ECONOMIC VIABILITY OF A SIMPLE

TIDAL STREAM ENERGY CAPTURE

DEVICE
CONTRACT NUMBER:
TP/3/ERG/6/1/15527/REP
URN NUMBER: 07/575

The DTI drives our ambition of

prosperity for all by working to
create the best environment for
business success in the UK.
We help people and companies
become more productive by
promoting enterprise, innovation
and creativity.
We champion UK business at home and
abroad. We invest heavily in
world-class science and technology.
We protect the rights of working
people and consumers. And we
stand up for fair and open markets
in the UK, Europe and the world.

ECONOMIC VIABILITY OF A SIMPLE TIDAL

STREAM ENERGY CAPTURE DEVICE

DTI Project No: TP/3/ERG/6/1/15527/REP

TP/3/ERG/6/1/15527/REP
URN 07/575

Participants:
LOG+1 Limited
ALSTOM Power Limited
WUMTIA

The work described in this report was carried out with an Applied Research grant under
the DTI Technology Programme: New and Renewable Energy, which is managed by
AEAT Energy & Environment. The views and judgements expressed in this report are
those of the participants and do not necessarily reflect those of the DTI or AEAT Energy
.
& Environment.

First published February 2007

Joint Contractor Copyright 2006

1.

ABSTRACT

This tidal stream energy project has compared the overall economics of two
horizontal axis tidal turbine devices: a fixed pitch, bi-directional, variable speed
turbine generator device with a variable pitch, variable speed turbine generator
device that rotates to face into the tidal flow.
The project has established, theoretically, the extent to which the loss in energy
conversion efficiency of the simpler to construct fixed pitch device is
counterbalanced by a reduction in capital and O&M costs and whether the system is
technically feasible and sufficiently economic to warrant further development.
The conclusion is that the simple fixed pitch, bi-directional device is competitive on a
life cycle cost basis and worthy of further consideration.

2.

EXECUTIVE SUMMARY

2.1

Objectives

The project compares a generic axial, fixed pitch, bi-directional, variable speed
turbine generator device with a generic axial, variable pitch turbine generator device
that rotates to face into the tidal flow. The principal objective is to identify the
circumstances and extent to which the simple bi-directional device may be
competitive with the variable pitch device on a lifetime cost basis.
2.2

Introduction

This collaborative project involves the Wolfson Unit for Marine Technology and
Industrial Aerodynamics (WUMTIA) of the University of Southampton, ALSTOM
Power Ltd - Technology Centre and LOG+1. While not party to the grant
arrangements with the DTI, E.ON UK Power Technology Ltd. on behalf of E.ON UK
Renewables Developments Ltd., has provided a utility perspective and Converteam
Ltd has provided information on generators and power conversion aspects.
The agreed project scope was limited to horizontal axis tidal turbines (HATT), and
did not include consideration of alternative approaches such as vertical axis turbines
or oscillating hydrofoil systems.
In order to control the performance of a turbine in variable flow conditions such as
are typical of a tidal stream, it is necessary to alter the angle of attack of the blades
relative to the flow. This can be achieved using a variable-pitch system, with
attendant problems of sealing, mechanical complexity, increased cost and reduced
reliability, or by using a simple rotor with fixed pitch blades, the rotational speed of
which is altered to achieve a specific relationship between current speed and rotor
rotational speed.
The hydrodynamic performance of a sub-sea device will only be one parameter in
the lifetime cost of each system. The marine environment is particularly aggressive
towards equipment operating within it for extended periods with limited access or
opportunity for routine inspection and maintenance.
The Operation and Maintenance (O&M) element of the lifetime cost of a tidal stream
concept may well be greater than the significant contribution of O&M costs to the
cost per kWh anticipated for offshore wind energy, and be a major determinant in
the commercial viability of tidal stream energy. A commercially successful HATT
system will need a very high level of reliability and accessibility, with the longest
periods between routine maintenance inspections consistent with optimum wholelife economics. The premise is that the lowest capital and operating costs are more
likely to be achieved if the marine components are kept simple.
The project has attempted to identify and quantify the trade-offs between reliability,
efficiency and cost of generic HATTs. The overall economic performance will depend

to a varying extent on all aspects of the installation, operation, maintenance and

decommissioning approaches adopted.
While revenue assumptions have been made to derive comparative financial returns
from the two approaches, the most important output from the project is likely to be
the estimated cost of electricity delivered to the grid over each system lifetime,
assumed to be 25 years. This will help to establish whether the fixed pitch, bidirectional, variable speed device warrants further development.
2.3

Work Carried Out

The individual contributions to the project are summarised as follows:

WUMTIA - hydrodynamic design and energy capture of fixed and variable
pitch blades for HATT
ALSTOM Power Ltd - mechanical and electrical concept solutions and lifetime
cost estimates
LOG+1 - lead partner, project management. O&M and electricity market input.
While not party to the grant arrangements with the DTI, E.ON UK Power Technology
Ltd. on behalf of E.ON UK Renewables Developments Ltd., has provided a utility,
end-user perspective and a peer review capability. ALSTOM has also enlisted
Converteam Ltd in considering the generator and power conversion options
appropriate to the two devices.
Converteam Ltd. was formerly ALSTOM Power Conversion Ltd., part of the ALSTOM
group, but has been sold and is now a completely separate and independent
commercial entity.
The project was divided into three work packages, which are closely linked to the
partners responsibilities. While the work packages were reported to AEA
Technology chronologically there was a significant amount of iteration between the
partners as the project advanced; this applied particularly to the performance of the
optimised blades and the development of the cost model and financial analysis of
the fixed pitch and variable pitch devices.
WUMTIAs theoretical comparison of the performance of the fixed pitch and variable
pitch devices required an understanding of the hydrodynamic behaviour of
horizontal axis tidal turbines through use of basic theory, blade element momentum
(BEM) analysis and finally a lifting surface panel (LSP) code.
ALSTOMs cost analysis involved estimates of the following for the two devices:
capital cost, energy yield (dependent on device performance and availability),
operating cost (dependent on device reliability, preventive maintenance activity etc.),
other farm costs (permits, leases, general overhead etc.). From this, ALSTOM
derived the financial reports for farms of each device (profit, cash flow etc.),
compared the financial performance of farms operating the two devices and
assessed the sensitivity of the results to changes in the variables/assumptions.

Results from Monte Carlo simulations of device availability, together with cost
estimates for capital and maintenance costs, were fed into the cost comparison
spreadsheet to provide a comparative cost and revenue performance for each
machine in the context of a farm of 30 devices, each rated at 1MW, having a 20m
diameter rotor, and operating in a tidal stream with a maximum flow speed of 2.5
m/s. Costs common to the two device types were included at this stage. For cost
modelling purposes two turbines were assumed to be attached to a common
mounting. This arrangement was chosen to reflect the fact that many proposed
devices feature one or more pairs of contra rotating turbines to minimise the net
torque transmitted to the mounting structure.
Fault Tree Analysis (FTA) was chosen in preference to Failure Mode and Effect
Analysis (FMEA) for the reliability elements of this study because it was considered
easier to construct a generic analysis. Use of the FTA technique allows a generic
failure to be described without detailed consideration of the precise nature of the
failure.
FTA is a top-down technique in which a specific top-level fault is identified (in this
case, lack of electrical output to the grid when such output would be expected), and
all combinations of lower-level fault causing the top-level fault are identified. In
contrast, FMEA is a bottom-up approach in which all possible faults of all
components are identified, and the implications of each fault are assessed, for
example providing the answer to the question What is the effect of a mechanical
failure of a main gearbox third-stage planet gear axle?, FMEA and FTA may in most
cases be regarded as providing equivalent information, but meaningful FMEA
requires the existence of a fairly detailed design definition of the system; implying
loss of generality for the study.
In order to assess overall failure probabilities for the study devices a generic fault
tree was created, and the individual fault conditions featured in the tree were
ascribed probabilities of occurrence, derived from published data for comparable
equipment in a North Sea Oil production context. The generic fault tree could be
reconfigured to represent fixed pitch or variable pitch devices and the overall failure
probability was then recalculated automatically.
2.4

Main Results

The principal objective of this study has been to identify any circumstances under
which simple but less efficient devices offer competitive life cycle performance to
complex but more efficient ones.
Comparison of the variable and fixed pitch design indicates that the energy yield
from the fixed pitch design is likely to be up to 10% less than that for the variable
pitch design. With a generator of a given capacity, this difference could be
accommodated by an increase of about 1.0 m in rotor diameter over the 20mbaseline diameter.

The comparative system cost analysis has demonstrated that the relative merit of
the two approaches depends upon the assessment criteria adopted and may be
summarised as follows:
The variable pitch, variable rotational speed machine (with an energy capture
performance notionally 10% better than that of the fixed pitch machine)
produces more energy in a given period than the fixed pitch, variable
rotational speed machine, unless the absolute reliability of both machines is
very low,
The fixed pitch machine always offers lower initial capital cost and unplanned
maintenance cost than the variable pitch machine,
The fixed pitch machine offers lower cost per unit of electricity generated
unless the absolute reliability of both machines is very high,
The percentage increase in energy capture using the variable pitch machine is
generally much less than the notional 10% difference given perfectly reliable
machines; relative performances encountered during the study implied that
the variable pitch machine would generally produce 4 7% more energy than
the fixed pitch machine, although its worst performance produced 1% less
energy.
The discrepancy between the notional and calculated performances arises because
the availability of the variable-pitch machine is lower than the availability of the
fixed-pitch machine, so some or all of the variable-pitch machines theoretical
energy capture advantage is lost. The availability depends upon reliability and
accessibility, but the relative importance of accessibility itself depends on overall
reliability. In general the higher the overall reliability of both devices the greater the
relative energy capture of the variable pitch machine. The 7% improvement quoted
was associated with the highest reliability assumed in the study, and the 1% worse
performance reflected the lowest reliability used. For reliability modelling purposes
the variable pitch machine is treated as a variant of the fixed pitch machine.
Although uncertainties exist in the absolute reliability values, it is considered that
the relative reliabilities of fixed pitch and variable pitch machines have been well
represented for study.
The baseline case consisted of a farm of 30 off 1 MW turbine units configured with
an active DC link and two network bridges, each serving 15 of the turbines. The
variable pitch option also allowed turbine axis rotation in yaw, so as to always face
into the tidal stream. The baseline case, using the calculated hydrodynamic capacity
factor of 22.8% for the fixed and 24.5% for the variable concept and mean failure
rates, shows an overall energy production of 53 and 55 GWh per year respectively.
Using these energy production figures the cost of electricity is 119 /MWh for the
fixed pitch concept, compared to 129 /MWh for the variable pitch concept over a 10
year operating period, and 94 /MWh and 104 /MWh respectively for a 15 year
period.
The results shown in the study are based on 110% of rated power being achieved at
peak spring current velocity. This is based on the assumption that (consistent with
diesel generator set practice) the generator may be operated at such a power output
without damage, providing the operation at over 100% rating is of limited duration,

and at sufficiently infrequent intervals that the design winding temperatures are not
exceeded. The economically optimum design of tidal stream turbine system will
probably not require peak spring tide current speeds to generate its rated power, but
the definition of the true economic optimum device may well depend upon the
precise stream parameters at its intended operational location. Therefore the rated
speed has been taken to be the peak spring current speed for the purposes of this
study.
It is recognised that the hydrodynamic capacity factors could be increased by using
a generator rated at below the maximum power available in the stream, and this
could lead to a lower cost of electricity. This aspect has not been considered in this
study.
2.5

Conclusions

This tidal stream energy project has compared an axial, fixed pitch, bi-directional,
variable speed turbine generator device with an axial, variable pitch, turbine
generator device that rotates to face into the tidal flow. The project has established,
theoretically, the extent to which the loss in energy conversion efficiency of the
simpler to construct fixed pitch device is counterbalanced by a reduction in capital
and O&M costs and whether the system is technically feasible and sufficiently
economic to warrant further development.
The project compared the best theoretical hydrodynamic solution to the capture of
energy from a tidal stream with a simple alternative solution that would cost less to
manufacture and should be expected to be more reliable during its operational life.
The conclusion was that the simple fixed pitch device was competitive on a life cycle
cost basis and worthy of further consideration.
A number of tidal stream concepts under development address the problem of
reverse tidal flow by including a mechanism to change the blade pitch by 180 deg or
more. While this will do away with the cost and reliability implications of a rotator
mechanism, it requires a compromise on the blade section that will result in a
reduction in the energy yield compared with the best hydrodynamic solution. This
would seem to reinforce the case for further consideration of the fixed pitch, bidirectional, variable speed turbine generator device.
Tidal stream devices are likely to be located well away from centres of population
and the established transmission and distribution network. The availability of grid
connections at a date and cost that does not prevent commercial deployment of tidal
stream and other renewable energy devices has to be urgently addressed by the DTI
and Ofgem.
Tidal power is arguably one of the more expensive in capacity terms of the available
technologies, certainly at this stage of its development. While commercial
deployment of devices will help to drive down the capital and operating costs, tidal
power will need additional support while the successful concepts become
competitive renewable energy options. The Government is proposing differentiated

support levels to different renewable technologies; such support will be necessary to

enable marine renewable technologies achieve their potential.
2.6

Recommendations

The following activities offer logical steps to the implementation of a commercially

successful tidal turbine device:
Further refinement of the economic analysis methods used to determine
parameters for an optimised device,
Develop design specifications for a tidal stream turbine system and apply
suitable processes to optimise the design point for a given set of tidal
conditions,
Undertake further reliability modelling to determine plant redundancy
requirements,
Undertake model testing of two dimensional bi-directional sections and then
three dimensional rotors, both for cavitation performance and overall
performance to confirm that predicted in the study,
Install a reduced scale version of the device (around 20 30% full scale) in a
marine environment, probably with a dump load rather than a grid
connection. Test results will be used to finalise the design of the prototype
unit,
Install a full-scale prototype at a suitable location with grid connection. This
would not be commercially optimised, but would demonstrate all the major
systems,
Install a multi-unit tidal stream turbine farm using the commercial design
incorporating developments from the testing of the prototype unit.

10

3.

CONTENTS PAGE

1. ABSTRACT .............................................................................................................................4
2. EXECUTIVE SUMMARY .................................................................................................5
3. CONTENTS PAGE ........................................................................................................... 11
4. GLOSSARY ......................................................................................................................... 12
5. PROJECT BACKGROUND ............................................................................................14
6.

BLADE DESIGN FOR FIXED AND VARIABLE PITCH TURBINES ........... 23

7. MECHANICAL AND ELECTRICAL SYSTEM DESIGN OPTIONS............. 37

8. COMPARATIVE SYSTEM COST ANALYSIS ......................................................49
9. BIBLIOGRAPHY ................................................................................................................ 60
APPENDIX A:

MONTE CARLO SIMULATION TECHNIQUES....................... 62

APPENDIX B:

CASES ASSESSED .............................................................................. 72

11

4.

GLOSSARY

Abbreviation
BEM

DNV
DSV

FTA

FMEA

HATT
LRU

LSP

OHL
O&M

OEM
OREDA
RIB
SCADA

TSR

Term
Blade Element
Momentum

Description
Hydrodynamic analysis code which
matches knowledge of the change in a
fluid's axial and swirl momentum with the
lift and drag performance of a blade
section.
Det Norske Veritas
A Norwegian based Classification Society.
Dive Support Vessel Class of ship associated with offshore
operations typically featuring a Dynamic
Positioning system, diver support facilities
and 150 tonne crane.
Fault Tree Analysis
A top-down technique in which specific
top-level faults are specified and all lower
level faults causing it are identified.
Failure Mode &
A bottom-up technique in which all
Effects Analysis
component faults are identified and the
implications of each fault is assessed.
Horizontal Axial
A tidal stream turbine with a horizontal
Tidal Turbine
rotor shaft.
Line Replaceable
An individual tidal stream turbine that can
Unit
be replaced in the event of failure or at the
end of its design life.
Lifting Surface
Hydrodynamic analysis code based on
Panel
quadrilateral boundary elements that solves
Laplace's equation for potential flow.
Overhead Line
As opposed to buried or submerged cable.
Operation and
Control of the day to day operation of the
Maintenance
tidal stream farm and the
maintenance/replacement of all equipment
during the life of the project.
Original Equipment Manufacturers of the equipment used in a
Manufacturer
tidal stream farm.
Offshore Reliability
A database developed by a number of
Database
leading oil companies.
Rigid Inflatable Boat Boat with rigid hull and large inflatable
tubes down each side.
Supervisory Control Controls and monitors all the equipment in
and Data
the tidal stream farm.
Acquisition
Tip Speed Ratio
Ratio of the rotational speed at the blade tip
to
the onset (undisturbed) speed of the tidal
current.

12

Term
Azimuthing

Description
The ability of a HATT to turn so that its axis is aligned with the
flow of the tidal stream.
Cavitation
The formation of vapour cavities when the local static pressure
(due to high local flow speeds) reduces to that of the fluid's
vapour pressure.
Hydrodynamic A measure of the efficiency of the rotor in capturing energy from
Efficiency
the tidal stream. It is calculated as the (total energy produced by
the rotor over a full tidal cycle) divided by (energy produced by
the rotor operating at its rated power over the same time). It
therefore does not take account of mechanical or electrical losses
in the system, or reliability and availability factors.
Monte Carlo
A system used to predict reliability and availability levels based
on the laws of probability.
Tidal Ellipse
Refers to a polar plot of the variation in tidal stream strength and
direction over a tidal cycle.

13

5.

PROJECT BACKGROUND

5.1

Introduction

A variety of approaches is being taken to the harnessing of tidal stream energy to

produce electrical energy for supply to the grid. The hydrodynamic performance of
the sub-sea device will only be one parameter in the lifetime cost of each system.
The marine environment is particularly aggressive towards equipment operating
within it for extended periods, with limited access or opportunity for routine
inspection and maintenance. The O&M element of the lifetime cost of a tidal stream
concept may well be greater than the already significant contribution of O&M costs
to the cost per MWh anticipated for offshore wind energy and be a major
determinant in the longer term whole life commercial viability of tidal stream
energy. To develop a commercially successful system it will be essential for the
marine components to have the highest possible reliability and accessibility,
commensurate with the trade off in reduced efficiency due to longer periods
between routine maintenance inspections.
The agreed project scope was limited to horizontal axis tidal turbines (HATT), and
did not include consideration of alternative approaches such as vertical axis turbines
or oscillating hydrofoil systems. The project compares a generic axial, fixed pitch, bidirectional, variable speed turbine generator system with a generic axial, variable
pitch turbine generator system that rotates to face into the tidal flow. The
comparison
establishes the theoretical extent to which the loss in energy
conversion efficiency of the fixed pitch device is counterbalanced by the estimated
reduction in capital, operational and maintenance costs.
This is a collaborative project involving the Wolfson Unit for Marine Technology and
Industrial Aerodynamics (WUMTIA) of the University of Southampton, ALSTOM
Power Ltd - Technology Centre and LOG+1, whose individual contributions to the
project are summarised as follows:
WUMTIA - hydrodynamic design and energy capture of fixed and variable
pitch blades for HATT
ALSTOM Power Ltd - mechanical and electrical concept solutions and lifetime
cost estimates
LOG+1 - lead partner, project management. O&M and electricity market input.
While not party to the grant arrangements with the DTI, E.ON UK Power Technology
Ltd. on behalf of E.ON UK Renewables Developments Ltd., has provided a utility
perspective and Converteam Ltd has provided information on generators and power
conversion aspects.
5.2

Project Objectives

Existing and proposed HATT designs employ either fixed-pitch or variable-pitch

blades, with each developer tending to adopt one of these approaches to the
exclusion of the other. The principal objective of this study is to identify the
circumstances in which one or other of these approaches is superior, on a whole life
14

basis. This will in turn require the project to identify and quantify the trade-offs
between reliability, efficiency and cost of generic HATTs through the application of a
system approach. The overall economic performance will depend to a varying extent
on all aspects of the installation, operation, maintenance and decommissioning
approaches adopted.
While revenue assumptions have been made to derive comparative financial returns
from the two approaches, the most important output from the project is likely to be
the estimated cost of electricity delivered to the grid over each system lifetime,
assumed to be 20 years.
5.3

Project Process

The project was divided into three work packages:

Work package 1:
The primary objective of this work package was for
WUMTIA to conduct a theoretical comparison of the performance of the fixed
pitch and variable pitch devices to establish whether they were sufficiently close
to warrant the continuation of the project. Other objectives for this work package
included the development by ALSTOM of the initial specification and generator
power conversion concept and data gathering by LOG+1 on other tidal stream
turbine concepts.
Work package 2:
The main focus of this work package was on the
production by ALSTOM of the concept design for the generator and power
conversion system, together with the development of the overall cost model, for
which LOG+1 would work up the O&M component. In parallel with this, WUMTIA
would finalise the design of the turbine blades for the fixed pitch and variable
pitch devices.
Work package 3:
The final work package included finalising the
hydrodynamic analysis, system validation and sensitivity analysis and
completion of the life cycle cost analysis and the final report.
5.4

Wind Turbine Analogy

When considering the Initial Project Specification, and the general concept of a tidal
current turbine, it is tempting to use modern wind turbines as models for
assumptions; some modern medium-sized wind turbines are operating routinely at
similar power and/or torque levels to the generic tidal device, so it might be
expected that certain major components or subsystems (e.g. gearbox) could be
translated into tidal current service with little or no modification.
This approach is of limited use; the conditions experienced by a tidal turbine are not
simply direct analogues of those experienced by a wind turbine, scaled by some
function of relative density and current speed. For example, tidal current velocities
are predictable over very long timescales and are not subject to random excursions
over these values, whereas wind velocity is an expression of a series of stochastic
processes, including random gusts. The tidal turbine arguably therefore needs
lower structural margins on rotor blades. Conversely, the tidal-turbine working
environment is corrosive, with suspended solids leading to at least the possibility of

15

erosive damage over the lifetime of the device. Another important consideration for
tidal turbines is the presence of marine life.
A comparison of some of the factors affecting wind and tidal stream turbines is
given in Table 5.1 below:
Feature

Effect / Implications of Feature on

Offshore Wind Turbine

Tidal Current Turbine

~1.25 kg/m3

~1025 kg/m3

~25m/s

2-5m/s

~12m/s

2-5m/s

50m/s +

As for normal operation

Variation of velocity
with time

Stochastic, variable in
magnitude and direction
over timescales of the
order of seconds to
years.

Variation in magnitude &

direction predictable for given
location over periods of years.

Rotor diameter
(typical)

90-120m

Fluid density
Max velocity during
normal operation
Velocity for rated
output
Max velocity during
life

15-30m based on current

designs
Mechanical integrity, cavitation
at tip, water depth. Speed
Mechanical integrity,
reduced as diameter increased.
Limitations on rotor
primarily fatigue life due
Limitation is on blade stress,
diameter
to self weight stresses
primarily due to thrust forces
being much greater resulting
from higher density of fluid
Variation of flow
Complex (turbulence + waves if
Complex (turbulence)
pattern
top of rotor near surface)
Corrosion
Salt spray conditions
Immersion in salt water will
require careful consideration of
material combinations
Unlikely to be a serious
Potential for serious problem;
Erosion
problem
may exacerbate corrosion
Depends on deployment
Maintenance access
Weather dependent
method but probably more
difficult than offshore wind
Could be important for
Marine growth
Not an issue
performance and maintenance
Rotor rotational
<15 rpm for large
7-20 rpm based on current
speed
machines
designs
Table 5.1
Comparison of some of the factors affecting wind and tidal stream
turbines

16

5.5

E nvironmental and Other Impacts

Unlike a wind turbine, the tidal stream turbine, being submerged, will have neither a
direct visual effect nor a noise effect on human beings in normal conditions. There
are, however, a number of areas in which submerged devices could have potentially
significant environmental impacts or raise safety considerations for other maritime
users and activities. Such areas include: ecology, pollution, coastal processes and
interference with the tidal flow, noise and navigation safety considerations. Such
issues will need to be taken into account in selecting a location for an individual tidal
stream turbine or an array of turbines, and are likely to have cost and time
implications. There is not likely however, to be a significant difference between the
environmental impacts of the two devices compared in this study; environmental
impacts are not considered further in this report.
5.6

UK tidal energy resources

As stated in Path to Power, the June 2006 BWEA report on wave and tidal power,
The UK possesses some 35% of Europes wave resource and 50% of its tidal
resource. BWEA, using data from the Carbon Trusts Marine Energy Challenge
(MEC), estimates that 3GW of wave and tidal capacity could be installed in the UK by
2020. .
In the long term, 3% to 5% of current UK electricity demand could be
met by tidal stream energy.
While there is clearly a substantial UK resource of tidal stream energy, by its nature
it tends to be located close to headlands in the west and north of the UK, which in
turn are less accessible from the existing grid infrastructure. Each location clearly
also has its own characteristics, with some such as Portland Bill being very
asymmetric in current speed and direction. The tidal ellipse for a specific location is
likely to be a major determinant of the technology to be applied, with the facility for
the devices to be rotated to meet the flow direction being much more important in
some locations than others.
5.7

Assumed tidal stream conditions

A tidal stream features a continuously changing current velocity, with a pattern of

change being in principle different for every location, but predictable for a given
location with reasonable accuracy over periods of the order of years. For the
purpose of this project, we have assumed a symmetric tidal ellipse with the tidal
stream varying in magnitude over the semi diurnal ebb and flow cycle and a
fortnightly spring/neap period, but have considered the sensitivity of the fixed pitch
devices to a limited range of off axis tidal flows.
It should be remembered that the variable pitch azimuthing machine features a rotor
which operates at low efficiency in a reverse flow; the azimuthing feature allows
optimum presentation of the rotor either to rectilinear flows or to flows in which the
reverse tidal current is off-axis with respect to the forward current.

17

5.8

Existing HATT tidal stream devices

Information was gathered on existing tidal stream energy capture devices and the
following were selected as offering the most direct comparison with the two HATT
devices being considered in the study.
Tidal Stream Turbines promotes a horizontal shaft two bladed turbine of 1 to 2 MW
for current design of 20m diameter rotors. The concept has undergone initial river
tests and water tank testing of a scale device is planned for 2006. Each turbine unit
is mounted on a horizontal structure attached to the main upright, the whole unit
being semi-submersible and attached to a gravity foundation on the seabed, which
will a llow the turbine to turn to follow the tidal flow. The turbine generator unit is
raised to the surface for maintenance purposes using the buoyancy of the nacelle.
Marine Current Turbines (MCT), probably the most advanced of the current
developers, offer 1MW horizontal shaft axial flow turbine units under the names
Seagen and Sea Array. Two full-size commercial units are due to be installed in
Strangford Narrows in Northern Ireland shortly. The turbine units, mounted in pairs
on a monopile, use variable pitch blades with a sufficient range of pitch change to
operate in both tidal flow directions. The structure includes a mechanism to raise
the turbines to the surface for maintenance purposes.
SMD Hydrovision has a horizontal shaft 2 bladed fixed pitch turbine with 18 m
diameter rotors device called TidEL, which comprises two 500 kW turbines mounted
on a frame designed to float clear of the seabed and to move in such a way as to
follow the tidal flow.

Figure 5.1
right)

Marine Current Turbines (above left) and 1/10 scale model TidEL (above

18

Swan Turbines Ltd are developing a horizontal shaft three bladed turbine. The
current programme is to develop a 350kW medium scale demonstrator unit,
together with modelling and cost optimisation by summer 2006. The plan is then to
install a pre-commercial system of 1 MW capacity between 2008 and 2010. Each
turbine unit is mounted on a frame attached to the seabed. The emphasis is on
simplicity and a gearless low speed generator, offering high efficiency over a range
of speeds, has been developed. The unit will be raised the surface for maintenance
purposes. It is not known whether the unit can rotate to follow the tidal flow or the
nature of its blade design, with fixed or variable pitch blades.
Hammerfest Strom AS is a joint venture between ABB, Rolls Royce, Statoil and
Sintcef. A 300 kW horizontal shaft axial flow turbine was installed off the coast of
Norway in 2003. The submerged structure weighs 120 tonnes and has gravity
footings of 200 tonnes. The three-bladed rotor is 20 m in diameter with a variable
pitch range that allows it to operate in both tidal flow directions. We do not have
any information on the performance of this prototype device.

Figure 5.2
right)

Swan Turbines (above left) and Hammerfest Strom devices (above

Tidal Generation Ltd is developing a 1 MW horizontal shaft three blade fixed pitch
turbine. A 1MW prototype unit is due to be installed at the European Marine Energy
Centre in Orkney in 2006. Commercial units are planned by 2010. Each turbine unit is
mounted on the frame attached to the seabed. The frame is a lightweight structural
design for easy installation and removal and is designed to follow the tidal flow.
Tidal Hydraulic Generators Ltd are developing a horizontal shaft axial flow turbine of
between 250 kW and 1 MW depending on stream velocity for an array of five 6mdiameter rotors. A prototype unit is planned for 2007 to be followed by a 10 MW
farm by 2009. The turbine units are mounted on a frame attached the seabed that
can be installed or lifted within a day. No information is available on whether the
blades are fixed or variable pitch to facilitate bi-directional flow.

19

5.9

Generic device for study purposes

This project is quite deliberately not specific to any existing or proposed device,
although it requires certain assumptions to be made. A conceptual generic tidal
turbine was defined, to the level of block diagrams, sketch layouts and some basic
performance and structural calculations, to provide a basis for cost and reliability
estimates, and to ensure that assumptions made in order to perform the study were
realistic. Some of the assumptions are also the subject of sensitivity studies.
The generic devices take the form of horizontal axis turbine assemblies each having
a single, un-shrouded rotor driving an electrical generator via an appropriate
mechanical transmission system. For the fixed-pitch machine, the rotor blades are
set at a fixed pitch, whilst for the variable-pitch machine, a mechanism and control
equipment is provided to allow the blade pitch to be continuously adjusted in order
to maximise energy extraction from the tidal stream and the nacelle is able to rotate
about the vertical axis so that the rotor axis is always in line with the current flow.
The designs for both the fixed pitch and variable pitch options use cage type
induction generators connected to power conversion electronics systems to provide
a 50Hz supply to the grid from the variable-frequency generator output, over a fairly
wide range of input rotational speeds. For the variable pitch option it would be
possible to operate the turbine at fixed speed and use a simple cage
induction generator connected directly to the grid. However, the directly-connected
asynchronous (cage) induction generator could only operate to generate power over
a narrow range of rotational speeds close to the synchronous speed, with little
opportunity to maintain optimum tip speed ratio over the range of tidal conditions
encountered. The hydrodynamic capacity factor for this type of installation would
be much lower than for the option including power electronics, and the directly
connected option was therefore not considered further for this study.
5.10

Initial Project Specification

The initial outline for the project proposed two variants; Option 1 being a fixed-pitch
device and Option 2 being a variable-pitch device with the following idealised
parameters:

Blade Pitch
Flow Direction
Rotor Axis
Rotor Axis Depth
Maximum Rotor Diameter
Nacelle Body
Current Speed
Starting
Table 5.2

OPTION 1
Fixed
0 and 180 degrees

OPTION 2
Variable
0 degrees

Horizontal
20m
20m
2.5m diameter by 6m long
0 to 5 m/s
Self-start at current speeds > 1m/s

Initial project specification

20

In order to frame the project, a specification was created for a generic device. The
values quoted below are considered reasonable initial assumptions for a device to
operate in some of the more attractive UK sites, but should not be interpreted as
representing the economically optimum device. The values quoted in Table 5.3
below should be considered to be generic and are not necessarily completely
consistent with each other.
The choice of 2.5m/s rated velocity was determined by study of Reference 1, with the
intention of making the study applicability as wide as possible; in terms of peak
spring tidal velocity Vsp, 2.5 m/s is exceeded by 41 of the 57 sites listed, together
accounting for more than 95% of UK tidal stream resource.
The partners identified the 20m rotor diameter and 1MW unit rated power per device
early in the project and, for the purposes of whole life cost analysis, deployment of
30 such devices to make a 30MW wind farm. These assumptions are consistent with
the prototype device and large array definitions in the 2006 BWEA report The Path
to Power. The Path to Power identifies four stages of marine renewable
deployment in the UK. The capacity definitions for these stages of deployment are:
Prototype devicesingle pre-commercial device up to 1MW in size
Small arraysmall arrays up to 5MW in total export capacity
Large arraylarge arrays up to 30MW in total export capacity
Significant projectcommercial projects in excess of 30MW
Parameter

Value

Unit

Generator rated power

Maximum tidal current spring tide
Ratio of peak spring tidal current speed : peak neap tidal
current speed
Rotor diameter

1
2.5

MW
m/s

2.0

20

Maximum nacelle diameter

Maximum nacelle length

4
10

m
m

Water depth
Number of blades

40
3

Transmission voltage
Cable distance from device to shore

33
5

kV
km

Number of devices per farm

30

Table 5.3

Device assumptions

The power conversion solution selected will comprise rectifier equipment located in
each turbine nacelle, which will convert the variable frequency AC output from each
induction generator to DC at about 3.3kV. The DC output from each unit is connected
via a common link to a single network bridge that converts the output from the tidal
stream farm to 50Hz AC. This network bridge may be located either on a dedicated

21

platform located centrally in the tidal stream farm, or onshore if the farm is close to
land. The output from this bridge is connected to the grid via a step-up transformer
to raise the output voltage to 33kV or 132kV.

22

6.

BLADE DESIGN FOR FIXED AND VARIABLE PITCH TURBINES

6.1

Hydrodynamic
Hydrodynamic Performance of HATT - Predictions

6.1.1 Summary
This WUMTIA study was the key element of Work Package 1, with the objective of
providing:
An understanding of the hydrodynamic behaviour of horizonta l axis tidal
turbines through use of basic theory, Blade Element Momentum (BEM)
analysis and finally a Lifting Surface Panel (LSP) code, and
An answer to the relevant question of by how much a mechanically complex
controllable pitch turbine outperforms a simple fixed pitch, bi-directional
system.
WUMTIA initially confirmed the basic theory, identified relevant computational tools
and developed a methodology to define the turbine blades and analyse tidal
behaviour. Validation studies were carried out using BEM and LSP code. The
developed tool for investigating tidal behaviour was used to investigate a number of
different concept HATTs. A study of comparative performance has shown that the
use of a variable pitch system has only a limited influence on the delivered energy.
However, the ability to vary the pitch allows machine power to be limited to its rated
value for higher current speeds.
6.1.2 Background
The performance of devices that use the kinetic energy associated with a current or
wind the performance is that of a low-head (wind or tidal) turbine. In this case the
available power P for a given device capture area A and associated wind/current
velocity V is

P = Cp 12 AV3
where the power coefficient Cp is a measure of the overall hydrodynamic efficiency
of the device. This will depend on the tip speed ratio (blade tip speed to current
velocity), Mueller, (2005). The theoretical (Betz) maximum efficiency is 0.59 with
practical values lower than this. The much greater density of water results in a tidal
turbine being able to generate comparable amounts of power to a wind turbine with,
for example, 20% of the diameter of a wind turbine and in a current of 30% of the
wind speed.
The gravitational attractions of the moon and sun, as the earth rotates, generate
local tidal currents converting potential into kinetic energy. The magnitude of the
local tidal range is influenced strongly by the local seabed bathymetry and shoreline
orientation. The local behaviour of tides is controlled by dominant semidiurnal
(period 12hrs and 25 mins) and diurnal periods and with monthly variations in
maximum and minimum range (spring and neap tides). Predicting the time of local
high and low tides and to a lesser extent their magnitude has occurred for many
centuries. It is this predictability that makes the use of tidal energy attractive. The

23

key is the behaviour of the local tidal range, back flow and secondary induced flows.
It is usual that tidal currents will be most strongly associated with close-to-shore
locations that ease the problem of connecting the generating system to the local
electricity network. Although tidal energy is the main source of marine currents,
additional oceanic (geostrophic) currents such as the Gulf Stream could also offer
possible locations although these will more usually be associated with deep water.
The amount of energy that can be extracted from a given location will be associated
with the appropriate selection of possible type/capacity of the extraction device. For
example, estuarine barrage schemes are able to extract significant amounts of
power; the La Rance scheme in France has a generation capacity of 240MW based
on a basin area of 22km2 and a tidal range of 8.55m. However, other types of device,
principally variations on vertical or horizontal axis rotating systems, will only be able
to extract energy associated with that of the mean current within their capture area.
For horizontal axis machines, this will be controlled by their blade diameter and for
vertical axis machines by their height and diameter. The complete reversal of flow
will result in times when the local current is below a critical value for viable
operation of the device. In addition the tidal range will possibly constrain a
minimum water depth in which the device can operate and likewise the magnitude
of the current will drop in rough proportion to the increase in local water depth.
6.1.3 HATT Concepts
There are four main strategies for operating HATT:
(1)
Fixed rpm, fixed pitch:
(2)
Fixed rpm, variable pitch: as tidal current increases, control pitch to
maximise energy capture within the rated power of the generator.
(3)
Variable rpm, fixed pitch: as tidal current increases, control rpm to
maximise energy capture within the rated power of the generator.
(4)
Variable rpm, variable pitch: as tidal current varies control both rpm and
pitch to:
a. Maximise energy capture within the rated power of the generator.
b. Maximise power within generator set limit.
The variable direction of the tidal current; primarily reversing by 180 degrees
through the tidal period gives the following design options:

Ideally to maximise power capture, variable pitch devices should be able to

azimuth the complete nacelle about the vertical axis to face the tidal flow.
Other options are to weathervane whole turbine system or have fixed
orientation at a given site, chosen to minimise yaw influence throughout tidal
cycle, and be able to change the pitch of the blade by 180 degrees or more to
maintain a single direction of rotation. The latter option will however require a
symmetrical blade section.

Fixed pitch devices can also either azimuth the complete nacelle about the
vertical axis to face the tidal flow or weathervane the whole turbine system.
Alternatively the nacelle could have a fixed orientation at a given site, to

24

minimise yaw effects over the tidal cycle, with the blade section shape
designed to work as a bi-directional device.
An interesting design choice for bi-directional devices is whether to use a high
performance asymmetric section with much poorer performance when working in
opposite direction flow or use a symmetrical section that performs the same in both
directions.
6.1.4 Available computational tools and analysis
The following computational tools were used by WUMTIA during the analysis:
Blade Element Momentum Code (BEM) - existing Fortran code developed and
validated for predicting performance of stall regulated wind turbine blades as
a series of DTI/EU contracts up to the mid-90s.
PropGen - an in-house code to automate the process of defining a rotational
propeller or turbine blade originally developed to produce marine propeller
geometry suitable for use with Palisupan and other codes, and modified to
generate a horizontal axis wind turbine/tidal turbine with a final prescribed
wake pitch specified by the user.
Palisupan - a surface panel code developed by Turnock initially to investigate
ship rudder/ propeller interaction which provides a reasonable compromise
between computational effort and physical accuracy in modelling the flow
interaction achieved by the use of a lifting surface panel method.
Adaptflexi - a powerful geometry manipulation system that its purpose
designed to generate high-quality surface panels and RANS (Reynolds
Averaged Navier Stokes) meshes. Palisupan is integrated within its
environment to automate a geometry definition and post-processing analysis.
Panvise - a visualisation tool matched with Adaptflexi and Palisupan.
Cp Comparison

Figure 6.1
0.5

0. 45

Fixed

0.4

0. 35

Variable

0.3

Cp

The study then analysed a

generic tidal cycle, derived a
simple turbine blade geometry
sufficient for this stage of the
analysis (blade optimisation
was
addressed in Work
Package 2), identified the
relevance of cavitation to the
performance of the selected
blades and carried out a high
level performance comparison
between the fixed and variable
pitch devices. The results of
this comparison are shown in
the
adjacent
Figure is
6.1.
[The
above figure
not

0. 25

0.2

0. 15

0.1

0. 05

0
0

10

TSR

25

12

14

16

18

6.1.5 Output of Study 1

The work demonstrated the ability to predict accurately the performance of a HATT
using a lifting surface panel method (LSP). This study included: an investigation of
the influence of the sensitivity to the numbers of panels used, developed
methodology for a fixed wake using BEM analysis and investigated sensitivity of
results and validated predictions against experimental (full and model) scale data for
comparable wind turbine tested in early 1990s.
The influence of blade number and pitch was analysed using an existing BEM code
originally developed for use with horizontal axis wind turbines and compared with
results of LSP. Good agreement was obtained when compared with published
theoretical and experimental results. A parametric approach was developed for
generating a computational model of a HATT suitable for use with LSP or with BEM.
A performance analysis tool was produced which is capable of using the nondimensional power performance curve of a HATT and calculating total power
delivered for particular tidal conditions eg spring maximum and ratio of spring/neap
tides. This has included evaluation of the performance of the following devices:
Unidirectional fixed pitch device;
A typical fixed pitch design or equivalent variable pitch device;
A theoretical device which can maintain constant Cp;
A theoretical device that reduces rpm to limit maximum power.
The interim conclusion was that the concept of using a fixed pitch bi-directional
device, when combined with generator system that can use a variable range of rpm,
offers comparable performance to that of a more sophisticated and inherently more
mechanically complex system that uses a controllable blade pitch and azimuths to
face the tidal flow.
6.2
Design of three bladed tidal turbine blades for bibi- directional fixed pitch or
azimuthing variable pitch operation
6.2.1 Optimisation Study
The second stage of analysis by WUMTIA was in essence an optimisation study for
the blade design of a 20 m diameter three bladed HATT, making further use of the
tools developed and validated in the first stage. The design goal was to maximise
power output over a complete tidal cycle for an assumed location with a maximum
spring mean current of 2.5 m per second. The turbine was to be sited in water of 40
m depth with an axis of rotation at 20 m depth. The analysis included blade
selection and development of the design strategy, taking account of the explicit
constraints imposed as a result of other structural and environmental
considerations, in addition to assumptions introduced by other partners. The results
consisted of two blade designs in terms of section shape and variations in blade
twist, thickness and chord with the radius from root to tip.

26

There are limitations imposed by the scale of the study and for the devices in
question there is only limited operational experience available so the design
constraints associated with, for example, marine growth had to be estimated. The
objective of the optimisation study is to maximise power output through a complete
tidal cycle. As shown in the previous study, power is proportional to the cube of
tidal current speed and so the highest possible power is only generated for a limited
amount of time.
One of the constraints of HATT arises from the relatively low stream kinetic energy
extraction efficiency. The most advanced devices operating today are targeting an
efficiency of around 40-45%; however the Betz theorem limits efficiency to 59.6%.
The following strategy was adopted for selecting viable blade designs: design and
select section, optimise planform/pitch/blade setting, evaluate selected blade
performance, evaluate the blade loading, modify section thickness, modify blade
root and, finally, check performance.
The main constraint on selected blade shape is the ability to withstand the applied
design load, the control of which is through selection of a suitable thickness/chord
ratio. Altering the blade taper can control the root bending moment. A further
constraint for the variable pitch device is the method by which the hydrodynamic
shape is merged into the necessary circular cross-section at the blade root to allow
for blade pitch change. It was only possible to make a rough assessment of the
likely performance degradation of the proposed new bi-directional sections with
cavitation number.
Experimental evidence suggests that tidal turbines may experience strong and
unstable sheet and cloud cavitation, and tip vortices at a shallow depth of shaft
submergence. Acceptable levels of cavitation on marine current turbines are not yet
clear. Another factor to be considered is how different blade materials will respond
to possible cavitation erosion.
The tidal cycle performance analysis program, developed as part of Work Package 1,
was further modified to include a final control strategy. This strategy assumes that
the rate of revolution of the blade can be modified within prescribed rpm limits,
typically plus or minus 50%, to find the tip speed ratio that gives maximum delivered
power. A further refinement is that this can be modified to limit power output at a
rated power, typically chosen to be 10% above rated value.
6.2.2 Blade Parameters and Optimisation Tools
The optimisation study required a series of suitable shape definition parameters. A
reasonable blade shape for a tidal turbine, based on a wind turbine, had been found
in the first series of investigations. The method of parametrically adjusting this
shape provided a reasonable approach for determining which blade shape
parameters should be chosen.

27

The basic BEM code, cwind, used in the first phase, was extensively reworked and a
series of 9 variants (tidal_bemv1:9) were developed to facilitate an automatic
optimisation approach.
A spreadsheet was developed to allow the full geometric shape definition required
for the surface panel code to be developed from the same parameters found in the
optimisation study. The shape was specified at 10 spanwise stations starting at an
initial hub/blade diameter ratio of 0.2. A later study confirmed that only small gains
in performance are obtained if smaller values are used (2%). In a real design this
region is primarily structural and contributes a limited amount to performance.
The overall pitch setting of the blade, along with the blade tip speed ratio (TSR),
determines the thrust and power developed. These were dimensionalised in terms
of, for this study, three blades with a maximum diameter of 20m. Blade chord, pitch
and thickness were required for the blade at each spanwise station. Together with a
specified section sha pe, known for the variable pitch blade and derived for the bidirectional blade, this allowed a ruled shape to be constructed. Variations in shape
were controlled by modifying the blade taper, area, pitch and thickness.
The optimisation process explored the geometric design space as well as the range
of TSR and blade pitch settings. The execution of the analysis required an iterative
solution to capture the necessary operating angle of attack for each section. This
was accomplished, resulting in a simple strategy of exploring each variable at equal
separation between a minimum and maximum constraint. Typically, such a global
search created of the order of 33,000 evaluations. Once an optimum had been found,
2 or 3 additional cycles of user specified progressive refinement was usually
sufficient to finalise the design. The large number of evaluations gives a clear
indication of the sensitivity of the design space to the various choices of design/flow
variables.
6.2.3 Bi-directional fixed pitch tidal turbine blade
The most important component of the bi-directional concept is the identification of a
section shape that works equally well in both forward and reverse tidal directions.
Such a shape requires 180deg rotational symmetry about its mid-chord. Previous
work at the University of Southampton has developed a similar concept for use with
rim driven marine thrusters.
In this work, the 2D section analysis code, Xfoil, was used to analyse the
performance of a range of angles of attack up to the initial onset of stall (separated
flow). This section lift and drag performance is required by the BEM code. A
sensitivity study was carried out for the effect of thickness/chord ratio (t/c) on lift and
drag. It was found that the section shape performed best at its design t/c. In
principle, a section shape could be optimised for each of the required spanwise t/c.
This was beyond the resources of this study but would be required for a final design.
The analysis carried out of this numerically developed bi-directional section does not
allow prediction of performance beyond stall. This information is important in being

28

able to predict such performance parameters as turbine starting torque and runaway
speed. However, section performance beyond stall can be extrapolated based on
knowledge of measured section performance. It is well documented for wind
turbines that experimentally measured 2D stall parameters under-predict the section
lift beyond stall on a real blade. This is because the rotating blade causes spanwise
flow within the stalled region on the downstream face of the blade. This effect was
measured in the early 1990s at the University of Southampton and was used to
make an estimate of the likely behaviour beyond stall. For a limited number of
cases, the sensitivity of this assumption was assessed by varying the values used in
the section performance curve.
Figures 6.2 and 6.3 show the performance (Cp vs. TSR and Ct vs. TSR) for the final
blade shape. The shape of the curve reflects the performance of the section with a
more limited range of operation without stalled flow. However, at the maximum
power condition (N=17rpm, U=2.5m/s) all of the sections are operating on or close to
their optimum angle of attack (eg max lift/drag ratio). Figure 6.4 shows the torque
generated with a 2.5 m/sec current.

Power Coefficient, Cp

Structural analysis was assessed using the surface panel code. Figure 6.5 shows a
typical output indicating local surface pressure variation. This can be imported
directly as a load map for use with a commercial finite element analysis code.

0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0

6
TSR

Figure
Figure 6.2 Cp performance of Optimum Bi-directiona l blade

29

10

Thrust Coefficient, Ct

0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

TS R

Figure 6.3 Ct Performance of optimum blade

0.7

Torque, MNm

0.6
0.5
0.4
0.3
0.2
0.1
0
0

10

15
RPM, N

Figure 6.4 Torque generated with 2.5 m/s current

30

20

25

10

A
simpler
beam theory
approach was used to find the
appropriate
spanwise
t/c
variation. It is worth noting
that the design loading case is
driven by the maximum
spring tide (rather than
extreme events as required
for
wind
turbines).
Hydrodynamic
loads
(primarily
in
the
flow
direction)
rather
than
centripetal loads dominate in
the radial direction.
This
allows greater mass to be
placed at higher radii and
reduces the need for small t/c
towards tip. As a result the t/c
Figure 6.5 Pressure map
distribution
for
the
bidirectional blade can have a much smaller variation and the section will be operating
in its known range.
The surface panel analysis also provides a check of performance against that
predicted by the BEM code. Definition of the correct downstream turbine wake is
important for obtaining satisfactory performance. Fortunately, BEM analysis can
supply most of the necessary information to construct such a wake. The
construction of a suitable LSP mesh requires considerable care to achieve a high
quality mesh with regular quadrilaterals that control panel size, aspect ratio, low
skew, lack of planarity and proximity. The low pitch of the blade and its intersection
with the assumed hub provided a considerable challenge.
6.2.4 Variable pitch turbine blade
A similar approach was followed for the variable pitch blade with the
planform\optimisation resulting in a significant improvement in performance giving
a maximum Cp of 0.49, Figures 6.6, 6.7 & 6.8. Figure 6.9 shows the optimum pitch
angle setting as the tip speed ratio varies.

31

Power Coefficient, Cp

0.6
0.5
0.4
0.3
0.2
0.1
0
0

10

15

20

25

30

20

25

30

40

50

60

TSR

Figure 6.6 Cp performance of Optimum Bi-directiona l blade

Thrust Coefficient, Ct

0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

10

15
TSR

Figure 6.7 Ct Performance of optimum blade

0.8

Torque, MNm

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

10

20

30
RPM, N

Figure 6.8 Torque generated with 2.5 m/s current

32

25

Blade Pitch, deg

20
15
10
5
0
-5

10

15

20

25

30

-10
TSR

Figure 6.9 Optimum Blade Pitch Setting

Note: Discontinuities in the performance curves are associated with the discrete
jumps of 0.5 degrees in the pitch setting used.
6.2.5 Yaw Characteristics
Many tidal sites are relatively bi-directional, however, some sites can have flow
reversal of 20o or more away from 180o such as the flow around islands and
headlands, for example Portland Bill where a swing upon flow reversal of around 35o
from rectilinearity is present. It is thought that if a yawing turbine, rather than a
fixed turbine, was to be used at unusual sites such as Portland Bill an extra 10% of
energy may be harnessed.
Experiments have been carried out to measure power and thrust on a model marine
current turbine under yawed flow conditions. Such experiments show that there is a
consistent decrease in power and thrust with an increase in yaw angle. Optimal
orientation for any turbine is based on the design flow speed.
Momentum theory suggests that the power is proportional to the square of the
cosine of yaw angle () and the thrust as a cosine of the yaw angle:
C Pow = 4a (cos a)2
C T = 4a(cos a )
Where:

CPow is the Power coefficient, C Pow =

CT is the thrust coefficient, C T =

T
0.5R 2U 02

a is an axial flow factor

Uo is the tidal speed (m/s)
33

QR
0.5R 2U 03

 is the rotation speed (rad/s)

is the density of water (kg/m3)
Q is the rotor torque (Nm)
T is the rotor thrust (N)
Batten et al demonstrated that the data attained from experiments fits to the
momentum equations and the cosine square rule. Thus the curve fits obtained may
be confidently used to compare the effect of fixed and yawing turbines at different
locations.
In this study the power comparisons have made an implicit assumption that no such
variation exists and that the two principal tidal directions are 180 degrees apart.
However, as over 60% of the power is developed when the tidal current is between
2.0 and 2.5m/s and is likely to be closely aligned with the axis of the tidal turbine (
small), this is deemed reasonable.
6.3

Comparison of variable and fixed pitch design

6.3.1 Blade power curves

Figure 6.10 plots the power curves of both optimised blades as well as the variable
pitch blade working at a constant pitch of -5o. It is worth noting that the bidirectional blade has an 8 % increase of blade surface area. This reflects the poorer
section performance and is then seen as a drop in power capture over a lunar cycle.
However, this drop in performance is nowhere near as bad as would be first seen
from the large decrease in area beneath the curve of the bi-directional device. This
is associated with how much of the tidal cycle a device can operate at a particular
TSR. If a suitable generator control can be implemented the fixed-pitch bi-directional
device can be made to operate at or near its optimum TSR. The range of effective
TSR operation is significantly lower than for the more robust aerodynamics of the
conventional variable pitch device but the power captured will not be. As an
example of this, for the variable pitch blade operating at a constant design rpm it is:
i)
TSR 30.0 (slow current, <0.55m/s) it is stationary for 21.4% of lunar tidal
cycle and generates no power
ii)
TSR 20.0 (modest current, <0.9m/s) it operates for 13.5% of time but
generates 1.2% of power
iii)
TSR 10.0 (reasonable current, <1.78m/s) it operates for 44.2% of time and
generates 36% of power
iv)
TSR<10.0 (high current, <2.5m/s) it operates for only 20.9% of time but
generates 62.8% of total power.
The control ability to move the operating rpm to give at or near optimum TSR will
allow more power to be generated in current regimes ii) and iii).

34

Power Coefficient, Cp

0.6
0.5
0.4
0.3

Variable Pitch

0.2

Bi-directional

0.1

Variablecons tant pitch

0
0

10

15

20

25

30

TSR

Figure 6.10 Comparison of blade power capture

Table 6.1 below summarises the energy capture based on these performance curves
for a series of control strategies. It is assumed that a 10% exceedance of the rated
generator capacity is acceptable.
Rated
BI-DIRECTIONAL
Diameter Power
DESIGN
+10%
m,
MW
Fixed RPM
Variable Fixed
[hub/D
(14.0)
RPM
RPM
=0.2]
(17.0)
Variable
Pitch
20
1.1
114
160
169
20
0.75
90
151
158
20
1.25
114
161
169
Table 6.1
days)

VARIABLE PITCH DESIGN

Variable
RPM,
Variable
Pitch
177
163
178

Fixed
RPM
(17.0)
Constant
Pitch
166
156
167

Variable
RPM
Constant
Pitch
177
163
177

Comparison of various control strategies (energy in MWh over 29.4

The total capacity is taken to be the rated power over 29.4 x 24 hours. For the three
rated generators of 1.14, 1.0 and 0.68 MW the energy they could have absorbed
would be 802, 704, 481 MWh respectively.
For the purposes of the economic
assessment hydrodynamic capacity factors of 23% and 25% were used for the bidirectional and variable pitch devices; based on a rated 1MW generator capacity (802
MWh) and 20m blades. Hydrodynamic capacity factors assume 100% reliability and
100% availability. It is worth noting that a 1m increase in blade diameter will allow
the bi-directional device to capture the same amount of energy as the best
performing 20m variable pitch devices.

35

6.3.2 Axial Load

One of the effects of a less efficient blade is an increase in axial load (more energy is
transferred into axial load rather than into rotating the blade). The maximum axial
load coefficient (Ct) is 0.83 for the bi-directional blade (and constant pitch setting of
variable pitch blade) but the ability to pitch the blade allows a value of less than 0.8
to be achieved at optimum TSR for the variable pitch device.
If both blades are assumed to operate at maximum Cp for a maximum current of
2.5m/s the increase in axial load will be 3.75% for the bi-directional device. For the
proposed 1m increase in diameter this would give an increase in axial load of 14.4%
overall. This would need to be reflected in greater strength of the support structure
and seabed foundations.
6.3.3 Summary
A number of areas of uncertainty arise from the exclusive use of numerical analysis
tools, which in general relate to the actual section performance when either stall or
cavitation occurs. Use of suitable model scale experiments will reduce this
uncertainty. The recommended strategy would be 2D section performance tests in a
cavitation tunnel to evaluate the drop-off in section performance with varying
degrees of cavitation. This would then need to be compared with results of three
dimensional blade tests in a circulating water channel or towing tank. Although
these would be at model scale they would allow the necessary modifications to the
2D performance curves to be made to give a calibrated 3D performance. Scale
effects can then be accounted for in the usual way.
A final area of uncertainty is associated with the blockage effects associated with the
turbine operating in a finite depth channel with a free surface. The interaction of
free surface waves (onset and radiated/diffracted) as well as the change in free
surface controlled by the turbines extraction of energy can again be approached
based on experiments. It is likely that blockage for the proposed design will have a
limited effect although this will depend on the tidal range at a given site as well as
the local site bathymetry. This is an area of considerable research interest.

36

7.

MECHANICAL AND ELECTRICAL SYSTEM DESIGN OPTIONS

7.1

Device Block Diagrams

Generic block diagrams have been developed to illustrate the main relationships
between the various elements of the fixed-pitch and variable-pitch approaches.
These relationships have been quantified to allow important design considerations
to be identified, e.g. cooling requirements and size of components. The purpose of
the block diagrams is to indicate the main generic influences exerted by each
element on the others, without causing the diagrams to be specific to a particular
solution. For example, the heat flow from the generator to the nacelle volume may
in practice take place as a fluid flow of air circulated by the generator cooling-fan, if
the machine is so equipped.

Atmosphere

SEA
Rotor
blade

Main
shaft
seal

Nacelle
volume

Pitch
joint

Rotor
shaft

Main
gearbox

Hub
Bearings

Main
shaft
brake

Nacelle
switchgear

Cable

Generator

Shore

Hydrodynami c ef fect

Ancillaries

Mechanical power
Force / t orque
E lectric power
Heat t ransf er
Fluid f low

Nacelle
structure
Figure 7.1

Rotator

Pylon

Base

SEABED

Variable--Pitch Machines
Initial Conce ptual Block Diagram for Fixed and Variable
Components specific to
variable pitch device

Components differing between fixedand variable pitch devices

The block diagrams do not represent all possible solutions, e.g. where a generator is
designed for direct drive. Each device includes components and systems not
directly part of the main power flow path, but important for practical operation.
Some of these features are listed below. The list is not exhaustive; some of the items
are optional, some complementary, or alternative means to achieve the same end. In
some cases the details of the equipment described by a particular heading will differ
between the fixed pitch and variable pitch systems. This study assumes common

37

components between fixed pitch and variable pitch designs unless commonality is
impractical or implicitly precluded.
Examples of ancillary components and systems that might feature in a tidal current
turbine include: rotor brake (capable of stopping rotor), rotor lock (prevents rotation
of already stationary rotor), bilge pump, nacelle pressurisation system, rotor shaft
seal, inflatable rotor shaft seal, cathodic protection system, anti-condensation
heater, navigation/anti-collision lights, control system, deployment/retrieval related
systems (e.g. winch, ballast pump), condition monitoring system, nacelle rotator
system and brake. Some of these components are indicated explicitly on the initial
system block diagram above, whilst others could be considered to be included in the
Ancillaries block.
An un-shrouded tidal turbine for 1MW power output is unlikely to turn faster than 20
rpm, and may well turn more slowly. Most conventional e lectric generators require
an input shaft speed two orders of magnitude higher. Appropriate arrangements
must therefore be provided to generate electrical power in a suitable form for
onward transmission. There are several ways in which this problem may be
addressed.
For both fixed and variable pitch devices, the main power flow is from the tidal
stream to the grid, via a rotor, transmission, generator and electrical interface
(generator and some form of power conditioning). These items are analogous with
early wind turbine experience and there are several approaches for these
components, provided the concepts for transmission, power conditioning and
generator are technically consistent. The concepts for a practical fixed pitch and a
practical variable pitch device considered in this report may not be strictly optimal
for a practical machine, but will be reasonable initial assumptions. The main options
for the transmission and generator type are listed below.
Possible Gearbox / Generator
Combinations

TwoTwo-stage gearbox
(ratio 1:6 4 max)

ThreeThree-stage gearbox
(ratio 1:512 max)

Unlikely
Possible
Unlikely

Feasible option
Feasible option
Feasible option

Induction (asynchronous)
DoublyDoubly-fed induction
Synchronous
Table 7.1

Possible Gearbox / Generator Combinations

Permanent magnet generators with direct drive or single stage gearbox options
were considered but discarded for the purposes of this project, but they may well
reappear as permanent-magnet generators become better developed. The main
issue related to the control of the permanent magnet generators under fault
conditions; necessary increases in rating for the power electronics to cope with the
higher voltages generated during overspeed could undo the economic premise of
the fixed pitch device.

38

The induction generator was chosen to avoid the need for regular brush
replacement, which would be an intrinsic feature of either a suitable synchronous
generator or the doubly-fed induction generator options.
The use of fixed speed, directly connected, induction generators for the variablepitch option was considered. However, the reduction in revenue (due to the lower
capacity factor) together with the need to purchase reactive power from the grid
made this option less attractive than using a variable pitch, variable speed
generation system with power conversion equipment. Hence for the variable pitch
turbine, the studies were conducted using variable speed turbines with induction
generators connected to the grid via power conversion equipment.
7.2

Generator/power conversion concepts

At the very outset of this project, it was thought that there would be a significant
difference between the generator and power conversion systems required for the
fixed pitch and variable pitch devices, with the system for the variable pitch device
likely to be analogous to the doubly fed induction generator systems used for
modern wind turbines. The system for the fixed pitch, bi-directional, variable speed
device was expected to require more complex, and expensive, power conversion
electronics.
In discussion with Converteam, however, it became clear that the same generator
and power conversion system approach could be applied to either the fixed pitch or
the variable pitch device. A number of different systems were considered in
conjunction with Converteam, including the following:

Option
The first option was a simple system consisting of a single
Option 1
turbine, AC generator and power converter with a step up transformer. This
arrangement is usually used in near to shore, low-power projects, and would
be suitable for a pilot project or technology demonstrator. It would be a poor
choice for a farm as there is no opportunity to take benefit from the common
costs of a farm installation.

Figure 7.2 Option 1 Single System

39

Option 2 The second option, involving multiple generators each with a dedicated
power converter, would provide a high level of system flexibility, allowing
generators to rotate at different speeds as well as individual generators to be
shut down. However the system is complex and it would be a high-cost solution
with no cable or transformer redundancy.

Figure 7.3 Option 2 - Multiple turbines, single system

Option 3 The use of multiple induction generators connected to a single power

converter is appropriate where there is little variation in turbine speed across the
farm. This is important; high circulating currents between the generators would
occur if the differences in speed were significant. In the absence of such large
circulating currents, this option offers a lower cost in comparison with the other
system configurations for a high-power farm. Again there is no cable or
transformer redundancy and in this case no inherent converter redundancy is
possible without a reserve converter.

40

Figure 7.4 Option 3 Multiple induction generators

Option 4
Under the active DC link option, each generator has a dedicated power
converter machine bridge that feeds a common DC link; a single network bridge then
feeds the transformer. This option is suitable for projects where generators are
located closely together but not necessarily rotating at the same speed. Individual
generator control is possible but if the converter network bridge needs to be shut
down, power generation ceases. The assumed farm capacity of 30MW would in
practice imply the use of two network bridges, each serving half the farm.

Figure 7.5 Option 4 - Active DC Link

Option 4 was selected as the basis of the study as representing the most economic
overall solution consistent with a practical layout of turbines on the seabed. In
summary, this will comprise rectifier equipment located in each turbine nacelle,
which will convert the variable frequency AC output from each induction generator
to DC at about 3.3kV. The DC output from each unit is connected via a common link
41

to a single network bridge that converts the output from the tidal stream farm to
50Hz AC. This network bridge may be located either on a dedicated platform located
centrally in the tidal stream farm, or onshore if the farm is close to land. The output
from this bridge is connected to the Distribution Network.
7.3

Farm scale architecture

A fundamental assumption is that tidal turbine devices will be deployed in farms

comprising a number of individual turbines sharing certain common facilities and a
common grid connection point. Whilst for the purposes of this study the assumption
is that a farm will include 30 turbines, it should be noted that no effort has been
devoted to determining an optimum farm size. The optimum farm size will be very
dependent on specific site conditions. It is, however, necessary to consider the
subject of farm scale operations briefly in order to establish the relative importance
of the differences between fixed-pitch and variable-pitch turbines.
7.4

Connection to Distribution System

Initial deployments of marine turbine devices are likely to entail up to 30 devices

with the nominal rating of devices constrained to about 1 MW per device by the
available water depth, although some developers are now considering 2MW
devices. The resulting farm capacity of some 30MW will be connected for export to
the local network either at 33kV (distribution) or 132kV (distribution in England and
Wales and transmission in Scotland). It has been assumed here that 33kV and 132kV
are part of the Distribution Network.
Normally up to 20 MW can be readily connected at 33kV, with up to 30MW possible
if the connection is to a particularly strong 33kV network. Due to the rural locations
likely to be chosen for marine turbine devices, it may be more appropriate to
connect the majority of farms over 20MW to the 132kV network. Connection at 132kV
will be more expensive than at 33kV and, as with all connections, the actual cost will
be location specific. As has been found with wind clusters and small wind farms, the
connection cost could become a critical item in the financia l viability of a marine
turbine project.
As typically described above, each marine turbine device will generate 1MW of
alternating current (AC) electricity and this will then be converted to direct current
(DC) as the AC will not be at fixed frequency. Cables operating at DC will then take
the output to a central, offshore hub where there will be an inverter to AC at a
frequency of 50 Hz. Also positioned at this point will be a transformer and associated
switchgear to raise the voltage to 33 kV or 132kV.
A 33 kV or 132 kV submarine cable will be installed to dry land where there will be
an export meter and synchronising switch allowing connection to the local
Distribution Network. At this point filters will be required to ensure voltage
fluctuation/unbalance and harmonics do not exceed levels laid down in National
Engineering Recommendations P28, P29 and G5/4. The operation of the generators
will also need to conform to G75 and G59/1. Each installation will need to be studied

42

by the relevant Distribution Operator for connection and the levels referred to in the
documents listed discussed and adhered to.
G75:

Recommendations for the Connection of Embedded Generating Plant to

Public Electricity Suppliers Distribution Systems Above 20 kV or with
Outputs over 5 MW
G59/1: Recommendations for the Connection of Embedded Generating Plant to
Public Electricity Suppliers Distribution Systems
G5/4: Limits for harmonics in the United Kingdom supply system
P28:
Planning limits for voltage fluctuations caused by industrial and domestic
equipment in the United kingdom
P29:
Planning limits for voltage unbalance in the United Kingdom

7.5

Deployment and installation considerations

The device mounting structures have been assumed to be either a monopile with
one or more devices per pile (sketches a & b below), or mounted on cross-arms or a
semi-submersible structure anchored to the seabed and allowed to swing with the
current direction (sketches c & d below) . The variable pitch devices mounted on a
monopile will have rotating mechanisms to turn the devices during slack water to
face the next tide tidal flow. The cost of the two installation methods has been
assumed to be similar, within the scope of this project.
There are numerous proprietary approaches to the provision of maintenance access
to tidal turbines, some the subject of patent applications or grants. Nothing in this
report should be regarded as a comment on any specific system. However, the
maintenance access systems may be divided into three broad categories, which will
have different costs relating to the type and size of boat/ship/crane and time taken to
carry out the operations:

43

Changing the de vice geometry

(a&b). The device may be mounted
on a tower fixed to the seabed, with
the tower extending above the water
level. The device is raised up the
tower and clear of the water (eg by
means of a jacking device) and may
then be accessed from a boat or
lowered onto a barge for transport
onshore. MCT is an example of this
approach.
Changing buoyancy (c&d). The
device is mounted onto a semisubmersible structure that can be
manoeuvred to the surface by
changing the buoyancy, where it may
be accessed from a boat or lowered
onto a barge for transport to shore.
The TidalStream device is an
example of this approach.
Detaching important
elements
from the seabed-mounted part of
the device (e&f). The device is
mounted on a pylon fixed to the
seabed, which does not extend above
the surface. The device is detached
from the pylon by divers or an ROV,
lifted clear of the water by a heavy lift
crane, and lowered onto a barge for
maintenance there or onshore. An
example is the Hammerfest Strom
device.
Figure 7.6 Maintenance access options

Maintenance of the tidal turbine with the device mounted in the operational position
will be affected by weather conditions, the mounting concept and resulting
accessibility. For the purposes of this project, we have assumed that it will be
necessary to remove the device from the mounting structure and either carry out
repairs to the device on a surface vessel, or return it to shore. In the latter case a
replacement device could be installed in the same operation.
A related issue is the provision of appropriate facilities to allow safe human access
to the interior of the installation. The project has recognised the importance from
both a cost and safety perspective of minimising the use of divers in the
construction and maintenance process but detailed consideration is outside the
scope of this project.

44

7.6

Operation and maintenance

In deriving a cost for the operation and maintenance (O&M) for a tidal stream
turbine device it has been necessary to make a number of assumptions since no
units have yet been installed anywhere on a commercial basis, and hence there are
no data bases or track records that can used. The scheduled O&M cost has been
calculated on the assumption that a tidal stream farm comprising 30 off 1MW
individual units is installed at a location 5km off the UK coast line, and that a suitable
overhead line connection is available close to the shore. Scheduled maintenance has
been based on an annual service visit to each turbine module with a major overhaul
every five years when the complete turbine nacelle would be removed from its
mounting and taken to a shore facility. Note, however, that costs of such operations
are modelled as constant annual fees that are applied as average per-period cash
flows.
The design of the turbine nacelle mounting system has not been defined as part of
this study, but the same support system is assumed for both fixed pitch and variable
pitch turbines, enabling a valid comparison between the maintenance costs of both
types of turbine to be made. To keep maintenance costs to an economic level, and to
minimise delays due to adverse weather, the turbine mounting support system
would be designed so that turbine nacelles can be brought to the surface and be decoupled from their mountings without the use of divers or a floating crane. However,
it is assumed that divers would carry out an inspection of each turbine support
structure every 5 years. This maintenance regime is seen as a conservative cost
basis. The intention would be to include sufficient monitoring equipment to enable
the annual maintenance visit and 5 year overhaul periods to be extended so that
scheduled visits to each turbine could be extended to once every 5 years and the
major overhaul period extended to 10 -15 years.
Control, monitoring and operation of the tidal stream farm is assumed to be from a
remote facility that is an existing generating plant and that the operators of that
plant could monitor/control the tidal stream farm without additional manpower
resources. A high level of SCADA equipment would be included in each tidal stream
turbine to ensure that all the important parameters were available both to operators
in the control centre and to maintenance staff.
For the purposes of this study the major mechanical and electrical components
housed in the nacelle, along with the rotor blades, were considered as a single
replaceable unit for the purposes of maintenance, and have been termed a Line
Replaceable Unit (LRU). In the event of a failure of a component in the LRU, or for
some aspects of scheduled maintenance, the LRU would be removed as a complete
unit by a support vessel and either repaired or returned to shore for more major
servicing. To maximise the availability of the complete system, the support vessel
could install a serviceable LRU to replace that removed for maintenance/repair.
The scheduled maintenance requirements for each turbine are based on typical
scheduled maintenance requirements for offshore wind turbine units with
appropriate modifications for tidal stream turbines. Estimates of the frequency and

45

costs of unscheduled outages have been calculated using the Monte Carlo
simulation method. This estimates the probability of failure for each type of turbine
and the time taken before the unit is returned to service (failures solely causing
degradation of output are not modelled). Using this system, and allocating an
average cost for each intervention to repair a failed component, the unscheduled
maintenance costs for fixed and variable pitch turbines can be calculated.
The total scheduled O&M cost for a tidal stream farm of 30 units is estimated at
1.125m p.a. for a fixed-pitch installation, and 1.209m p.a. for a variable-pitch
installation. The costs have been derived from maintenance cost data for offshore
wind farms, which operate in a similar environment. This equates to an annual cost
per LRU of 5% of initial capital cost. The unscheduled outage costs for a tidal stream
farm have been estimated using North Sea Oil installation reliability data. The data
were used to provide an estimate of the number of failures, and thus number of
interventions, per period.
Unscheduled maintenance costs are modelled as a percentage of LRU cost (to reflect
an average cost of repair) plus a fixed fee per intervention (to reflect any special
charter/hire provisions or specialised crew fees which may be applicable). These
figures include the cost of intervention for unscheduled outages and the loss of
electricity production for the unit downtime. Overall availability is modelled as
scheduled availability (95% used for all cases) multiplied by unscheduled availability,
which is derived from the Monte Carlo simulation. This depends on device reliability
and repair probability, which in turn was based on the probability of suitable sea
conditions (when the significant wave height was less than 1m).
7.7

Decommissioning considerations

It is almost inevitable that the landlord for any of these devices around the UK will
be the Crown Estate, who will include as a requirement of the lease or licence an
obligation to decommission the offshore installation, but not necessarily the cables,
on completion of its useful life. This requirement is in the process of being
reinforced through a DTI consultation. On the assumption that the mounting
structure will consist of a monopile, the requirement is likely to be to remove the
structure to a level some distance below that of the surrounding seabed. This
requirement will of course be common to both the fixed pitch and variable pitch
devices. A notional sum has been included in the cost model.
7.8

System design
design assumptions

The list below details the main assumptions used in the performance of this project,
in addition to those in the Background above.

46

Assumption
Fundamentals
Output takes the form of
electricity.

Output is compatible with

the UK grid system
Devices are located in
identical conditions in
Crown Estate waters.
Tidal Stream Conditions
The tidal stream will vary
over the semi-diurnal ebb
and flow cycle with a period
of 12.4 hours and a
fortnightly spring /neap
period of 353 hours
Device in tidal stream with
uniform velocity over the
area swept by the blades.
The maximum tidal stream
velocity at the site is 2.5 m/s
Turbine axis is aligned in
both directions with the
tidal direction
System Design
Shore equipment 100m
from the shore high water
mark
Electrical connection to
33kV 200m from shore
equipment
Array configuration does
not affect device
performance
No location constraints
Device performance
identical for forward and
reverse flow
Identical accessibility
Corrosion protection free

Remarks
Exclude outputs in the form of mechanical
power, compressed gas, hydraulic power, heat,
electrolysed hydrogen, desalinated water,
chemical reaction product or other energy
vector or commercially valuable commodity
50Hz, appropriate voltage & frequency limits etc.
for compliance with Grid and Distribution Code.
Neglects variation that may occur between
individual sites in a farm. Implies that applicable
data for UK conditions are readily for the study.

Simplifying assumption intended to make the

study conclusions useful for a wide range of
potential UK sites. Ignores effect of solar
precession cycle.

Simplifying assumption that will generally not

be true in practice owing to vertical velocity
gradient and wave circulation effects.
Assumption intended to make study conclusions
useful for a wide range of potential UK sites.
Simplifying assumption that will generally not
be true in practice but will provide a reasonable
first approximation for many potential UK sites.
To isolate report conclusions from effects of
unusual site-specific features such as high cliffs,
extensive areas of marshland, etc.
Relates to strong 33kV connection for assumed
farm size of 30 devices of 1MW each
Simplifying assumption, meaningful analysis of
actual array effect requires site-specific
information.
Simplifying assumption, constraints sitespecific.
Simplifying assumption.

Maintenance access issues are identical for fixed

and variable pitch devices.
No separate allowance for cost of replacement

47

of sacrificial anodes or impressed-current

protection.
Business & Commercial
Cost elements are identical
for fixed and variable pitch
systems except where
explicitly defined otherwise.

Cost estimates for series

production rather than
development
No special offset

Analysis & Mathematical

Modelling
Cavitation effects negligible
Reliability data applicable

Overall system economics will have sensitivities

to maintenance method, scale of farm etc., but
fixed and variable pitch systems will use same
values for each variable unless an intrinsic
feature of the fixed or variable design precludes
this.
Development costs several times higher; likely
to preclude viable business case.
Business model does not include special
sources of finance e.g. DTI Wave and Tidal
Stream Energy Demonstration Scheme.

Detailed analysis of these effects is beyond the

scope of the study.
Assumed that any reliability data used are
representative of performance of component or
system in tidal turbine context

48

8.

COMPARATIVE SYSTEM COST ANALYSIS

ANALYSIS

8.1

Cost Analysis Model Methodology

The development and use of tidal turbines will require economic justification at a
variety of levels. The development of the tidal turbines and their installation
methods will require Original Equipment Manufacturers (OEM) to be satisfied that a
viable market exists for their products. At the other extreme the electrical utilities
that will buy the output and potentially develop, own and operate the tidal farms
will need to be satisfied that the electricity produced by the tidal turbines can be
utilised for an economic return in an auditable manner. If this utility business
rationale is not satisfied then the market will not develop and the OEMs will not be
attracted. For the purposes of this study the business model has been created from
the electrical utility viewpoint, with the sole revenue source for the business deriving
from the sale of electrical energy into a grid system and the relevant renewable
financial support/incentive mechanism.
The purpose of this study is to assess the relative economic performance of two
broadly similar devices that are physically different in detail. It is thus clearly
necessary to consider all factors having a significant impact on the cost of the
devices and on the revenue earned by the devices. Since the devices only generate
revenue in the context of a working farm, it is also important to appreciate costs
and benefits that are common, or broadly similar between the two devices. These
additional costs and benefits affect the sensitivity of the study result to the
differences between the machines.
Reliability and availability simulations, together with cost estimates for capital and
maintenance costs, are fed into a spreadsheet, which provides a comparison of cost
and revenue performance for each machine in the context of a farm of a specified
number of machines installed according to a user-defined schedule and (if so
desired) automatically replaced at the end of their useful life (which may differ
between the fixed- and variable-pitch machines). Costs common to both machine
types are included at this stage. The spreadsheet model run duration was chosen to
be 30 years to encompass the assumed 25-year mounting life and so to allow costs
for decommissioning to be included.
Fault Tree Analysis (FTA) was chosen in preference to Failure Modes and Effect
Analysis (FMEA) for the reliability elements of this study because it was considered
easier to construct a generic analysis. To use FMEA, it is necessary to consider in
detail each failure mode of each important component, which implies a level of
detail design definition inconsistent with the aims of this study. The FTA technique
allows a generic failure to be described without detailed consideration of the precise
nature of the failure.
The block diagram at Figure 8.1 below describes the solution process followed.

49

Figure 8.1

Block Diagram of main information flow during study

VARIABLE
FTA1
Reliability

RMC1

BOTH
Gearbox
Generator
Block diagram 2a

FIXED
FTA1
Reliability

RMC1

Accessibility
Assessment

Yield
Availability
Maintenance

Costs & business

assum ptions

Yield
Availability
Maintenance

Cost Comparison
GREEN items refer
to data for
VARIABL E pitch
m achine

PURPL E items refer to data

common to both m achines

KEY:
FTA1 = Fault tree analysis spreadsheet
RMC1 = Monte Carlo analysis spreadsheet
Blockdiagram2a = Generic tidal turbine system
initial design tool
Gearbox1 = Epicyclic gearbox initial design tool

50

RED items
refer to data for
FIXED pitch
m achine

Accessibility
Assessment
Installed
machines will not
always be
accessible/repaira
ble owing to
weather
conditions

8.2

Reliability and Availability Simulation Methodology

The simulation model used to predict reliability and availability levels for tidal
stream turbine farms is based on the Monte Carlo method, described in Appendix
A. The result of any Monte Carlo activity is always a simulation of the process, not
an analysis of it. In principle, two successive Monte Carlo simulations of the same
event with identical initial conditions will produce different answers. These answers
will, however, be such as to permit the estimation of a true value within some
confidence interval.
Monte Carlo techniques are generally most suitable for situations where the
probability equations for the process are intractable or, possibly, not capable of
explicit formulation. The complexity of this approach increases geometrically with
the probable number of failures during the device lifetime. The explicit calculation
method also becomes more complex if the probabilities of each event are
themselves functions of some other variable.
A situation that was considered as part of this study was the effect of seasonal
weather conditions on the repairability of the device, assessed for the purposes of
the study by assuming that repair probability PUA is a function of elapsed time. The
underlying assumption is that whilst the possibility of repair depends on the
availability of appropriate spare parts and other resources, an overriding
consideration is simply whether the device is accessible in order for the repair to be
performed. For example, a device which fails at the onset of winter is likely to be out
of commission longer than one which fails identically at the height of summer,
simply because weather conditions preclude maintenance access to the former
device.
For this study, several mode lling approaches were considered before construction of
the actual models. All the models use the concept of time measured in periods,
with values chosen to be of significance to the desired outcome. No specific
timescale is explicitly defined for a period; it needs to be long enough to allow
modelling of a meaningful proportion of the device lifetime, but short enough to
capture detail. An instance of unserviceability can never be shorter than a complete
period, so use of excessively long periods will lead to unrealistically low estimates of
device availability. It is vital that values of transition probabilities used in the
simulations are scaled to match the period chosen for the series of experiments. For
this reason, use of excessively short periods not only results in extra computer time,
but also implies that failure probabilities (as entered) become negligibly small.
Typical period and experiment durations used for this study included:
240 periods (20 years with 1 period 1 month)
312 periods (6 years with 1 period 1 week)
1560 periods (30 years with 1 period 1 week)
The reliability of the fixed and variable concepts was assessed using data taken from
the OREDA handbook (Ref. 10). This handbook is a DNV publication providing
statistical data from Offshore Oil & Gas operations (ENI, BP, Exxon, Norsk Hydro,
Phillips, Statoil, Shell, TotalFinaElf). The data are considered to be High quality

51

reliability data for both topside and sub sea equipment, giving failure modes,
observed number of failures, failure rate, uncertainty limits etc. Data are given as
number of failures in 106 hours, for mean, upper and lower values (90% confidence).
An example of the OREDA failure rates and the overall calculated failure rates for the
concepts is shown in Table 8.1 below.
OREDA Data
Fixed
52
143
3

Mean
Max
Min

No of failures *
Variable
71
194
5

Unplanned Availability
Fixed
Variable
92.6%
90.1%
81.1%
74.5%
99.6%
99.4%

*Mean number of failures per device over 30 years, determined from the Monte
Carlo simulation. This was converted, assuming a constant failure rate, to a number
of failures per turbine per unit time for cost modelling purposes.
Table 8.1
8.3

Failure rates and availability

Life cycle cost comparison of fixed and variable pitch devices

8.3.1 Financial Assumptions

Assumptions

Units

Rated Plant capacity

Capacity factor
Plant life
Year cost base
NPV discount rate

MW
%
Years

Fixed Pitch
Farm
30
23
15*
2006
10%

Variable Pitch
Farm
30
25
15*
2006
10%

* 15 years for principal LRU life, other elements have longer assumed lives.
Table 8.2

Financial assumptions

8.3.2 Cost Analysis

The schematic at Figure 8.2 below illustrates the major categories to which costs
were allocated for the purposes of the study.

52

Figure 8.2

Initial cost: Incurred at the start of the project, and covers elements such as land
purchase, construction of roads, buildings, transformer bunds and similar items
which could be considered to have infinite life within the context of a tidal
turbine farm.
Farm-level capital equipment (FLCE): Paid once at start of project and upon FLCE
life expiry. Includes items associated with farm operation but having finite life,
such as grid connection equipment, workboats and so on.
Shore-based equipment (SBE): Covers purchase and replacement of items
associated with individual turbines but located ashore, the prime example being
the network bridge equipment.
Mounting: A device designed to hold one or more turbines, having a finite life but
modelled as having zero failure rate. The variable pitch mounting includes
provision for rotating the nacelle.
Line replaceable unit (LRU): Essentially, the turbine plus directly associated
power generation and conversion equipment (gearbox, generator, device bridge
and ancillaries), all located subsea and having a finite life, finite failure rate and
weather-dependent access restrictions.

The costs have been estimated for the fixed and variable concepts. For the purposes
of the cost modelling results these costs have been factored to produce an overall
cost of 1.250m/MW for the fixed concept, which then produced a cost of
1.318m/MW for the variable concept. This represents a 5.4% increase in cost for
the variable pitch concept compared to the fixed pitch concept (consideration of the
original cost estimates showed the components added or increased in cost relative
to the fixed pitch items to represent a premium over the fixed pitch prices of about
7.5%; the overall effect including the items common to both solutions was then
about 5.4%).
Cost of connection to the grid was modelled using the assumption that the
connection would cost 120,000 per megawatt capacity, leading to a connection cost
for the farm as modelled of 3.6 million. This assumption requires caution in its
application because it is only a typically average cost for a 10MW to 20MW

53

installation. For smaller sites the cost of connection could be understated and for
larger sites, including the 30MW farm modelled, it could be overstated. It should be
recognised that the minimum cost of a connection at the present time (2006) is likely
to fall within the range 1 million - 2 million, and may be greater if the particular
circumstances dictate the installation of a significant length of overhead line. In
some circumstances, the connection cost could preclude the financial viability of an
otherwise attractive tidal farm. Annex E of Reference 15 discusses this topic further.
The cost estimates for a farm with 30 off 1MW units, with 2 units per mounting is
shown in Table 8.3 below. Operation & maintenance assumptions and estimates are
shown in Table 8.4.
Cost Item

Number per farm

Farm cost

Cost

Fixed

Variable

Initial set-up cost

3,750,000

3,750,000 3,750,000

Farm Level Capital Equip,

4,500,000

4,500,000 4,500,000

Shore Based Equip.

150,000

15

15

2,250,000 2,250,000

Mounting - Fixed

300,000

15

Mounting - Variable

322,500

Line Replacement Unit - Fixed

750,000

Line Replacement Unit - Variable

806,250

Table 8.3

Cost estimates

54

Fixed

Variable

4,500,000
15

30

4,837,500
22,500,000

30

24,187,500

Total

37,500,000 39,525,000

Cost/MW

1,250,000 1,317,500

Operation and
Maintenance
Routine O&M
(per MW/year)
Major servicing

Intervals
p.a.

Fixed Pitch Total

Cost (2006 )
37,500

Variable Pitch
Total Cost (2006 )
40,300

5 yrs

included in above

included in above

Unscheduled
Interventions
(/intervention)

as required

24,000

25,700

Fixed annual
farm running
cost

p.a.

320,000

320,000

Rates
DeDecommissioning
costs, per
mounting

p.a
At yr 25 after
mounting
commissioning

included in above
25,000

included in above
25,000

Table 8.4 Operation & maintenance assumptions and estimates

8.3.3 Cost model results and cost of electricity
A number of cases have been analysed using the Cost Model; a list of the cases
analysed is included as Appendix B. Using the unscheduled availabilities from the
Monte Carlo simulations, the calculated hydrodynamic capacity factors and a fixed
scheduled availability of 95%, the Cost Model calculates the annual energy
produced, as shown in Table 8.5. For the base case, assuming the mean failure rates
predicted using the OREDA data the annual energy production is 53 GWh for the
fixed concept and 55 GWh for the variable concept, thus the variable concept
produces 4.6% more energy. Note that the actual energy produced by each variable
pitch concept is 20% of the total energy potential of a 1 MW generator. For the
maximum failure rates predicted by the OREDA data the energy production falls by
12%, whereas for the minimum failure rates it increases by 8%.
The calculated cost of electricity are shown in Table 8.6 for the cases discussed. The
Base case has a calculated cost of electricity over a 10 year operating period of 119
/MWh for the fixed pitch concept, compared to 129 /MWh for the variable pitch
concept. These reduce to 94 /MWh and 104 /MWh for a 15 year period. Again the
increase for the worst and reduction for the best OREDA failure assumptions can be
seen. It must be noted that these latter figures assume a best case for the rate of
failure, and are very sensitive to this assumption.

55

Case

Description

1.0
1.1
1.2
6.4
9.0
9.1
9.2
9.7

Base case, mean OREDA data

as above, Max OREDA (high Failure rate)
as above, Min OREDA (low Failure rate)
as 1.2, modified maintenance schedule
as 1.2, with 50% annual running cost
as 1.2 with 50% initial farm cost
as 1.2 with 50% FLCE cost
as 1.2 with 50% LRU cost

Table 8.5

Number of failures
Annual Energy Total Energy Production
Hydrodynamic
per device over
Production
(25 yr)
Reliability
farm life
Capacity Factor
Availability
MWh
MWh
OREDA Fixed Variable Fixed Variable Fixed Variable Fixed Variable
Fixed
Variable
Mean
Max
Min
Min
Min
Min
Min
Min

1766
4838
112
112
112
112
112
112

23%
23%
23%
23%
23%
23%
23%
23%

25%
25%
25%
25%
25%
25%
25%
25%

88%
77%
95%
91%
95%
95%
95%
95%

86%
71%
94%
88%
94%
94%
94%
94%

52,641
46,108
56,663
54,340
56,663
56,663
56,663
56,663

55,044
45,557
60,721
56,738
60,721
60,721
60,721
60,721

Annual energy production

Case

Description
Discount rate + 10%

Reliability
OREDA

1.0
1.1
1.2
6.4
9.0
9.1
9.2
9.7

Base case, mean OREDA data

as above, Max OREDA (high Failure rate)
as above, Min OREDA (low Failure rate)
as 1.2, modified maintenance schedule
as 1.2, with 50% annual running cost
as 1.2 with 50% initial farm cost
as 1.2 with 50% FLCE cost
as 1.2 with 50% LRU cost

Mean
Max
Min
Min
Min
Min
Min
Min

Table 8.6

1288
3564
66
66
66
66
66
66

Cost of electricity

Annual Energy Production

MWh
Fixed
Variable
52,641
46,108
56,663
54,340
56,663
56,663
56,663
56,663

55,044
45,557
60,721
56,738
60,721
60,721
60,721
60,721

Cost of Electricity /MWh

Fixed
Variable
10 years 15 years 10 years 15 years
118.68
178.09
91.63
94.73
88.35
88.68
87.68
70.93

94.00
150.00
68.67
71.09
65.46
66.72
66.10
55.23

128.95
218.89
91.44
97.05
88.37
88.63
87.74
70.50

103.97
188.82
68.73
73.04
65.74
66.91
66.34
55.15

1,316,031
1,152,692
1,416,579
1,358,503
1,416,579
1,416,579
1,416,579
1,416,579

1,376,109
1,138,916
1,518,019
1,418,458
1,518,019
1,518,019
1,518,019
1,518,019

8.4

Conclusions and Recommendations

In some respects the tidal turbine industry of 2006 is comparable to the wind
industry of the early 1970s, with a large number of possible permutations of the
various design options, a diversity of approach and a shortage of hard evidence as
to which approaches are likely to have a commercial future.
The principal objective of this study has been to identify any circumstances under
which simple but less efficient devices offer competitive life cycle performance to
complex but more efficient ones. The results demonstrate that the relative merit of
the two approaches depends upon the assessment criteria adopted and may be
summarised as follows, within the range of performance parameters studied:
The variable pitch machine (with an energy capture performance shown to be
notionally 10% better than that of the fixed pitch machine in the rotor
hydrodynamic study) produces more energy in a given period than the fixed
pitch machine, unless the absolute reliability of both machines is very low,
The fixed pitch machine always offers lower initial capital cost and unplanned
maintenance cost than the variable pitch machine,
The fixed pitch machine offers lower cost per unit of electricity generated unless
the absolute reliability of both machines is very high,
The percentage increase in energy capture using the variable pitch machine is
generally much less than the notional 10% difference given perfectly reliable
fixed and variable pitch machines; relative performances encountered during the
study implied that the variable pitch machine would generally produce 4 7%
more energy than the fixed pitch machine, although its worst performance
produced 1% less energy.
The cost analysis shows that to be economically viable the reliability of a real device
must equal or better the best reliability assumed for the study. This would seem to
be a challenging target, but it should be noted that the study configurations
incorporate no redundancy; it is possible that the overall reliability of a real device
could be significantly improved given careful detail design and the incorporation of
redundant features in specific areas.
It is recognised that the hydrodynamic capacity factors could be increased by
limiting the power output from the rotor, and using a generator rated at below the
maximum potential power available in the tidal stream. This would require a rematching of the rotor diameter to the generator, but should have the effect of
reducing the cost of electricity. Further consideration would need to be given to the
matching of the power conversion equipment, fault cases where the power output
exceeded the limit, and the impact on capital cost.
As indicated above, the variable-pitch variable-speed machine is superior in terms of
capital cost per MWh to the fixed-pitch variable-speed machine when the reliability
of both machines is very high. However, within the study, the maximum
improvement calculated for the variable-pitch machine was a cost per MWh less
than 4% better than for the fixed-pitch machine. Within the limitations of the study,
the fixed-pitch approach offers the more robust design concept, with the

theoretically better performance of the variable pitch device only being delivered
when high absolute reliability is consistently achieved. This reinforces the view that
it may be beneficial for designs to start simple (more robust and reliable) and later
on to become more sophisticated. Wind turbine development followed a similar
path.
Additional remuneration for renewable generation over the commodity price of
electricity generated is currently via an obligation on electricity suppliers to purchase
a specified portion of their supplies as renewable generation in the form of
Renewable Obligation Certificates (ROCs). At present there is no differentiation
between technologies but as electricity suppliers all face the same cost pressures
they tend to follow each others choices. Currently this choice is predominantly
wind.
The Government proposes to amend the Renewables Obligation (RO) following the
recent Energy Review and has issued a consultation proposing differentiated
support levels to different renewable technologies and give additional certainty on
long-term ROC prices. These changes to the RO will require new primary legislation
and so will not be introduced until towards the end of this decade at the earliest.
Tidal power, whilst having a popular appeal in terms of green-ness perception and
environmental impact, is arguably one of the more expensive in capacity terms of
the available technologies, certainly at this stage of its development. Its merits as
part of a renewable portfolio lie in the predictability of the energy source allied to the
perception that it is a development area whose time has come. While commercial
deployment of devices will help to drive down the capital and operating costs, tidal
power will need additional support while the successful concepts become
competitive renewable energy options. Therefore higher value ROCs should be
introduced for marine renewable technologies and be sustained while successful
concepts achieve their potential.
Although the options analysed produced a variety of outcomes, there is sufficient
encouragement to further investigate the fixed pitch device. The following activities
offer logical steps to the implementation of a commercially successful tidal turbine
device:
Further refinement of the economic analysis methods used to determine
parameters for an optimised device,
Develop design specifications for a tidal stream turbine system and apply
suitable processes to optimise the design point for a given set of tidal conditions,
Undertake further reliability modelling to determine plant redundancy
requirements,
Undertake model testing of bi-directional sections both for cavitation
performance and overall performance to confirm that predicted in the study,
Install a reduced scale version of the device (around 20 30% full scale) in the
field, probably with a dump load rather than a grid connection. Test results will
be used to finalise the design of the prototype unit,

58

Install a full-scale prototype at a suitable location with grid connection. This

would not be commercially optimised, but would demonstrate all the major
systems,
Install a multi-unit tidal stream turbine farm using the commercial design
incorporating developments from the testing of the prototype unit.

59

9.

BIBLIOGRAPHY

Identifier

These references have been used in the study and are included for further
information.

Title

Phase 2 UK Tidal
Stream Energy
Resource Assessment
Design of Direct-driven
Permanent-magnet
Generators for Wind
Turbines
(PhD Thesis)
Calculation of Different
Generator Systems for
Wind Turbines with
Particular Reference to
Low-Speed
Permanent-Magnet
Machines
(Dr.-Ing Technical
Dissertation)

Author / Originating
Organisation

Remarks

Black & Veatch Ltd

Carbon Trust report

submission
107799/D/2200/03
Revision 2 15/07/2005

Anders Grauers
Chalmers University of
Technology
Department of Electric
Power Engineering

Technical Report No.

292
October 1996

Stephan Jckel
Technical University of
Darmstadt
2002

ISBN 3-8322-1512-3
Pub. Shaker Verlag
2003

Trials & Evaluation 1

Procurement Executive
Ministry of Defence
RARDE
Chobham Lane,
Chertsey
Surrey KT16 0EE
ed. Eberhard Zeidler
Max Planck Institute for
Oxford Users Guide to
Mathematics in the
Mathematics
Sciences
Leipzig
1996
Robert Harrison
Large Wind Turbines
Erich Hau
Design & Economics
Herman Snel
2000
Reliability
Mathematics and
Methods
(supplement to Report
85004)
Unclassified

60

HMSO London 1985

Pub. Oxford University

Press
ISBN 0-19-850763-1
2004
translated & typeset
Bruce Hunt
Pub. John Wiley & Sons
Ltd.
ISBN 0-471-49 456-9

Pub. John Wiley & Sons

Ltd.
ISBN 0-471-49972-2
Reprinted with
corrections 2003

Wind Energy
Explained

J.F. Manwell
J.G. McGowan
A.L.Rogers
2002

Principles of Electric
Machines and Power
Electronics

P.C. Sen
Professor of Electrical
Engineering
Queens University
Kingston, Ontario,
Canada
1989

Marine Energy
Challenge Marine
Energy Glossary

Richard Boud
Entec UK Ltd

002 glossary Issue 4.doc

B/14459/C001/002
The Carbon Trust
July 2005

10

OREDA Offshore
Reliability Data 4th
Edition

SINTEF Industrial
Management for the
OREDA Consortium

Distributed by Det
Norske Veritas
ISBN 82-14-02705-5

11

Wind and wave

frequency distributions
for sites around the
British Isles

Fugro GEOS for the

Health & Safety
Executive

HSE Offshore
Technology Report
2001/030

Pub. John Wiley & Sons

Ltd.
ISBN 0-471-85084-5

13

Waves,Tides and
Shallow Water
Processes, 2nd Edition

Open University course

team
(S330 Volume 4)

Pub. Butterworth
Heinemann in
association with the
Open University
ISBN 0-7506-4281-5

14

Total Design
Integrated Methods for
Successful Product
Engineering

Stuart Pugh

Pub. Prentice Hall

ISBN 0-201-41639-5

15

The Energy Challenge.

Energy Review Report
2006

Department of Trade &

Industry

61

APPENDIX A:
A1.

MONTE CARLO SIMULATION TECHNIQUES

Explanation of Monte Carlo simulation methods

The term Monte Carlo is used to describe methods of solution using the laws of
probability to reach a statistical estimate of a desired quantity. The basic principle is
that a process may be simulated by an average of a series of random events if
certain characteristics of the events are appropriate to the process being simulated.
The Monte Carlo simulation runs the process a large number of times, assigning
random values to the events and taking an average of all the outcomes. The result of
any Monte Carlo activity is always a simulation of the process, not an analysis of it.
In principle, two successive Monte Carlo process runs of the same event with
identical initial conditions will produce different answers. The average result for the
large number of processes run converges to a common solution. Monte Carlo
techniques are generally most suitable for situations where the probability equations
for the process are intractable or, possibly, not capable of explicit formulation.
For example, if (as is required for this study), it is desired to estimate the proportion
of its lifetime for which a device will be operating correctly, and the number of
occasions during that period when it will become necessary to repair the device, it is
in principle possible to construct a tree showing the probability of each of the
following circumstances:
1. Device operates without failure for whole design lifetime
2. Device operates for most of design lifetime, fails, but cannot be repaired
before design lifetime expires
3. Device operates for most of design lifetime, fails, is repaired, and operates for
remainder of life
4. Device operates for most of design lifetime, fails, is repaired, fails again and
cannot be repaired
5. (and so on...)
Figure A.1 below illustrates the situation for a device, which may be in either of two
mutually exclusive states, state A (available) or state U (unserviceable). The
transition AU, with probability PAU , represents failure, and the transition UA, with
probability PUA, represents repair. The overall lifetime of the device TT comprises N
equal time intervals I. The time during which the device is available is TA. The
availability of the device overall is given by TA/T T. The device starts in state A, and
in any given interval I its probability of failure is constant and is described as (PAU)I.
The probability of device failure over its complete lifetime TT is denoted by (PAU) T
note the single subscript.
The diagram gives expressions for the probability of each outcome. By definition,
the sum of the probabilities of all possible outcomes (i.e. including those not shown
in the diagram) is unity. It is apparent that the complexity of this approach increases
geometrically with the probable number of failures during the device lifetime.
Note that the probability expressions given in the diagram cannot in general be
evaluated without knowledge of the intermediate times t1...tn. By definition, these are
62

unknown, because they represent the times to random failure events, and
(potentially random) repair times following each failure event. The explicit
calculation method also becomes more complex if the probabilities of each event
are themselves functions of some other variable. A situation considered as part of
this study is the effect of seasonal weather conditions on the repairability of the
device, assessed for the purposes of the study by assuming that P UA is a function of
elapsed time. The underlying assumption is that whilst the possibility of repair
depends on the availability of appropriate spare parts and other resources, an
overriding consideration is simply whether the device is accessible in order for the
repair to be performed. For example, a device which fails at the onset of winter is
likely to be out of commission longer than one which fails identically at the height of
summer, simply because weather conditions preclude maintenance access to the
former device.

Final State

Overall probability

Availability

A
(no failure)

(1-PAU ) T = (1-PAU) I N

TT /TT (100%)

U
(one
unrepaired
failure)

(PAU ) t1*(1-PUA) T-t1

t1/TT

(PAU ) t1*(PUA) t2-t1 *

(1-PAU ) T-t2

[TT -(t2 -t1)]/TT

A
(one
repaired
failure)

(PAU ) t1*(PUA) t2-t1 *

(PAU ) t3-t2*(1-PUA) T-t3

[t1+(t 3-t 2)]/TT

U
(one
repaired
failure,
one
unrepaired
failure)

A
(two
repaired
failures)

(PAU ) t1*(PUA) t2-t1 *

(PAU ) t3-t2*(PUA) t4-t3 *
(1-PAU ) T-t4

[t1+(t 3-t 2)+

(TT -t4)] /TT

START
A

A
A

(PAU ) t1
(PUA) t2-

First failure

t1

t1

First repair

U A

(PAU ) t3-

t2

t2

(PUA) t4-

t3

Second
failure

A
U

t3
U

U A
Second repair

t4
(etc.)

Figure A.1

illustration of device states

63

The general approach adopted to modelling these problems is that a computer

simulation of the device may exist on one of several mutually exclusive states, with
probabilities of each possible transition between each pair of states defined either as
a constant or as some arbitrary function, with, in general, the probability of
transition from state X to state Y PXY not equal to the reverse transition probability
PYX. The model is run for a large number of experiments, with specified values of the
various probabilities and/or specified constants for the various probability equations.
For each experiment, the model iterates through a predetermined number of time
periods by generating a random variable in each simulated time period. By
comparing the random variable with the appropriate transition probabilities, the
appropriate destination for the transition is chosen (which can include the
transition from state X to state X, i.e. preservation of status quo). A record is then
incremented of number of periods in a given state, number of transitions between
particular states, or other interesting information.
The records of each experiment in a series may then be analysed statistically to
produce estimates for the model, such as averages and standard deviations for
availability, number of repair events required in a given lifetime and so on.
A2.

Explanation of Monte Carlo simulation models

For this study, several mode lling approaches were considered before construction of
the actual models used. Consideration of the various conditions in which a tidal
turbine might exist led to the concept of an eighteen state model, representing the
machine throughout its life cycle. The eighteen state model was considered to be
comprehensive, but as a result included conditions which would probably have little
effect on the outcome of the comparative study. The complexity of the model
increases rapidly with increasing number of states (in part, because the number of
possible transitions increases geometrically with the number of possible states), and
it was considered that the additional information provided by this model did not
justify the extra complexity relative to a simple system. The eighteen state model
encouraged the development of a notation to identify the various states, which was
preserved (with one modification) for subsequent work.
Elimination of the (assumed) least frequently encountered states from the eighteen
state model led to the development of a five state model. Contemplation of this
model led to the conclusion that it still represented excessive detail for the purpose
of this study (although either the eighteen state or five state models might be
worth further study in the context of an investigation focused on the detailed
economics of a particular device).
All the models use the concept of time measured in periods, with values chosen to
be of significance to the desired outcome. No specific timescale is explicitly defined
for a period; it needs to be long enough to allow modelling of a meaningful
proportion of the device lifetime, without the creation of excessively large computer
files, but short enough to capture detail. An instance of unserviceability can never be
shorter than a complete period, so use of excessively long periods will lead to
unrealistically low estimates of device availability.

64

It is vital that values of transition probabilities used in the simulations are scaled to
match the period chosen for the series of experiments. For this reason, use of
excessively short periods not only results in extra computer time, but also implies
that failure probabilities (as entered) become negligible. As an example, a device
with a failure probability of 0.0001 per hour has a 50% probability of failure after a
period of 6931 hours (0dp), or about 9 months. To model the same device at
periods equivalent to one second, the failure rate per period must be entered as
0.00000003. A failure rate of 0.0001 per hour is possibly too high for economic
viability of a tidal turbine. Typical period and experiment durations used for this
study included:
240 periods (20 years with 1 period 1 month)
312 periods (6 years with 1 period 1 week)
1560 periods (30 years with 1 period 1 week)
A2.1 Eighteen-state model
Figure A.1 below shows the main features of the eighteen state model. The various
states are identified and described fully in the table below. An important feature of
this model was that each state was associated with a three-by-three matrix
expressing the output as scaled full (i.e. whatever would be expected given the
conditions at the time), degraded or zero, and the accessibility of the device as
irrelevant, accessible for repair or inaccessible for repair.
The CD & DC transitions (and corresponding pairs between other states) were to be
determined by a random or partially random variable; these transitions representing
occasions when the weather and/or sea state became unsuitable or suitable to
permit safe access to the device..

65

Scaled full
output

Accessibility
irrelevant

Available for repair

Unavailable for repair

Degraded
output

F
H

Zero output

I
K
M

L
N

Y
Figure A2

Main features of Eighteen-state model

Certain transitions (shown with dashed arrows in the diagram) were solely the result
of operator action. It was contemplated that operator strategy might affect overall
farm economics, by (for example) only permitting limited degradation before
shutting the machine down. However, a machine having been repaired or
reconditioned was assumed to be returned to service as soon as possible without
operator action. It was assumed that the probability of destruction would always be
lower from a non-operational than from an operational state.
The effect of operator strategy, whilst possibly of importance to the operator of a
tidal turbine farm (and an interesting topic for study in its own right), is not directly
y
relevant to the aims of this study, so for this purpose represents an undesirable
complication.

66

Notation

Description

Accessible?

Output

It was assumed that each instance of degradation (state E or F) would be

associated with a reduction in device yield, modelled as a simple percentage of
original output, as a percentage of output at the time of degradation or as a random
variable such that the effect of each degradation of output would vary from incident
to incident.
One reason for moving away from this model was that exploration of the full
solution space would require very large amounts of computer time to correctly
represent the effects of those transitions having a low overall probability of
occurrence.

Remarks

Serviceable & operating

100%

Most of elapsed time in this

state for economic viability

Serviceable but deliberately shut

down

0%

Faulty but operating

Faulty but operating

Faulty, output (further?) degraded

Faulty, output (further?) degraded

Faulty, deliberately shut down

0%

Faulty, deliberately shut down

0%

Failed completely

0%

Failed completely

0%

Undergoing repair / refurbishment

0%

Interrupted repair / refurbishment

0%

Commissioning

0%

Interrupted commissioning

0%

Decommissioning

0%

Interrupted decommissioning

0%

67

0100%
0100%
0100%
0100%

Output not explicitly defined

Output not explicitly defined
CE transition implies reduced
output
DF transition implies reduced
output
To allow repair during
scheduled maintenance
period?
As an alternative to
destruction?

Remove marine growth etc. or

repair ?

Once per life cycle

Once per life cycle (if not

destroyed)

Decommissioned

0%

Out of water?

Destroyed

0%

Beyond economic repair

Table A1

Monte Carlo notation

Whilst the above forms a comprehensive representation of the life cycle of a device,
most of the time must be spent in state A for economic viability. It is questionable
what value may be had from the fine detail of the minority of the time spent not
operating.
A2.2 Five-state model
A significant difference between the eighteen state and the five state models was
that for the five state model, accessibility was no longer considered as an explicit
feature of the solution. Seven of the eighteen states in the eighteen state model
directly corresponded with seven of the other states, the sole distinction being
whether repair operations were possible. The model could be simplified by replacing
the duplicated states with variable transition probabilities, such that (for example) a
low value for repair probability could correspond to winter conditions and an
inaccessible device, whilst a higher value might reflect summer conditions.
It was also recognised that some of the states were trivial for the purpose of
calculating device availability and maintenance. For example, logically, a device
having been repaired would indeed spend a short time in state B (serviceable but
shut down) before being returned to full operation, but in practice the amount of
time spent in B would be short (of the order of a few hours) compared to a time
step of the Monte Carlo simulation (of the order of a week, or perhaps a month).
The result of these considerations was the model shown at Figure A3 below.
Operating states are as described for the eighteen state model above, except that
the accessibility of the device is not explicitly known from the operating state.
The diagram includes notations for the probabilities of various transitions (omitted
from the eighteen state diagram for clarity).
Note that PVW where V=W represents the probability that the machine will remain in
its present state. Also, PKK may be related to the mean time to repair the machine. PII
and PCK may be related to the accessibility of the machine owing to weather
conditions (although these also depend on the availability of parts and labour to
conduct a repair).

68

Scaled full output

PAA

PAZ
PAC

Degraded
output
Zero output
Figure A3

PCC

PAI

C
PCZ

PCK
PCI

PII

PKA

PKK

PIK

PKZ

I
PI Z

P ZZ=1

Five-state model

Probabilities of transitions between the various states are also shown in Table A2
below.

Initial state

Final state

Table A2

PAA

PAC

PAI

PA Z

PCC

PCI

PCK

PCZ

PII

PIK

PI Z

PKA

PKK

PKZ

Probability of transition between states

The total number of possible sequences for a run of N time periods is thus 4N,
although some of these sequences are trivial, and if the objective is to estimate the
availability and number of repair events, it is likely that many sequences will be

69

equivalent. However, it should be remembered that the above model, run with time
periods representing one week for 52 periods, i.e. representing one full year of
operation, could produce in principle any of approximately 2x1031 possible
outcomes, ignoring the fact that, as implied above, some of the outcomes are
impossible.
This example highlights how complex even an apparently simple model such as the
above may be in practice. It was considered desirable to simplify the model even
further for the purposes of this study.
A2.3 Two-state model

PAA

Figure A4

A
PAU

Zero output

Scaled full
output

The five-state model may be simplified further by assuming that the probability of
device destruction is identical for each device, and in any event sufficiently small to
have little effect on the overall farm economics. It may be argued that, for the
purpose of calculating device availability, it matters little whether a device has
failed (state I) or is under repair (state K), because in neither case is it
generating. The foregoing arguments imply a three-state model, which is working
normally, working at reduced output or not working. As a final simplification,
ignoring the state of degraded operation results in a two-state model (with a
change of notation), which is either in state A (available and operating) or state U
(unserviceable). This forms the basis of the model used for most of this study, Figure
A4.

PUA
U
PUU

Two-state model

Again, PVW V=W represents the probability that the machine will remain in its present
state. Probabilities of transitions between the various states are also shown in the
table below.
For this model, PA U (failure probability per time increment) may be constant, or could
in principle be a function of some other variable. For example, PAU could be a
function of elapsed time chosen to simulate general wear and deterioration or
perhaps to replicate the well-known bathtub failure rate curve.

70

Initial state

Final state
A

PAA (=1(=1-PAU)

PA U

P UA

P UU (=1(=1-P UA)

Table A5
PUA is perhaps best represented as a function of elapsed time. It may then represent
the effects of bad weather precluding access to the device for part of the year.
For much of the work conducted in support of this study, PUA was modelled as:
PUA =PUAmax-(A1[sin(2(A4t +A3)/TT)]M )
where
PUAmax = maximum value of PUA allowed for the given series of experiments
A1
= Arbitrary constant, controlling the minimum value of P UA for the series of
experiments
A3
= Arbitrary constant, controlling the phase shift of the PUA curve
A4
= Arbitrary constant, controlling the period of the PUA curve
t
= Elapsed time
TT
= Total time (in periods)
M
= Exponent (positive integer) Note that the value of A4 differs by a factor of 2
for a given P UA curve period, according to whether the exponent is even.
This function was chosen to give a reasonable degree of control over the form of the
PUA curve without the user having to input a large number of arbitrary constants.
For example, the idea was considered of allowing the user to input a separate PUA for
each period of the experiment, or at least for a recurring subset of the experimental
periods, so allowing (for example) repair probabilities to be defined for each
calendar month. This would add seven degrees of freedom to the model (versus the
sinusoidal model described above), but would not necessarily clarify the results.
Reference 11 was used to estimate a range of access probabilities for a notional
device in likely UK locations. Correspondence between the power-sine PUA function
incorporated in the Monte Carlo simulation and the access probabilities calculated
from Reference 11 over the course of a calendar year varied; in some cases a close
match was obtained whilst in others the power-sine formulation could replicate the
maximum and minimum or the mean values, but did not otherwise display good
correspondence with the Reference 11 function.

71

APPENDIX B:

CASES ASSESSED

The various cases assessed are summarized in the table below. The complete run
plan (and results, when available) was populated into a spreadsheet containing full
details of assumed costs etc. In general, all run inputs matched those for Case 1.0
unless otherwise specified, with the main exceptions that overall device availability
and expected number of failures vary according to failure rate, accessibility and (for
those cases modelling no unscheduled maintenance) maintenance interval.

Case

For all runs, equipment was ordered so as to produce maximum farm output as
soon as possible, and automatic re-ordering of all equipment except mountings was
permitted, so each run modelled a farm generating for a twenty-five year period
within an overall run duration of thirty years, including decommissioning of all
installed mountings (but not including any other site decommissioning cost
element). Note that the individual cases were not calculated in the order listed in
Table B1 below.
List of Cases Assessed

1.0
1.1
1.2
2.0
2.1
2.2

Description

Remarks

Baseline case, two LRU per mounting,

mean failure rate. Variable speed for
both types.
As 1.0, but maximum failure rate

PUA represents Reference 11 node 16357

[50.413N, 1.650W English Channel] for
Hsig = 1m

As 1.0, but minimum failure rate

Alternative case, two LRU per
mounting, mean failure rate. Fixed
speed for both types.
As 2.0, but maximum failure rate
As 2.0, but minimum failure rate

3.1

As 1.0, but four LRU per mounting

3.2

As 1.0, but one LRU per mounting

4.1
4.2

As 1.0, but 50% of initial farm cost

As 1.0, but 200% of initial farm cost

4.3
4.4

As 1.0, but 50% of FLCE cost

As 1.0, but 200% of FLCE cost

5.1

As 1.0, but 50% of LRU life

Questionable whether this assumed

specification would be acceptable for
connection to the UK grid.

Changes in mounting and SBE cost to

reflect four LRU per mounting. Farm
capacity limited to 30MW installed, to
retain comparison basis. Mounting order
quantities changed.
Changes in mounting and SBE cost to
reflect one LRU per mounting. Farm
capacity limited to 30MW installed, to
retain comparison basis. Mounting order
quantities changed.

Results reflect the changed number of

replacement LRUs needed (calculated by
model)

72

5.2

As 1.0, but 200% of LRU life

Results reflect the changed number of

replacement LRUs needed (calculated by
model)

6.0

As 1.0, but no unplanned maintenance,

1 year planned maintenance interval

6.1

As 6.0 but two year maintenance

interval

6.2

As 6.0 but maximum failure rate

(Case 1.1 assumptions apply)

6.3

As 6.1 but maximum failure rate

(Case 1.1 assumptions apply)
As 6.0 but minimum failure rate
(Case 1.2 assumptions apply)
As 6.1 but minimum failure rate
(Case 1.2 assumptions apply)

6.4
6.5
7.0

As 1.0 but maintenance by DSV

7.1
7.2

As 1.1 but maintenance by DSV

As 1.2 but maintenance by DSV

8.0

As 1.0 but worse access probability

8.1

As 1.1 but worse access probability

8.2
9.0

As 1.2 but worse access probability

As 1.2 but 50% annual running cost

9.1
9.2

As 1.2 but 50% initial farm cost

As 1.2 but 50% FLCE cost
As 1.2 but 50% initial farm cost & 50%
FLCE cost
As 1.2 but 50% initial farm cost & 50%
annual running cost
As 1.2 but 50% FLCE cost & 50%
annual running cost
As 1.2 but 50% initial farm cost, 50%
FLCE cost & 50% annual running cost

9.3
9.4
9.5
9.6

Cost results should be regarded as

OPTIMISTIC, as failure rate
underestimated for this specific
circumstance by calculation process
Case not calculated as other Case 6 runs
indicated this situation would be
uneconomic
Case not calculated as other Case 6 runs
indicated this situation would be
uneconomic

9.7

Table B1

Same location as 1.0 but Hsig= 4m to reflect

DSV capability and intervention fixed fee
150000 (normally 1500)
As 7.0
As 7.0
PUA represents Reference 11 node 14824
[59.579N, 4.315W West Shetland Shelf]
for Hsig = 1m. PUA function chosen
overestimates access probability for about
55% of year but preserves overall average.
May not be sensible to use RIB in this
location.
As 8.0
As 8.0

Maintenance cost allocation preserved by

doubling percentage maintenance
allocation.

As 1.2 but 50% LRU cost

Cases assessed

73