Content uploaded by Andrew Henderson
Author content
All content in this area was uploaded by Andrew Henderson on Oct 26, 2014
Content may be subject to copyright.
FLOATING OFFSHORE WIND ENERGY
Andrew R. Henderson and Minoo H. Patel
Department of Mechanical Engineering, University College London, Torrington Street,
London, WC1E 7JE. 0171-380-7220. E-mail:a_henderson@meng.ucl.ac.uk
SYNOPSIS
The location of wind turbines on large floating structures offshore offers the obvious advantages of no
land usage and a probably more reliable wind resource. However, there are potentially significant
technical and cost drawbacks.
This paper describes the preliminary results of an EPSRC project aimed at developing analytical and
numerical design tools for evaluating the performance of offshore wind farms. The principal problems
that have been addressed include the determination of an optimum hull-form for the floating structure
and of developing analysis tools for the interaction of the motion in waves of the platform with the
turbine aerodynamic performance as well as the blade and hub loads. The project also addresses
secondary issues such as platform weathervaning, its mooring system and the means for energy-
transmission to shore. The possible integration of wind turbine output with power from associated gas
turbines in the North Sea’s offshore fields is also described.
The paper presents some of the theory and results arising from the work to date and identifies the
kinds of engineering and cost problems that are expected to be encountered.
1 INTRODUCTION
Over the last decade, interest in offshore
windfarms has grown slowly, with the current
handful of small experimental windfarms
likely to be joined by full-scale commercial
schemes in the near future. All current and
planned windfarms are in shallow water and
based around structures resting on or piled to
the seabed.
This paper describes preliminary results from
an EPSRC funded project aimed at developing
analysis tools for investigating the behaviour
and performance of wind turbines mounted on
large floating structures. The feasibility of
such floating windfarms will depend on their
cost, which will be influenced greatly by the
kinds of construction and sizes that will be
technically acceptable. The first stage in
investigating this is the development of
reliable tools for sizing and for performance
prediction.
In a wider context, the use of floating offshore
wind energy will depend critically on two
factors - whether costs can be brought down
and whether land-use pressure in shallower
waters will encourage the utilisation of deeper
water regions.
It has also to be recognised that the offshore
waters off the British Isles are currently one of
its principal energy sources, through its
hydrocarbon deposits. Some of the gas
accumulations in these offshore fields are
difficult to extract and utilise economically.
The combined use of small uneconomic gas
deposits for power generation together with
wind turbines on a floating structure could
offer a more economic and strategically
worthwhile source of significant amounts of
electrical power.
The focus of this research-project is the
development of tools for evaluation and
optimisation of offshore windfarm
configurations. This paper describes the first
two main tasks in this - namely:
• Calculation of the motion response of
vessels suitable for floating offshore
windfarms.
• Calculation of the effect this motion will
have on wind turbine performance and loads.
Submerged Pontoon
Turbine Tower
Figure 1
The Semi-submersible Concept
The floating hull concept under investigation
here is already well-proven in the oil-industry,
namely the semi-submersible (Figure 1)
design. The main structure of the vessel is
located below the ocean surface giving a
number of advantages over traditional
structures with hull forms close to the water
surface. These include reduced wave loads,
(since the wave kinematics decay
exponentially with depth) and longer natural
periods of motions (hence reduced response
motion).
The deep submergence of the pontoons
combined with a structure made up of
pontoons, columns and bracing yields the
above characterisation of low motion response
to waves. This feature has made the semi-
submersible a work-horse for drilling,
production and well servicing in the offshore
oil industry. These vessels have typical
dimensions of from 80 to 120 m and
displacements of from 12,000 to 40,000
tonnes. However the floating wind farm
application requires considerably larger semi-
submersible structures with deeper drafts and
larger displacements. At the same time, the
structural configuration has to be easy to build
and of intrinsically low cost. This scale up and
a characteristic of the resultant technical
performance is one of the principal objectives
of the work reported here.
The problem is not as daunting as it sounds
initially. It has to be remembered that
considerable work has been done around the
world on a variety of large structures for use as
floating airports, industrial sites and other
applications. See (1) and (2) for details.
2 MOTION OF A FLOATING VESSEL
2.1 Wave Loading
In previous work (3), it was shown that waves
contribute the majority of the rigid-body
motion-inducing dynamic loads of a large
floating structure.
Wave loading, F
Figure 2 - Morison’s equation
A simple and accurate method of calculating
the normal forces, F, is the Morison equation.
This is an empirically based equation and was
developed in the 1960s for wave loadings on
piles in shallow water (4).
dF C dV U C dS U U
MD
=+.. .
&... . .
ρρ
1
2
where CD, CM = coefficients of drag and
inertia loading (experimentally
derived)
S = projected area (to waves)
U = wave velocity
V = volume
ρ = fluid density
However, because of its relative accuracy and
ease of use, its use has been gradually
extended to many other applications including
the free-floating horizontal and vertical
columns modelled here. Since it can be shown
that the inertia forces dominate, the second
drag term has been ignored; this enables
general analytical forms of the load to be
derived, without further simplifications.
The wave particle acceleration and pressure
equations can be obtained from the linear Airy
wave potential flow function, φ, for water of
finite depth (5):
Note that an exact solution of the
wave/structure interaction problem can be
obtained by using linear diffraction theory for
infinitesimal gravity waves. A solution using
this approach is computationally complex and
expensive and yet gives results that are only a
few percentage points different from the
Morison equation based approach described
above (6). It was decided for the purposes of
the parametric interaction and optimisation to
use the simpler approach described above.
2.2 Vessel Response
For a vessel experiencing wave loads only, the
response can be characterised by an RAO
(Response Amplitude Operator), i.e. the ratio
of amplitude of vessel motion to wave motion.
Since the form of analysis is consistently
linear, this will not depend on the wave height.
The RAO transfer function for translational
motion can be defined as:
() ()
km
F
RAO +−
=2
.
ω
ω
ω
where F = total wave force
m = mass (or I for inertia) including
added mass
k = buoyancy stiffness (only present
for heave, roll and pitch)
with the rotational-form being equivalent.
This analysis generates a wave-frequency
dependent RAO curve for each degree of
freedom. For a full description of the theory
see (7).
2.3 Analysis Method
Two vessel designs were examined: a weather-
vaning and a non-weather-vaning
configuration depending on whether a more
expensive design with rotating mooring joint
would be used. The design criterion used in
the analysis is to minimise velocity and
acceleration RAOs at the nacelle.
For simple comparisons to be made between
the performance of different configurations,
the motion-response data were averaged as a
single number, the performance parameter.
This could now be compared with the value for
a suitable base vessel. Hence a negative
percentage change in the performance
parameter represents reduced motion and
therefore a better design.
2.4 Weathervaning Vessel
A weathervaning vessel will be able to rotate
so that it is always facing into the wind. The
turbines should therefore be arranged so that
none are in each others wakes, which suggests
a design based on a line.
Two main aspects were examined:
• the effect of altering the orientation of each
inter-turbine pontoon section individually,
• the effect of adding pontoon abutments
(with stability columns) at the centre and
wings of the design.
2.4.1
Re-design of Main Pontoon
-30%
-20%
-10%
0%
10%
20%
30%
450 500 550 600 650
Total Pontoon Length (m)
Change in Perfomance
Parameter
Layout
V
Layout
AAA
Layout
WW
Layout
UU3
Figure 3 - Response of Weathervaning
Designs (angled main structure)
Figure 3 shows the effect of varying the
pontoon orientation angle on the overall
response performance (as defined by the
performance parameter described above). The
independent variable used is the total pontoon
length, as this allows different designs to be
compared at similar pontoon costs.
It can be seen that Layout WW seems to be the
most efficient at minimising nacelle motion. A
schematic diagram of that layout is shown
below (Figure 4), which shows how this
particular design is defined. Each inter-turbine
section of the pontoon can be orientated either
bowwards (section Nr. 3 counting from the
centre) or sternwards (sections Nr. 1, 2 and 4
from the centre) with the section orientation
angle defined as a multiple of the nominal
vessel design angle, as 1x (sections Nr. 2 and
3) and 0.5x (sections Nr. 1 and 4). The other
designs represented in the chart: AAA and
UU3, have slightly different layouts.
α
α/2
α/2
α1.3 x diameter
turbines
submerged
pontoons
anchor
Figure 4 - Layout WW
2.4.2
Added abutments to Linear Layout
-45%
-40%
-35%
-30%
500 600 700 800 900
Total Pontoon Length (m)
Change in Perfomance
Parameter
Layout
111
Layout
101
Layout
101
x0.5
Figure 5- Response of Weathervaning
Designs (lineal main structure)
Figure 5 shows the affect of adding pontoon
abutments of various lengths at the centre and
wings of a linear vessel. Layouts 111 and 101
have abutments at the centre plus wings and at
the wings only respectively. The best
performing design seems to be Layout
101x0.5, which has the stability towers located
half way along the pontoon-abutment (Figure
6).
turbines
abutment
pontoons
stability
columns
1.3 x diameter
anchor
Figure 6 - Layout 101 x0.5
Further analysis shows that:
• the motion response can be improved for
this design by about 3.5% if the abutment
pontoons are angled outwards at 34o from the
perpendicular shown,
• if similar abutment-pontoons are added to
Layout WW, the overall response becomes
similar (but such a design would be more
difficult to construct).
Therefore, it can be concluded that for
weathervaning designs, Layout 101x0.5
(Figure 6) is the most suitable.
2.5 Non-Weathervaning Vessel
A non-weathervaning vessel cannot rotate to
face into the wind, hence turbines will
inevitably operate in another turbines wake at
times. If the distribution of wind direction is
uniform (unlikely), a symmetrical design is
required, i.e. the turbines should be located in a
ring and analysis undertaken for how they
should be connected. The designs investigated
here were based on polygon, star and fractal
shapes.
-65%
-60%
-55%
-50%
-45%
-40%
-35%
-30%
500 1000 1500
Total Pontoon Length (m)
Change in Perfomance
Parameter
Star
Fractal
MPL
Fractal
RA
Octagon
+Ab
Octagon
Figure 7 - Response of
Non-weathervaning Designs
Figure 7 shows the effect of increasing the
turbine separation on the performance
parameter for the various designs. Clearly the
octagon shape gives the best performance.
This design was therefore investigated further,
namely the effect of adding inward-pointing
and outward-pointing pontoon-abutments
attached to the main polygon structure at the
turbine locations (Octagon + Ab curve). It can
be seen that this did not improve the
performance as greatly as increasing the
polygon diameter (Octagon curve).
Therefore, it can be concluded that for non-
weathervaning designs, the Polygon Layout
(Figure 8) is the most suitable.
2 x diameter
turbines
pontoons
Figure 8 - Octagon Layout
3 TURBINE LOADS AND
PERFORMANCE
3.1 Model
A model was developed to calculate the effect
that the motion has on the turbine loads and
power output. These are predominantly due to
two sources:
• Blade aerodynamics,
• Blade, nacelle, tower inertia and gravity.
The loads are initially calculated in a two-
dimension state-space domain, representing
blade azimuth angle and vessel motion
respectively.
The main features of the model are:
• Rigid structures,
• A simple dynamic stall model (8),
• Otherwise steady-state aerodynamics,
• Turbulence (i.e. stochastic) effects
ignored,
Because of the vessel motion, there are
additional important axes systems:
• Global axis (defined by wind-
direction),
• Mean Vessel axes (takes account of
vessel misalignment to wind),
• Dynamic Vessel axes (defined by
instantaneous vessel orientation),
• Nacelle axes (takes account of yaw,
pitch and cone of turbine),
• Blade axes.
3.1.1
Aerodynamic Loads
The aerodynamic loads are calculated using
standard aerodynamic-momentum theory, as
described widely including in (9).
3.1.2
Inertia and Gravity Loads
Inertia and gravity loads are generally of an
equal or greater magnitude than aerodynamic
loads, and in the case of a floating turbine, this
tendency is extended. The loads are found by
applying Newton’s II law to the acceleration
vectors.
Consistent and disciplined application of the
axes systems means all inertia loads, including
gyroscopic loads, will be present.
3.1.3
Frequency Analysis
Since stochastic features have been ignored,
analysis in the faster frequency domain is
possible. This is done by Fourier-transforming
the state-space loads twice, with respect to
each of the state-space axes results in discrete
frequency domain load spectra.
3.2 Turbine Loads and Performance
Using axes transformations and translations,
the loads can be calculated at any location and
in any axes, with the following locations
selected as critical design loads:
• Blade root edgewise-axial fatigue stress
loads,
• Blade root flapwise-axial fatigue stress
loads,
• Rotor shaft fatigue axial stress loads,
• Nacelle yaw loads,
• Tower root sideways-axial fatigue stress
loads,
• Tower root fore-aft axial fatigue stress
loads.
The zero frequency components of the rotor
shaft axial-direct-load and roll-bending-
moment give the average turbine thrust loads
and power output respectively.
The fatigue damage due to a frequency domain
spectrum is estimated using a simple
cumulative method (SCM), shown here, and
the more complex Dirlik (10) method.
Although the damage predicted varied between
the methods, the relative changes tended to
agree very well. All charts display the fatigue
damage as a percentage of a mean value of the
first two charts.
3.2.1
Results
SCM Fatigue Damage Estimation
0%
20%
40%
60%
80%
100%
120%
140%
Ba se Yaw Til t Co ne Y, T & C
Resolution
Fatigue Damage
(as % of average)
Blade Root Edgewise Blade Root Flapwise Roto r Sha ft Bend ing
Tower Top Yaw Tower Bas e Side -side To wer Base Fore-aft
Figure 9
Typical Fixed Turbine Loads
Three sets of results are shown here. The first,
Figure 9, shows how the critical design loads
of a fixed-base turbine might vary depending
on turbine orientation. It can be seen that yaw
drive and tower base loads are minimal as
expected.
SCM Fatigue Damage Estimation
0%
50%
100%
150%
200%
250%
300%
350%
400%
450%
500%
None Surge Sway Heave Roll Pitch Yaw
Resolution
Fatigue Damag
e
(as % of average)
Blade Root Edgewise Blade Root Flapwise Rotor Shaft Bending
Tower Top Yaw Tower Base Side-side To wer Base Fore-aft
Figure 10
Effect of Unit RAOs
Figure 10 shows the effect that nacelle motion
unit RAOs would have on these same design
loads. It can be seen that vessel motion has a
very significant effect on the nacelle yaw and
tower base loads and that the location and
value of the loads varies greatly with the
direction of the motion.
SCM Fatigue Damage Estimation
0%
1%
10%
100%
1000%
Max RAO V (67.5 deg) 101x0.5 Octagon
Resolution
Fatigue Damage
(as % of average)
Blade Root Edgewise Blade Root Flapwise Rotor Shaft Bending
Tower Top Yaw To wer Base Side-sid e Tower Bas e Fore-a ft
Figure 11
Optimised Vessel Layouts
Figure 11 displays the loads for the optimised
vessel designs. As expected, the optimised
designs suffer significantly reduced loads
compared with the original V-shaped design.
4 DISCUSSION
The main disadvantage of the floating wind
energy concept is cost. While land-based
windfarm generation costs are nudging
2p/kWh in Ireland and the first experimental
offshore windfarms in Denmark came in at
around 6p/kWh several years ago, previous
analyses have suggested that for floating wind-
farms, the costs will be in the region of
10p/kWh. Clearly a reduction of almost an
order of magnitude is needed and significant
innovative thinking is needed.
These and other vessel design aspects will be
looked into in greater detail in the remaining
period of the project, hence some of the
comments here are brief but some ideas that
might possibly allow floating offshore wind
energy to become viable include:
4.1 Gas generation
Many oil fields also have a small quantity of
gas present, which is uneconomical to pipe to
the shore. Money is spent on re-injecting this
into the oil-filed (as regulations do not allow it
to be burnt off). If a wind-farm were to be
located alongside, gas generators could be used
ensuring a more-valuable steady supply of
power. A floating wind-farm could be moved
on when the gas field was exhausted.
4.2 Wave Power
In the last couple of years, there has been a
renewal of interest in wave energy around the
world. Many of the concepts instinctively
suggest that wind turbines could be
constructed on the structure and indeed the
original plans for the ill-fated OSPREY device
did include two wind-turbines.
4.3 Transmission
The greater distances from the electricity-user
means that the power transmission is another
cause for the overall greater power-costs. As
an alternative, hydrogen could be generated
and delivered ashore using the existing oil
pipelines.
4.4 Mooring
Moorings enable station-keeping and in the
case of
• turret mooring, allow the vessels to rotate
into the wind
• and tensioned moorings, greatly reduce
vertical motion.
However, moorings generally make up a very
significant part (typically 20%) of the total
cost.
ACKNOWLEDGEMENTS
The authors would like to thank EPSRC,
who are funding the research, being
undertaken jointly with the Energy Research
Unit at CLRC, Rutherford Appleton
Laboratory.
REFERENCES
(1) Kobayashi, K. et al; Large-scale on-sea
experiment of mega-float structures,
Sumitomo, 17th international conference on
Offshore Mechanics and Arctic Engineering;
May-1998.
(2) Iijima, K. et al; Structural response
characteristics of very large semi-submersible
and design considerations; 17th international
conference on Offshore Mechanics and Arctic
Engineering; May-1998.
(3) Henderson, A.R.; Motion of a Floating
Offshore Windfarm. MSc Thesis,
Loughborough University, September 1997.
(4) Morison, J.R. et al; The forces exerted by
surface waves on piles. Petroleum
Transactions. 189 (TP 2846), 149. 1950.
(5) Airy, Sir G.B.; Tides and waves; Encycl.
Metrop., Art 192, p 241; 1845.
(6) Leblanc, L. et al; A mixed panel-stick
hydrodynamic model applied to fatigue life
assessment of semi-submersibles; Conference
on Offshore Mechanics and Arctic
Engineering; 1993
(7) Patel, M.H.; Dynamics of Offshore
Structures. Butterworth. ISBN 0-480-01074-6.
1989.
(8) Øye, S.; Dynamic stall - simulated as time
lag of separation, Proc. of the 4th IEA
Symposium on the Aerodynamics of Wind
Turbines; January 1991.
(9) Freris, L.L., Wind Energy Conversion
Systems, Prentice Hall. ISBN 0-13-960527-4.
1990.
(10) Dirlik, T.; Application of computers to
fatigue analysis; University of Warwick
Thesis; 1985.