103)

Figure 5. Simplified plot showing change in frequency for dry

sand, saturated sand and clay

Initial response

Response after 5000 cycles

60 Response after 10 000 cycles

Initial KL
KL ⫽ 30·12 KL ⫽ 11·06 KL ⫽ 5·95
50 KL after 5000 cycles

KL after 10 000 cycles

40
Horizontal load: N

30

20

10

0
0 2 4 6 8 10 12 14 16 18 20 22 24
Pile head horizontal displacement: mm

Figure 6. Foundation stiffness degradation in clay

horizontal directions. To gain a better understanding of the tion, as expected, the foundation provides additional flexibility to
observed results, the model system was replicated in SAP 2000, the structure.
and the dynamics were calibrated. From this computer model it
was found that different vertical stiffness between individual The change in natural frequency of the system due to cyclic
caissons could produce the secondary modes recorded. In addi- loading is shown in Figure 11 for two types of soil for a constant
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
165
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

11 shows the results of five tests carried out on dry fraction E

sand in which up to 1.2 million cycles were imposed, and three
tests carried on a mixture of fraction B and fraction E for the
same period. The large gap in the results (between 0.5 million
cycles and 1.05 million cycles) corresponds to a weekend break
in recording results. The smaller interval in the data represents
the overnight period when the snap-back test could not be
conducted.

The test results suggest that the natural frequency of a wind

turbine founded on suction caissons has the potential to change
over time when it is subjected to cyclic loading. From the results
obtained, the change in natural frequency follows a non-linear
relationship, first increasing and later stabilising. The sand sample
with a lower initial relative density (LED) showed a greater
densification response; this is probably because it had a greater
contraction potential. Following the initial increase in natural
frequency, the latter apparent decrease could be attributed to the
sand dilating in accordance with critical-state behaviour.

KL 5. Discussion and conclusion

KR
Offshore wind turbines are relatively new structures, the first one
built in UK waters being at North Hoyle (2001). Very little is
known about the long-term behaviour of these structures, but it
has become apparent that soil–structure interaction is changing
KV the dynamic behaviour of the structures over relatively short time
spans. In an extreme case this could lead to the structure
becoming unusable.

Considering these factors, the following conclusions can be

Figure 7. Simplified structural model
drawn from the experimental study.

(a) Offshore wind turbines are dynamically sensitive structures,

(F/GD2 ), which corresponds to a particular strain level very because the natural frequency of the overall system is very
similar to the parameter (P/GD2 ) for monopiles. The definition of close to the forcing frequencies. The choice of foundation
(F/GD2 ) is given in Figure 12, where F is the net foundation type and design alters the overall system frequency. The soil
shear load – that is, the total lateral load acting on the wind foundation makes the wind turbine system more flexible than
turbine due to wind, wave and current (this parameter can also be the turbine tower would otherwise be on its own.
used for scaling the strain field produced by the vertical load, but (b) Tetrapod foundations, in comparison with monopile or
owing to the complex nature of load distribution this was monopod foundations, enhance the complexity of the
neglected). While plotting the data in Figure 11, the lower of the dynamic response of the overall system. This increased
two peaks in the frequency domain was used, as this represents complexity is due to the width of the dynamically critical
the first natural frequency of the structure (see Figure 10). Figure zone and the existence of two response peaks, a result of the

Number of Measured lateral Predicted change in Observed change in Near-field strain

cycles, N foundation stiffness, Kl : frequency based on frequency of system: % level, P/GD2
N/mm Figure 6: % (see Figure 5)

1 (Initial cycle) 30.12 0.0 0.0 3.4 3 103

5000 11.06 37.0 4.0 3.4 3 103
10 000 5.95 55.0 30.0 3.4 3 103

Table 4. Observed frequency of system and predicted change in

frequency based on foundation-only degradation model

Delivered by ICEVirtualLibrary.com to:

IP: 131.227.131.74
166
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

Blades

Rotor

1000 mm

Accelerometer

Force sensor
Actuator
Tower

Actuator 200 mm

55 mm
200 mm
75 mm
Tetrapod
400 mm

1200 mm

Sand Figure 9. Schematic diagram of tetrapod set-up

(a) Sand type

Leighton Buzzard Mixture of Leighton

fraction E Buzzard fraction B
and fraction E

Test ID LED LBED

% LB fraction E 100 15
Tetrapod
% LB fraction B 0 85
Specific gravity of 2.647 2.647
solids, Gs
łcrit 32 –
emin 0.613 0.289
emax 1.014 0.614

Table 5. Experimental sand properties used in suction caisson tests

Suction caissons
variation of the vertical support stiffness of the individual
caissons.
(b) (c) The frequency of the system can change with the number of
loading cycles, owing to changes in foundation stiffness. This
Figure 8. (a) Experimental set-up of 1:100 scale model of suction change depends on the type of soil and the nature of the
caissons; (b) details of foundation foundation. The change in frequency of the overall system
cannot be predicted from the degradation of foundation
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
167
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

0·030
Fixed base Acknowledgements
Tetrapod suction caisson frequency
Power spectral amplitude

0·025 The contributions to the project from Dr David Nash and

foundation supported
on dry sand Professor Colin Taylor are duly noted. The work presented in this
0·020 paper is part of the ongoing research into dynamic stability of
0·015 offshore wind turbines being conducted at the University of
Bristol.
0·010
REFERENCES
0·005
Adhikari S and Bhattacharya S (2011) Vibrations of wind-
0 turbines considering soil–structure interaction. Wind and
0 2 4 6 8 10 12
Frequency: Hz Structures 14(2): 85–112.
Adhikari S and Bhattacharya S (2012) Dynamic analysis of wind
Figure 10. Typical test result from snap-back test on wind turbine turbine towers on flexible foundations. Shock and Vibration,
supported on suction caisson, plotted in frequency domain 19(1): 37–57.
Andersen OJ and Løvseth J (2006) The Frøya database and
maritime boundary layer wind description. Marine Structures
35 F/GD2 ⫽ 1 ⫻ 10⫺4 19(23): 173–192.
Bhattacharya S, Lombardi D and Muir Wood D (2011) Similitude
30 Overnight break
% change in natural frequency

Weekend break
relationship for physical modelling of monopile-supported
25
LED wind turbines. International Journal of Physical Modelling in
20
LBED Geotechnics 11(2): 58–68.
15 Bladed (2008) Wind Turbine Design Software. GL Garrad
10 Hassan, Glasgow, UK.
15 Bye A, Erbrich C, Rognlien B and Tjelta TI (1995) Geotechnical
10 design of bucket foundations. Proceedings of Offshore
5 Technology Conference, Houston, TX, USA, OTC 7793.
0 Byrne BW and Houlsby GT (2003) Foundations for offshore wind
0 0·2 0·4 0·6 0·8 1·0 1·2 1·4 turbines. Philosophical Transactions: Mathematical, Physical
Number of cycles (⫻106) and Engineering Sciences 361(1813): 2909–2930.
Carter J (2007) North Hoyle offshore wind farm: design and
Figure 11. Long-term suction caisson results
build. Proceedings of the Institution of Civil Engineers –
Energy 160(1): 21–29.
MSL Dihoru L, Taylor CA, Bhattacharya S et al. (2010) Stiffness design
for granular materials: theoretical and experimental
considerations. Proceedings of the 7th International
F Conference on Physical Modelling in Geotechnics
(Springman SM, Laue J and Seward L (eds)). CRC Press,
Seabed London, UK, pp. 129–134.
G (soil modulus) German R (1989) Particle Packing Characteristics. Metal Powder
D
Industries Federation, Princeton, NJ, USA.
Hardin BO and Drnevich VP (1972) Shear modulus and damping
in soils: design equations and curves. Journal of the Soil
Figure 12. Definition of F/GD2 used for data analysis of suction Mechanics and Foundations Division, ASCE 98(7): 667–692.
caissons Houlsby GT, Ibsen LB and Byrne B (2005) Suction caisson for
wind turbines. Proceedings of Frontiers in Offshore
Geotechnics, ISFOG 2005 (Gourvenec S and Cassidy MJ
stiffness alone. The results thus present an intriguing (eds)). Taylor & Francis, London, UK, pp. 359–366.
challenge for the future design of offshore wind turbines, if Ibsen LB and Brincker R (2004) Design of a new foundation for
the observations made in these scale model tests are offshore wind turbines. Proceedings of the 22nd International
replicated in operational prototype systems. Any change of Modal Analysis Conference (IMAC XXII), Detroit, MI, USA.
the dynamic properties of a wind turbine will only increase Kühn M (2000) Dynamics of Offshore Wind Energy Converters
the sensitivity of such a system, limiting the range of safe on Monopile Foundations: Experience from the Lely Offshore
operation of such structures. Further research is required to Wind Farm. Engineering and Physical Sciences Research
understand more clearly the long-term performance of Council, Didcot, UK, EPSRC Workshop Report on OWEN
offshore wind turbines. (Offshore Wind Energy Network).
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
168
On: Sun, 07 Apr 2013 17:51:53
Geotechnical Engineering Dynamics of offshore wind turbines
Volume 166 Issue GE2 supported on two foundations
Bhattacharya, Cox, Lombardi and Muir Wood

Leblanc C, Byrne BW and Houlsby GT (2010) Response of stiff Welch PD (1967) The use of fast Fourier transform for the
piles to random two-way lateral loading. Géotechnique 60(9): estimation of power spectra: a method based on time
715–721. averaging over short modified periodograms. IEEE
Pierson WJ and Moskowitz L (1964) A proposed spectral form Transactions on Audio and Electroacoustics 15(2): 70–73.
for fully developed wind seas based on the similarity theory Zaaijer MB (2006) Foundation modelling to assess dynamic
of SA Kitaigordskii. Journal of Geophysical Research 69(24): behaviour of offshore wind turbines. Applied Ocean Research
5181–5190. 28(1): 45–57.

WHAT DO YOU THINK?

To discuss this paper, please email up to 500 words to the
editor at [email protected]. Your contribution will be
forwarded to the author(s) for a reply and, if considered
appropriate by the editorial panel, will be published as a
discussion in a future issue of the journal.
Proceedings journals rely entirely on contributions sent in
by civil engineering professionals, academics and students.
Papers should be 2000–5000 words long (briefing papers
should be 1000–2000 words long), with adequate illustra-
tions and references. You can submit your paper online via
www.icevirtuallibrary.com/content/journals, where you
will also find detailed author guidelines.
Delivered by ICEVirtualLibrary.com to:
IP: 131.227.131.74
169
On: Sun, 07 Apr 2013 17:51:53