Anda di halaman 1dari 227

Impacts of Tidal Stream Devices on Electrical Power Systems.

Andre Garth Bryans, BSc, MSc

A thesis presented on application for the degree of Doctor of Philosophy


School of Electronics, Electrical Engineering and Computer Sciences Faculty of Engineering and Physical Sciences The Queens University Belfast

September 2006
Supervisors: Dr. B. Fox, Prof. Peter Crossley, Prof. T. Whittaker and Prof. M. OMalley

Abstract
The increase in fuel price, concerns over energy security and global warming have fuelled a global drive towards renewable power generation. The majority of renewable generation investment in Ireland like many countries is in the form of wind generation. However concerns about the feasible level of wind generation are leading to development and investment in other forms of renewable generation, such as tidal stream generation, a near market ready form of variable but predictable generation.

A review of the tidal stream systems under development was undertaken to determine the operational limitations of the systems most likely to reach the market first. The tidal stream resource was modelled within a 2D ocean model and viable areas identified based on the operational limitations of the tidal energy device seen as being closest to market readiness. The viable areas were analyzed producing the profile and magnitude of the currently viable resource.

The impact of tidal generation on system operations was studied with consideration to its effect on the system ramp rate, demand profile, capacity / availability factor, generation capacity credit, unit commitment, net system emissions, net generation cost and cost based price received by tidal generation. Tidal generation was found to be manageable on the system considering the currently viable resource.

The benefits that tidal generations predictability may offer in comparison to wind generation was quantified in terms of the effects on emissions, market aspects, and system operations. This involved the development of a unit commitment model with methods for providing reserve against uncertainty in wind forecasting. Tidal generations predictability was found to offer benefits over wind in most aspects considered.

The grid connection of the viable tidal resource was studied in terms of transmission loss adjustment factors, short circuit ratings, capacity of the transmission system and cost of 33/38 kV grid connection. As a result of this analysis the most attractive tidal resource around Ireland was identified. The

impact of a prototype scheme on voltage control for a lower voltage 11 kV system was also considered, and voltage was found to be within tolerable limits.

Acknowledgments
This work has been financed by Northern Ireland Electricitys SMART program (Sustainable Management of Assets and Renewable Technology), the UK Department of Enterprise, Trade and Investment (DETI) and Sustainable Energy Ireland. Thanks is owed to Dr. B. Fox, Prof. M. OMalley, Prof. P. Crossley & Prof. T. Whittaker who provided supervision and technical guidance in carrying out this research. The School of Electronics, Electrical Engineering and Computer Sciences, Queen's University of Belfast and the Electricity Research Centre, University College Dublin provided technical support, and an administrative hub for the project. Thanks is also owed to Kirk McClure & Morton Ltd., Danish Hydrographical Institute, British Oceanographic Data Center, Prof. Jenkins, Marine Current Turbines and The Engineering Business Ltd. who provided technical support, software or data.

List of Acronyms Used


ACCC ADCP CCGT CFL DETI DFIG DHI EAMC EDR ESB EU GR LMP LOLE MCT MIC MIP MSL NAP NECD NI NIE PPA REFIT ROCs RoI SMART SMC SONI SOP TED TG TLAF TMPC TSO TUoS UC UK Alternating Current Contingency Calculation Acoustic Doppler Current Profiler Combined cycle gas turbine Courant Friedrichs Lewy criterion Department of Enterprise, Trade and Investment Doubly fed induction generator Danish Hydrographical Institute Energy averaged marginal cost Electrical down rating Electricity Supply Board European Union Generation reduction Locational Marginal Pricing Loss of load expectation Marine Current Turbines Marginal incremental cost Mixed integer programming Mean sea level National allocation plan National Emissions Ceiling Directive Northern Ireland Northern Ireland Electricity Power purchase agreement Renewable Energy Feed In Tariff Renewable obligation certificates Republic of Ireland Sustainable Management of Assets and Renewable Technology System marginal cost System Operator Northern Ireland Scheduled Outage Periods Tidal Energy Device Tidal generation Transmission Loss Adjustment Factor Total marginal plant cost Transmission System Operator Transmission Use of System Unit commitment United Kingdom

Table of Contents

1. Introduction
1.1. 1.2.
1.2.1. 1.2.2.

The Need for Renewable Energy Incentives for renewable generation


Northern Ireland Republic of Ireland

1 4 4 4 5 6 6 6 7

1.3. 1.4. 1.5. 1.6. 1.7.

Existing renewable generation Incentive for tidal energy Available tidal resource Tidal generation and system operations Tidal generation and grid connection

2. Energy extraction from tidal stream


2.1.
2.1.1. 2.1.2.

The Origin and Nature of Tidal Energy


The Tidal Generating Forces The concentration of tidal energy on shelf seas

8 8 18 20 21 23 30 30 31 32 33

2.2.
2.2.1. 2.2.2.

Generation from tidal stream


TED systems currently in development Marine Current Turbines

2.3.
2.3.1. 2.3.2. 2.3.3. 2.3.4.

Impact of tidal energy extraction


Effect on tidal regime Effect on the local eco-system Effect on human actives Global scale

3. Resource assessment
3.1. 3.2. Introduction Shelf sea models
34 35

3.2.1. 3.2.2. 3.2.3. 3.2.4. 3.2.5. 3.2.6. 3.2.7. 3.2.8. 3.2.9.

The definition of closed boundaries and land cells Selecting the Grid Scale Size The setting of grid cell depths Selecting a Time Step The forcing of open boundaries The inclusion of a nested grid Bed friction Eddy viscosity Model Calibration

38 40 43 44 44 46 46 47 47 48 49 49 50 51 51 54 54 56 59 63 63 64

3.3.
3.3.1. 3.3.2. 3.3.3. 3.3.4. 3.3.5. 3.3.6. 3.3.7. 3.3.8.

Development of an oceanographic database


Seabed depth Seabed slope Maximum spring tidal current velocity Maximum neep tidal current velocity Wave height Tidal Phase Distance from Ireland Database integration

3.4. 3.5. 3.6. 3.7.

Determination of Power Output Accessible Resource Energy in the Irish Sea Conclusion

4. System operation with tidal generation


4.1. 4.2. Introduction The potential methods of development and operational control
4.2.1. 4.2.2. Tidal Superposition Electrical Down Rating (EDR) 66 67 70 73 65 65

4.3. 4.4.

The effect of tidal generation on demand profile The effect of tidal generation on the system ramp

rate
4.4.1. 4.4.2. 4.4.3. 4.4.4. Increasing amounts of tidal generation Ramping on the system during a spring neap cycle Comparison between wind and tidal energy Effect of EDR on the system ramp rates 74 76 78 78 79 81 82 83 85 87 89 89 94 95 97 99 104 107

4.5. 4.6.
4.6.1. 4.6.2. 4.6.3.

Capacity & availability factors Capacity Credit


Scheduled outages Calculation of loss of load expected Identification of the capacity credit of tidal generation

4.7.
4.7.1. 4.7.2. 4.7.3. 4.7.4. 4.7.5. 4.7.6.

Frequency response of a TEDs to disturbances


Inertia constant Generator type Ability of generators to provide inertial response Method of calculating turbine inertial constant Model of frequency response Frequency response provided by MCTs TED

4.8. 4.9.

Power quality Conclusion

5. Impact of renewables on thermal plant


5.1. 5.2.
5.2.1.

Introduction Irish case study characteristics


Electricity markets

109 110 110 112 112 114 114 116 118 118

5.3.
5.3.1. 5.3.2. 5.3.3. 5.3.4.

Unit commitment
Operating costs Technical constraints Ancillary services Optimisation and unit commitment

5.4.
5.4.1.

Method of adding wind to the dispatch


Fuel saver approach

5.4.2.

Forecast approach

118 124 126 128 129 132 130 134 135 140 140 142 143 144 145

5.5.
5.5.1.

PLEXOS Model set-up


Small test case

5.6.
5.6.1. 5.6.2. 5.6.3.

Effect on emissions
Carbon dioxide Sulphur dioxide Nitrogen oxides

5.7. 5.8. 5.9.


5.9.1. 5.9.2. 5.9.3.

Effect on operation of generation units Effect on the market The effect of tidal generation on plant usage
Effect of tidal penetration on existing generation Effect of tidal penetration on new generation Effect of tidal energy on the use of storage systems

5.10. 5.11.

Effect on carbon emissions Conclusion

6. Grid connection
6.1. 6.2.
6.2.1.

Introduction Transmissions loss adjustment factors


Method of calculating the TLAFs for the North Coast and Larne.

146 147 147

6.2.2.

TLAFs at Coleraine and Larne

148 151 152 153 164 166 168 170 173 174

6.3. 6.4.
6.4.1. 6.4.2. 6.4.3. 6.4.4. 6.4.5. 6.4.6. 6.4.7.

Short circuit levels Connection capacity and cost


The northeast cost Maiden Islands Arcklow Carnsore Point Malin Head Shannon Strangford

6.4.8.

All sites

175 176 182

6.5. 6.6.

Strangford case study Conclusion

7. Conclusion
7.1. 7.2. 7.3. 7.4. Tidal stream energy devices Tidal stream resource available to Ireland Impact of tidal generation on system operations Grid connection and site development of tidal resource 7.5. Recommendations for future research
187 184 184 185 186

1. Introduction
1.1 The Need for Renewable Energy
Societys demand for energy is increasing at a rapid pace (see Figure 1-1) with developing countries such as China rapidly increasing their demand for energy. The majority of the energy used to meet this demand is in the form of fossil fuels. These are fuels that originated as organic material having captured their energy from sunlight and converted it to chemical energy. The organic material initially formed a spongy peat, which underwent sedimentation to form fossil fuels such as coal, oil and gas. The majority of the peat was formed during the Carboniferous Period about 360 to 286 million years ago. Therefore fossil fuels represent a finite resource that once used will not naturally replenish in a feasible time frame.

Figure 1-1. Annual global energy demand record and prediction (Energy Information Administration, 2006), note this uses the US definition of Quadrillion (1015).

The rapid increase in demand (see Figure 1-1) in conjunction with concern over the security of supply from nations acting as net exporters of fossil fuels has resulted in a rapid increase in price of fossil fuels such as oil (see Figure 1-2).

Figure 1-2. Average annual crude oil price in $ per barrel of oil (1 BBl = 159 litres) in the United States from 1920 to 2006, not corrected for inflation (Oilnergy, website).

The effect of burning such large amounts of fossil fuels to meet the energy demand is to release green house gases which act to trap heat around the earth by enabling high energy radiation to heat up the earth but impeding the escape of low energy thermal radiation, resulting in global warming (see Figure 1-3).

Figure 1-3. Change in surface temperature in the northern hemisphere over the last 1000 years (Intergovernmental Panel on Climate Change, 2001). 2

Studies have indicated that if green house gas emission rates are not abated society will face dangerous climate change (Schellnhuber et al., 2006). Whilst it is difficult to perform an exact cost benefit analysis for avoiding dangerous climate change due to the uncertainties, it is believed that the costs of reducing emissions is much less than facing the consequences of climate change (Schellnhuber et al., 2006).

The Kyoto Protocol was signed in 1997 by nations agreeing to the United Nations Framework Convention on Climate Change (United Nations, 1997). This agreement sets binding emissions reduction targets for developed countries (or groups of countries such as EU). A burden sharing agreement between EU member states (EU, 2002) has required the United Kingdom (UK) and the Republic of Ireland (RoI) to limit their CO2 emissions to 87.5% and 113% of their 1990 levels, respectively. Also, the European National Emissions Ceiling Directive (NECD) (EU, 2001a) aimed at reducing acid rain has set out legally binding national limits for emissions of nitrous oxides (NOx), sulphur dioxide (SO2), volatile organic compounds and ammonia. Of these three, the power generation sector contributes towards the production of NOx and SO2. Within the NECD the UK must limit its SO2 and NOx emissions to 585 kt and 1,167 kt respectively, and Ireland must not exceed 42 kt of SO2 and 65 kt of NOx emissions per annum from 2010. However, Ireland is finding the NOx target particularly challenging, with NOx emissions of 135 kt in 2001 due to large economic growth and enlargement of the transport sector over the past decade (Department of Environment, Heritage and Local Government (Ireland), 2003). A large part of the emissions reduction is expected to be made through the power generation sector through methods such as the introduction of different fuel types (limited options left), exhaust gas cleaning methods, control methods and the introduction of renewable generation. The development and introduction of renewable generation is seen to have a double benefit, because in developing renewable generation forms to a commercial stage, developing countries such as China can also begin to install renewable generation on a much larger scale, resulting in a much greater global saving in emissions. Therefore to incentivise the development of renewables the European Union has set a target of 22.1% renewable generation by 2010, with the RoI and the UK expected to provide 13.2% and 10% respectively of their generation by renewable means (EU, 2001b). Northern Ireland

contributes towards the UK renewable generation target and has been allocated a local target of 6.3% by 2012 (Department of Enterprise Trade and Investment, 2004).

1.2 Incentives for renewable generation


A number of mechanisms have been established to encourage investment in renewable forms of generation both in NI and RoI.

1.2.1

Northern Ireland

In NI a Non-Fossil Fuel Obligation was established in 1993, which placed an obligation on the distribution company to provide 16 MW by installed capacity of renewable generation. By 2005 this was increased to 45 MW. Further investment was incentivised by introducing renewable obligation certificates (ROCs) in April 2005. The scheme works by requiring supply companies to obtain a given percentage of their energy from renewable sources. For non-renewable generation this may be achieved by buying ROCs from another company. If a company has insufficient ROCs it is charged a buy-out fee - 32.33 per MWh in the first year (Ofgem, 2005). The buy-out fund is then distributed to the suppliers in proportion to the ROCs provided. The ROCs scheme provides an incentive to produce up to 6.3% of electricity by renewable sources by 2012 (Department of Enterprise Trade and Investment, 2004; Department of Enterprise Trade and Investment, website; Department of Communications Marine and Natural Resources, 2005). However, if this is exceeded, the spare ROCs can be traded with mainland UK, which has a target of 12% by 2012.

1.2.2

Republic of Ireland

Renewable generation in RoI has been incentivised through the Alternative Energy Requirement scheme since the 1990s. This involves a bidding process for 10 to 15 year contracts to provide renewable generation at a fixed bid price under a power purchase agreement (PPA). Each type of renewable generation has a cap price which the applicants in the bidding process can submit up to. This scheme was replaced in 2006 with the PPA REFIT (Renewable Energy Feed In Tariff) (Department of Communications, Marine and Natural Resources, 2006), which replaces the process of bidding with a fixed reference price for each type of renewable generation (shown in Table 1-1).

Generation

Reference price (eurocents per kWhr) 5.7 5.9 7.2 7 7.2

Large Scale Wind Small Scale Wind Hydro Biomass Landfill Gas Other Biomass

Table 1-1. The fixed reference price for renewables within the PPA REFIT (Department of Communications, Marine and Natural Resources, 2006).

1.3 Existing renewable generation


Renewable generation currently makes up 2.6% of the electricity supplied in RoI (Sustainable Energy Ireland, 2006) and over 3% in NI (Department of Enterprise Trade and Investment, website), with 778 MW of renewables installed on the All-Island system (Department of Communications Marine and Natural Resources, 2005). Wind energy represents most of the renewable generation, augmented by hydro generation in RoI. Whilst there is still a large untapped wind resource, most of the hydro resource has been realised.

Figure 1-4. The installed capacity of renewable sources of generation on the All-Island system.

1.4 Incentive for tidal energy


The main thrust of Irelands response to the EU target has been provided to date by wind generation. However, there is concern about the feasible level of wind generation that can be absorbed by the system (Garrad Hassan, 2003), and there is a desire to achieve greater diversity of renewable energy supply. Tidal generation (TG) offers an energy source which, unlike wave generation, is not linked to the velocity of the wind, and is largely predictable within the generation scheduling time-scale. Also, tests on a medium-sized prototype have confirmed that the technology can deliver renewable energy at little extra cost to consumers (Whittaker et al., 2003).

1.5 Available tidal resource


To understand the impact of tidal stream devices on the system, it is necessary to first establish the magnitude and profile of the viable resource and the characteristics of the generator. Therefore a review of the technology was conducted and the devices which are near market ready are identified (see Chapter 2). To determine the tidal resource, an ocean model is developed using the Mike-21 software, indicating the tidal flows around Ireland (see Chapter 3). This work was conducted with the guidance of Kirk McClure & Morton Ltd. An oceanographic database is developed using the Mike-21 data and data provided by the British Oceanographic Data Centre which is in turn questioned to give the viable resource and generation profile (see Chapter 3).

1.6 Tidal generation and system operations


The impact of tidal generation on system operations is studied with consideration of its effect on: system ramp rate; demand profile; capacity / availability factor, generation capacity credit; unit commitment; net system emissions; net generation cost; and cost-based price received by tidal generation (see Chapter 4). Methods of controlling the effect of tidal generation are identified as including installation at different locations with opposing times of peak generation and reducing the tidal generation. The effectiveness of these methods is considered with each of the operational issues studied.

Tidal generation is used in conjunction with wind generation to clearly demonstrate the operational difference between variable (wind) and variable but predictable (tidal) forms of generation (see Chapter 5). To do this it is necessary to include wind and tidal generation in the unit commitment model. This involves the development of methods of providing reserve for the uncertainty in wind forecasting.

1.7 Tidal generation and grid connection


Given the location and magnitude of the resource it is necessary to quantify the issues for grid connection. These include: transmission loss adjustment factors, short-circuit ratings at the point of transmission system connection; capacity of the transmission system at the point of connection; and cost of grid connection, including the necessary upgrades to the 33/38 kV system - see Chapter 6. Whilst most of these are available for the Republic of Ireland in the Forecast Statement, it was necessary to model each to obtain compatible results for the system in Northern Ireland. The impact of the prototype scheme in Strangford Lough on the 11 kV system was also considered in relation to voltage control.

2. Energy extraction from tidal stream


2.1 The Origin and Nature of Tidal Energy
The development of renewable sources of energy, which are variable in nature, has increased the importance for electrical engineers to develop an understanding of the energy source and the forces driving the turbine. The following explanation will demonstrate that tidal energy is a variable but, unlike wind, is accurately predictable. This is a summary of a more extensive report produced as part of a project for Sustainable Energy Ireland (Bryans, 2004).

2.1.1

The Tidal Generating Forces

The earth and moon orbit each other about a common centre of gravity, which is much closer to the earth because the earth has a much greater mass than the moon. Therefore all points on the earths surface follow the same circular path each lunar month (ignoring the earths daily rotation), and experience the same centrifugal force (see Figure 2-1). The gravitational force exerted by the moon on the earth is proportional to the distance from the moon, so the gravitational force on the lunar side of the earth is greater than on the far side of the earth. The sum of the gravitational force and the centrifugal forces produces forces acting away from the earth at the equator whilst the two forces tend to cancel each other out at the poles (assuming the moon to orbit above the equator) (Pugh, 1987).
d

Centrifugal force Gravitational force Centre of gravity


R
R

Figure 2-1. The gravitational and centrifugal forces in the earth moon system.

The tidal generating forces at any point on the surface of the earth can be given in both the vertical and horizontal directions (see Figure 2-2).

Earth

Fh

Fv

Moon

Figure 2-2. The tidal generating forces in the earth moon system. Gravitational force Fg = G Centrifugal force
m m me d2

Newton, 1687

(2-1) (2-2)

Fc = me R 2
Where: mm = the mass of the moon me = the mass of the earth

= the angular velocity of the moon and earth about


the common centre of gravity.
G = gravitation constant =

ga 2 me

In the centre of the earth the centrifugal force equals the gravitational force:
me R 2 = G R 2 = G m m me d2

(2-3) (2-4)

mm d2

Therefore given the centrifugal force is constant over the entire earth:
Fc = G mm d2

(2-5)

So the tidal generating force on the surface of the earth, directly under the moon can be found in the vertical direction:
mm FTG = G (d a )2 mm G 2 d

(2-6)

a = radius of the earth FTG = Tidal generating force

FTG

mm = 2g m e

a d
3

(2-7)

3 mm Fh = 2 me m Fv = 3 m me

a g sin (2 ) d a 2 g cos ( 1) d
3

(2-8)

(2-9)

The resulting effect is to generate a tidal bulge on both sides of the earth, on the moon side due to the gravitational attraction of the moon and on the far side due to the centrifugal force of the earth and moon orbiting a common centre of gravity. It is now necessary to add the earths spin, which is anti-clockwise looking down from the north pole. So an island would be pulled though the two tidal bulges, demonstrating a simidiurnal tide. However, because the earth is offset to the moon by 23o, one tidal bulge will seem greater than the other (see Figure 2-3), displaying a diurnal tide. Moon N

S Figure 2-3. The tidal bulge established at 23o in relation to earths tilted axis.

The gravitational and centrifugal forces between the earth and the moon must be added to those of the earth-sun system. Although the tidal generating force of the earth-moon system is 2.17 times greater than that of the earth-sun system, the effect of the earth-sun system is still very significant in driving the spring / neep cycle.

The moon revolves around the earth over a period of one lunar month (27.32 days), during which the superposition of the lunar and solar tidal generating forces can be constructive, resulting in large spring tides, and destructive, resulting in smaller neep tides (see Figure 2-4).

10

Full moon Neep tide Half moon Force exerted by moon Force exerted by sun Sun New moon Spring tide Spring tide

Sun

Sun

Figure 2-4. The formation of spring and neep tides through the superposition of the lunar and solar tidal generating forces.

The earth takes 24.0 hours to complete one full rotation in relation to the sun, therefore the simi-diurnal solar tidal constituent S2 will have a period of 12.0 hours, due to there being a tidal bulge on each side of the earth. However the moon is orbiting the earth, so in the 24 hours taken for the earth to complete one rotation the moon will have moved forward in its orbit slightly so the earth will have to rotate for a further 0.84 hours to complete one rotation in relation to the moon. Therefore the simi-diurnal lunar tidal constituent M2 will have a period of 12.42 hours (see Table 1-1). The equilibrium tide can be explained through the superposition of a number of these tidal constituents (harmonics based around the lunar day, the sidereal month, and the tropical year, see table 2-1). The M2 and the S2 can be considered as being the major tidal constituents used in representing the equilibrium tidal forcing.

11

Tidal constituent M2 S2 Simi diurnal N2

Period (hours) 12.42 12.00 12.66

Origin 2 1

Name

Principal Lunar

2( 1+ 2+ 3) Principal Solar 2 1- 2+ 4 2( 1+ 2) Larger Lunar Elliptic Lunar-solar semi-diurnal Lunar-solar diurnal Larger Lunar Larger Solar

K2

11.97

K1 Diurnal O1 P1 Where Frequency

23.93 25.82 24.07

1+ 2+ 3 1- 2 1+ 2- 3

Period (solar days) 1.035 27.32 365.24 6797.3

Name Lunar Day Sidereal Month Tropical Year Moons Node

1 2 3 4

Table 2-1. The major tidal constituents recognised in the equilibrium tidal forcing (adapted from lecture notes, School of Ocean Science, University of Wales, Bangor).

The equilibrium tidal theory ignores the effect of flow around landmasses, the frictional effect of the bed on the flow and the establishment of harmonic amplification. The tidal range in many lakes and basins is observed and calculated to be very small, in the order of a few centimetres, while in shelf seas the tidal range is observed to be in the order of metres. The reason for this discrepancy is that most of the tidal range experienced in shelf seas is forced from the tidal wave generated in the deep ocean.

The establishment of a sea surface gradient due to the tidal generating forces results in a current flow due to the pressure gradient at all depths down the water column. Such a current flow in the deep ocean results in a large net flux of water in the horizontal plane.

12

When the water depth is reduced the same flux of current must flow therefore current speed increases and the tidal range increases (see Figure 2-5) (Simpson, 1998). Tidal

Shelf Sea

Current Velocity

Ocean Figure 2-5.a. The amplification of tidal range and current velocity due to the oceanic forcing of shelf seas. If the shelf sea is longer than the wavelength of the tidal wave then multiple standing waves are set up with nodal points of zero tidal range (see Figure 2-5.b.). .

Figure 2-5.b. The propagation of an oceanic forced tide in a shelf sea longer than one tidal wave length. Such a theory can be applied to the tidal forcing in the North Sea, as demonstrated in Figure 2-8a, to show entire lines across the North Sea, which should experience zero tidal range.

The effect of the earths rotation on any object in motion or fluid flow is described well by Pond & Pickard (1983) Consider the following hypothetical situation. A long-range gun mounted at the North Pole is aimed along a meridian directly at a target fixed on earth and some distance to the south. In plane view, a projectile fired from the gun will travel in a plane fixed relative to the fixed stars but the target will be carried to the east by the rotating earth during the flight of the projectile. From the point of view of the

13

gunner, also rotating on the earth, the projectile will appear to curve to the left. See Figure 2-6.

Actual path

Targeted path

Target point

Earths rotation

Figure 2-6. An object travelling from the North Pole to the Equator. To represent this effect an imaginary force is used termed the Coriolis force, which acts anti-clockwise of any object or fluid moving in the Northern Hemisphere and to the left in the Southern Hemisphere.

Figure 2-7a. The effect of Coriolis force on moving particle (image is from wikipedia, website).

The effect of the Coriolis force acting to the right of the current flow in a shelf sea or gulf means that the current is forced to one side, depending on which way it is flowing, as shown in Figure 2-8.b. This results in a sea surface slope on the nodal line (amphidrome) at 90o to the current flow (Pugh, 1981; Pugh,1987). Therefore when one side of the nodal line is experiencing high water (to the right of the current flow) the opposite side is experiencing low water. So the only point that retains a tidal range of zero along the nodal line is the point in the centre around which the tidal wave can be

14

described as rotating. This point is known as the amphidromic point. The effect of friction on the current flow means there is a weaker current reflected back from the basin than entering the basin. Therefore the current flow on each side of the amphidrome is not equal and so the amphidrome is displaced left of the current entering the basin, as shown in Figure 2-8.d.

Figure 2-8.a. The theoretical establishment of a standing wave in the North Sea (Doodson & Warburg, 1941).

Figure 2-8.b. The theoretical establishment of a tidal current in the North Sea (Doodson & Warburg, 1941).

15

Figure 2-8.c. The theoretical relative times of high water around amphidromic points in the North Sea (Doodson & Warburg, 1941).

Figure 2-8.d. The measured relative times of high water around amphidromic points in the North Sea demonstrating the effect of friction in terms of amphidromic displacement (Doodson & Warburg, 1941). The current velocity in the vertical plane is acted upon by the frictional resistance of the

seabed (see Figure 2-9), causing the currents near to the bed to decrease in velocity according to the relationship shown in Eqn. 1-2 (Department of Energy, 1990).

Figure 2-9. The current velocity profile approaching the seabed.

16

U (z )

z 7 = u 0.32 h

for 0 < z < 0.5h

(2-10)

U ( z ) = 1.07u

for 0.5 < z < h

(2-11)

u = depth mean current velocity


z = height above sea bed
h = water depth

U ( z ) = current velocity at height above the bed z


These give results accurate to within +15%. However, they do not apply well to the bed layer. The bed layer extends a few centimetres from the sea bed, the flow within this layer will be smooth if the Roughness Reynolds Number is less than 3.5 or turbulent if it is greater than 68. Therefore within the bed layer the following relationships should be utilized.
U (z ) = U (z ) = u ln ( z / z ob ) ln ( / 2 z ob ) / 2h u ln ( / 2 z ob ) ln ( / 2 z ob ) / 2h

for z ob z 0.5 for 0.5 z h

(2-12)

(2-13)

z ob = seabed roughness length, determined by the nature of the sea bed

= thickness of the boundary layer.


The energy removed from tidal motion in the form of friction occurs to a large extent in shelf seas, where the water depth is relatively shallow (~200m compared with ~3,000m) and the current speeds are high. Therefore the shelf seas account for 74% or 2.6 TW (Munk & Wunsch, 1998) of global tidal friction, which acts to reduce the speed of the tidal wave and slow the rotational speed of the earth. The action of reducing the speed of the tidal wave causes the tidal bulge to lag, as shown in Figure 2-10. The phase lag of the tides causes the gravitational pull of the tidal bulge to occur ahead of the moon in its lunar orbit (see Figure 2-10). This transmits some of the energy lost from the rotational speed of the earth into the angular momentum of the earth moon system. The increase in angular momentum of the earth moon system increases the orbital distance from the common centre of gravity (Lambeck, 1980).

17

No friction present

Friction present

Figure 2-10. The effect of tidal friction.

2.1.2

The concentration of tidal energy on shelf seas

Tidal energy is amplified on shelf seas. However, there are areas on shelf seas where this energy is concentrated even further. Such areas are of extreme interest to the developers of TEDs (Tidal Energy Devices), because the greater the concentration of the energy, the lower the capital cost per kW of power extracted. Such areas include narrow channels and the entrances to gulfs, estuaries, loughs and seas. To raise the sea level of the area contained within the gulf, estuaries, etc., there must be a much higher flux of water at the entrance than at the landward end of the sea, as shown in Figure 211. However, not all estuaries experience this phenomenon; estuaries that have been exposed to long periods of erosion have undergone sedimentation inland and erosion at the mouth of the estuary, leading to a more uniform flux along the estuary.

Sea Flux

Land
Figure 2-11. High current velocities at the entrances to gulfs, estuaries, Loughs and

seas.

The tidal streams are also concentrated off headlands due to the establishment of a coastal current. This current is set up due to the effect of the Coriolis force. As the sea level drops, the coastal current is forced to the right and during flood it is forced to the left. When current is forced against a head it is driven offshore, regardless of whether the tide is flooding or ebbing, as shown in Figure 2-12.

18

Offshore current during ebb tide Headland

Offshore current during a flood tide

Headland

Figure 2-12. The establishment of an offshore current from a headland.

Such areas may appear to be excellent areas to establish TEDs. However, headlands also act as focal points for wave action and so must be considered with great care. This is because the shallow water waves (waves in water depth less than their wave length) travel faster in deeper water than in shallow water and so bend towards areas of shallow water (see Figure 2-13).

Headland

Increasing seabed depth Wave propagation

Figure 2-13. The propagation of waves onto a headland.

Waves involve the movement of the water particles under them in a circular motion which decreases in size with depth (see Figure 2-14). In shallow water the circular motion becomes elliptical near the seabed. The passage of waves over and through a

19

tidal turbine would result in a net increase of energy from the turbine because of the increase in velocity. However, it would increase the velocity gradient across the turbine and would impose serious structural strain that would drive up the cost per MW.

Figure 2-14. The motion of particles under a surface water wave in shallow water.

2.2 Generation from tidal streams


People have been harnessing the power of the tides for millennia; a good example of this in Ireland can be seen on Mahee Island in Strangford Lough where the early monastic community of Nendrum constructed a tidal barrage system to drive wooden turbines that powered millstones (see Figure 2-15) for milling grain from 618AD (McErlean et al., 2002). The flood tide would fill man-made lagoons, then during the ebb tide the water would run out through turbines driving the millstones.

Figure 2-15. Artists representation of what a tidal mill would have looked like at

Nendrum (McErlean et al., 2002).

In the last century tidal power has been used for the generation of electric power using barrage schemes such as La Rance, in France (Frau, 1993). These harness tidal energy

20

by capturing high water behind a barrage and then exploiting its potential energy by allowing it to flow out during low water. However, this method prolongs the period of high water in the estuary, thus endangering the original estuarine ecosystem. It is also only able to generate power during low water; otherwise the estuary would fill up with silt. The capital cost of developing a barrage system is very high, forcing investors into an all-or-nothing gamble. Hence in the last few years there has been a focus on generation from tidal streams, which would remove the need for a barrage, high capital costs and the environmental impacts that barrage schemes incur. This report will therefore focus on tidal stream rather than tidal barrage schemes.
2.2.1 Tidal Energy Devices (TEDs) currently in development

The development of TED systems would appear to be taking a similar route to that of wind turbine development, with a large number of initial concepts being reduced in time to produce a few viable solutions. Therefore the TED systems in development have been studied and divided into four classes:

Class of TED development

Stage of development

Class A

Development and installation of a full sized prototype, close to being market ready.

Class B

Development and installation of a full sized prototype, not close to being market ready.

Class C Class D

Development of a scale prototype. System patented but only in the design phase.

Table 2-2. Ranking method used during assessment of TED systems.

In classes C and D no distinction has been made between projects with ongoing research and those which are in financial difficulty, because it has been recognized that systems in the early stages of development may have difficulty attracting funding and therefore go through periods of inactivity.

21

Class

Company TED

developing Concept

Class A

Marine Current Turbines

Twin 2 bladed horizontal axis turbine, pile mounted (Marine Current Turbines, website).

Class B

The Engineering Business

Undulating wing mounted on the seabed (The Engineering Business, website; Trapp, 2004).

Class B

Hammerfest Stromas

3 bladed horizontal axis turbine, mounted on the seabed (Hammerfest Stromas, 2002a; Hammerfest Stromas, 2002b; Hammerfest Stromas, website).

Class C

Open Hydro

Constrained flow system with horizontal blades mounted from the circumference, leaving a hole in the center. The blades have a permanent magnet around them so they can act as a rotor and the constraint funnel as a stator (OpenHydro, website).

Class C

Blue Energy Limited

Enclosed vertical axis turbine (Blue Energy Ltd., website).

Class C

JA Consult Tidemill

Twin 2 bladed horizontal axis turbine, mounted on a hinged pile (J.A. Consultants, 2004).

Class C

Hydro Venturi

Constrained flow system uses the pressure change on the primary flow to pull in either air or water from a secondary flow, which is used for generation (Hydroventuri, website).

Class C

SMD Hydrovision

Twin 2 bladed horizontal axis turbine, mounted on a buoyant support chained to the seabed (SMD Hydrovision, 2004).

Class D

Van den Noort Innovations BV

Enclosed horizontal axis turbine with many fine blades (Van den Noort Innovations BV, website).

Class D

Hydraulic Current Turbines Ltd.

3 bladed horizontal axis turbine, mounted on the seabed, using hydraulic power aggregation (Hydraulic Current Turbines, 2003).

22

Class D

Ocean Tecs

Twin Savonius rotors, mounted on a pile or existing wind turbine (OceanTecs, 2004).

Class D

Sea Energy Ireland

Constrained

flow

vertical

axis

turbine

(Callaghan, 2003). Class D Hydrohelix Energies Constrained flow horizontal axis turbine, mounted on the seabed (Hydrohelix Energies, website). Class D Edinburgh University Vertical axis turbine, mounted from a floating, tethered ring. Class D Lunar Energy Limited Constrained flow horizontal axis turbine, mounted on the seabed (Lunar Energy, website). Class D Verdant Power Horizontal axis turbine, mounted on the seabed, with a wire screen to protect sea life (Verdant Power, website).
Table 2-3. Companies developing TED systems.

The only company developing a system which could be described as being close to market ready is Marine Current Turbines. Whilst a report has been prepared containing a detailed description of all the systems listed in Table 2-3 (Bryans, 2004), this thesis will only describe the system being developed by MCT.

2.2.2

Marine Current Turbines

MCT (Marine Current Turbines) are developing twin horizontal axis 2 bladed turbines mounted on a monopile in such a way that they can be jacked up for servicing (see Figure 2-16). Currently MCT has limited the feasible installation sites to have spring peak current velocity of > 2 m/s and with a depth of 20 m to 40 m. MCT have successfully installed a 300 kW prototype off the coast of Lynmouth which has been in operation since 16/6/03, dumping power into a load bank. MCT are currently expected to be the first UK company to provide a full-sized (1.2 MW peak) grid connected TED during 2006 / 2007. Funding has been secured for this system, planning permission has been granted in the Strangford Narrows, and Northern Ireland Electricity (NIE) have offered grid connection options for it. Following successful installation and operation in

23

Strangford, the next stage will be a semi-commercial venture with the installation of about 10 turbines at a site which has not yet been announced.

Figure 2-16. The current system design for a ~1.2 MW system to be located in

Strangford Lough, utilizing a 15 m blade diameter (Wright, 2004).

The 300 kW prototype off the coast of Lynmouth was found to produce better energy conversion efficiency than expected (Wright, 2004). The model used to predict energy output was based on a wind turbine model which has a maximum theoretical efficiency of 0.59 known as the Betz limit. The Betz limit is dependent on the velocity difference between the front and rear of the turbine (see Eqn. 1-14 - 23).

The energy contained in the flow (wind or tidal) is:


E area = 1 1 V 2 E k = mV 2 2 2 Where: E area = Energy per unit area E k = Kinetic energy per unit volume
m = mass V = velocity before the turbine

(1-14)

= density

24

The power output is: P = E k AV = 1 AV 3 2 (1-15)

Where: A = the area covered by fluid entering the turbine P = the wind / tidal stream power output
m = AV = A0V0

(1-16)

Where:

m = the mass flow rate A0 = the area covered by fluid exiting the turbine V0 = the velocity of fluid exiting the turbine
m = A

V + V0 2

(1-317)

P =

1 2 1 2 1 V + V0 2 m V m V0 = A V V02 2 2 2 2
2 V0 1 V 2 V0 1 V

(1-18)

1 V P = AV 3 1 + 0 4 V

(1-19)

1 1 V P = AV 3 1 + 0 2 2 V = wind power C p Where:

(1-20) (1-21)

C p = Betz coefficient
1 V C p = 1 + 0 2 V
2 V0 1 V

(1-22)

x (the ratio of V : V0 ) can range from 0 to 1. The maximum value is when


= 0 = 3 x 2 2 x + 1 dx 3x 2 + 2 x 1 = 0 dC p

(1-23)

x = 1/3, or 1 (not possible)

Cp = 0.59.

25

As wind passes through a wind turbine, kinetic energy is extracted reducing the wind speed and causing the air beyond the turbine to expand. However behind a tidal turbine the water cant spread out because of the sea surface and the seabed. Therefore the mass transport behind the turbine is lower than the mass transport in front, causing the water level in front of the turbine to rise. This increases the pressure differential across the turbine until the mass flow behind the turbine is the same as that in front of it, and the energy is extracted through maintaining the sea level difference across it. In effect this establishes a natural expansion funnel around the turbine, increasing the efficiency. However, the pressure difference also drives water round the turbine, limiting the benefit. Therefore the maximum efficiency is dependent on the number of turbines in a row and the water depth to turbine diameter ratio.

This phenomenon has been demonstrated in a non-quantitative manner using a twodimensional model in which the row of turbines have been simulated by dramatically increasing the bed friction (from a Manning number of 32 to 1) along the line of turbines. The line of turbines was located in a completely uniform estuary of width 600 m and length 39.8 km, with a uniform depth of 50 m (a beach with a high bed friction was located at the end of the estuary to prevent wave reflection) as shown in Figure 217.

Figure 2-17. The uniform estuary modelled to study the effect of turbines with

constricted side flow.

26

Figure 2-18. The surface elevation along the centre of the estuary with no turbines

installed and with three rows installed during mid flood.

These results (shown in Figure 2-18) demonstrate the establishment of a head of water across the turbine, the reduction of the tidal range across the turbine and the reduction in peak current speed. Similar findings have been made by Bryden (2003).

Previous work (Bryden, 2003) demonstrated that in a channel with a constant flow, the velocity after the turbine would be greater than the velocity before the turbine. The reason for this is, the water depth beyond the turbine is reduced and to maintain the same mass transport the velocity must increase. To review the work of Bryden (2003) the experiment described above was repeated. However, the beach was removed and one end was set to have a surface elevation of zero, whilst the other end was elevated above zero to induce a constant flow along the channel. One row of turbines was placed in the channel at 16 km from the mouth. Figure 2-19 supports the work of Bryden (2003) demonstrating the depth to decrease behind the row of turbines, demonstrating a head of water is setup across the row causing the velocity increases behind the turbines.

27

Figure 2-19. The effect of a row of turbines completely across a uniform channel with a

constant stream speed.

The study described above does not permit water to flow around the turbines. However, in reality, as the sea level rises in front of the turbine, the increased pressure gradient would also act to force the water around the turbine. Therefore installing the turbines in long rows at 90o to the current flow will reduce the flow around the turbines and would increase efficiency. The distance between each row is dependent on two factors: the turbulent wake from the previous row; and the distance required to recover the peak velocity (in areas where the turbine rows are being installed at the edge or over a small section of the main flow) through the velocity shear stress set up between the main flow and the flow behind the turbine (see Figure 2-20). Some models have attempted to represent the velocity and turbulent wakes (Thomson, 2004). However, until actual measurements are taken from the turbine in Strangford Lough, these models cannot be calibrated or verified.

Figure 2-20. The formation of velocity and turbulent wakes behind turbine rows.

The economic predictions made by MCT indicate that tidal energy will become more financially attractive than off-shore wind (see Figure 2-21). However, when looking at

28

these Figures it must also be remembered that the power extracted from the tide and hence the financial return is a cubic function of the velocity.

Figure 2-21. The economic predictions made by MCT (Wright, 2004).

Whilst MCT are devoting much of their efforts towards developing a system capable of harnessing tidal stream in the most financially attractive areas (water depths of 20 40m with peak current speeds of greater than 2 m/s), they have also considered some designs for future deeper water systems once the most viable areas are developed. One of these designs includes mounting a number of large turbines on a horizontal support beam which is lowered onto multiple mono-piles at each end, and can be raised for servicing, which may be feasible for depths up to 50 m. A second design aimed at accessing the resource in depths up to 100 m is based around mounting multiple turbines above large submerged buoys. However another developer (SMD Hydrovision) has a design which may surpass MCTs 50 - 100 m system. This is a class D system (see Table 2-3) and comprises of a twin 2 - bladed horizontal axis turbine, mounted on a buoyant support chained to the seabed (SMD Hydrovision, 2004). The big technical challenge developers face in scaling up blade size is constructing a gearbox to convert very low rotational speeds to a usable speed for generators. However, better gearboxes are being designed for wind turbines and multi-pole synchronous generators capable of operating at lower speeds are also being developed.

29

The choice of a 2-bladed turbine is in keeping with the strategy of developing a system for operation in areas of high current speed. Research done on wind farms indicates that 2-bladed turbines are more efficient at high current speeds than 3-bladed turbines (see Figure 2-22). Therefore in the future when developing large turbines for deep water in areas of lower current speed it may be more efficient to switch to a 3-bladed design.

Figure 2-22. The efficiency in relation to the tip speed ratio found for wind farms

(Twidell & Weir, 1986).

2.3 Impact of tidal energy extraction


In recent years man has come to understand that the earths climate depends on a delicate balance between a number of equally opposing forces. It is therefore important to check the significance which tidal power extraction has for the energy balance as well as the environment as a whole.

2.3.1

Effect on tidal regime

Extracting energy from the tidal stream in effect increases the amount of seabed friction. The effect on the tidal regime of increasing seabed friction can be seen in Figure 2-8, which depicts the movement of the amphidromic point (point of zero tidal range). It will

30

also result in a reduction of the tidal range down-steam of the turbine and a reduction of the peak tidal stream speed.

2.3.2

Effect on the local eco-system

The impact of TEDs on the eco-system can be considered during the installation and during the operation of the TED.

During installation there is first a need for a jack-up barge to drill a number of sample bore holes to asses the geophysics of the seabed to be certain the bed is capable of holding the pile, and therefore ascertain the depth the pile needs to be driven or drilled in by. The second jack-up barge is then used at a later date to drive or drill the pile into the seabed and to install the turbine on to the pile. The action of a jack-up barge lowering its pile feet on to the seabed will destroy anything under them, whilst the drilling will result in a discharge of suspended particulate material into the water column. Therefore the jack-up barge coming in will crush any coral, sponges, rock features, etc. and those near to the drilling point will almost certainly be smothered by particles settling on them. However much greater damage is caused to the benthic environment through fishing nets being trailed over the seabed (often illegally), and after the installation of the turbine this area will become physically protected from such action. The discharge of the particulate material from the drilling process is not thought to be a major environmental risk because it can be pumped to the surface where the water current speeds are greatest (bear in mind these turbines will only be installed in areas of high current speed), allowing the particles to be dispersed to such a concentration that the resulting settling rate on the benthic organisms is negligible when compared to an influx from a river following heavy rainfall.

During operation the current behind the turbine will be mixed so the current speed at the seabed will be greater that before. Therefore it is probable the native benthic species will be replaced with a gradient of benthic species approaching the turbine, in the same way that a gradient of coral types are found across a coral reef, with the organisms favouring very high current speeds next to the turbine and those favouring lower current speeds further away from the turbine. The monopile used by the turbines is known (from its use in offshore wind turbines) to cause seabed scaring in areas of soft sediment such as sand banks. To prevent this the area around the base of the turbine often has to
31

be filled with rocks. The impact on the organisms that live or pass through the water column is through to be very small because of the slow rotational speed of the turbines. The tip speed of a 15 m 2-bladed device in Strangford Lough is estimated to reach ~30 knots (~15 m/s). When this is compared to the speed of a mackerel or a diving bird, which can reach speeds of about 35 knots (~18 m/s), it is seen that not only can fish such as mackerel swim though the turbine but, technically speaking, they could actually swim around in front of the blades. Slower fish such as cod may not be able to swim in front of the blades, but could certainly swim between the blades. Mammals are larger and therefore the window available for a marine mammal to swim through is much smaller. It is much more likely that marine mammals will avoid the structure due to their higher level of intelligence and acoustic awareness. Should it be found to the contrary, it is possible to install an acoustic warning device on the turbine developed for fishing nets to scare away marine mammals. However, sick or injured animals may stray into the turbine in a confused and disoriented state. Concern has, however, been raised about seal populations, and whether seals may try to swim through turbines. Experts have advised that it is unlikely, but that the situation should be monitored to confirm this.

2.3.3

Effect on human activities

Tidal stream turbines are expected to have a minimal effect on human activities. The visual impact is relatively small compared to that of a wind turbine (see Figure 2-23). However, the installation will require the establishment of an exclusion zone around the turbine for all maritime use. Diving and drift diving (a sport which involves a diver using strong currents to cover large distances) near the turbines would also be prohibited.

32

Figure 2-23. Scaled comparison of a tidal turbine and a wind turbine (Whittaker et al.,

2003).

2.3.4 Global scale

On the larger scale the installation of TEDs will result in the earth moon distance increasing at a greater rate (~1 cm per year per 1 TW year extracted). Much larger deviations in the rate of the earth moon separation are believed to have occurred in nature due to the formation of polar ice caps, and the geological movement of continents and coast lines.

33

3. Resource assessment
3.1 Introduction
To study the effect tidal generation may have on a power system it is essential to understand the resource magnitude, the nature of the variability the resource may bring on the system and the location at which it may connect to the system. To gain an understanding of the resource, an oceanographic model was developed, the output of which was combined with data from wave buoys to populate an oceanographic database (see Figure 3-1). The oceanographic database was interrogated by specifying the feasible range of each parameter to determine the viable areas. The total resource magnitude was calculated using each of these feasible areas and superimposing the results to give the entire resource magnitude and profile.

Chart of tidal constituents

Chart of seabed depth

Data from wave buoys Wave height

Mike 21 Shelf sea model

Seabed slope Seabed depth Distance from Ireland

Current flux

Water depth

Max. spring current velocity

Max. neep current velocity

Tidal Phase

Feasible resource Input data Ocean model Oceanographic database

Figure 3-1. Method of resource assessment.

The following chapter explains the setup and development of each step detailed in Figure 3-1, and presents the resulting feasible resource according to the operational range of the MCT system.

34

3.2 Shelf sea models


Models aimed at reproducing the observed flows in coastal waters are extremely challenging due to the sharp gradients in physical parameters such as water depth, velocities, turbulence and salinity. For these reasons, and in combination with the limitation of computing resources, modellers make simplifications to the relationships defining coastal currents. However, as computing power increases the simplifications which have been made in the past can be reduced to produce more realistic and accurate predictions (Haidvogel and Beckmann, 1998).

Models of coastal waters function by describing the problem both spatially and in time. Each of these divisions can be represented in a number of different ways. The grid which divides the area up geographically in the x and y directions, can take the form of structured regular grid cells known as an Arakawa-C grid (see Figure 3-2) (e.g. DieCAST, GBM, GFDLM, HAMSON and GHERM: referenced in Table 3-1) or structured curvilinear grid cells (SCRUM and SPEM: referenced in Table 3-1). The grid cells can also be created in an unstructured format, in the form of triangles (QUODDY: referenced in Table 3-1) or quadrilaterals (SEOM: referenced in Table 3-1), which produces a better representation of the sharp depth gradients seen in coastal waters. Some models only use two dimensions (the x and y) (Flather, 1993; Flather & Heaps 1975; and Falconer & Owens, 1987) whilst other models divide the grid further in the z direction. The z dimension can either be at intervals of constant geopotential (e.g. DieCAST, GBM, GFDLM and HAMSON: referenced in Table 3-1), or it can follow the bottom topography, with a constant number of divisions in each grid square (GHERM, POM, SCRUM, QUODDY and SPEM: referenced in Table 3-1). The latter approach in particular can be difficult to implement in areas of sharp depth gradients because the vertical layers may not line up with their counterparts in neighbouring grid squares. This problem is greatly exaggerated as the depth nears zero (which happens when introducing drying banks).

35

i = the position in the x direction j = the position in the y direction U = the current in the x direction V = the current in the x direction = surface elevation

Figure 3-2. The Arakawa-C grid (modified from Arakawa & Lamb, 1977)

Model Reference DieCAST Reviewed by Haidvogel & Beckmann (1998) GBM Dippner (1993) GFDLM Bryan (1997) HAMSON Reviewed by Haidvogel & Beckmann (1998) GHERM Beckers (1991) POM Reviewed by Haidvogel & Beckmann (1998) and Webb (2002) SCRUM Song & Haidvogel (1994) SPEM Haidvogel et. al. (1991) QUODDY Reviewed by Haidvogel & Beckmann (1998) SEOM Reviewed by Haidvogel & Beckmann (1998) Table 3-1. Coastal water models reviewed by Haidvogel and Beckmann, 1998.

Different models include differing selections of geo-physical and hydro-physical relationships describing fluid transport. However, there are a core set of governing equations which must be included within all these models before considering other locally important physical factors, such as the effect of wind and drying banks. The governing equations make three basic assumptions that the fluid is incompressible, hydrostatic, and boussinesq (density differences are sufficiently small to be neglected).

36

The governing equations (Flather & Heaps, 1975). Continuity equation: + (Hu ) + (Hv ) = 0 t x y Equations of motion: u u u ku u 2 + v 2 +u +v fv + t x y H v v u kv u 2 + v 2 +u +v + fu + t x y H were: u = the current velocity in the x direction. v = the current direction in the y direction. f = the Coriolis parameter. k = drag coefficient of bottom friction. H = total water depth + h . g = the acceleration due to gravity. (3-1)

) )

1 2

+g

=0 x =0 y

(3-2)

1 2

+g

(3-3)

= the sea surface elevation.

The continuity equation states that if the volume of water entering an area is different to that exiting the area then the volume of the water contained in that area will change, resulting in a change in sea surface height.

The equations of motion state that acceleration is proportional to the barotrophic effect of the sea surface slope, the friction, the Coriolis force and the rate of change in velocity in the x and y directions.

On starting the research two models existed of the shelf sea surrounding Ireland the first was a three dimensional model developed commercially by the Metoffice, in conjunction with the Proudman Oceanographic Laboratory and further refined by the Marine Institute (Marine Institute, website), however this remained inaccessible. The second was a 2 dimensional model (Mike-21) developed by DHI (Danish Hydrographical Institute) which had been implemented in previous studies by Kirk McClure Morton Ltd. to include; the Irish Sea, the North coast and the West coast of 37

Ireland. Kirk McClure Morton Ltd. kindly agreed to make the results of these models available, leaving only the southeast coast of Ireland uncovered. DHI provided a 1-year license for Mike-21 enabling the development of a model to cover the southeast coast of Ireland. Following the development of the southeast model, Kirk McClure Morton Ltd. was awarded a contract to develop an ocean model for all the sea around Ireland (Kirk McClure Morton et al., 2004), and agreed to make these results available. These compared well to the model developed for the southeast and offered a set of results for the entire island, mitigating the problem of joining data sets from different models. Therefore the results from the Kirk McClure Morton Ltd. model were used to populate the oceanographic database, whilst the development of the southeast coast model has been presented in the following section as an example of methodology.

The development of the model for the southern coast is broken up into a number of different stages, these include:

The definition of closed boundaries and land cells The setting of grid cell depths Selecting a time step The forcing of open boundaries Selecting a bed friction parameter Selecting an eddy viscosity parameter

3.2.1

The definition of closed boundaries and land cells

All the data regarding the land sea boundary was input from 1:50,000 scale digital charts (produced by C-Map Norway) apart from a short stretch of coast line on the south of Ireland for which there was no 1:50,000 scale chart held (highlighted in Figure 3-3). Therefore this section of coastline was covered at 1:100,000 scale. The data from digital charts is output to file in a format of latitude and longitude with the use the software package Mike C-Map (access to both Mike C-Map and the C-Map Norway data base was available under the terms of the software license of Kirk McClure Morton Ltd.).

38

Figure 3-3 The area of coastline covered at the lower resolution of 1:100,000 highlighted against the total study area.

The land sea boundary is defended with a series of polygons which overlap each other (Figure 3-4) to form the complex outline of the coastline, seen on the final map (Figure 3-5). The areas between the polygons, on the landside have to be covered with either one large polygon or a series of larger polygons, in order that land can be defined as being completely covered in polygons. A graphical user interface is provided within the Mike Zero software for this.

Figure 3-4. Land / sea boundary defined by a series of overlapping polygons.

39

Figure 3-5 Resulting land / sea boundary translated from the polygons in Figure 3-4. 3.2.2 Selecting the Grid Scale Size

Choosing the grid scale to use with a model is a balance between providing sufficient resolution to define the nature of the flow in the area against the time taken to process the model. With these considerations and with the maximum charted depth resolution of the area being to 1:25,000 the scale was set to a 3600 m x 3600 m grid scale for a coarse model of the entire area. This grid will be used to enable the bad data generated next to boundaries to dissipate before forcing a nested finer grid of 1200 m x 1200 m which in turn forces an even finer grid of 400 m x 400 m (see Figure 3-6).

40

41

42

3.2.3

The setting of grid cell depths

The water depth in chart datum (the lowest predicted water level) is also provided in the C-Map Norway database with irregular spacing. Unlike the data points defining the land sea interface the bathymetry data points can be superimposed on data points from charts of different scales. Therefore to obtain as accurate a representation of the bathymetry as possible data was output from scales ranging from 1:1,000,000 to 1:20,000 of selected areas. The accuracy of the data was maintained in coastal areas by limiting the bilinear search radius. Therefore, it was necessary to provide a few manually interpolated data points in the sparser areas.

The Mike C-Map software also provides the difference between chart datum and MSL (mean sea level). Correction grids were generated within Mike-21 to correct the data to mean sea level at the same resolution as each of the bathymetry grids, using the same method as that used to generate the bathymetry grids (see Figure 3-8).

Figure 3-8. Correction grid used to account for the difference between chart datum and mean sea level for the 3600 m x 3600 m bathymetry grid.

The water depth data was superimposed in the Mike Zero software with the land sea interface and bilinearly interpolated to provide a grid of Mean Sea Level (MSL) and land squares (given the value of 10 m above MSL).

43

3.2.4

Selecting a Time Step

The time step taken between each cycle was determined using the CFL (Courant Friedrichs Lewy) criterion, which states the maximum time step that can be chosen for the model to remain stable.

The method used to determine the maximum time step.

t <

x
gD

(Richtmyer, 1967)

(3-4)

where:

t = the time step. x = the grid spacing (smallest of the two directions).
g = acceleration due to gravity (9.81 ms-2). D = the grid square depth (m).

To be certain that stability is maintained within the model it is common practice to divide the maximum possible time step by 4 (see Eqn. 3-5).

For example:

t <

400 4 9.81 200

(3-5)

t < 2.3s
3.2.5 The forcing of open boundaries

The open boundaries are the boundaries along which the sea / ocean are present and therefore experience a forcing from outside the model. Without this forcing there would be no movement of water within the model (unless wind influence is included). There are five open boundary areas within the model:

The north boundary of the Atlantic (labelled as TA in Figure 3-6) The north boundary of the Irish Sea (labelled as TB in Figure 3-6) The east boundary of the Celtic Sea (labelled as E in Figure 3-6) The south boundary of the Celtic Sea (labelled as B in Figure 3-6) The west boundary of the Atlantic (labelled as L in Figure 3-6)

44

The open boundary surface elevations where forced using tidal height predictions for each open boundary cell at every time step. The tidal prediction was performed at five points along each open boundary or as near to the boundary as could be accurately measured on the contour plot of tidal constituents (Howarth, 1990). The values between these points where interpolated using a matlab routine developed for the purpose of achieving a smooth curve along the boundary (see Figure 3-9) rather than a linear interpolation which was found to cause instabilities in the model. These tidal

predictions were generated in Mike 21 from the M2, S2, K1 and O1 tidal constituents using the admiralty method (Doodson & Warburg, 1941). The amplitude and phase of each tidal constituent was obtained from Howarth (1990) which provides contour maps of the phase and elevation for the British Isles based on tidal analysis of data from buoys and ships.

Figure 3-9. An example of sea surface elevation determined along a boundary with the use of curved interpolation.

The depths of the grid squares on the open boundaries and those in the adjacent row / column were set to be greater than 10 meters in depth (chart datum) in order to prevent these squares from drying out. In doing this is must be noted that in these boundary areas of shallow water there will be unrealistically high current velocities. However the boundaries have been set sufficiently far outside the area of interest to account for this effect and for the effect of not permitting along boundary flow. Again as long as the

45

area directly adjacent to the boundary is not an area of interest the direction of flow will deviate to a more realistic direction over a number of grid squares.

3.2.6

The inclusion of a nested grid

The nested grid of resolution 1200 m x 1200 m was included within the original grid of resolution 3600 m x 3600 m as seen in Figure 3-6. Within the 1200 m x 1200 m grid was nested a further 400 m x 400 m grid model. This gird will provide a more accurate representation of the current velocities within the area of interest around the south coast of Ireland. To join the nested grid into the original grid a multiple of three nested grid squares must fit exactly into the original, therefore a grid spacing of 400 m was chosen to fit into the original gird spacing of 1200 m, which fits within the 3600 m. The depths around the boundaries must also match exactly. Mike21 provides a routine which smoothes the edges of the child grid and takes an average over 3 grid squares, it then smoothes and averages the area where the child grid will nest into the parent grid so the depth values are exactly the same in the two grid files, preventing any leakage between the two grids when the model is run.

The errors adjacent to the boundaries in the original grid are not expected to occur near the boundaries of the nested grid because only a small amount of smoothing has been applied to the bathymetry. Also, the external grid forces at the boundaries of the nested grid are obtained much more accurately and smoothly. The nested gird also enables along boundary flow generated from the direction of the flux into the boundary from the external grid cells. However there will be some localised error along these nested boundaries because the bathymetry has been smoothed.

3.2.7

Bed friction

The bed friction was approximated over the entire model because the bed friction is largely dependent on the seabed roughness and there is little information available for this at the chosen resolution of the model. The frictional effect was given as a manning coefficient of 32 which is related to the drag coefficient, shown in Eqn. 3-6.

46

1 Cd H 3 n= g

(3-6) (lecture notes, School of Ocean Science,

Where:

University of Wales, Bangor). n = the Manning number Cd = the drag coefficient H = water column depth g = acceleration due to gravity (9.81)

3.2.8

Eddy viscosity

The eddy viscosity was determined using the relationship given in Eqn. 3-7.

Ev = 0.04

(dx )2
dt

(3-7)

where
dx = grid square width dt = time step Ev = eddy viscosity

3.2.9

Model Calibration

The model was calibrated against both locally measured sea surface elevation data where possible (see Figure 3-6), and predictions generated from local tidal harmonics (see Figure 3-7). When using the locally measured data, it is important to realise that there will be deviations due to the effect of weather, and at low water the tidal gauge often runs dry so the values for low water must be extrapolated from the rest of the curve. To calibrate the model to match these points as best as possible the phase and amplitudes of the tidal constituents used to generate the boundary conditions were slightly altered, until the best comparison was achieved.

A detailed description of how each record compared to the expected value has been given in Bryans (2004).

The model has been calibrated to be within tolerable limits, further and finer calibration would have been possible, by altering the boundary conditions and by creating a grid of
47

the bed friction and adjusting it. However the data which the model is being calibrated against is currently not believed to be sufficiently accurate to warrant undertaking this work. The Irish Geological Survey have indicated concern over the accuracy of the data from which the local tidal predictions are made in the Republic of Ireland, creating a major problem in model calibration. They are sufficiently worried about this data that they have commissioned a complete resurvey of tidal constituents around southern Ireland.

The velocity profile at a few locations over the model was inspected to confirm there are no large velocity peaks. A few small peaks in velocity were observed in areas where there are large areas of drying and flooding such as the Bristol Channel and the Welsh coast. This can be expected in drying and flooding areas and will be confined locally to those areas (Bryans, 2003).

3.3 Development of an oceanographic database


The oceanographic database was established to provide a source of data for each factor which may restrict the development of a site for tidal generation. The database was structured to store each of the limiting factors in individual grids comprised of grid squares the same size as the all-island Mike-21 model (405 x 405 m). This database was populated using the outputs from the Mike-21 shelf sea model, and data from wave buoys. Each of the limiting factors listed below were calculated and stored in the appropriate gird. These grids could then be tested with limitations to determine the feasible areas of tidal resource. Seabed depth Seabed slope Maximum spring tidal current velocity Maximum neep tidal current velocity Wave height Tidal Phase Distance from Ireland

48

3.3.1 Seabed depth

The seabed depth data was taken directly from the Mike 21 model in grid form as the uncorrected chart datum (the lowest predicted tide) as shown in Figure 3-10. The depth within the model has been limited to a minimum of 300m to reduce the calculation time within the model.
Depth (m)

Figure 3-10. Depth of the seabed in chart datum (lowest predicted tidal level).

3.3.2

Seabed slope

The seabed slope was calculated according to the height difference between the adjacent grid square depths using the least squares method. The maximum slope experienced around the point was taken to be the value of the seabed slope (see Figure 3-11).

49

(degrees)

Figure 3-11. The calculated slope of each grid square within the study area.

3.3.3

Maximum spring tidal current velocity

The current velocity in the X and Y directions was determined from the Mike21 model output of current flux (in m3s-1m-1) between grid squares according to eqn. 3-8.

V =

F d +

(3-8)

where F = current flux d = depth V =current velocity

= surface elevation

The velocity vector was then converted to a scalar measure of speed. This process was repeated for each 15-minute model time step around a spring tide to find the maximum value for each grid square (see Figure 3-12).

50

Figure 3-12. The maximum tidal current speed in each grid square during a spring tidal

cycle.

3.3.4

Maximum neep tidal current velocity

The maximum neep tidal current was calculated in the same way as the spring tide, however it was derived from a series of Mike-21 model outputs based around a neep tide rather than a spring tide.

3.3.5

Wave height

The data available to determine the wave height around Ireland is very sparse, especially on the west coast. The data used in this study was combined from two sources the first being the British Oceanographic Data Centre (BODC) and the second being the Department of Energy, 1990. The data from the Department of Energy, 1990 was aimed at determining the maximum wave height expected during 50 years however the A&B modal frequency parameters were also provided so it was possible to convert this back to the maximum height per year and to the significant wave height (the mean height of the highest 1/3 waves) using the relationship given in eqn. 3-9.

51

H s = A + pB where A,B = modal frequency parameters p = probability of the wave Hs = the significant wave height

(3-9)

The data provided by the BODC was in the format of significant wave height. Each data set was positioned on a grid of resolution 16.2 km and the rest of the grid was interpolated using a linear interpolation in Mike21. Three grids were produced the 50 year significant wave height (Figure 3-13a) which demonstrates the wave height the structure must be capable of withstanding, the 1 year significant wave (Figure 3-13b) which demonstrates the wave height the turbine must be capable of operating during and the significant wave height (Figure 3-4c) which indicates the normal effect waves can be expected to have on the turbine.
(m)

Figure 3-13a. The 50 year significant wave height interpolated to a 16.2 km grid

resolution.

52

(m)

Figure 3-13b. The 1 year significant wave height interpolated to a 16.2 km grid

resolution.
(m)

Figure 3-13c. The significant wave height interpolated to a 16.2 km grid resolution.

53

3.3.6 Tidal Phase

The time of peak current speed in each grid square was identified by asking if the speed both before and after each time step was less than at the current time step, indicating the point of maximum current speed. Upon identifying the time of maximum current speed a note was made that it had already been found and to exclude the grid square from further searches. The final result would have given an answer ranging from 0 to 6.2 hours for each grid square however it is more conventional to give the tidal phase in the range from 3.1 hours to +3.1 hours, therefore this value was subtracted from the array.

Figure 3-14. The tidal phase patterned around Ireland.

3.3.7

Distance from Ireland

To calculate the distance from Ireland it was first necessary to distinguish between Ireland and other land. This was done manually by selecting the land within manually specified grids and setting its depth to a greater value than the remaining land. Two approaches were identified to calculate the distance from Ireland:

54

Start a search from each grid square to find the nearest point of land, this was done by searching out from the grid square in an ever expanding search radius. When land is found it is sensible to assume the adjacent grid square will find a land square at a near similar search radius and so the search can be narrowed down. However to conduct such a search from every single grid square would take longer than week of processor time on a P4 2.8 GHz.

The distance from the centre of all the squares less than one squares distance from a land square can be determined because they will be half the square width away, whilst those diagonally next to the land can be found by simple trigonometry. This procedure can then be repeated for the squares next to those which have just been determined, given that the distance to the centre of each square will now be a full squares width. This method can be completed in less than 1 hour using a P4 2.8 GHz.

The second method was chosen to optimise on processor time. The data was recorded in a grid file and can be used to produce a contour plot or as a limiting factor when selecting feasible sites (see Figure 3-15).
Distance from Ireland (m)

Figure 3-15. The calculated distance from Ireland.

55

3.3.8

Database integration

To determine the feasible areas with the study area, each grid representing a limiting factor was tested to determine the grid squares that were within the feasible. This was repeated for all the limiting factors and only the squares which remain viable in all of them have been identified as being viable.

The database was tested against MCTs current design assuming the feasible ranges for development of a site given in table 3-1.

Limiting factor

Feasible range for development of site

Seabed depth Seabed slope Max. spring tidal current speed Significant wave height Distance from Ireland MCTs 1st generation of TED.

Between 20 m and 40 m. N/A (only used in gravity mounted TEDs) 2.25 m/s Less than 2 m. Less than 15 km

Table 3-2. The conditions required to assign a site as being feasible for development by

Figure 3-16. The areas identified as having a current velocity >2.25 m/s and a depth

between 20 and 40m.

56

It is believed by the author that following technology advances the number of feasible areas will increase. This statement was made based on the fact that larger turbines in deeper water will be capable of harnessing greater amounts of energy for a lower current velocity as shown in Figure 3-17. Therefore to provide cases that may anticpate the resource that may be accessible by 2nd and 3rd generation turbines the database has also been tested for peak spring current speed of 2.0 m/s and 1.8 m/s, whilst the depth range has been increased to 20 to 50 m (Figure 3-18) and 20 to 70 m (Figure 8-13).

Figure 3-17. The power output in relation to current velocity and blade diameter, based

on a horizontal axis turbine with an efficiency of 40%.

Figure 3-18. The areas identified as having a current velocity >1.8 m/s and a depth

between 20 and 50 m.

57

Figure 3-18. The areas identified as having a current velocity >1.8 m/s and a depth

between 20 and 70 m.

The interrogation of the database found the majority of feasible sites to be off the north coast of Ireland (see Figure 3-16 to 3-18). There are also a few sites off the east coast of Ireland on the Arklow and Codling banks (see Figure 3-16 to 3-18) and a few possible sites off the southwest headlands (see areas of high current speed in Figure 3-12). The sites off the southwest headlands would almost certainly experience extreme wave climates for reasons explained in chapter 2 and would therefore not be feasible to develop. There are also a few small sites, which lie outside the grid resolution of this model. Such sites may include the Shannon estuary, Strangford Lough and areas on the west coast sheltered from wave action behind Islands. There was also some resource seen to open up off Malin Head when the feasible depth range was set to 20 to 70 m (Figure 3-18).

58

3.4 Determination of Power Output


According to the literature the general power available per square meter can be determined according to Eqn. 3-10.

The power available per square meter of sea surface (Fraenkel, 2002).

P=

1 hk s k n k ef V 3 2 Where: h = water depth k s = Daily availability factor (0.424) k n = neap / spring availability factor (0.57)

(3-10)

k ef = efficiency V = Max. Current velocity

Including k s and k n will give the average power output over both spring neap cycle and over a daily cycle. The values of k s and k n given in 3-8 were presented in Fraenkel (2002). The efficiency of the TED can also be included to appreciate the total amount of energy available at any given location. However the TEDs cannot be installed infinitely close behind and in front of each other because there is a need to enable the tidal current to both recover in velocity behind them and return to a laminar flow (see chapter 2). In chapter 2 it is theorized that the most efficient method of installing the turbines would be in rows at right angles to the direction of current flow, with the turbines in these rows being infinitely close to each other (side by side). The area seen by each row can therefore be determined according to eqn. 3-11.

The area seen by each turbine row. (h c )x 4 Where: (3-11)

h = water depth c = blade clearance depth

x = width of the grid square, assumed length of the row

59

The number of rows in each grid square depends on the spacing between each row, chapter 2 explains the difficulties associated with choosing an appropriate distance. However an estimate was made to use 15 blade diameters between each row. Therefore the number of rows in each grid square could be found by:

The number of turbine rows per square. Number of rows = Where:

y
2rs

(3-12)

y = breadth of the grid squares


r = the turbine radius s = spacing in blade diameters

It is therefore possible to determine the number of rows of turbines in each grid square, and hence the area seen by the turbines in each grid square:

The number of turbine rows per square. y Area = (h c )x 4 2rs (3-13)

Power output for a given grid square can be determined as follows:

The power output per square.

P=

1 Ak s k n k ef V 3 2 Where A = Area seen by turbines, within the square

(3-14)

However the values of ks and kn were tested against values calculated from the verified model output and were found to vary according to location. Therefore when determining the power output from each grid square for the resource assessment these values were dropped and an approximation of the velocity was used to calculate the power at 15 min time steps throughout a spring neap cycle. The velocity of the current follows a sinusoidal pattern with a period of 12.4 hours, during which time the power

60

output would peak twice, once in each direction. The tidal velocity also varies according to the spring neap cycle over a period of 14.75 days. To determine the power at a given time the availability factors were removed from equation 3-8 and the velocity at the given time was approximated from the maximum velocity during a spring tide and the maximum velocity at a neap tide by superimposing a spring-neap sin function to account for this velocity range on a second semi-diurnal sin function to account for daily variation (see Figure 3-19).

The method of determining an approximate tidal velocity for each square at any given time. 2 (24 + T ) Vsn = (Vs Vn )sin 14.75 2 (T + ) Vt = (Vs Vsn )sin 12.4224 Where (3-15)

(3-16)

Vsn = Velocity according to a spring neap cycle Vt = Velocity according to a spring neap cycle and
the semi-diurnal cycle.

Vs = Maximum velocity during a spring tide Vn = Maximum velocity during a neap tide

= The tidal phase difference


T = Time (hours from start time of model data)

61

Figure 3-19. An example of a tidal current velocity approximated from only semi-

diurnal, and spring-neap constituents. The power extraction from the velocity assumes an efficiency of 40% according to experimental data obtained by MCT.

To economise on processing time, an array was created consisting only of the viable grid squares. This array can contain data including the: Maximum power output from the grid square Mean power output from the grid square Maximum spring current velocity Maximum Neap current velocity The location of the Square The water depth of the square The distance from Ireland The group the square is in, (every cluster of squares is automatically assigned a group number).

The total tidal generation at any given time can be established, taking into account the different tidal phases between the squares with the use of the array given above. A routine was developed to do this but also to include the ability to selectively remove generator clusters (or groups) from the dispatch. Although the average tidal phase is

62

given for the group when selecting which generators to remove the individual tidal phase is used to calculate the total tidal generation at a given time.

3.5 Accessible Resource


The average power extractable from feasible areas around Ireland out to a distance of 15 km, assuming a 40% efficiency, and a row spacing of 15 blade diameters by MCTs first generation TED has been shown to be 72 MW or 374 MW at peak output. This was following the removal of 40% of the sites to account for local obstacles not seen on the 405 m grid scale.

Table 3-3. The feasible resource in maximum depths ranging from 40 m to 70 m, with

peak current speeds ranging from 2.2 to 1.8 m/s and with a low speed cut of 0 to 1 m/s. The tidal resource, which is feasible for the current technology, would be capable of supplying approximately 2% of the annual power system demand. This finding compares very well with (Kirk McClure Morton et al., 2004) which estimated the resource at 2.18% of the system demand.

3.6 Energy in the Irish Sea


The Irish Sea contains very large amounts of energy at any given time in the form of kinetic energy as the tide travels in and out, or potential energy during times of high and low water. It is not yet possible to model the exact quantitative effect large amounts of

63

tidal generation will have on the tidal regime in the Irish Sea, due to a lack of data regarding flow around a turbine. Therefore to develop a basic understanding of the amount of energy involved in the natural frictional dissipation occurring in each grid square was calculated using the output from the Mike-21 model to provide the tidal current velocity (see Eqn. 3-17).

D = AV 3 k
were

(3-17)

D = frictional dissipation A = Area of grid square V = tidal current velocity k = sea-bed friction co-efficient (0.0025)

= density of sea water


The natural frictional dissipation of all the grid squares in the Irish Sea was found to peak at 133 GW during a spring tide, assuming a constant bed friction. This compares well with previous studies by Taylor (1919) and Munk (1997). Therefore the extraction of a few gigawatts of tidal power from the Irish Sea could therefore not be expected to have a significant effect on the over all tidal regime of the Irish Sea.

3.7 Conclusion
The Island of Ireland was estimated to have a viable tidal resource capable of supplying 2% of electrical energy demand. It was also determined that removing a few gigawatts of tidal power from the Irish Sea would not significantly affect the tidal regime. However it has not been possible to develop an accurate model of the wake effects of the tidal turbines due to the lack of real data to verify such a model. MCT are planning to collect accurate measurements of the wake effects from the first twin turbine unit to enable the development of such models. It is expected that following the development of turbine wake models the resource assessment can be improved in accuracy by optimizing the turbine row separation. Wake models will also enable a much more accurate impact study on the local tidal regime.

64

4. System operation with tidal generation


4.1 Introduction
The transmission system on the island of Ireland is operated as two areas: the NIE system in Northern Ireland (NI) and the ESB system in the RoI. The NIE system comprises a 275 kV double-circuit ring with an underlying 110 kV system. The NIE demand ranges from 550 MW to 1750 MW. The ESB transmission system consists of a 400 kV grid with underlying 220 kV and 110 kV systems. ESB demand ranges from 1600 MW to 4400 MW. An interconnector with a maximum trading limit of 400 MW links the systems, with plans for reinforcement. The combined system is inter-connected to Scotland through the NIE area by a HVDC 500 MW interconnector with a maximum trading capacity of 400 MW. Generation on both systems is in large units, relative to system size, of up to 480 MW.

When considering a new form of generation for development and investment, it is important for TSOs (Transmission System Operators) to understand the challenges and advantages for system operation. Methods of deployment and control will be considered to minimize the effect of a fluctuating generation source on the grid system and to provide the best possible return for capital investment. The effect on grid system operations will be studied with respect to: The effect of TG on demand profile The effect of TG on system ramp rate Capacity & availability factors of TG Capacity credit of TG Frequency response of TEDs to disturbances Power quality of TG

4.2 The potential methods of development and operational1 control


Two methods of controlling the effect of tidal energy on the power system have been identified as being the use of tidal superposition and the use of electrical down rating (EDR) and generation reduction (GR).

The use of the word operational in this context refers to the operation of TEDs and not the grid system.

65

4.2.1

Tidal Superposition

It is possible to greatly reduce the diurnal variation in the total power output from a number of turbines, or turbine groups, by locating them at different times of high water around the coast. The examples in Figure 4-1 show two similarly sized groups located 3.1 hours apart, or 4 groups located at 1.55 hours apart.

Figure 4-1. The total power output profiles for identical turbines situated at 6.2, 3.1, 2.07 and 1.55 hours apart.

The resource assessment presented in Chapter 3 indicated that with the technology currently available, there is a maximum separation time of 1.25 hours between areas of feasible resource around Ireland. This prevents the effective use of tidal superposition around Ireland. However, the development of renewable energy systems by developed countries has a two-fold approach: firstly to reduce to the emissions of the country in question but perhaps, more importantly, to provide a mature technology for developing countries such as China and the Philippines to use on a much larger scale, providing global benefits to a global problem. Therefore this aspect of tidal energy has been included within the study so that other countries may benefit.

66

4.2.2

Electrical Down Rating (EDR)

The power output from a turbine or group of turbines will only reach its maximum output during a spring tide, which occurs for a short time twice a month. Therefore it is not envisaged that developers would consider it economically viable to rate the electrical equipment to harness all of the energy available at a spring tide. Instead, the maximum power from the turbine would be down rated by altering the pitch of the blades (see Figure 4-2). This has been termed Electrical Down Rating (EDR). If a TED were to undergo 40% EDR of the maximum power available from the turbine, only 10% of the total energy available would be lost (see Figure 4-3). It is therefore envisaged that developers will balance the cost of the grid connection and the rating of electrical components against the capacity saving from spilling energy at higher tidal flow rates.

Figure 4-2. EDR of one turbine by 50% of the maximum energy extractable from the tidal steam resource.

67

Figure 4-3. The energy lost as a result of reducing the rated power output from a group of turbines.

In areas of the world where it is possible to control the semi-diurnal variation in power from a TED, it may also seem worth introducing EDR to reduce the capital cost. However, EDR of two opposing turbines will result in a very peaky net power output because with no EDR, the two power output variations are in opposite directions. When EDR is applied, one output is clamped at the rating for lengthy periods, thus exposing the system to the full ramp rate of the second cluster (see Figure 4-4). This will increase the necessary ramp rate of conventional generation.

Figure 4-4. An example of the spiking that will occur in the total power output from two opposing turbines with EDR. It should be noted that the spring-neap cycle has been reduced from 14.75 days to 4 days in order that the effect can be clearly seen.

68

It is possible to reduce this spiking by having the TSO impose power limiting in a operational sense to reduce the effects of tidal energy during times when it may compromise system security (within Ireland at the levels of penetration available, this would be very rare). This curtailment of generation has been termed as generation reduction (GR). Using GR it would be possible to achieve a steady form of generation during periods of spring-neap cycle (Figure 4-5). The energy cost of doing this can be seen in Figure 4-6. However, it is envisaged that GR would be implemented only when required and therefore the energy cost of doing this would lie between the energy cost of EDR and GR to prevent spiking in Figure 4-6.

Figure 4-5. The effect of EDR on the maximum output of a group of four 8.5 MW turbine groups located at opposing 1.6 hours apart, with the same power output.

69

Figure 4-6. The energy lost as a result of EDR from the maximum power output.

The application of GR to a TED group would mean that this group is naturally carrying reserve (depending on the turbines response time to the event) that could be called upon when there is a system shortfall of generation. Depending on the market this energy may therefore not be lost but may be sold as capacity on the reserve market.

4.3 The effect of tidal generation on net generation profile


The electricity market is somewhat unique in that production must match demand on a second-to-second basis. If this balance is not maintained the voltage and frequency will fall outside acceptable levels and customers will have to be shed from the system to protect their equipment from damage and the system from collapse. Therefore the rate and predictability of change for electricity demand is of extreme importance to TSOs. Tidal energy can be considered to be extremely predictable, however by its very nature it is known to fluctuate.

Tidal generation has its greatest effect during a spring tide when the tidal flows are strongest. Although the time of high water moves forward by 50 minutes each day, the time of high water during a spring tide occurs at about the same time of day. Thus the average annual energy depends on the time of day, as shown in Figure 4-7. This correlation of tidal current with time of day, and hence system demand, reflects the suns influence on the tidal pattern.

70

Figure 4-7. Annual average tidal power and average winter demand over a day.

The seasonal demand profiles for 2006 and 2010 were determined from 15-minute spot data from ESB for 2003 and 30-minute metered data from NIE, again for 2003. The house load was assumed to be 4% of this and was removed accordingly. The data were scaled up by assuming a growth in load in NIE of 1.8% per year and a growth of 3.2% per year on the ESB network. The 30-minute NIE data were used to produce 15-minute data by means of linear interpolation. This was considered to be possible because the data is based on metered data and not spot data. The two sets of data were then added to give an all-island demand for 2006 in MW every 15 minutes (see Figure 4-8). 30 min. NIE 2003 demand 15 min. ESB 2003 demand

-4% for house load

-4% for house load

Increase 1.8% per year

Increase 3.2% per year

Linearly interpolated to 15 min. data Add both data sets

Figure 4-8. Method of scaling system demand up to 2006 and 2010 data.

71

The seasons were divided according to the dates used by both TSOs that are representative of times of similar demand on the system (see Table 4-1).

Season Autumn Winter Spring Summer

Start Date 1st September 1st November 1st March 1st June

End Date 31st October 28/29th February 31st May 31st August

Table 4-1. The seasons as defined by both SONI and ESBNG

During a neap tide the tidal generation does not have much impact on demand profile. However, during a spring tide (time of maximum resource availability) the tidal generation will reduce the point of maximum generation, and hence the most expensive form of generation from the profile (see Figure 4-9). However, tidal generation is also seen to reduce the point of minimum generation and delay and hence steepen the morning rise seen by other generators (see Figure 4-9). Reducing the point of minimum generation may in some cases force a large thermal plant to turn off due to its minimum generation constraints (described in more detail in Chapter 5).

Figure 4-9. The average net generation profile for each season in 2006.

72

4.4 The effect of tidal generation on the system ramp rate


Generation must ramp up and down to match system demand. A concern of TSOs relating to variable forms of generation is the maximum ramp rate to which the system is exposed see the demand variation in Figure 4-9 between 5 and 10 h. To quantify this effect, the 2006 seasonal demand profiles (obtained as described above) were considered.

Comparisons were made against the normal system ramp rates under the following case studies: Case 1 The effect of increasing amounts of tidal generation on system ramp rates Case 2 Ramping on the system during a spring neap cycle Case 3 The comparison between tidal and wind energy in terms of the effect on system ramp rate Case 4 Effect of EDR on the system ramp rates

In all of the cases above it was necessary to scale the tidal energy production. This was preformed simply by multiplying the original tidal energy output (from chapter 3) by a factor to get the desired output. This method was checked against tidal power profiles generated from the resource assessment model for inconsistencies and was found to produce a good representation of the tidal generation (see Figure 4-10). The use of such a scaling technique for further ramp rate analysis will reduce the amount of processor time required enormously with little effect on the outcome. Although the viability method would at first seem a more accurate method, it must be remembered that even with this method the order in which the viable sites are to be developed is difficult to ascertain.

73

Figure 4-10. The difference between scaled and modelled tidal generation ramp rates.

4.4.1

Increasing amounts of tidal generation

The TSO is concerned with the worst-case scenario during any given season and the frequency of these events. The reason for the TSOs concern about the season lies in the amount of generation that may be available to respond to high levels of ramp rate. During summer for example, a lot of plant is normally out on maintenance.

The addition of tidal generation scaled to 72 MW of average power production (374 MW of total capacity) has been found to have an asymmetric effect on the system ramp rate, with a much greater influence on maximum ramp rate (rate of increasing demand on the system) than on the minimum ramp rate (rate of decreasing demand on the system), as shown in Figure 4-11. To determine the long-term worst-case scenario on the system the tidal profile was phase shifted in steps of 1 day from 0 to 7 days. Each of these cases were re-run and the worst possible scenario for each season has been identified and presented in Figure 4-11.

74

Figure 4-11. The worst-case effect on the 2006 demand ramp rate following the subtraction of 72 MW of average tidal generation (374 MW installed, with no EDR) phase shifted in steps of 1day from 0 to 7 days.

It is also apparent from Figure 4-11 that the system naturally experiences an asymmetric effect with demand increasing at up to 24 MW/min whilst the fastest rate of reducing demand is 18 MW/min. The frequency of these events and the effect of increasing tidal generation can be understood by studying Figure 4-12 in which, again, the existing asymmetry of the system ramp rate can be seen. The generation of an average of up to 100 MW of tidal generation onto the system can be seen to have limited effect on the net system ramp rates (see Figure 4-12).

Figure 4-12. The frequency of the net system ramp rates for 2006 when considering an average tidal generation of: 0 MW, 50 MW, 100 MW, 200 MW and 300 MW average generation. 75

The source of the asymmetry in system ramp rates can be understood when considering the average system demand for each season (see Figure 4-9). The demand can be seen to ramp up sharply during the early morning. However, it tails off more gradually in the evening. It is noticeable that when the tidal generation profile is netted off the system demand, the morning rise in generation can be seen to steepen because the tidal generation has delayed the onset of the morning rise and then has stopped due to slack water (although the time of high water moves forward by 56 minutes every day, the time of high water on a spring tide is almost constant).

4.4.2

Ramping on the system during a spring-neap cycle

The time of maximum tidal stream power moves forward by approximately 56 mins every day. Therefore, although during the day of the spring tide the ramp rate on the system increases, it is conceivable that there may also be times when the tidal generation reduces the ramp rate on the system. To test this argument the maximum ramp-up and ramp-down rates were considered for a 60-day period at the beginning of each season. In Figures 4-13 & 4-14 there are two obvious patterns. Firstly, the ramp rates decrease during the weekends. Secondly, the system ramp rates increase every second week due to the spring-neap cycle. Whilst there are days during which the maximum ramp rate has been reduced by tidal generation, on the whole tidal generation causes the maximum system ramp rate to increase (see Figures 4-13 & 4-14).

76

Figure 4-13. The maximum ramp-up rates experienced by the system each day during the first 60 days of each season, with an average of 100 MW tidal generation.

Figure 4-14. The maximum ramp-down rates experienced by the system each day during the first 60 days of each season, with an average of 100 MW tidal generation.

77

4.4.3

Comparison between wind and tidal energy

To compare the effect that wind and tidal energy have on the system ramp rates, the tidal energy was scaled to have the same average output as that predicted for wind generation in 2006 (300 MW) and 2010 (692 MW). The tidal energy with no EDR in both 2006 and 2010 was found to have a much greater impact on system ramp rate than the wind energy (see Figure 4-15). The maximum system ramp down rates where seen to increase in magnitude from 13 MW / min to 25 MW / min in 2006 and out to 40 MW in 2010. However it must be remembered that both these values are assuming a level of tidal energy which is unrealistically high for the purposes of comparing it to wind generation.

Figure 4-15. The all island system ramping scaled to 2006 with a predicted average wind generation of 300 MW and a matching average tidal generation (unrealistically large, simply used to compare against wind).

Tidal energy is seen to have an extremely heavy impact on the system ramp rate when compared with wind energy. Whilst this effect on the system is predictable and can be planned for in the system dispatch, it may be seen as a factor which would limit the percentage of tidal generation in areas which can not benefit from generation from opposing tidal flows.

4.4.4

Effect of EDR and GR on the system ramp rates

The potential benefits from EDR in terms of reducing ramp rate were studied by determining the net generation ramp rate from the theoretical outputs of two similar 78

TED clusters with identical resource sited 3.1 hours out of phase (superposition). With tidal superposition the ramp rate would decrease (see Figure. 4-16). However, the level of EDR required to cause this was considerable because the point of maximum ramp rate occurs mid way down the power curve (see Figure. 4-2). The decrease occurs in a stepwise manner because no one point remains the maximum point continually as the EDR is increased. Instead a particular point during the season is found to have the maximum ramp rate. If the EDR is increased until this point is removed, then a new point is found, so the ramp rate is decreased in steps. Further studies of the effect of EDR on the system ramp rate for the predicted Irish tidal resource indicate that this stepwise effect is very prominent due to variations in the demand profile.

In systems with tidal superposition the ramp rate will initially be much lower because the diurnal variation has been reduced. However, if EDR is applied to these systems, the resulting spikes in the net power output, as seen in Figures 4-4 & 4-5, cause the combined ramp rate to increase from the profile seen for two TED areas under GR to that seen for one TED area under ERD (Figure. 4-16).

Figure 4-16. The effect of EDR on the rate of power output ramping from the turbines.

4.5 Capacity & availability factors


The capacity factor of a generator is a measure of the time it is available to the system to provide the maximum constant rated output. Historically the term capacity factor has 79

been used for two purposes: firstly by investors to understand the expected rate of return for their investment; and secondly by TSOs who wish to understand how often the plant will be providing generation onto the system. Whilst with conventional thermal generation the two were the same and could be described in one parameter, with tidal stream generation the two should be treated as two separate parameters. The capacity factor will refer to the rate of return for investment purposes, whilst the availability factor will refer to the amount of generation the TSO can expect the unit to provide for the system. The capacity and availability factors were determined with the use of the following equations.

cf =

PTurbine PTurbine

Mean Max

af =

PTotal PTotal

Mean Max

(4-1)

PTurbine

Mean

= the mean power output from an individual turbine following EDR and GR

PTurbine

Max

the maximum power output from an individual turbine, following EDR and GR

cf = PTotal PTotal
Mean

capacity factor = = the mean power output from all turbines following EDR and GR the maximum power output from all turbines following EDR and GR

Max

af =

availability factor

The capacity factor from a tidal turbine permitted to generate the maximum power available to it from the tidal flow during a spring tide will be very poor (~0.18) due to the very short period that the turbine will operate at the maximum output. However, this will increase by increasing the EDR of a single turbine or turbine group (see Figure 417). Under 50% EDR a turbine can be expected to give a capacity factor of 30% (a similar value to that for wind generation). In Figure 4-17 it can be seen that the use of GR has much greater benefits to the TSO in terms of the units availability factor. The cost of these benefits can be seen in Figure 4-6.

80

Figure 4-17. The availability factor profiles with relation to EDR for a single turbine and two turbines at opposing tidal phases, EDR firstly as separate turbines and secondly as a single generator.

EDR will offer limited operational benefits when applied to turbines at opposing tidal phases (see Figures 4-1, 4-5). The reason for such a limited benefit is that when the power outputs of the turbines at opposing phases are superimposed, following EDR, it results in a very peaky output (see Figure 4-5), increasing the maximum output and therefore decreasing the availability factor. Although fine-tuning the level of EDR can reduce this effect, this can only be done if the magnitude of tidal output remains constant, which is not the case between spring and neap tides.

In scenarios where it is possible to develop tidal energy at opposing tidal phases, the operational regime would be envisaged to be a balance between EDR and GR according to the needs of the TSO to control the fluctuation of tidal generation on the network, and the economic constraints of the site developer.

4.6 Capacity Credit


The term capacity credit is used to assess the contribution a generator will provide towards ensuring there is sufficient generation on the system to prevent a loss of load occurring when generator units fail. Whilst there must be spare generation to cover the system during times when generator units fail, there must not be too much spare

81

generation because this would be too expensive. It is therefore very important whilst developing long-term policy to clearly understand in a quantitative manner the benefit that a particular form of generation, such as tidal generation, will contribute towards the capacity credit of the system.

4.6.1

Scheduled outages

Generators not only fail (for which there is a probability for each unit), but they are also removed for scheduled maintenance. Most of the maintenance is conducted over the summer when the demand is low, with lesser amounts occurring during autumn and spring. However, no maintenance is scheduled during winter when the demand is high. The maintenance times are provided as the percentage of time the unit is out for scheduled maintenance in each season. A routine was developed to remove each unit for the time that it requires each season all in one block in such a manner that the maximum amount of generation is maintained on the system at all times (see Figure 4-18). The maintenance schedule includes all the thermal and hydro plant on the system. However, wind generation has not been included to clearly identify the capacity credit associated with tidal generation.

Figure 4-18.a. The 2005 generator power available to the system following the removal of generators for scheduled maintenance. Each generator is represented by a different colour / shade. 82

This routine functions by adding the generator SOP (Scheduled Outage Periods) in order of size (capacity multiplied by time of outage). On adding each SOP it tests every possible location in time for the SOP on top of the sum of the existing SOPs. The optimal time for the SOP is selected based on which combination results in the minimum total scheduled outage at any given time and the lowest standard deviation of the total SOPs over the season.

Figure 4-18.b. The 2007 generator power available to the system following the removal of generators for scheduled maintenance. Each generator is represented by a different colour / shade.

4.6.2

Calculation of loss of load expectation

The method of determining the LOLE (loss of load expectation) can be explained simply for a three-generator scenario (Transmission System Operator Ireland, 2003), as shown below.

83

Capacity (MW) Unit A Unit B Unit C 10 20 50

Forced outage probability 0.05 0.08 0.1

Table 4-2. Example of three generator units, modified from example in (Transmission System Operator Ireland, 2003).

Case

Force outage state A B 0 0 1 1 0 0 1 1 C 0 1 0 1 0 1 0 1

Capacity (MW)
0 50 20 70 10 60 30 80

Probability

Ability to supply 55MW

Probability of lost load 0.0004 0.0036 0.0046 0 0.0076 0 0.0874 0 0.1036

1 2 3 4 5 6 7 8 Total

0 0 0 0 1 1 1 1

0.0004 0.0036 0.0046 0.0414 0.0076 0.0684 0.0874 0.7866 1.0000

No No No Yes No Yes No Yes

Table 4-3. Example of how to calculate the expected loss of load for a three-generator system (Transmission System Operator Ireland, 2003).

In the three generator system proposed as an example in Tables 4-2 and 4-3, the system could be expected to fail to meet the 55 MW of demand 10.36% of the time, or 907.5 hours a year. However this assumes a constant demand of 55 MW for every time step over the year. With a varying demand the calculation would have to be preformed for each time step and the number of hours of lost load for each time step could then be added to give the total LOLE over the year. Whilst this system works well for a threegenerator system, it is not practicable with a larger system such as the Irish grid system with up to 50 generators. The calculation becomes extremely large with 250 or

(1.13 10 ) possible states per time step. The computational time required to perform
15

this and the amount of information which would need to be stored in RAM is very high. For this reason a Monte-Carlo simulation (Fitzgerald, 2004) was utilized in which each generator was determined to be on or off according to a normalised random number

84

generator weighed to the forced outage probability. The year was then re-run a number of times (until the error margin failed to improve much further) and the average forced outage time was found.

4.6.3

Identification of the capacity credit of tidal generation

In previous studies the capacity credit has been taken to be the amount of the generation that can be substituted with the form of generation being studied (e.g. tidal or wind). However the difficulty with taking this approach is the capacity credit will depend on the capacity credit of the generator being substituted. Therefore to mitigate this problem the capacity credit has been defined as being the amount of extra demand that can be supplied by a determined amount of tidal generation whilst maintaining the specified LOLE of 8 hours for the Irish system (TSO Ireland, 2003). Table 4-19 demonstrates the logic behind the routine used to perform this calculation.

85

Find LOLE Multiply demand up or down accordingly Capacity of tidal generation Is LOLE ~8 hours per year? Yes Subtract tidal gen. from demand Add the addition-supported demand to the demand (0 on first step) Find LOLE Increase / decrease the additionsupported demand No

No

Is LOLE ~8 hours per year? Yes

Addition supported demand / capacity of tidal generation

Figure 4-19. Method of determining the benefit of tidal generation in terms of the additional demand that can be supplied given the same LOLE criteria.

The routine described in Figure 4-19 was run, incrementing the installed capacity of tidal generation. The use of this random technique results in a noisy relationship. This noise was smoothed out using the following running mean technique: Data i = (0.25Datai 1 ) + (0.5Datai ) + (0.25Datai +1 )

(4-2)

The average additional load that could be supplied by tidal energy was found to be about 17.9% of the installed capacity of tidal generation. The methodology was repeated for a tidal pattern which was subjected to an EDR of 50%. This was found to increase the load that can be supplied to 19.2% of installed TG. However it must be remembered

86

that as EDR is implemented, the level of installed capacity decreases, and the capacity / availability factors increase. The tidal record has been phase shifted forward by 3 hours to quantify the benefit to the tidal generation of having the peak tidal resource during a spring tide occurring during a time of peak demand. Under these circumstances the additional load that can be supplied is 11.7% (of installed TG), indicating that through the good fortune of the time of peak spring tidal flow at the feasible areas for development the capacity credit of the tidal generation is increased by 53%.

Figure 4-20. The Capacity credit offered the Irish grid system by tidal generation during 2005.

The long-term capacity credit was studied by phase shifting the tidal record by up to 7 days in steps of 1 day. This covers both the effect of the 14.75 day spring neap cycle and the effect of the demand pattern over the course of a week. It was found that when the spring tides occurred during weekends over the period of peak of winter demand the additional demand suppliable drops from 17.9% to 10%.

4.7 Frequency response of TEDs to disturbances


Under normal operating conditions the load on the Irish system would not be expected to change by more than a few MW per minute. However, disturbances occur on the system that cause much greater changes in demand / supply. Such disturbances can be caused by the failure of a large generator, a system fault, or the loss of a large amount of

87

load. It is the normal policy of system operators to carry sufficient spare generation and interruptible load on the system at any time to cover for the loss of the single largest unit. However, during the time required to ramp up the remaining generators (between 5 to 15 seconds (Electricity Supply Board National Grid, 2005)) on the system there is a power deficit. The same is true for a system fault. If one or more phases of a circuit are permitted to come into electrical contact with each other or ground, a large amount of fault current (reactive power) flows across the point of contact until the system protection isolates the fault.

For the system to remain stable and ride through a disturbance, it must have energy storage that can respond immediately. In traditional systems this energy is contained within the kinetic energy of the large, heavy generators and turbines that are spinning at high speed. If the load should suddenly increase, energy would be extracted from the rotating generator and turbine by causing it to slow down slightly, reducing the system frequency. On the other hand, should the load suddenly decrease, the excess energy could be absorbed by using it to increase the kinetic energy of the generator and turbine (see Figure 4-21).

Figure 4-21. Example demonstrating the inertia response characteristics of synchronous generators on an electrical network (Mullane, 2003).

In traditional thermal plant the generator and turbine are on the same shaft and spin at the system frequency (50 Hz), resulting in a large amount of kinetic energy. To quantify the amount of kinetic energy that is contained within a generator turbine system the inertia constant H is given (see Equation 4-3), which is length of time the generator could supply its rated output solely from its own kinetic energy.

88

1 J 2 H= 2 S

(4-3)

H = inertia constant (s) J = inertia (kgm2)

= angular velocity (rads-1)


S = rated power output (VA)

The ability of generation to provide this short term energy buffer on the system depends on the inertia constant of the generator and turbine, but also on the type of generator used and its ability to access the store of kinetic energy quickly.

4.7.1

Inertia constant

The inertia constant provided by new forms of generation such as wind turbines has been found to be smaller than that of conventional thermal plant (Mullane, 2003). Although a wind turbine looks very large and has a large (50 m) rotational diameter compared to that of a conventional thermal plant, it must be remembered that a thermal plant rotates at high speed (e.g. 3000 rpm for a 2-pole alternator at 50 Hz), whilst the turbine on wind plant rotates between 20-25 rpm and is made of light-weight materials. Further more, to reduce the cost of the gearbox induction machines with 4 or 6 poles have been commonly installed in wind turbines, reducing the high-speed end to 1000 rpm or 1500 rpm at 50Hz. This reduces the amount of kinetic energy stored in the generator (Mullane & O'Malley, 2005; Lalor, 2005). Therefore it is conceivable that a tidal turbine which has a much smaller blade (15 m) and slower speed of rotation (10 rpm) (Whittaker et al., 2003) would contain even less kinetic energy than a wind turbine.

4.7.2

Generator type

There are two basic types of generator that can be used; a synchronous or asynchronous generator. The synchronous generator consists of a magnetised rotor that rotates within three coils (the stator), subjecting them to a changing magnetic field, hence inducing voltage and current flow within them. The induced current flow within the stator results in the establishment of a complementary magnetic field within the stator, which rotates between the poles of the stator to follow the rotating field of the rotor. Whilst the rotor

89

field is leading the stator field the machine is acting as a generator, with the stator applying torque to the rotor field. When the rotor field is lagging the machine is acting as a motor, with the rotor applying torque to the stator field (see Figure 4-22).

Figure 4-22. An example of a synchronous machine with two poles.

The magnetic field within the rotor can be generated either with a permanent magnet or with use of a wound rotor supplied with a direct current through slip rings (see Figure 423). For practical and economic reasons this type of rotor is often used in large generators. However, this means that to start such a generator requires external power either from the grid or, in event of a black start, from a small diesel generator to energize the rotor.

Figure 4-23. An example of a wound rotor synchronous machine with two poles.

The configuration of machine shown in Figure 4-22 is termed a two-pole machine because the rotor has two poles, north and south. For this machine to produce 50 Hz, it must rotate at 3000 rpm or 50 rps. However, as the number of poles is increased the speed of rotation decreases. So, for the 4 pole machine shown in Figure 4-24 to produce a 50 Hz supply, it must rotate at 1500 rpm or 25 rps.

90

Figure 4-24. An example of a synchronous machine with four poles.

The second basic type of machine is an asynchronous machine, which can have either a wound rotor or a squirrel cage rotor. The squirrel cage induction machine replaces the rotor with a series of conducting aluminium rods (see Figure 4-25). As the field from the stator (generated by the system) rotates around these rods, large currents induced within these rods produces a magnetic field which rotates around the rotor to follow that of the stator. Although the field can rotate around the stator, it applies a torque on the rotor. Therefore the rotor rotates to follow the rotating field on the system. Under these circumstances it is lagging and acting as a motor on the system. If the rotor is rotated faster than the system frequency, currents will again be established within the rotor rods producing fields, which will apply torque to the stator, causing the machine to act as a generator.

Figure 4-25. A squirrel cage rotor (Danish Wind Industry Association, website).

The wound rotor induction machine consists of a rotor which contains coils (see Figure 4-26). These coils can be connected to slip rings, as they would be in a synchronous machine. However, rather than supplying them with a direct current, they are shorted out. They may simply be shorted out on the rotor with no slip rings, reducing the maintenance. Therefore they act in much the same way as the squirrel cage rotor. However, on starting a squirrel cage induction machine large amounts of current are drawn due to the rotor being stationary and stator field rotating around it at the system 91

frequency. With a wound rotor machine, because the rotor circuit is accessible, resistance may be added into the rotor circuit during starting to reduce this large starting current. This can be done either with resistors and power electronics on the rotor, or it can be done externally through slip rings (see Figure 4-27b.).

Figure 4-26. An example of a wound rotor induction machine with four poles.

In traditional thermal plant the turbine provides a steady, controllable source of torque to a synchronous generator in which the rotor and stator are strongly linked. However, in renewable generation where the energy source may fluctuate, this is not as easy. It is for this reason that induction machines have become more common on the system. The induction machine can be used in three basic modes of operation: a standard squirrel cage machine can operate with a speed variation of up to 2%. However, with a wound rotor design it is possible to operate it in either a narrow-band or variable-speed mode (with a speed variation up to 15%). When the rotor in a squirrel cage induction machine (see Figure 4-27a) moves away from synchronous speed, the current within the rotor coils increases to a point where it would reach the coils rated limit. Under normal conditions this would be the point of maximum speed variation. However, under narrow-band operation, power electronics and resistors can be fitted to the rotor to control the amount of current that is permitted to flow within the rotor windings as the rotor moves away from synchronous speed (see Figure 4-27b). The variable-speed operation can be achieved by adding power electronics externally to the rotor through slip rings (see Figure 4-27c). The current induced in the rotor, which would have exceeded the rating, can now be used to provide power output to the system rather than being dissipated in resistors. However slip rings require regular maintenance, which is expensive, especially when used in a turbine at sea.

92

A.

B.

C.

Figure 4-27. Connection configurations of a squirrel cage induction machine (A.), a wound rotor induction machine in narrow band operation (B.) and a DFIG (doubly feed induction generator) (C.) (Mullane, 2003).

A very wide speed range can also be achieved by placing the generator behind a full AC/DC/AC converter rated for 100% of the power provided. In this way the generator can rotate at any speed and produce power at any frequency which can then be converted to DC and back to AC at the desired frequency. This system can be operated with either an asynchronous machine, such as a squirrel cage induction machine (Figure 4-28B.) or a synchronous / fixed-pole machine (Figure 4-28A). A B

Figure 4-28. Connection configurations of a synchronous (A) and an induction machine (B) operating through a AC/DC/AC connection (Mullane, 2003).

93

4.7.3

Ability of generators to provide inertial response

The ability to provide internal response on the system following or during a disturbance depends on the generators ability to transfer the change in speed of the system frequency into the rotor and primemover. In narrow-band operation, although the response is slower due to the freedom of the rotor and stator to change relative speed within a given range, the kinetic energy contained within the rotor and prime mover is accessible. However, whilst operating in a variable-speed mode it is not possible for the system to access the kinetic energy (this may change with improvements to the control loop). Refer back to the previous analogy of the system being compared to a rotating bar to which all the loads and generators are connected. Whilst the synchronous machine can be symbolised as being connected with a drive chain in this analogy, the induction machine would be better represented as being connected with a rubber belt, and the DFIG as being connected with a rubber belt and not adding any inertia to the system (see Figure 4-29). A DFIG enables accurate control over the rotational speed of the rotor using the power electronics, so by introducing a supplementary control loop which links the rotor speed to the system frequency it is possible to reintroduce inertia response (Ekanayake & Jenkins, 2004; Lalor et al., 2005). Machines that are behind an AC/DC/AC converter are completely disconnected from the system frequency. However, the DC/AC connection can be programmed to output more power when the system frequency is dropping. It is only able to do this if the machine is able to provide the power in the first place. Therefore the inertial response behind a correctly configured AC/DC/AC converter could be described as being the same as the machines response to a disturbance without the converter present (ignoring the different operating frequencies).

Figure 4-29. Example demonstrating the inertial response characteristics of synchronous, induction and DFIG generators on an electrical network (Mullane, 2003). 94

4.7.4

Method of calculating turbine inertial constant

It is expected that the inertia of a tidal turbine is much lower than that of a wind turbine. Therefore this study will estimate the inertia contained within the turbine and the response that this may provide to the system in the event of a fault.

In the absence of detailed structural data relating to the design of the turbine blade and hub, inertia constant was estimated using the available information. The combined weight of the turbine and hub for a turbine blade of radius 11 m is 42.4 tonnes (Whittaker et al., 2003). It has been estimated that out of this the hub would weigh about 12.8 tonnes and the blades 29.6 tones (see Figure 4-30).

Figure 4-30. The hub and blade assembly used on the 300 kW prototype installed in the Bristol Channel (Marine Current Turbines Ltd., 2003).

In order to estimate the inertia within the turbine blades, the depth and width of blade profile given in Figure 4-31a was measured and the shape of the cross-sectional area was assumed to remain constant along the blade as suggested by Figure 4-31b. It has therefore been possible to divide the blade into 1000 segments and calculate the relative volume of each segment. The inertia was determined using the distance to each segment and the mass of the segment (Equation 4-4).

95

Figure 4-31a. Blade profile (modified Figure 4-31b. The blades under construction from Binnie Black & Veatch, 2001). that were used on the 300 kW prototype installed in the Bristol Channel (Marine Current Turbines Ltd., 2003).

r 2 MDrWr J blade = D W r r

(4-4)

J = inertia (kgm2) M = mass (kg) W = blade width (m) D = Blade depth (m) r = distance from the centre of the hub to the blade section (m)
The inertia of the hub was determined again by assuming it to have a constant density, by dividing it into 1000 rings and by calculating the relative volume of each segment. The inertia can then be calculated using the distance to each segment and the mass of the segment (equation 4-5).

J hub

2rn2 M 2r 2 2rn2 = 2 2rhub n =1000


n =1

(4-5)

r = length of each segment (m)

96

The total inertia on the low speed end of the shaft is therefore given by summing

J rotor = J hub + J blade . However, the inertia constant of the generator on the high-speed
end was taken from literature to be 0.5 s for a 660 kW generator (Mullane, 2003). The inertia on the low speed end of the turbine was converted to the high-speed end according to the following relationship:

Jh =

Jl n2

(4-6)

n = the gearbox ratio Jl = inertia at the low speed end Jh = inertia at the high speed end
Following the above method the inertia constant of the rotor on the high-speed end of the turbine was found to be 0.27 s for a turbine rated to 1.7 MW (0.448 MWs) in a peak flow of 3 m/s and 0.46 s for a turbine rated to 1.0 MW (0.448 MWs) in a peak flow of 2 m/s. In calculating the rating of each of these machines the EDR was taken to be 20% (Whittaker et al., 2003). Therefore the total inertia of the generator and the rotor can be taken to be between 0.77s and 0.96s.

4.7.5

Model of frequency response

A model has been developed (Mullane & O'Malley, 2005; Lalor, 2005), which was aimed at determining the effect of increased penetration of wind turbines on the inertial response of the system to disturbances. The model simulates the 2010 Irish grid system on a single bus-bar, ignoring any transmission effects (see Figure 4-32), and bases the unit commitment on a merit-order dispatch, which has been verified against historic dispatches where possible. The system was developed from the 2004 system, with the assumption that all new plant would be open- or combined-cycle gas turbines and that an addition of 500 MW HVDC interconnector to Wales has been built.

97

Figure 4-32. Frequency response model flow chart (adapted from Lalor, 2005).

The induction machines have been included in the model using a verified fifth order d-q model. The DFIGs used the same model for their induction machine, with a fieldorientated controller added to control the electromagnetic torque and hence the speed of the rotor. The fifth order d-q model was run twice once with the assumption of constant torque, and once with the assumption of constant power output. The constant torque model is based on the concept that the turbine would be operating at the optimal point on the efficiency performance ( C p ) versus tip speed ratio ( ) curve (see Figure 4-33). Therefore as the shaft slows to keep constant torque, the power input from the turbine decreases. This may be the case for most of the time when tidal generation is operating below its electrical rated output. However, if the turbine is required to spill water by feathering the blades, it would move away from the optimal point on the C p versus ( ) curve and, under these circumstances, it would be better to consider the use of the constant power assumption.

98

Figure 4-33. The efficiency performance ( C p ) versus tip speed ratio ( ) curve for variable speed wind turbines (Mullane, 2003).

The model was run during both the summer night valley (minimum inertia on the system) and during the winter peak in demand (maximum inertia on the system). A frequency event was triggered through the tripping of 422 MW of generation, following which the system frequency was monitored in conjunction with the inertial response provided by tidal generation. This approach was repeated, incrementing tidal generation in steps of 100 MW from 0 MW to 500 MW, and changing the type of generator used to examine the effects of a standard induction machine, a DFIG and a DFIG with a supplementary control loop. From these results it is also possible to determine the effect of full AC/DC/AC conversion because, if the AC/DC/AC conversion is set up to respond to a frequency change on the system, it will simply export the generators inertial response characteristic to the system. If it is not set up to respond to a frequency change then it will simply act in the same way as a DFIG.

4.7.6 Frequency response provided by MCTs TED Assuming constant torque, the addition of tidal generation using induction machines caused the frequency nadir to fall further than it would with conventional synchronous generation. In the case of the summer night valley, this is sufficient to cause a pumped storage unit to trip from pumping (see Figure 4-34). However, to understand the effect on the system frequency, the point at which the pumping unit trips was decreased. It can

99

now be seen that whilst all three types of generator will cause the frequency nadir to drop further, the standard induction machine will offer some contribution to inertial response, whilst the DFIG will not offer any. The DFIG with supplementary control loop will offer a large inertial response initially. However, the effect is to slow the turbine down, moving it away from the point of optimal power extraction on the C p versus ( ) curve. This reduces the power output of the turbine (see Figure 4-35) and aggravates the effect on the frequency nadir (see Figures 4-36 to 4-38).

Figure 4-34. Frequency trace following a loss of 422 MW of generation on the system during the summer night valley with 500 MW of tidal generation on the system. Dotted lines indicate the frequency response provided by a pump storage unit tripping from pumping.

100

Figure 4-35. Inertial response following a loss of 422 MW of generation on the system during the summer night valley with 500 MW of tidal generation on the system. Dotted lines indicate the response provided by a pump storage unit tripping from pumping.

Figure 4-36. The effect of up to 500 MW of tidal generation on the frequency nadir during the summer night valley following a loss of 422 MW of generation.

101

Figure 4-37. Frequency trace following a loss of 422 MW of generation on the system during the winter peak with 500 MW of tidal generation on the system.

Figure 4-38. The effect of up to 500 MW of tidal generation on the frequency nadir during the winter peak following a loss of 422 MW.

The constant power model When the model was run with the constant power assumption the DFIG with the supplementary control loop provided a better inertial response than without the supplementary control loop, but was not as good as the squirrel cage induction machine (see Figures 4-39 & 4-41).

102

Figure 4-39. Frequency trace following a loss of 422 MW of generation on the system during the summer night valley with 500 MW of tidal generation on the system.

Figure 4-40. Inertial response following a loss of 422 MW of generation on the system during the summer night valley with 500 MW of tidal generation on the system.

103

Figure 4-41. The effect of up to 500 MW of tidal generation of the frequency nadir during the summer night valley following a loss of 422 MW of generation.

The inertial response of the DFIG with the supplementary control loop remained below that of the standard induction generation because of the timing of the response and the interaction with other machines on the system.

4.8 Power quality


Power fluctuation is a problem which has been experienced with wind turbines, due to tower shadow, velocity shear and velocity gusting. Tower shadow results in an area of turbulence around the tower, so as the blade passes the tower the available energy to the turbine is reduced for that instant. Velocity shear describes the increase in wind speed with height from the ground. It has recently been described as playing a large role in contributing to flicker and is caused by the blades passing through areas of differing wind speeds. Hence producing a cyclic fluctuation in the power output from the turbine. Finally, velocity gusting is more random and the result of the rapidly fluctuating nature of the primary energy source. If the system is operating in a down-rated mode (EDR) and therefore not at the optimal tip speed, it is possible to maintain a constant power to the generator by altering the blade angle. However, whilst it is operating at an optimal power output, these fluctuations in velocity will be transmitted to the drive train and must be absorbed either within the stored kinetic energy of the turbine and generator or by the kinetic energy stored on the grid system from other plant. For a fixed speed generator, power fluctuations will result in torque fluctuations on the drive train, which

104

decreases the life span of the machine. However, by using a variable speed machine, it is possible to absorb the fluctuations in power as changes in turbine speed, thereby reducing the fatigue on the machine. This will result in a more stable power output. Also in a large farm it can be expected that the power flicker from individual turbines will to a certain extent cancel each other out. However TEDs have been found to have a very low inertia constant. The MCT design has only two blades, which may amplify the effect of both velocity shear and tower shadow.

Both the effect of the velocity gradient and the tower shadow are very difficult to model, because they are very much dependent on the current speed, the turbulence in the water, the water depth, the point the tidal cycle and the salinity, i.e. they are very much site dependent. It will be necessary to monitor a grid-connected TED to determine the effect of flow conditions on power quality. However, in order to demonstrate the power quality issues associated with adopting a two-bladed design over a three-bladed design, a simple model has been developed which assumes the velocity profile to follow the seventh power characteristic (see Chapter 2).

The power output from the turbine at any given angular position was determined by dividing the blades into elements, each of which has a sweep area around the turbine. The further out the section is on the blade the greater the swipe area. The power output of each blade section was calculated. However, rather than using a depth averaged velocity, the depth of the blade section was determined by simple trigonometry and the velocity at that depth used to calculate the power output (see Equation 4-7).

105

0.5 As kV 3 (h(t )) Pt = B s =1
N

(4-7)

P = power output t = time N = total number of sections s = blade section h = depth of blade section

= density of water
As = area swept by the blade section s k = efficiency of the turbine Vh = velocity at height h off the bed (see eqn. 2-10 & 2-11) B = number of blades

The power from a turbine with two blades was compared against an identically sized turbine with three blades in the same velocity profile (see Figure 4-42).

Figure 4-42. Velocity profile for a turbine in 35 metres of water with a depth average current speed of 2.5 m/s according to the seventh power relationship (see Chapter 2).

The fluctuations in power input to the drive train are observed to be three times greater for the two bladed turbine than for the three bladed turbine (see Figure 4-43), due to the extra smoothing gained from the additional blades.

106

1.05 Two bladed turbine Three bladed turbine Power output (MW) 1

0.95

0.9

0.85 0 50 100 150 200 250 300 Turbine rotation (degrees) 350 400

Figure 4-43. The comparison of the power output from a two bladed and a three bladed turbine with 4 m clearance to the seabed in 35m of water with an average current speed of 2.5 m/s.

Fluctuations in the order of 7% of the power rating of the turbine is considered acceptable, given that the magnitude of fluctuations experienced by wind turbines is in the order of 12% (Larsson, 2000). However, the use of variable speed turbines will reduce the physical fatigue on drive train and dampen the fluctuations in the output power, improving the power quality.

Tower shadow can be expected to produce very similar results, with the two-bladed turbine producing a much larger fluctuation than the three-bladed turbine. However, developing a simple model of tower shadow is not practical due to the lack of measurements and models of the current flow over the support wing in both directions.

4.9 Conclusion
Tidal generation was found to offer a generation capacity credit of 11.7% to 17.9%. However, further work should be conducted to identify how the increasing penetration of wind generation on the system may affect this and how tidal generation may in turn affect the generation capacity credit of wind generation.

The frequency response of tidal turbines was modelled using an inertia constant calculated from pictures of a turbine and estimates of the components proportional

107

mass. When more accurate structural design data is made available the estimation of the inertia constant should be refined.

108

5. Impact of renewables on thermal plant


5.1 Introduction
Thermal plant has the luxury of an unrestricted and reliable fuel supply (in most cases). This is not the case with most renewable generation such as hydro, tidal, wind, solar and wave, which harness an energy source that can vary over time. This means that such forms of generation must be augmented by thermal plant to reliably supply demand (Sustainable Energy Ireland, 2004; Garrad Hassan, 2003). Concern has been expressed about the effect of increased starts and ramping of thermal plant on the net system emissions (SONI, 2004), and the effect that generation bidding into the market will have on prices. In an attempt to reduce the cost of accommodating variable renewable generation on the system and to improve the emission reductions achieved by renewable generation, industry has started to invest in improving the forecast of renewable generation, in particular wind generation to improve the efficiency of the system (Giebel, 2003).

Tidal generation has been promoted as being a variable but predictable form of renewable generation (Department of Trade & Industry, 2004). This chapter uses the case studies of tidal and wind to develop an understanding both of the impact of tidal generation on thermal plant and the cost and benefits predictability can have on a power system. The benefits will be considered in terms of the effects on emissions, market aspects, and system operations. Such an understanding is important when assessing the benefits of wind forecasting models and national policy for future generation mix.

An initial study of system emission reductions facilitated by tidal or wind generation were initially estimated based on a simple merit order dispatch and did not account for start costs, start-up times and ramp rates. To improve on these results a 1-year unit commitment based study will be developed here, to appreciate the net impacts after annual cycles. The following chapter will present some of the characteristics of the Irish system including a brief description of the electricity market and trading rules and consideration will be given to the nature of the unit commitment problem and the most appropriate way to solve it. To develop an appreciation of the benefits predictability can deliver from a variable source of generation (tidal), wind generation will be included within the unit commitment as a less predictable form of generation to form a 109

comparison against. The benefits will be considered in terms of the effects on emissions, market aspects, system operations and the effect on the usage of other generation.

5.2 Irish case study characteristics


The All-Island power system in Ireland comprises of the NIE and Eirgrid systems as described in Chapter 4, these systems have approximately 58 large thermal generators between them which range in size from 5 MW up to 480 MW operating on peat, natural gas, coal, oil, distillate and diesel. These units participate within a competitive electricity market. The focus here will be on the market and trading rules that are expected to be implemented for the All-Island system, however the underlining unit commitment and dispatch should be very similar in most market systems.

5.2.1 Electricity Market An electricity market can be described as either gross pool or a net pool. In a gross pool market the dispatch of all generators is determined according to their offers, market rules and technical limitations. In a net pool market the majority of the dispatch is determined through bilateral contracts between generators and distributors / large customers, and only the demand not met through the bilateral contracts is settled through a market bidding process in the net pool. Therefore in a gross pool all the demand is settled within the market and any bilateral trading is a financial contract outside of the pool. The market (either gross or net) can be settled in a number of different ways, including; offer price (sometimes called pay as bid), locational marginal pricing (LMP), regional (load weighted price), regional (reference nodal price) and single market price. The single market price pays all the generators within the market the same price which is dependent on the highest cleared offer.

In NI there is currently a central market, whilst in ROI there is currently a bilateral market. However a single electricity market including both NI and ROI is being designed which should be implemented in 2007. This market will be in the form of a gross pool with an ex-post centralised unit commitment and single market-clearing price based on the price of the next available MW (marginal price). However details of

110

capacity payments, no load costs, start costs and constraint payments have yet to be agreed on and finalized. There are a number of rules within the market place to introduce incentives to minimize the constraints on the system, and obtain socialized goals. In almost all markets gaming of these rules and the market structure is possible and must be guarded against, however here the possibility of gaming is not considered. Other important market issues include: Constrained on & off payments: A generator may be instructed to remain on while its offer price is above the single market price, or off whilst its offer price is below the single price, due to transmission or operational constrains. Some markets compensate constrained generators, whilst others do not. Market dominance: Markets in which a single company owns a large amount of the generation have a greatly reduced level of competition. Such a company may at times have the ability to manipulate the single market price through strategically setting the offer prices of all its generation. This is a problem in Ireland (Commission for Energy Regulation, 2004), however it is not considered here. Capacity payments: Investment in new generation plant can be incentivised by providing a capacity payment for plants which are available to generate, but whose offer price is too high to be regularly cleared in the market (peaking plants) (Transmission System Operator Ireland, 2003). Locational factors: With a vertically integrated power generation, transmission and distribution company it was possible to centrally plan when to build generation, and decide the type and the location to optimise the utilisation of the transmission system. However with the introduction of liberalized markets there is a need to incentivise generation location. There are a number of ways to incentives the location; such as an LMP (Locational Marginal Pricing) market, or to base the TUoS (Transmission Use of System) charge on congestion at each point in the system so that those in sections of greater congestion have to pay more. The generation location on the transmission system also affects the transmission system losses, therefore to incentives generation to consider the effect its location will have on system losses, they can be charged for the transmission system losses that they incur according to the TLAF (Transmission Loss Adjustment Factor). 111

5.3 Unit commitment


Balancing of supply and demand is divided into a number of time frames; power system dynamics in the millisecond range, automatic generator control in the second range, economic dispatch in the minute range, unit commitment in the hour range, operational planning in the month range, and finally the network and generation planning is in the year range. Here the focus will be placed on the economic dispatch and unit commitment time frames. Economic dispatch involves setting the generators levels economically with respect to constraints, whilst unit commitment deals with the on / off decisions. There are a number of factors which must be considered when studying a unit commitment and economic dispatch for a system; operating costs, technical constraints and ancillary services.

5.3.1 Operating costs The operating cost for a generator will consist of both fixed and variable costs. Energy is required to heat a plant up before it can start generating. For a large coal plant a considerable amount of energy is needed to get the whole plant up to generating temperature whilst for a diesel generator the amount required would be small. This energy is termed as the start cost and is often given incrementally as the energy required to start a machine from hot, from warm and from cold. The time a machine takes to cool to each point is also important to determine the amount of energy required to start it (see Figure 5-1).

112

Figure 5-1. An example of the energy require to start a coal fired plant and a distillate generator.

Once a machine is generating its fuel requirements in terms of energy can be represented by the following equation (see Figure 5-2). The cost (/h) can then be calculated by multiplying the energy requirement (J/h) by the energy cost (/J).

Eth (E el ) = P 2 + P +

(5-1)

were: Eth (E el ) = Thermal energy requirement for electrical energy output E (J/h).

J /h = Incremental heat rate squared 2 MWh

= Incremental heat rate

J /h MWh

= No load energy requirement (J/h)


E el = Electrical energy output (MWh)

113

Figure 5-2. An example of the heat rates for a coal fired plant and a distillate generator.

5.3.2 Technical constraints The operation of the generation units on the system is dependant not only on cost but

also by the technical limitations of the units. To start a large thermal unit first requires that the boiler and turbine are heated up, this can take up to 18 hours in some cases, whilst other plant such as small diesel generator and aero derivates can start in an few minuets. Therefore whilst it may seem cheaper to turn a plant off in the next time step due to a reduction in load, consideration must be given to when that plant will be required again and the potential cost of not having that plant available due to its minimum start time, as well as the cost associated with starting it. The other major technical limitations include the ramp rate, the minimum level of power output that a generator can operate at before it must be switched out and the maximum capacity of the generator (Dillon, 1998; Walsh, 1998; Wood & Wollenberg, 1996). The transmission system also imposes technical constraints on the system in terms of the line rates between the nodes, whereby the cheapest combination of generation may not be possible due to a congested transmission system (Ongsakul & Petcharaks, 2003).

5.3.3 Ancillary services Ancillary services are also offered by generators, and can be paid for in a number of

ways. These services can include:


114

Reactive power control can be provided in a number of ways such as from synchronous generators and capacitors. The reactive power can be paid for, either directly to generators offering reactive power control (per MVAr) or as part of the transmission cost in cases where capacitors have been installed in critical areas of the system.

Reserve is supplied as the capacity available to respond to a change in the supply demand balance. It may be provided as the remaining capacity on a generator running below its rated capacity or it may be the amount of load being consumed by an interruptible customer (Wood & Wollenberg, 1996). There are two costs associated with this, the first is the cost of providing capacity payments and the second is the increased cost of the dispatch required to carry the amount of the reserve. Normally the reserve is always greater or equal to the largest single supply to the system, however in interconnected systems it may be possible to share reserve with the adjacent system (System Operator for Northern Ireland, 2003; Transmission System Operator Ireland, 2003).

Black start Only some units on the system are capable of starting without the use of an external gird supply. In the event that the entire grid system is lost due to an extreme set of circumstances it would be necessary to call upon these units to reenergise parts of the grid system in an attempt to restart the system. Machines which provide this facility to the system are paid a black start capability payment (All Island Project, 2005).

When the operator is solving the dispatch problem according to the cost and technical constraints of each unit, the cost of the ancillary services should also be considered. If the ancillary services were considered after the energy optimisation has been performed, the ancillary services would only be available from those machines committed. This could result in a very expensive choice of additional ancillary services, or even be infeasible. Other machines may need to be started to fulfil the reserve requirements. This would change the unit commitment due to the minimum stable generation level of the additional machine. Therefore the ancillary services constraints introduces further complexity into the unit commitment problem faced by the system operator.

115

5.3.4 Optimisation and unit commitment There are many factors and limitations to consider when finding the optimal unit

commitment, and there are a number of different methods that can be employed to solve the unit commitment problem. These include:

Complete enumeration: is only applicable on very small systems where it is possible to calculate all combinations of on / off state for a period of time and identify the most cost effective combination. This method is guaranteed to give the optimal solution (when used ex-post) but rapidly becomes computationally heavy as generators are added to the system.

Merit order: is a very simplistic method of producing a unit commitment and is implemented by compiling a list of generators based on the average price per MW at full loading (ignoring start costs) and commits them in this order until the demand and reserve is met.

Genetic algorithm: operates by generating a number of parent unit commitment sequences, these are then cut and spliced with each other and the most costly ones are removed and replaced and the process repeated. Therefore the efficient sections of the unit commitments are joined together and the inefficient parts are removed.

Lagrangian Relaxation: simplifies the unit commitment problem by ignoring or relaxing the constraints between units, optimising each unit individually before reintroducing the constraints. Therefore there is only a linear increase in processing time with an increasing number of units. However it does not handle large numbers of complex coupling constraints easily, such as those associated with combined cycle plants, or cascading hydro plants. It is also common for it not to converge on a feasible solution (Dillon, 1998; Walsh, 1998; Wood & Wollenberg, 1996; Ongsakul & Petcharaks, 2003).

Neural network: involves either the modelling or construction of a large network of interconnected processing nodes neurons. The cost function is then mapped to the energy function of the network (Walsh & O'Malley, 1997; Dillon, 1998; Sasaki et al., 1992).

Linear programming: involves defining the linear interdependency between constraints and variables. This method has the advantage that it is computationally fast and will always find a solution. However the unit commitment problem is not linear and therefore it cant find the optimal
116

solution. However it can be used in combination with other solvers to reduce the computational time.

Dynamic programming: is similar to complete enumeration, however it does not follow through with every possible combination. Instead it looks at the viable options that are open in the next few time steps; at each time step it will only continue to map out a given number of the lowest cost combinations. However the problem with this method is by leaving out possible commitments over time, the chance of finding a near optimal solution reduces, especially when there are plants on the system with long start times (Dillon, 1998; Wood & Wollenberg, 1996).

Mixed integer programming: uses a combination of linear programming to find a upper and lower boundary to the solution area, and a branch & bound or branch & cut technique in which the problem is divided up and an integer solutions found for each sub-problem or it is identified as falling outside the solution area (Takriti & Birge, 2000; Streiffert et al., 2005; MOSEK, 2002).

Most commercial packages make use of either lagrangian relaxation or mixed integer programming to get close to the optimal solution, whilst those used by planners are often solved using linear programming for quick estimations. In NI the central market, is dispatched according to the PK unit commitment software (developed by Peter Kellen, Power Optimisation Ltd.), whilst in ROI the bilateral market, is balanced by experienced loading engineers. The new all island electricity-trading market is being studied using PLEXOS (product of Drayton Analytics) by both system operators and both regulatory bodies. The quality and speed of the solution depends on the manner in which it is defined and presented to the solver. Most commercial packages define the problem to a secondary solver such as CPLEX (developed by ILOG inc.), or MOSEK (used by PLEXOS). Packages such as PK have a very elegant way of defining the problem which results in fast accurate answers, however this has been possible because they have been written and refined to define a set problem in a given system within a given market. PLEXOS on the other hand has been developed as a generic study tool which enables the user to input any system and any market with any generators and to test different strategies and scenarios both in the short and long term. Therefore the problem is defined in a much more flexible and less elegant way, compromising solution time and performance. Given the need for a flexible study tool and in order to
117

remain compatible with the research being undertaken into the future all island market, PLEXOS has been chosen as the software package to use for this study (following testing and verification of its solutions). PLEXOS uses the MOSEK solver which enables the optimisation to be solved quickly as a linear program by relaxing the integer constraints, with greater probability of determining the optimal solution using the mixed integer programming (MOSEK, 2002).

5.4 Method of adding wind to the commitment


The addition of wind to the unit commitment is complex due to the unpredictability of wind. There are two basic approaches to dispatching wind; the fuel saver approach and the forecast approach.

5.4.1 Fuel saver approach The fuel saver approach backs off conventional generation to accommodate power from

wind generation when it happens to be there. However no conventional plant will be switched off under this scheme and the minimum generation level of the conventional plant becomes a limiting factor on the amount of wind that can be accepted onto the system. Until recently SONI operated on the fuel saver approach due to the limitations of the wind forecast model used to operate the system. This model (termed as TREE) is based on persistence (the most reliable short term forecast) and involves the TSO looking out the window; if the leaves on the tree are blowing, then the TSO can expect wind generation for the next 3 hours! SONI upgraded this method of forecasting in 2006 to ANEMOS (Giebel, 2003; ANEMOS, website).

5.4.2 Forecast approach The forecast approach allows conventional plant to be switched off as long as there is

sufficient reserve to hedge against uncertainty in the wind forecast. Previous work has found the error margin of the forecasted wind to increase with time (Bryans et al., 2005b). Therefore to ensure against the unpredictability of wind generation requires less reserve in the near future than in longer time frame. Therefore when carrying reserve on fast start machines such as gas turbines, more reserve can be supplied as standing reserve instead of spinning reserve, saving fuel. However as the forecast horizon increases so does the reserve requirement for wind, therefore whilst the fast start plant may be able to provide sufficient reserve for the first hour it may be necessary to start a larger plant to ensure there is sufficient reserve capacity 2 or 3 hours out.
118

There are a number of statistical and physical methods of forecasting wind speed; there is the persistence method, the numerical weather predictions model, and a combination of the two. The persistence method works well in the first few hours but after that it is surpassed by the numerical weather predictions. The meteorological offices (met office) operational model (produced every 6 hours by the met office at a grid square resolution of 12 km) has an error associated with the modelling simplifications (see Figure 5-3), e.g. the reduction of the topography to a resolution of 14km. Therefore even if the current data from satellites and met stations around the country is used to force the model, there would be an interpolation error at time zero (see Figures 5-3 to 55). Therefore by combining the two techniques it is possible to predict the weather in the first couple of hours according to persistence and after that according to the numerical weather predictions. A project which aims to do this is the ANEMOS project (see Figure 5-4) which is what Eirgrid and SONI are considering using to manage large amounts of wind on the Irish system.

Figure 5-3. The error reported between the forecast and recorded wind speed for the

Aldergrove met station in Northern Ireland (Bryans et al., 2005b).

It is also possible to improve the error by nesting a second model inside the 12 km grid square model to more accurately represent the topography around the wind farm, an example of this is the WAsP model (ANEMOS, website). Secondly the prediction can be compared with what is actually happening. If it is found the forecast has the right profile but is slightly out-of-phase it is possible to shift its phase to give a more accurate forecast. Similarly with the amplitude, if it is found to have the correct profile, but to be slightly out on amplitude it is possible to make the appropriate corrections and improve the forecast, an example of this is the Model Output Statistics (MOS) model (ANEMOS, website).
119

The ANEMOS project includes both persistence and physical model techniques to improve the forecast (Figure 5-4). The numerical model is prepared using the data from a physical model such as HIRLAM, which is passed through the WAsP model to account for local topography and finally through the PARK model to account for turbine wake effects and power curve (collectively termed as HWP). The results can be improved using statistical techniques both with the WAsP model and with the PARK model, by applying the MOS model.

Figure 5-4. Comparison of persistence against modelled wind speed (Giebel et al.,

2003).

It has been possible to generate a similar curve to ANENOS in Figure 5-5 which is based on data from Ireland. This was achieved using actual wind speed data (recorded at Aldergrove weather station) and the met office predicted wind speed for Aldergrove, both of which were provided by SONI in support of this study. The error limits given in Figure 5-5 contain the wind speed error 90% of the time. A confidence level of 90% was chosen based on the reliability of conventional thermal plant. However, the confidence level can be increased or decreased to provide more or less security as required.

120

Figure 5-5. The absolute wind speed error limits, containing 90% of the errors (i.e. a

confidence level of 90%) from the Aldergrove (Northern Ireland) met. station and numerical forecast assuming.

The reserve necessary to cover for the uncertainty of wind generation depends on where the turbine is operating on the power/speed curve (see Figure 5-6). If its operating at high wind speeds near the top of the curve, a change in wind speed will have minimal effect on the power output. However, if operating mid-way up the curve (7 - 15 m/s), a change in wind speed would have a large effect on power output. Such effects can be seen in data recorded from wind turbines (Sustainable Energy Ireland, 2004). The predicted operating point and the wind speed error are used to determine the power error from the power curve (see Figures 5-6 & 5-7).

121

Figure 5-6. The wind power curve selected to represent the entire islands wind

generation, based on a 1.3 MW bonus turbine (Idaho National Laboratory, website).

Figure 5-7 depicts the reserve required for a 1.3 MW turbine with 3 hours uncertainty, assuming a confidence level of 90%. This was determined by looking up the wind speed error for 3 hours ahead on Figure 5-5 and applying it to the power curve in Figure 5-6 to calculate the power range. Note, it falls off to zero at high wind speeds because at the upper end of the power curve a change in wind speed will have little or no effect on the power output (see Figure 5-6). However when wind generation is the largest in-feed to the grid system there is a risk it could trip out during a storm due to high wind speeds. This work has not provided reserve for this but has made the assumption that TSOs would dispatch the system in a more secure manner when storms are forecast.

Figure 5-7. Reserve requirement for a 1.3 MW turbine with 3 hours uncertainty,

assuming a confidence level of 90%.

122

PELXOS does not provide a method to include a forecast that can be updated at regular intervals. So it is not possible to run the model and to update the wind forecast after each time step. The ideal way to solve this problem would be to develop a custom written unit commitment model, developed to define the stochastic optimization to a solver. However the development of such a model is outside the scope of this project.

The unit commitment for wind was solved in two stages - first assuming wind to be fully predictable, then to account for the uncertainty in predication by modifying the first solve. The initial predictable wind case was simulated by subtracting the annual wind profile from the demand, and solved in PLEXOS for the new demand. However, to account for unpredictable wind power, extra reserve must be carried, so a minimum reserve target was set to cover for the uncertainty of wind generation with a three-hour horizon (see Figure 5-5). The three-hour reserve horizon enables most upper merit plant on the system to start. This was calculated for each time step according to the methodology described previously. The method of provided this reserve is shown in Figure 5-8, enabling most upper merit plant to start and extending over the reserve timeframe with the greatest change in forecast error (see Figure 5-5). Accurate data describing the number and type of wind turbines installed, and to be installed is not available on which to base a custom power curve. Therefore the power curve of a 1.3 MW Bonus turbine (see Figure 5-6) was scaled to represent all wind turbines on the system.This neglects geo-spread from a reserve point of view, studies have indicated the maximum time difference in output between wind farms in Ireland is 3 hours although it is normally less than 1 hour (Bryans, 2006) however there have been no accurate studies conducted on the of geo-spread on the system reserve requirements. The additional reserve requirement for wind uncertainty means at times the original unit commitment from PLEXOS has a shortfall in the reserve it can provide. Therefore a logic routine was developed to take the unit commitment output from the first solve in PLEXOS and modify it to provide the additional reserve for wind uncertainty. A merit order of plants on the system which can be used to provide the reserve was established for each 12 hour period based on the cost they would incur to run at minimum generation during reserve shortfalls, considering the start-up costs and base-load cost. The reserve was then dispatched in this merit order until there was no reserve shortfall. The reserve requirements for 1 hour and 2 hour forecast horizons were also calculated. The units committed were de-committed to standing reserve during periods when the units
123

providing reserve could be started in less than 2 hours whilst sufficient reserve was still available on the system to cover the 1 or 2 hour reserve horizon. The resulting unit commitment was then fed back into the PLEXOS model as a pre-committed case, apart from the hydro plants which were fixed to the original dispatch, giving the final unit commitment and the economic dispatch.

Figure 5-8. Method of including wind generation within existing unit commitment

(UC) software.

This method provides an approximation of a rolling 3-hour unit commitment for wind reserve, and solves the problem with an aim to minimize the cost of running the reserve carrying units at or above their minimum stable levels.

The main disadvantage with this method of accounting for the uncertainty in wind is, it does not perform a complete optimisation of the entire problem, but rather solves it in two steps. However given this is only a small level of additional generation on top of the unit commitment problem this is considered acceptable (Bryans, 2005). It also does not quantify the effect of actually using the reserve.

5.5 PLEXOS Model set-up


PLEXOS is continually being developed by Drayton Analysis with new versions being released on a regular bases. Whilst a licence was purchased which enables the project to upgrade to these new releases it was decided to test a version rigorously and then freeze the version i.e. not upgrading due to the risk of bugs being introduced in later versions. The version, which was tested and utilised throughout the project, was 4.857 R1. The model of the All-Island system was constructed in PLEXOS using a combination of data sources as indicated in Appendix 1.

124

The dispatch model was initially developed to include a description of the transmission system, which was solved using a DC load flow. This allowed the inclusion of transmission system constraints. However initial simulations found transmission system congestion to be very rarely a limiting factor. To confirm that PLEXOS was solving the transmission system flows in a realistic manner the model was converted to PSSE format (using a routine developed for the purpose) and solved within PSSE as a steady state model using the generation profile selected by PLEXOS during the period of transmission constraints and the same lines indicated overloading. Therefore having confirmed this constraint has a negligible effect, it was removed to simplify the problem and reduce computation time.

The generator units were described within the model as generating between the maximum and minimum stable levels, while not exceeding the maximum ramp up or down rates when changing generation level. The times to cool from hot to warm and from warm to cold were also provided, with the associated energy required to bring the plant up to synchronization from each state. This enabled the energy cost to be determined as a linear interpolation based on the unit off time. The minimum start-up time to synchronize the plant from hot was also provided as a constraint.

The system reserve was determined on each time step as being the largest infeed, and the reserve provision that each generator can make towards this was included as a constraint. The energy cost of thermal plants was based on each units heat rates. Hydro plant were not considered to have any energy costs. The energy inflow to the head storage area was specified, limiting the plants total energy production. In the case of pumped storage the efficiency of the pumping/generation cycle was specified along with the storage size. The fuel type that each generator uses to start and then to run was specified, enabling a cost to be determined based on the fuel price and its calorific value. The fuel prices used in this study are based on 2007 predictions (ESRI, website).

The most rigorous solver that PLEXOS has to offer is the mixed integer solver (MIP), this took approximately 15 hours to solve for 12 months at a resolution of 1 hour running on a 3 GHz XEON dual processor, dual core machine. Therefore to solve for the 26 cases studied for one year took about 2.5 weeks.

125

5.5.1 Small test case To prove and demonstrate the performance of the routine to add wind into the

commitment according to the methodology described in Figure 5-8 a small test case was developed which was tested rigorously. A few of the test results are presented below to provide a demonstration of the methodology because it would not be practical to present the results of the actual system in such a comprehensive manner due to the size of the system.

The test case included the following generators and demand (see Table 5-3 and Figure 5-9) on a single bus bar system and was solved using a mixed integer solver. This particular case includes a constant system reserve of 100 MW rather than the standard provision of the largest single infeed and does not impose any constraint on the amount of reserve any machine can provide, simply to make it easier to understand the performance of the routine. No load Incremental Min up Min down Start time (h) cost cost Cost time (h) G1 50 120 0 2 400 300 12 G2 80 200 0 2 650 500 10 G3 25 60 0 6 160 500 14 G4 60 120 0 6 75 120 11 Wind 1 50 0 0 0 0 0.1 Table 5-3. The generator properties of the small test case used to test the model. Gen name Min gen (MW) Max gen (MW)

Figure 5-9. The demand curve used within the test case.

126

Figure 5-10. Wind profile used with the error margins based on the curve in Figure 5-5

to account for 90% of the errors in wind speed.

The extra reserve required at various times, according to this method can be seen in Figure 5-10 and in Table 5-3. It can also be seen in Table 5-3 that to provide this reserve there was a need to change the unit commitment on 4 of the hours during the simulation.

127

First run
Unit 1 Unit 2 Unit 3 Unit 4 Wind Wind res. need Available res. System res. Res. Shortage After wind res. Unit 1 Unit 2 Unit 3 Unit 4 Wind

1 50 173 0 60 10 8 58 100 0 1 50 173 0 60 10

2 50 155 0 60 5 5 75 100 0 2 50 155 0 60 5

3 50 145 0 60 0 0 85 100 0 3 50 145 0 60 0

4 50 128 0 60 10 8 103 100 0 4 50 128 0 60 10

5 50 115 0 60 15 11 115 100 0 5 50 115 0 60 15

6 0 157 0 60 27 17 3 100 14 6 0 132 25 60 27

7 0 156 0 60 32 18 5 100 14 7 0 131 25 60 32

8 0 153 0 60 50 0 8 100 0 8 0 153 0 60 50

9 50 140 0 60 50 0 90 100 0 9 50 140 0 60 50

10 50 177 0 60 43 18 53 100 0 10 50 177 0 60 43

11 50 195 0 60 33 18 36 100 0 11 50 195 0 60 33

12 50 200 0 79 12 9 11 100 0 12 50 200 0 79 12

First run
Unit 1 Unit 2 Unit 3 Unit 4 Wind Wind res. need Available res. System res. Res. Shortage After wind res. Unit 1 Unit 2 Unit 3 Unit 4 Wind

13 50 198 0 60 30 18 33 100 0 13 50 198 0 60 30

14 50 192 0 60 28 17 38 100 0 14 50 192 0 60 28

15 50 172 0 60 44 17 58 100 0 15 50 172 0 60 44

16 50 166 0 60 50 0 64 100 0 16 50 166 0 60 50

17 50 181 0 60 50 0 49 100 0 17 50 181 0 60 50

18 50 200 0 75 50 0 15 100 0 18 50 200 0 75 50

19 50 200 0 89 32 18 1 100 17 19 50 200 25 64 32

20 50 200 0 90 24 16 0 100 15 20 50 200 25 65 24

21 50 200 25 66 12 9 60 100 0 21 50 200 25 66 12

22 50 200 25 70 0 0 55 100 0 22 50 200 25 70 0

23 50 200 0 76 0 0 14 100 0 23 50 200 0 76 0

24 50 194 0 60 0 0 36 100 0 24 50 194 0 60 0

Table 5-4. The unit commitment of the small system described in Table 5-3, prior and

post processing by the routine described in Figure 5-8. The units altered by the routine are highlighted in red.

5.6 Effect on emissions


The tidal record was scaled from that of the feasible resource (chapter 3) in increments of 100 MW from 0 MW to 500 MW and then subtracted from the demand. The wind record was taken from the All-Island project data set and was scaled in increments of 50 MW from 0 MW to 450 MW and in increments of 200 MW from 600 MW to 3000 MW. The unit commitments for both certain and uncertain wind forecast have been

128

included to determine the benefit of predictability. The case was solved for the 2007 system (All Island Project, website; Transmission System Operator Ireland, 2005) using the MIP solver within PLEXOS. Wind generation and tidal generation have different capacity factors: ~0.19 for tidal and ~0.35 for wind (Sustainable Energy Ireland, 2004; Garrad Hassan, 2003). Therefore to make a comparison of them as predictable and less predictable forms of generation, the net system emissions have been presented against the energy delivered by each as a percentage of the 2007 All-Island energy demand.

The generation which is most likely to be displaced by wind, or tidal generation is the mid and upper merit plant such as oil and gas (see Figure 5-11). Therefore it is the emissions from these plants that will be reduced.

Figure 5-11. The generation fuel mix and average cost of the plant on the Irish gird

system for 2007 (ESRI, website).

5.6.1 Carbon dioxide The amount of CO2 emitted is dependent mainly on the amount of carbon contained in

the fuel and the amount of fuel burnt to produce a unit of energy (see Figure 5-12). Therefore as the usage of fuel containing carbon is reduced so will the carbon dioxide emissions (Hunter, 1982). This does not however imply that fuel consumption will decrease linearly with decreasing demand, due to changes in generator efficiencies at different loadings.

129

Figure 5-12. The mass of carbon contained within the fuel required to produce a joule

of energy.

Tidal generation and certain wind provided a net reduction in CO2 emissions, there was an average reduction of 2.65 %CO2 per TWh produced by tidal (or certain / predictable wind), however uncertain wind only offered a reduction of 1.6 %CO2 per TWh (see Figure 5-13). The reason tidal generation offered a greater reduction in CO2 emissions is that no additional reserve was being carried for tidal generation. Note that in the following figures the effects of tidal generation have been depicted up to an installed capacity of 500 MW (2.7 % of system energy demand) because with the current technology there is only a viable resource of 374 MW installed (see Chapter 3). Whilst the resource may increase over time as the technology progresses, the generation portfolio of the system will also change and would invalidate any results above this level of capacity.

Figure 5-13. Effect of wind and tidal generation on net system CO2 emissions.

130

5.6.2 Sulphur dioxide Much like carbon dioxide sulphur dioxide emissions depend on the amount of sulphur

in the fuel (see Figure 5-14). However the burning efficiency of SO2 is also very important because, whilst sulphur is obtained only from the fuel, it can react with calcium carbonate or calcium oxide to produce calcium sulphate instead of sulphur dioxide (eqn. 5-2). The calcium sulphate can then be removed within the ash rather than being emitted as a harmful gas. This plays a much greater role than initially expected in the new fluidized bed peat plants at Lough Ree Power, Lanesboro, Co. Longford, and West Offaly Power, Shannonbridge, Co. Offaly (Connolly & Canning, 2005;). The emissions model, developed previously, was updated to account for this (Denny & O'Malley, 2006b).

Figure 5-14. The mass of sulphur contained within the fuel required to produce a joule

of energy.

2CaCO3 + 2SO2 + O2 -> 2CaSO4 + 2CO2 (calcination of limestone followed by sulphation) or CaCO3 + H2O + SO2 -> CaSO4 + H2O + CO2 (calcination of limestone followed by hydration and sulphation)

(5-2-1)

(5-2-2)

Unlike the carbon dioxide emissions uncertain wind generation and tidal offered approximately equal savings of SO2 (see Figure 2-19), there was average annual reductions of 2.0 %SO2 per TWh from tidal and 2.75 %SO2 per TWh from wind. When additional generation was started to supply the reserve necessary to cover for the unpredictability of wind generation, it was necessary to reduce, in general, the

131

percentage loadings. This has a two-fold effect: first, it can shift some generation from plant that are high in SO2 emissions, such as peat, to much cleaner plant which are more expensive such as gas, offsetting the SO2 produced from having additional plant on the system (see Figure 5-15).

Figure 5-15. Effect of wind and tidal generation on net system SO2 emissions.

5.6.3 Nitrogen oxides Nitrogen oxides (NOx) is the collective term for nitric oxide (NO), nitrogen dioxide

(NO2) and nitrous oxide (N2O). NOx is formed during combustion by the reaction of nitrogen present in the fuel (see Figure 5-16) and in the air in the combustion chamber (Delabroy et al., 1998) as shown in Figure 5-17. The proportion of nitrogen converted to NOx depends, among other variables, on both the concentration of air in the burner and the temperature within the burner: the higher the temperature the more oxides are produced (Delabroy et al., 1998). However, a higher temperature also leads to more fuel-efficient generation. The problem becomes complicated further when both combined and open cycle gas turbines are considered. When a gas turbine is run at full load it can be fed with a lean mix of fuel which leads to reduced emissions. However, when loaded <70% and <60% for combined and open cycle turbines respectively, this is no longer possible and the NOx emissions increase (Denny & O'Malley, 2006).

132

Figure 5-16. The mass of nitrogen contained within the fuel required to produce a joule

of energy.

Figure 5-17. The NOx emissions from CCGTs and OCGTs at various running levels

(Denny & O'Malley, 2006).

Predictable generation such as tidal offered greater NOX savings. There was an average annual reduction of 3.24 %NOx per TWh from tidal, compared to average annual reductions of 1.4 %NOx per TWh from uncertain wind generation. To provide the reserve for wind, gas turbines are run more often part-loaded, because they are the marginal units and so must run back to accommodate the minimum generation of the new unit started. Whilst the new unit running at minimum generation is often also a gas turbine, compounding the problem further. Therefore there are high levels of NOx emissions (see Figure 5-18), and reduced NOx savings at higher levels of wind penetration.

133

Figure 5-18. Effect of wind and tidal generation on net system NOx emissions.

5.7 Effect on operation of generation units


An analysis of the average number of units running and started for both the tidal and wind generation revealed that on average the number of units running decreases with increasing penetration of wind or tidal. However, the number of starts performed does not decrease within the penetration of tidal generation studied, whilst for wind, there is an increase in the number of units started. When considering dispatches of uncertain wind generation there is on average more units running (Figure 5-19) than tidal and wind generation requires more units to start (Figure 5-20). Whilst wind generation that was not dispatched to account for uncertainty would require the system to respond in much the same way as for tidal generation with a similar number of units running and (Figure 5-19) a lower number of starts (Figure 5-20).

Figure 5-19. The average number of units running (excluding wind generators) each

hour for tidal and for wind that is assumed to be certain.

134

Figure 5-20. The average number of units started (excluding wind generators) each

hour during dispatches for tidal and for wind that is assumed to be certain.

5.8 Effect on the market


The details of the new All-Island market have yet to be agreed on and finalized. Therefore the following approach was developed to identify the impact that wind generation can have on a market. This is aimed to inform market developers / operators of the generic issues surrounding certain and uncertain variable generation.

The marginal unit during each time step was identified as being the thermal unit with spare capacity (taking into account both the generation and the reserve provision of the unit) that has the lowest short run marginal cost. The short run marginal cost of this unit was then taken as the system marginal price for that time step.

Increasing the wind penetration on the system decreases the net fuel cost of the system (Figure 5-21), however there is a fuel cost associated with providing the reserve to cover for the uncertainty of wind generation (Figure 5-22). The no load and start costs always increase with increasing wind, however on some time steps the incremental costs decrease due to the starting of a large plant with a low incremental cost but high start cost.

135

Figure 5-21. The system fuel cost to generation with increasing tidal and wind that is

assumed to be certain, and that which carries reserve for forecast uncertainty.

Figure 5-22. The system fuel cost to provide the reserve against forecast uncertainty.

Wind and tidal generation are assumed to be price takers (they bid into the market at zero), therefore the SMC (system marginal cost) is depressed with increasing penetration because the marginal plant setting the SMC is substituted (Figure 5-23). Wind as a less predictable form of generation was found to depress the SMC to a greater extent than tidal generation (Figure 5-23), because to provide the additional reserve on the system, units which are cheaper than the marginal units must be run back to accommodate the minimum stable generation level of the extra units. Therefore the marginal units are committed on, and the SMC is taken from a cheaper unit (see Figure 5-23).

136

Figure 5-23. Effect of tidal or wind generation on the weighted Averaged Marginal

Cost (AMC) and the revenue that either wind or tidal generation receive based on the Marginal Cost (MC).

Wind and tidal generation are found to depress the revenue they receive at a greater rate than they decrease the SMC (at rates of 1.4 times greater for wind and 2.7 times greater for tidal), because they only affect the price while they are generating and hence when they are receiving money from the market. This would be typical of all variable generation which bid into the market at 0 /MWh. The revenue received by wind or tidal generation through a single price based on the SMC depends on the SMC during the time of generation. Wind generation has a diurnal pattern of greater wind speeds during the day (Figure 5-24) when the SMC is high. This initially results in greater revenue for wind than for tidal (Figure 5-23), which has a semi-diurnal pattern and so takes price equally during both the night and the day. However this is very quickly reversed due to much greater effect wind has on the SMC (Figure 5-23).

137

Figure 5-24. The average time of tidal [6] and wind generation over a year compared

with the average system marginal price and demand over a year.

The single market price was calculated post unit commitment based on the marginal units incremental cost, and the start and base load costs. The single market price has been set at the short run marginal cost of the marginal plant and assumes the generators receive payment for the base load costs and start costs they incur on an individual basis (see Eqn. 5-3). Method of payment Rgen ,t = (Pmarginal,t Ggen ,t ) + S gen ,t + Bgen ,t + C gen ,t R gen,t = Generator revenue in interval t. Pmarginal,t = System marginal cost / MWh in interval t. G gen ,t = Generation (MWh) in interval t. S gen,t = Start costs incurred by that generator equally spread over the total up-time. B gen,t = Base load cost (no load cost) in interval t. C gen,t = Constraint payments in interval t.

(5-3)

A generator which is constrained on due to technical constraints is paid the difference between its operating cost and its SMC based revenue. However there were no payments made for transmission constraints as the transmission system was removed from the model in a simplification (explained earlier). Capacity payments and reserve provision payments were also not included, and because the system only carried the necessary reserve for wind and did not actually use it there was no payment for the use of reserve.

138

Payment to generation in a market place is based on a number of factors such as the SMC, capacity payments, ancillary services, constraint payments and start costs. If a market were to base the single market price on the SMC paying constraint payments and start costs when incurred (see Eqn. 5-3), then increasing levels of wind generation would cause the single price to fall and constraint payments to rise, resulting in a significant decrease in wind generations gross market revenue as shown in Figures 525 and 5-26. It is expected that the All-Island market will avoid the problem of wind generation reducing the market price by not using the actual operational unit commitment to set the price but rather to set the price in a simplified unit commitment with no constraint payments.

Figure 5-25. The cost of generation and market revenue with increasing wind

generation and the effect of wind management (WM) on these.

139

Figure 5-26. The difference in the management cost of wind (constraint payments, no

load costs and the annual system start costs) with wind generation assumed to be certain and when carrying reserve to account for the uncertainty of wind generation.

5.9 The effect of tidal generation on plant usage


Tidal generation does provide some capacity credit to the generation portfolio, however there must be generation available to supply the load during the times of slack water. The aim here is to identify if the installation of tidal generation will significantly affect the operation of other types of plant. The effect of increasing tidal generation was considered in terms of its effect on existing generation and new generation (coal, additional nuclear and additional distillate generation). The effect on existing hydro storage and potential for other types of storage was also considered.
5.9.1 Effect of tidal penetration on existing generation The generation on the system modelled within PLEXOS for 2007 was divided up into

generation type. These types include oil, peat, coal, open cycle gas turbine, and combined cycle gas turbine. The level of generation from each type of plant was recorded as the level of tidal generation increased from 0 MW installed capacity to 500 MW in steps of 100 MW. Tidal generation displaced combined cycle gas turbines, oil, open cycle turbines and peat (in that order), however it did not affect coal which was base loaded or distillate which is used as expensive peaking plant (see Figures 5-27 which shows the percentage change and Figure 5-28 which shows the energy change). No single type of generation increased its net annual output to support tidal generation. This is because the marginal plants were gas turbines (in particular open cycle gas

140

turbines), which are suited to fast starts and high ramp rates, so there was no need to replace them with another form of generation.

Figure 5-27. The reduction in generation as a percentage of the total energy output from

each generation type, following the installation of tidal energy in increments of 100 MW.

Figure 5-28. The reduction in energy output from each generation type, following the

installation of tidal energy in increments of 100 MW.

141

5.9.2 Effect of tidal penetration on new generation To understand the effect tidal generation may have on the entrance of a new plant to the

market, the 2007 PLEXOS model was run with and without tidal generation for additional CCGT, coal, nuclear and distillate plant on the system, all of which were run independently. The CCGT, coal and distillate plants were modelled using the values for similar plants on the system whilst the following values where assumed for the nuclear plant:

Property Value Max Capacity 540 MW Min Stable Level 300 MW Min Up Time 10 hrs Min Down Time 12 hrs Start Cost 100000 euro/start Fuel Price 1 euro/GJ Heat Rate 1 GJ/MWh Table 5-5. Values used for the nuclear plant

The tidal generation had no effect on the amount of generation from the base loaded nuclear and coal plant. In support of the findings of the first method the installation of tidal generation reduced the amount of generation from the CCGT plant and peaking distillate plant see Figure 5-29.
Effect of tidal generation on fuel usage

Distillate

Figure 5-29. The generation produced with and without tidal from the new forms of

generation independently added to the 2007 PLEXOS model.


142

5.9.3 Effect of tidal energy on the use of storage systems Use of storage systems has on many occasions been suggested as a method of removing

variability (Leonhard & Grobe, 2004). For tidal energy to produce energy continually with the use of a storage system it would need to be capable of storing up to 25 times the installed capacity of the turbine (with no electrical down rating) e.g. a pump storage facility the size of Turlough Hill (1.5 GWh) would be capable of removing the variability from 53 MW of tidal generation capacity (assuming 70% efficiency). The reason such a large store is needed is due to the time between spring and neap tides (see chapter 2). Therefore the use of storage schemes to remove the variability of tidal generation is considered impractical. The storage of energy is a service which must be able to make a return on its capital cost within the market place as a separate entity to tidal generation. To determine if the installation of tidal generators would change the operation (and hence market opportunity) of the storage facilities currently on the Irish system the total annual pump load of the storage facility was compared with increasing amounts of tidal generation on the system. However the installation of up to 500 MW installed capacity of tidal generation was found to have a minimal effect on the operation of the pump storage facility (see Figure 5-30). The storage of energy results in losses which increase the cost of the energy, therefore it is only viable where there are large differences in SMC, and tidal energy was not found to create a large enough differential (see Fig. 5-23). The unit commitment of the hydro plant was not changed between the certain and uncertain wind cases. A separate study was conducted into the effect of wind on storage systems using the same method to calculate the reserve requirement for each time step, but specifying the wind reserve requirement in advance to PLEXOS (i.e. not updating it on each time step as depicted in Figure 5-8). This enabled the hydro plant to provide reserve but also let the solver plan for the reserve in advance.

143

Figure 5-30. The usage of the pump storage facility at Turlough hill as tidal and wind

generation is increased on the system. Wind Gen. refers to the results from the mixed integer solution using the method described in Figure 5-8..

5.10 The effect on carbon emissions


The impact of renewable generation on emissions is limited because it substitutes the most expensive plant on the system which are also the cleanest plant (see Figure 5-11 & 5-12), whilst leaving the plant which emit the highest levels of emissions largely unaffected. It is however possible to change this by attributing a cost to carbon emissions, or a carbon tax which increases the cost of worst polluting plant relative to the much cleaner plant.

The European Union has introduced a carbon-trading scheme (EU, 2003). Each European state submits a national allocation plan (NAP) for each 3-year period (Environmental Protection Agency, 2004) to the European committee for approval which outlines the amount of CO2 that is allocated to each sector, based on the CO2 production from the previous year. Companies within each state are allocated CO2 production allowances from the NAP which may then be used or traded. However if a company produces CO2 without owning an allowance it is fined at the rate of 100 /tonne (EU, 2003).

Increasing the value of carbon caused an increase in the use of the gas fuel, whilst peat, oil and coal decreased (see Figure 5-31) in usage due to their high carbon content. Diesel oil initially increased in usage however with increasing carbon value it dropped

144

down again. Hydro storage plant was also found to decrease in usage because the cost of the base load plant had increased reducing the price difference between peak and base load conditions.

Figure 5-31. Effect of increasing carbon tax on the fuel usage.

The unit commitment problem is primarily driven by minimizing the fuel cost and is sensitive to changes in fuel prices. This can be seen in Figure 5-31 in which the fuel prices were altered to account for the cost of carbon. Such changes in fuel prices would also have knock on effects for the emission levels, number of units running and the number of starts.

5.11 Conclusion
An appreciation of the benefits predictability can deliver from a variable source of generation such as tidal over a variable and uncertain source of generation such as wind have been quantified in terms of the effects on emissions, market aspects, system operations and the effect on the usage of other generation. However further work is being undertaken within the 2020 All-Island grid study group to develop a rolling unit commitment which will provide a better estimation of how the TSOs will optimize generation on the system whilst accommodating the necessary reserve for the uncertainty of wind generation. The effects on market aspects should also be revisited once the market design has been completed and any outstanding market change requests at the go live date have been processed.

145

6. Grid connection
6.1 Introduction
The areas of viable tidal resource have been divided into seven separate regions. These include the North Coast, the Maiden Islands, Strangford, Arklow, Carnsore Point, Shannon, and Malin Head (see Figure 6-1). The grid connection study will present an understanding of both the resource and the existing network in each area. The network will be considered in terms of the effect that generation in each area would have on system losses, the short circuit level in each area, the existing transmission capacity and an approximation of the cost associated with connecting various amounts of generation in each area. However it should be noted from the outset that whilst this will give an estimate for the cost and technical obstacles, all parts of this problem are extremely dynamic, both in terms of project costs and network capacity. Therefore this study has been restricted to a high level study aimed at identifying the most economic sites by rank order and not aimed at determining the exact cost of connection.

Figure 6-1. Location of the 110 kV+ buses where viable tidal sites could connect.

146

6.2 Transmission loss adjustment factors


As power is transmitted from points of generation to points of load some power is lost due to the impedance of the transmission system. Therefore, locating generation near to the load can reduce losses. A system of charging each generator for the losses that it incurs on the system is in place to both pay for these losses and to incentivise new generation to balance its location between the fuel source and the load. Traditionally this has been calculated by incrementing the generation slightly at each node and comparing the difference in total system losses. This assumes that a generators effect on losses is linear. This procedure is repeated for different loading scenarios, from which an estimation is made of the Transmission Loss Adjustment Factors (TLAFs) for each season, based on the seasons demand profile.

The TLAFs for the 110 kV system in ROI are published and are available for use in this study, however the TLAFs on the NI system are not published and therefore had to be calculated for use in this study as described below.

6.2.1 Method of calculating the TLAFs for the North Coast and Larne Two methods were considered for calculating the TLAFs at Coleraine and Larne. The first method was conducted in accordance with the traditional method described above, using cases provided by SONI for calculating the losses from 600 MW demand up to 1700 MW demand in steps of 100MW. Each of these cases was solved to determine the total NI system losses without any additional generation and then with an additional 10 MW at either Coleraine 33 kV bus or at the Larne 33 kV bus. This was repeated to consider the effect a generator can deliver with a power factor of 1.0, 0.95 to 1.0 and 0.85 to 1.0. The difference in losses was then estimated for each hour of each season through interpolation of the difference of losses versus system demand. However this method does not take into account the effect that a liberalised generation market may have on the system losses. Therefore a second study was conducted to account for this by taking the generation unit commitment from the PLEXOS model (chapter 5) for 2007 and setting the generation and load within the PSSE NI model to that for each hour in the unit commitment, resulting in 8760 cases. Each case was then solved using the ACC (Alternating Current Calculation)

147

solver within PSSE. The results of each were divided into seasons / day or night according to those defined by both TSOs (see table 6-1.) Season Winter 1 Winter 2 Spring Summer Autumn Start Date Nov 1 Jan 1 Mar 1 Jun 1 Sep 1 End date Dec 31 Feb 28/29 (last day) May 31 Aug 31 Oct 31

Time Start time End time Day 08:00 23:00 Night 23:00 08:00 Table 6-1. The definition of the seasons, day time and night used by the system operators in both NI and RoI.

6.2.2 TLAFs at Coleraine and Larne As the generators ability to provide reactive power was increased, the benefit to losses both at Coleraine and Larne was reduced (see Figures 6-2 & 6-3). This is because the total current increased, resulting in increased losses. Under all cases a greater reduction in losses was offered through adding generation at Coleraine as opposed to Larne, because Larne is electrically closer to large generators. Therefore losses between the generator and the demand at Larne are initially smaller.

The SONI base cases indicated that at Larne there was very little benefit to night time losses from additional generation, because the load at Larne is lower and the base load plant is electrically very near. However in the un-constrained market, more base-loaded plant was located in ROI, reducing the generation at Larne and increasing the scope for reduction in system losses, through generation at Larne.

In the SONI base cases for NI demand between 600 and 1700 MW generation at Coolkeeragh never exceeded 150MW, however in the All Island unit-commitment from PLEXOS the CCGT at Coolkeeragh has a maximum generation of 404MW with a further 53MW available on a distillate turbine. Therefore when using the PLEXOS unit

148

commitment, the reduction in losses at Coleraine was found to be less than estimated using the traditional technique (see Figure 6-2).

Figure 6-2. The reduction in losses resulting from increased generation at Coleraine.

Figure 6-3. The reduction in losses resulting from increased generation at Larne.

The results for the traditional method implemented with 12 cases and the method based on a full unit commitment have been presented for generation with a power factor of 1.0 (see

149

Table 6-2). Tidal generation is likely to be connected through IGBTs capable of four quadrant operation (European Commission, 2005), however this will be at the end of a 33 / 38 kV line and therefore it is unlikely to be able to greatly affect the voltage at the 110 kV bus without causing very large voltage fluctuations on the 33 kV line.

Bus

Winter (Jan/Feb) Day Night Wexford1 1.015 1.007 1 Arklow 0.981 0.978 Letterkenny1 1.056 1.018 Moneypoint1 0.961 0.982 1 Tarbert 0.964 1.003 1.071 1.031 Coleraine2 3 1.143 1.056 Coleraine 2 1.044 1.025 Larne 1.053 1.005 Larne3

Spring (Mar/May) Day Night 1.013 1.005 0.982 0.980 1.040 1.032 0.965 0.981 0.968 1.003 1.072 1.030 1.120 1.055 1.047 1.025 1.038 1.005

Summer (Jun/Aug) Day Night 1.007 1.002 0.980 0.982 1.035 1.003 0.968 0.984 0.979 1.004 1.072 1.030 1.120 1.055 1.047 1.025 1.038 1.005

Autumn (Sep/Oct) Day Night 1.012 1.000 0.978 0.976 1.026 1.020 0.970 0.987 0.981 1.019 1.071 1.031 1.143 1.056 1.044 1.025 1.053 1.005

Winter (Nov/Dec) Day Night 1.013 1.003 0.981 0.974 1.032 1.023 0.965 0.986 0.976 1.023 1.070 1.031 1.144 1.056 1.044 1.025 1.053 1.005

Table 6-2. The TLAFs for points of potential tidal connection.

Looking at the average TLAFs for the year it is possible to see that if tidal generation were to develop on the Shannon or Arklow they would be penalised for causing transmission system losses to increase, whilst those developed on the North Coast, the Maiden Islands or Malin head would be rewarded for reducing system losses (see Figure 6-4).

1 2

Published by Transmission System Operator Ireland (2005). Calculated using 8760 cases based on the output of the PLEXOS model. 3 Calculated using the traditional method for 12 standard load cases.

150

Figure 6-4. Yearly average TLAF applied to each point of potential tidal connection at the 110 kV level.

6.3 Short circuit level


During fault conditions the impedance to ground or between phases is greatly reduced, thus increasing the reactive power flow. It is this increase that is used to operate protection equipment which trips to isolate the fault. The amount of the maximum current that can be drawn at any given location on the system during a fault is termed as the short circuit level of that busbar. As more generation is added at each bus the short circuit level increases. This means that switchgear around these bus bars may need to be replaced with equipment capable of interrupting a larger current. Fortunately as the AC/DC/AC converter technology being used on the MCT turbines accurately controls the amount of energy being extracted from the turbine, it wouldnt contribute much towards the short circuit level at the local bus-bar. However for the benefit of other developers who may choose not to connect through a full AC/DC/AC system, the short circuit levels have been presented in Table 6-3 based on PSS/E model results for the NI system and the forecast statement for the ROI system (Transmission System Operator Ireland, 2005). The fault current comprises of both an oscillating AC and a decaying DC component. The rate of decay of the DC component is dependent on the ratio of the reactance to resistance between the generator and the fault (X/R ratio), as this ratio decreases the DC component decays faster (see Table 6-3). The current which the circuit breaker must be capable of breaking during the fault is the root

151

mean square of the AC component plus the remaining DC component (the total RMS break current). However the worst-case scenario occurs if the circuit breaker is closed onto an earthed piece of equipment because there will be no time for the DC component to decay. This is termed as the RMS make current (see Table 6-3).

Bus Wexford4 Arklow 4 Letterkenny 4 Trillick 4 Moneypoint 4 Tarbert 4 Coleraine 5 Larne 5

Single phase to earth fault Three phase fault Peak Tot RMS Peak Tot RMS Switch Make Break Make Break gear fault X/R Ratio (kA) (kA) X/R Ratio (kA) (kA) rating (kA) 4.2 3.2 11.1 5.2 10.3 4.7 25 9.7 8.8 21.2 8.3 25.6 10.2 26 5.3 4.1 15.9 7.1 14.3 6.3 25 4.3 3.2 6.4 3.1 3.7 1.7 25 3.9 2.6 4.5 2.3 3.3 1.6 25 14.1 12.7 49.3 17.9 60.6 23.3 25 3.1 6.7 4.5 3.8 7.9 5.3 39 4.5 9.3 6.1 5.8 9.3 6.0 18

Table 6-3. The maximum peak make and total root mean square break currents given both three phase and single phase fault conditions in winter peak and summer minimum conditions.

6.4 Connection capacity and cost


At each location a detailed study of the possible methods and cost of connecting the resource to the transmission system (110 kV+) has been conducted. This normally involves both building new infrastructure to get ashore and connecting into the existing distribution system which may require reinforcement.

Once the TED is connected to the transmission system there are still limitations to the amount of generation that can be accommodated at any point on the transmission system. These limitations will represent the firm connection offer that a developer is likely to receive at these points. A firm connection offer is what the TSO is prepared to guarantee that the transmission system can accept at any time, however the generator can also be offered a non-firm connection which will enable the generator to produce above the firm connection offer as long as it does not exceed the transmission system rating. The TSO
4 5

Published by Transmission System Operator Ireland (2005). Calculated using PSSE.

152

does not currently offer wind generators firm connection, so as to maintain the ability to curtail wind, when necessary to manage its variability and uncertainty. In the future the TSOs may decide to offer wind and tidal generation a firm connection with regard to transmission constraints but retain the ability to reduce their generation with regard to operational reasons. The transmission system limitations change rapidly as more applications for grid connection are received, old generation is decommissioned and the transmission system is reinforced to alleviate the points of congestion. Therefore whilst a generator may connect in an area with a very low firm connection offer, in a few years time this area may be re-enforced by the grid operator and the firm connection capacity revised upwards. When considering the capacity on the 110+ kV system it has been assumed that no changes to the existing system apart from those announced by the TSO would be made.

6.4.1 The northeast coast There are potentially hundreds of mega watts of feasible tidal generation identified off the northeast coast. The areas determined to be viable at maximum current speeds of 2.25m/s and 2.0 m/s between 20 and 40 m depth (Chapter 3) have been outlined in Figure 6-1. This resource has been divided into three groups: Benbane Head, Rathlin Island, and North Coast. The Benbane Head group is far off shore and only contains 3.6% of the areas resource in areas with current speeds greater than 2.25 m/s and 9.4% in areas with current speeds greater than 2.0 m/s. Rathlin Island contains the majority of the resource with 56.1% and 46.5% of the resource in areas of with current speeds greater than 2.25 m/s and 2.0 m/s respectfully. The North Coast contains the remaining 40.3% and 44.1% in areas of current speeds greater than 2.25 m/s and 2.0 m/s respectfully.

153

Figure 6-5. The available tidal resource off the North Coast.

Transmission system connection The nearest 110 kV (transmission system) points of connection are at Loguestown and Coleraine substations (approximately 60 km from the point of generation, as shown in Figure 6-6 and 6-7). Studies conducted by the author for NIE in relation to the capacity for additional generation onto the 110 kV system at Coleraine (Bryans, 2004a) indicate the system is currently capable of absorbing only a further 2.5 to 7.7 MW of generation at Coleraine before thermal limits are reached during the worst case n-1 scenario (see table 61) at the summer minimum load of 550 MW (worst case). The exact value within this range is dependent on the distribution of currently uncommitted wind generation.

154

Figure 6-6. The NIE transmission and 33kV distribution system near to the feasible tidal resource on the north coast (not to scale). Wire only diagrams of each scenario are available in appendix 2. Each line has been labelled (N1, N2, R1 etc.) so that it may be referred to from the following text. In an effort to improve system security and to provide a system capable of utilising a greater amount of renewable generation, NIE is considering a number of possible system upgrades. These possible upgrades include reinforcement of the existing 110kV Kells Coleraine circuit (upgrade A), and the addition of a further 110 kV circuit (upgrade B) and finally the reinforcement of the northeast 275 kV system (upgrade C). A full NI system transmission map is available from SONI upon request, however it cant be reproduced here to depict upgrades A to C due to data confidentiality restrictions.

155

System Upgrade Current system upgrade A upgrade B upgrade C

Range of feasible generation onto the system at Coleraine 2.5 MW to 7.7 MW 6.2 MW to 32.9 MW 10.1 MW to 95.2 MW 50.6 MW to 187 MW

Table 6-4. The feasible range of generation (depending on location of new wind farms) that can be added to Coleraine for each upgrade scenario before thermal limits are reached, during a summer minimum load of ~550MW.

The limits imposed on additional generation at Coleraine by the transmission system during the worst case of a summer minimum load, only exist for a short period of time. There are four periods of peak tidal generation per day. On the north coast only one of these periods of generation occurs during a time of very low system load during a spring tide. Therefore it would seem logical to install a larger amount of tidal generation onto the system than can be accepted under summer minimum load conditions. Then in the event that both the maximum tidal generation and the minimum predicted load are scheduled to occur at the same time the tidal generation could be reduced to prevent the system operating without the ability to withstand the worst case fault. Given that scheduled curtailment would occur during summer months, it may be favourable to conduct routine maintenance during these times when the sea tends to be calmer (depending on the operators ability to do such work during a spring tide).

Accepting the possibility that tidal generation may have to undergo some curtailment during periods of low system load and high transmission system loading, it is possible to utilize the greater transmission capacity at Coleraine such as that experienced during an 800 MW system load (see table 6-5).

156

System Upgrade Current system upgrade A upgrade B upgrade C

Range of feasible generation onto the system at Coleraine 13.9 MW to 44.8 MW 20.4 MW to 116.9 MW 24.8 MW to 210.7 MW 71 MW to 206 MW

Table 6-5. The feasible range of generation (depending on location of new wind farms) that can be added to Coleraine for each upgrade scenario before thermal limits are reached, during a system load of ~800MW.

Even if NIE were to make absolutely no changes to the current transmission system there would be a range of between 13.9 MW to 44.8 MW.

Distribution connection / transmission extension There is a 33 kV distribution system out to Ballycastle and Bushmills, with plans in place to put a cable from Bushmills to Rathlin Island to supply customers on Rathlin. However this cable is being installed as a 33 kV insulated cable, operating at 11 kV in the short term in the knowledge that Rathlin Island will probably develop wind, tidal or both forms of generation and will need greater connection capacity in the near future. Within this study the relevant 33 kV system has been added to the All Island PSSE model for the summer night valley 2006 2007 case and the winter peak 2006 2007 case.

157

Figure 6-7. The current transmission and distribution system in the north coast area.

A generator was added to the 33 kV system in Ballycastle and in Bushmills (see Figure 66). The system was tested for circuit overloads on the 33 kV system following incrementing the generator either at Bushmills or Ballycastle in steps of 1 MW with a power factor of 1.0 and then again with a power factor of 0.85. On setting up the case the system was solved using a solve from a flat start, permitting the transformers to tap change, then following each 1 MW increment of generation the system was resolved again using a ACCC solve but this time not from a flat start and not permitting transformers to tap. The reason the taps were locked on the transformers is a tidal generator will go from 0 MW to full output 4 times a day and some of the transformers on the distribution system have to have their taps operated manually at the substation (these would be unrealistic to use regularly). The line voltage was inspected to be within +0.1 pu.

The system was retested as described above following a number of potential reenforcements to the 33 kV system as described in appendix 2 and table 6-6.

158

Scenario Scenario A (Fig.4) Scenario B (Ap.2, Fig.A-2-2) Scenario C (Ap.2, Fig. A-2-3) Scenario D (Ap.2, Fig. A-2-4)

System Upgrade No upgrade


2

Total Upgrade of line BC1 to a 200 mm , 33 kV line. Total 2 Upgrade of line BC2 to a 200 mm , 33 kV line. Total 2 Upgrade of line BC1 to a 200 mm , 33 kV line. Upgrade of line BC2 to a 200 mm2, 33 kV line. Total Scenario E Upgrade of line BM1 to a 200 mm2, 33 kV line. (Ap.2, Fig. A-2-5) Total Scenario F Upgrade of line BM2 to a 200 mm2, 33 kV line. (Ap.2, Fig. A-2-6) Total Scenario G Upgrade of line BM1 to a 200 mm2, 33 kV line. (Ap.2, Fig. A-2-7) Upgrade of line BM2 to a 200 mm2, 33 kV line. Total 2 Scenario H Upgrade of line BC2 to a 300 mm , 110 kV line. (Ap.2, Fig. A-2-8) 33:110 transformer Total Table 6-6. The potential 33 kV grid reinforcements investigated.

Cost 0 0 1.37 M 1.37 M 1.23 M 1.23 M 1.37 M 1.23 M 2.61 M 400 k 400 k 400 k 400 k 400 k 400 k 800 k 5.15 M 600 k 5.75 M

The costs of each upgrade have been based around the price information contained in table 6-7. This data has been collected and averaged from a number of sources (which are not in the public domain and cant be named according to terms of presenting the data). The price of the sub-sea cable in particular is open to a large degree of error; this is because sub-sea cable may be installed in many different ways depending on the type of seabed and the distance it has to travel.

159

Over head lines Voltage Rating 33 kV 12 MW 33 kV 30 MW 33 kV 60 MW 100 kV 95 MW Sub-sea cables Voltage Rating 33 kV 30 MW 33 kV 60 MW 110 kV 95 MW Transformers Ratio Rating 690 : 33,000 1.6 MW 33 : 110 Structures off shore equipment pad

Description 3 x 100 mm2 3 x 200 mm2 3 x 200 mm2 double circuit 3 x 300 mm2

Cost 35 k/km 40 k/km 70 k/km 135 k/km

Description 3 x 240 mm2 3 x 800 mm2 3 x 1000 mm2

Cost 173 k/km 220 k/km 437 k/km

Cost 15 k 400 k

2.4 M

HVDC Converters (IGBTs) Voltage Rating Volume 80 kV 150 MW 4500 m3 80 kV 250 MW 4500 m3 150 kV 150 MW 8000 m3 150 kV 250 MW 8000 m3 Table 6-7. The cost of equipment used within the study.

Cost 26.6 M 30.3 M 34 M 37 M

The results from scenarios A-H indicated it was possible to connect up to 53 MW onto the existing system (table 6-8) between Bushmills and Ballycastle. However this does not take into account the issues affecting connection from either Bushmills or Ballycastle to the TEDs. Therefore a further list of scenarios was created outlining the connection possibilities from Ballycastle and Bushmills to the areas of resource (see Table 6-9 and appendix 2).

160

PF

1.0 PF

0.8 5

Wm ax

Wm ax

SN V

SN V

Scen. A (Bushmills) Scen. A (Ballycastle) Scen. B (Ballycastle) Scen. C (Ballycastle) Scen. D (Ballycastle) Scen. E (Bushmills) Scen. F (Bushmills) Scen. G (Bushmills) Scen. H (Ballycastle)

27 26 32 32 55 33 39 61 94

32 31 36 36 62 38 44 67 104

32 32 37 34 49 36 42 65 100

35 35 39 39 60 41 46 69 102

27 26 32 32 49 33 39 61 94

Table 6-8. The capacity at either Bushmills or Ballycastle for additional generation.

Scenario Scenario T-A (Ap.1, Fig.9)

Scenario T-B (Ap.1, Fig.10)

Scenario T-C (Ap.1, Fig.11)

Scenario T-D (Ap.1, Fig.12)

System Upgrade Use of planned R1 & R6 100 mm2, 33 kV line (no cost). Construction of R4 200 mm2, 33 kV line 3.6 km. Desk study, survey, EIA, switch gear Sub-sea cable ~240 mm2, 33 kV, 0.5 km. Total Use of planned R1 100 mm2, 33 kV line (no cost). Upgrade of R6 to 200 mm2, 33 kV line double circuit. Construction of R4 200 mm2, 33 kV line double circuit. Construction of R2 200 mm2, 33 kV line sub-sea cable. Sub-sea cable ~800 mm2, 33 kV, 0.5 km. Upgrade / second cable into Bushmills Desk study, survey, EIA, switch gear Total 2 Use of planned R1 & R6 100 mm , 33 kV line (no cost). Construction of R4 200 mm2, 33 kV line. Construction of N1 200 mm2, 33 kV line. Sub-sea cable ~240 mm2, 33 kV, 1 km. Sub-sea cable ~240 mm2, 33 kV, 0.5 km. Desk study, survey, EIA, switch gear Total Use of planned R1 & R6 100 mm2, 33 kV line (no cost). Construction of R4 200 mm2, 33 kV line. Construction of R5 200 mm2, 33 kV line. Construction of N1 200 mm2, 33 kV line double circuit. Sub-sea cable ~800 mm2, 33 kV, 1 km.

Fir

mc

161

a pa city

PF

0.8 5

1.0 P

Cost 02 144 k2 30 k 86 k2 260 k 0k2 700 k2 252 k2 1.9M22 110 k2 125 k 65 k 3.2 M 0 k2 144 k2 360 k2 172 k2 86 k2 85 k2 842k 0 k2 144 k2 1.6 M2 630 k2 220 k2

86 k2 85 k Total 2.7 M Scenario T-E 2 Use of planned R1 & R6 100 mm , 33 kV line (no cost). 0k2 (Ap.1, Fig.13) Construction of R4 200 mm2, 33 kV line. 144 k2 Construction of R3 200 mm2, 33 kV line. 256 k2 R5 Sub-sea cable ~1000 mm2, 110 kV. 4.0 M2 Construction of N1 300 mm2, 110 kV line. 1.1 M2 Sub-sea cable ~1000 mm2, 110 kV, 1 km. 437 k2 Sub-sea cable ~240 mm2, 33 kV, 0.5 km. 86 k2 Desk study, survey, EIA, switch gear 85 k 2 x offshore 33:110 transformers + platforms 6.4 M Total 12.5 M Table 6-9. The connection scenarios considered for connecting areas of viable resource to either Ballycastle or Bushmills.

Sub-sea cable ~240 mm2, 33 kV, 0.5 km. Desk study, survey, EIA, switch gear

All the sub-sea cables given in the study above involve the use of 3 phase AC cabling, rather than a DC cable and power converters. This was simply because pricing information remained un-available for DC cabling and low power IGBTs. However scenario T-A makes the assumption that a 33 kV 3-phase AC cable will be installed to Rathlin Island as part of an existing plan to connect Rathlin to the Irish grid system. Whilst SONI and NIE have considered such plans, no firm commitment has yet been made to this. It would provide very little benefit to convert the operation of this cable to DC because one of the cores would have to be discharged (environmental legislation prevents the use of a ground return) resulting in only a 13% increase in transmission capacity.

It was not considered necessary to model the scenarios connecting from Bushmills and Ballycastle to the areas of resource because the voltage level is not critical on these lines as there is only a generator on the end and no demand customers. Furthermore by simply using the rating for the line provided by the manufacturer it has been possible to construct a table of different possible combinations of scenarios (see tables 6-10 and 6-11).

162

T-A

T-B

T-C

T-D

ari o

ari o

ari o

Sc en

Sc en

Sc en

Sc en

Scen. A (Bushmills) Scen. A (Ballycastle) Scen. B (Ballycastle) Scen. C (Ballycastle) Scen. D (Ballycastle) Scen. E (Bushmills) Scen. F (Bushmills) Scen. G (Bushmills) Scen. H (Ballycastle)

Capacity (MW) 26 27 32 32 49 33 39 61 94

15 15 15 15 15 15 15 15 15 15

45 27 27 27 27 27 33 39 45 27

45 41 41 45 45 45 41 41 41 45

75 41 41 47 47 64 41 41 41 75

110 41 41 47 47 64 41 41 41 109

Table 6-10. The capacity at both Bushmills or Ballycastle for additional generation.

-A

-B

-C

-D

Sc en

ari o
oT

ari o

T-E
12,466,121 304,052 304,052 294,470 291,509 235,545 313,808 313,808 323,564 167,142
Sc ena ri oT -E

oT

oT

oT

ari

ari

ari

Scen. A (Bushmills) Scen. A (Ballycastle) Scen. B (Ballycastle) Scen. C (Ballycastle) Scen. D (Ballycastle) Scen. E (Bushmills) Scen. F (Bushmills) Scen. G (Bushmills) Scen. H (Ballycastle)

Cost of scenario 0 0 1,373,960 1,234,800 2,608,760 400,000 400,000 800,000 5,752,350

259,829 17,322 17,322 108,919 99,642 191,239 43,989 43,989 70,655 400,812

Sc en

3,152,000 116,741 116,741 167,628 162,474 213,361 107,636 91,077 87,822 329,791

Sc en

841,486 20,524 20,524 49,232 46,140 76,672 30,280 30,280 40,036 146,530

Sc en

2,726,912 66,510 66,510 87,253 84,292 83,370 76,266 76,266 86,022 113,057

Table 6-11. The cost of the possible reinforcement scenarios and the connection scenarios. The green squares indicate the most economic progression of site development.

Based on the possible capacity of each feasible combination of scenarios and their cost it has been possible to estimate the connection cost associated with any given amount of generation up to 109 MW for the North Coast.

163

Sc en

ari

Figure 6-8. Approximation of the grid connection cost for tidal generation off the North Coast.

6.4.2 Maiden Islands The Maiden Islands are located near an area which has traditionally been a large source of thermal generation, and therefore has a high capacity at the transmission level (see Figure 6-9). At the Larne-Main 110 kV substation the generation capacity was found to be 116 MW during the summer night valley and 138 MW during the winter maximum based on an N-1 contingency analysis carried out on the 2006 to 2007 PSSE system model. This level of capacity exceeds the feasible resource in the area (see Figure 6-9).

164

Figure 6-9. The viable resource, current transmission and distribution system off the Maiden Islands.

As with the North Coast a series of grid reinforcements have been considered for the 110 kV Larne-Main substation (Appendix 2) the existing 33 kV network (scenarios A to D) and for the grid connection of the Maiden Islands to the 33 kV system (scenarios T-A to T-B). Each of the grid re-enforces was constructed in PSSE on the Irish grid system and tested to determine the new rating (see table 6-12).

165

-A

oT

ari

Scenario A Scenario B Scenario C Scenario D

Capacity (MW) 26 32 32 56

30 26 30 30 30

Sc en

60 26 32 32 55

Table 6-12. The capacity at Larne North 33 kV substation for tidal generation.
-A -B

oT

Sc en

ari ari

oT

-B

Scenario A Scenario B Scenario C Scenario D

Cost of scenario 0 180,000 180,000 360,000

3,088,338 118,782 108,945 108,945 114,945

Sc en

3,944,000 151,692 128,875 128,875 78,255

Table 6-13. The cost of the possible reinforcement scenarios and connection scenarios.

The cost of each scenario was also considered with the connection capacity that it offered, however connection at the Maiden Islands involves a long sub-sea cable which is expensive forcing up the price of connection (see Table 6-13, and Appendix 2).

6.4.3 Arklow The resource located near to Arklow on the codling bank is not viable with the current generation of technology, however the resource may become viable in the near future (see Figure 6-10). The site is difficult to assess for grid connection because there are a number of other developments which are and may take place near to this site which would have a large influence on the method and cost of grid connection. These include the development and expansion of the wind farm on the Arklow bank, the installation of an inter-connector to Wales from Arklow and the possibility of a wind farm development on the Codling bank. Given that the technology needs time to develop further and the possibility of

166

Sc en

ari

oT

significant changes to the system in that area any cost assessment will be open to a high level of error. To account for this, scenarios have been developed connecting the areas of resource directly to land, or to the wind farm at Arklow bank (see Figure 6-11 and table 614). The larger resource, which is further offshore (termed as Irish Sea in Figure 6-10), was not considered because of high connection costs, however, if an interconector was being installed this may drop considerably.

Figure 6-10. The viable resource near to Arklow.

167

Figure 6-11. The current transmission and distribution system near Arklow.

Cost Capacity (MW) Scenario Connection point Scenario A Connection at kilcoole 1,423,851 30 Scenario B Ballybeg 3,965,294 60 Scenario C windfarm 3,657,059 60 Table 6-13. The cost the possible connection options.

Cost / MW 47,461.72 66,088.24 60,950.98

Eirgird has indicated in the 2005 2006 transmission forecast statement that there is no spare capacity for generation at Arklow and that it would be unable to offer a firm connection. However considering proposed developments such as an interconector to Wales and further development of wind farms it is likely that this area will receive investment in the transmission system.

6.4.4 Carnsore Point The SEI report (Kirk McClure Morton et al., 2004) for tidal resource around Ireland indicates there is tidal resource off the Carnsore point (average output of 20 MW), this has 168

not been verified in this study due to the sea bed depth remaining outside the range of operation for the MCT design in the areas of high current speed. However, it has been included in the grid connection study because it is very near to the operational range and Kirk McClure Morton Lmt. identified the area in the SEI report.

On the Carnsore peninsula there is a 12 MW wind farm which is connected to the 38 kV distribution system into Wexford and onto the 110 kV transmission system. As with the previous studies (above) the relevant 38 kV system was added to the PSSE All-Island model and scenarios developed (see Appendix 2) to reinforce the 38 kV system to accommodate additional generation at Carnsore point. The cost of these was also based on table 6-7.

Figure 6-12. The current transmission and distribution system out to Carnsore point. Where steel core aluminum is SCA and all aluminum alloy conductor is AAAC.

169

-A

ari

Sc en

Scenario A Scenario B Scenario C

Capacity (MW) 20 24 37

30 20 24 30

60 20 24 37

Table 6-14. The capacity at Carnsore 33 kV substation for tidal generation.

Sc en

ari

oT

oT

-B

-A

oT

ari

Cost of scenario Scenario A Scenario B Scenario C 0 284,000 1,282,500

832,459 41,623 46,519 70,499

Sc en

1,040,000 52,000 55,167 62,770

Table 6-15. The cost of possible reinforcement scenarios and connection scenarios.

Eirgird has indicated in the 2005 2006 transmission forecast statement (Transmission System Operator Ireland, 2005) that there is no spare capacity for generation on the transmission system at Wexford before 2011 when it is expected that capacity for between 60 to 100 MW will be created depending on the upgrade strategies undertaken on the transmission system. However, as stated earlier the transmission system is dynamic and this could change rapidly depending on applications for grid connection.

6.4.5 Malin Head An area of viable resource was identified off Malin head with a peak output of 44 MW (see Figure 6-13). The nearest 38 kV point of connection on Malin head is the 38 kV substation at Carndonagh, which is connected into 110 kV system at Letterkenny, through a 110 kV line from Trillick. The 38 kV system was added to the all-Island PSSE system and tested

170

Sc en

ari

oT

-B

with the two wind farms on Malin head operating at peak output for capacity following a number of possible upgrade scenarios (see table 6-16 and 6-17).

Figure 6-13. The available tidal resource off Malin Head.

171

Figure 6-14. The current transmission and distribution system in the Malin Head area.

-A

oT

ari

Scenario A Scenario B Scenario C

Capacity (MW) 5 20 49

30 5 20 30

Sc en

60 5 20 44

Table 6-16. The capacity at Carndonagh 33 kV substation for tidal generation.

172

Sc en

ari

oT

-B

-A

oT

ari

Sc en

Scenario A Scenario B Scenario C

Cost of scenario 0 840,000 1,575,000

1,044,812 208,962 94,241 87,327

1,606,000 321,200 122,300 72,295

Table 6-17. The cost of possible reinforcement scenarios and connection scenarios.

Eirgird has indicated in the 2005 2006 transmission forecast statement (Transmission System Operator Ireland, 2005) that there is no spare capacity for generation on the transmission system at Letterkenny until 2011, when it is possible that up to 40 MW may become available depending on the upgrade strategies undertaken on the transmission system.

6.4.6 Shannon The resource within the Shannon estuary was not modelled within the ocean model because it is beyond the models resolution. Whilst it would have been possible to develop a finer resolution model of just the Shannon estuary to accomplish this, such a model was not developed because a good understanding of the current velocities within the Shannon already exist on charts of the estuary. Also the viable resource is likely to be small because of the need to maintain an un-restricted channel for shipping purposes. The areas believed to have viable depths and current speeds have been shown in Figure 6-15, in accordance with the work of Kirk McClure Morton (2004).

173

Sc en

ari

oT

-B

Figure 6-15. The current transmission and distribution system in the Shannon estuary.

There is a large number of potential sites within the estuary and the large number of possible connection points to the grid system, leading to a very large range in possible connection costs. However the main deciding point within the Shannon estuary is that of planning permission and where, if anywhere along the estuary it would be permitted to install turbines. Therefore given that any study of the Shannon at this stage would be open to such a large degree of error, and is < 10 MW, so it has been deemed as of no worth at this stage and has not been taken any further.

6.4.7 Strangford The resource within Strangford Lough has been modelled and measured in detail in Whittaker (et al., 2003) because this is the site MCT has chosen for the first full sized version of the turbine. Whilst there is a large resource within the narrows of Strangford Lough further development of this resource is limited due to the need to maintain the shipping channel into the Lough. There is an 11 kV line supplying the town of Strangford, from the 33 kV substation in Bishopscourt (see Figure 6-16). It is proposed to connect the

174

turbine to this line with a grid connection rated to 1.2 MW, any further development would require the installation of a 33 kV line from Bishopscourt.

Figure 6-16. The current transmission and distribution system in the Strangford estuary.

It is unlikely that sufficient space will be found for additional turbines within the Strangford narrows, therefore the grid connection cost has not been assessed further turbines. However it should be noted that following the installation of the MCT turbine the 11 kV line to Strangford will be operating near maximum capacity and any further development would require an extension of the 33 kV system.

6.4.8 All sites The cost of connecting incremental amounts of tidal generation at each site has been determined (see Figure 6-17) and used to depict which is the most economic site to choose for development of a given amount of tidal capacity. A development below 15 MW would favour Rathlin Island followed by development of the North Coast up to 41 MW, at which

175

point the resource at Arklow may become more attractive depending on the stage of technological development and the development of neighbouring projects.

Figure 6-17. Approximation of the grid connection cost for tidal generation at each site.

6.5 Strangford case study


The main concern displayed over the grid connection of renewable generation on the end of an 11 kV line such as that supplying the town of Strangford, is the impact it would have on voltage. Induction machines have not traditionally been used on systems because they do not provide voltage control unless combined with capacitors. The ability to control voltage is linked to the machines ability to produce and absorb reactive power. Reactive power may be described as the power produced at 90o to the real power and is used to support voltage and facilitate the transportation and use of real power. An analogy for reactive power is to consider the use of a wheelbarrow, to move the wheelbarrow forward you must lift the handles and then walk forward. The lifting of the handles could be seen as the necessary reactive power at 90o to the real power, which was to move the wheelbarrow forward (see Figure 6-18).

176

Figure 6-18. Analogy for reactive power and real power using a wheelbarrow.

Reactive power can therefore be defined as: Q = P2 + S 2 P = Real Power (Watts) S = Apparent Power (VAr) Q = Reactive Power (VAr) 6-1

A synchronous machine can be used to absorb or produce reactive power by either controlling the voltage on the rotor. However induction machines absorb reactive power both when motoring and when generating. Therefore it is necessary to provide power factor correction with the use of capacitors. If the system uses an AC/DC/AC converter between the generator and the grid system it may have the ability to control reactive power depending very much on the technology used. With the IGBT based converters it is possible to control power in both the real and reactive components (see Figure 6-19).

177

Figure 6-19. The active versus reactive power range for the ABB HVDC light converters at 1.1, 1.0 and 0.9 pu voltage (ABB, 2004).

MCT have not released any information as to the electrical configuration of the system they intend to install at Strangford Lough. However, the system in the Bristol Channel uses a squirrel cage induction machine connected through a AC/DC/AC converter (see Figure 620). Such converter systems may enable voltage support depending on the technology and the configuration (see Figure 6-18). Therefore, the impact on the voltage along the line was considered based on both; the use of a standard squirrel cage induction machine connected simply through a transformer and with a machine connected with full control over active and reactive power.

178

Figure 6-20. Electrical layout of MCTs prototype turbine in the Bristol Channel (European Commission, 2005).

The 11 kV line to Strangford was simplified from having over 100 separate lines and buses to just 19 buses (as described in Figure 6-21 and Appendix 3 table A-3-1) so that it could be modelled in Dig-Silent under the academic license (limited to 50 buses). The squirrel cage induction machine was based around the specifications of those used in the Lendrums Bridge wind farm (see Table 6-18) for which a model existed in PSCAD that enabled model verification. The simulation of the generator connected through the AC/DC/AC converter with full active and reactive power control was achieved using an external grid module in Dig-Silent, which enabled the manual adjustment of the active and reactive power components.

179

Figure 6-21. Diagram of the simplified circuit used to represent the 11 kV system to Strangford. The line sections (S1 to S19) properties are given in (Appendix 3 table A-3-1).

Voltage 690 volts Power output 660 kW Frequency 50 Hz Stator resistance 0.0048 ohms Stator reactance 0.068 ohms Rotor resistance 0.004 ohms Rotor reactance 0.0897 ohms Magnetizing reactance 2.81 ohms No. of poles 4 Table 6-18. Generator parameters of the squirrel cage induction machines

The voltage was studied at each node along the line, with the power from generator increasing from 0 MVA to 1.32 MVA. This study was repeated with the load at each node decreased from 100% to 0% in steps of 20%, to consider the effect of tidal generation during times of reduced loading on the line.

The use of the squirrel cage induction machine causes the voltage to rise to a maximum of 1.055 pu when there is no loading on the line and tidal generation is at its maximum (see Figure 6-22). This increase is within acceptable levels (+ 0.1 PU) according to distribution grid code.

180

Figure 6-22. The impact tidal generation using a squirrel cage induction generator on the 11 kV line to Strangford.

The use of an AC/DC/AC converter between the induction machine and the grid enables control over the voltage and maintains a much more uniform voltage along the line than without the converter (see Figures 6-22 and 6-23).

181

Figure 6-23. The impact of tidal generation using a generator connected through an AC/DC/AC converter on the 11 kV line to Strangford.

Either if a squirrel cage induction machine is connected directly to the 11 kV line through a transformer, or if it is connected through an AC/DC/AC converter the voltage at all points on the line will remain within acceptable limits. However the use of the AC/DC/AC converter would enable better voltage control of the line and would ease the voltage concerns of the line rather than exaggerate them.

6.6 Conclusion
Grid connection possibilities have been considered for the North Coast, the Maiden Islands, Strangford, Arklow, Carnsore Point, Shannon, and Malin Head. At each of these locations issues such as the effect on system losses, the short circuit level in each area, the existing transmission capacity and an approximation of the cost associated with connecting various amounts of generation have been presented. However the system is very dynamic and once additional generation or load is connected to the system the study will need to be updated. The cost of cables and lines also change in price rapidly depending on the cost of copper,

182

therefore whilst this study gives a comparison of costs between sites it will need to be updated to get accurate costs when being considered by developers.

183

7. Conclusion
7.1 Tidal stream energy devices
Fifteen tidal energy device (TED) designs were investigated including: Stingray in Scotland (The Engineering Business, website; Trapp, 2004); Hammerfest Strm in Norway (Hammerfest Stromas, 2002a; Hammerfest Stromas, 2002b; Hammerfest Stromas, website); TidEl in England (SMD Hydrovision, 2004); Verdant Power in New York (Verdant Power, website); and Hydroventuri in San Francisco (Hydroventuri, website). Only one developer was found to have a system with a full-size prototype, with further funding to develop a commercial system that is scalable, generates power during both the ebb and flood tide and utilizes power from the entire water column. This TED, developed by Marine Current Turbines Ltd. (MCT), is based on a two-bladed, variable-pitch, horizontal-axis turbine. Each unit consists of a pile driven into the seabed, with a horizontal support wing on the pile with a turbine on each end. This wing can be raised out of the water so that the turbines can be removed and serviced. MCT have developed this system to be viable in the areas of greatest energy density and minimum technical challenge to support the technology through its infancy. The MCT TED is considered viable in areas with a spring tidal current over 2.2 m/s, in depths between 20 and 40 m.

7.2 Tidal stream resource available to Ireland


The power available for the island of Ireland from tidal generation was found to be 72 MW on average over a year, with a maximum output of 374 MW during a spring tide. This is about 2% of the current Irish electrical energy demand. This resource is located off Rathlin Island, the North Coast, the Maiden Islands and the Coddling Bank. Other studies (Kirk McClure Morton et al., 2004) have indicated some resource off Carnsore Point (not confirmed), in the Strangford narrows, and in the Shannon Estuary. It is expected that, as the technology matures, developers will increase rotor diameter and deploy the TEDs in deeper water. Such developments would significantly increase the power output from a single turbine and may make operation in a lower current speed viable.

184

7.3 Impact of tidal generation on system operations


The impact of the currently viable tidal generation on the system ramp rate has been found to be manageable, whilst its effect on the demand profile would be beneficial during times of peak demand, which also coincide with the times of maximum average tidal generation.

Methods of managing the operational impacts of tidal generation have been found to include: placing tidal generation at different locations around the coast to avail of opposing times of peak generation (not possible in Ireland); electrical down rating (EDR) the turbine during installation so as to reduce the connection cost per MWh; and to provide a mechanism for occasional generation reduction. The capacity factor of tidal generation is 0.19 with no EDR, increasing to 0.29 with an EDR of 50% whilst only spilling 15% of the available energy. The high generation during the time of peak demand also leads to tidal generation having a generation capacity credit of 18%. However, during years when spring tide occurs over the weekend, this would drop to 10% during the time of peak system demand.

The tidal energy device being developed by MCT was found to have a low amount of stored energy, corresponding to rated output for 0.27 s. Therefore using the turbines inertia when it is not spilling water will result in the turbine slowing down rapidly, moving it away from the optimal point of power extraction. However, if it is connected through an AC/DC/AC converter, inertia response can be controlled and limited.

Tidal generation, as a certain but variable form of generation, offers a greater reduction in CO2 and NOX emissions than the uncertain wind generation. There are average reductions of 2.65 %CO2 per TWh from tidal (1.7 times greater) and 3.24 %NOx per TWh from tidal (2.3 times greater), compared to reductions of 1.6 %CO2 per TWh and 1.4 %NOx per TWh from uncertain wind generation. However, uncertain generation (wind) offered approximately equal savings of SO2, with reductions at 2.0 %SO2 per TWh from tidal and 2.75 %SO2 per TWh from wind. This is due to cleaner, more expensive plant running at base load to provide the reserve. The same phenomenon also resulted in wind generation

185

undermining the price it receives to a greater extent that tidal generation, based on system marginal price.

7.4 Grid connection and site development of tidal resource


Additional generation from the tidal resources at Rathlin Island, the North Coast, Malin Head, Carnsore Point and the Maiden Islands would result in a decrease in system losses, with the greatest reduction caused from generation at Rathlin Island and the North Coast. By contrast, realisation of the resource at the Coddling Bank and the Shannon Estuary would result in additional losses.

It is understood that the MCT system uses a full AC/DC/AC converter to connect to the grid and therefore would not contribute to the existing short circuit levels at the point of grid connection. However, if a different configuration is chosen resulting in a short circuit contribution from tidal generation, the existing transmission switchgear is over rated at all the areas of resource, with the exception of the Shannon Estuary. Rathlin Island and the North Coast have the greatest capacity in terms of short-circuit level.

The capacity on the transmission system varies, with firm offers being made only for capacity which is always available. There is significant firm capacity for generation at the Maiden Islands and the Shannon Estuary. There is very little firm capacity currently available at Rathlin Island, the North Coast, Carnsore Point, Malin Head and Arklow. However, the firm capacity at each of these locations increases following transmission system upgrades that are planned or being considered. It is envisaged that tidal generation may have to connect with a large non-firm connection until the transmission system is upgraded in the relevant area to make a firm offer possible.

The Rathlin Island resource is the most attractive in terms of connection cost for an initial development of about 15 MW, followed by further development along the North Coast. The Maiden Island resource has been found to be expensive to connect compared with the North Coast resource. The Codling Bank resource depends on further developments in the technology, but may benefit greatly from neighbouring offshore wind farms.

186

7.5 Recommendations for future research


To compare tidal generation to wind generation a unit commitment model was developed that provided the reserve for wind by modifying a unit commitment based around a perfect wind forecast. Ideally a rolling unit commitment model would be developed to solve the problem to include reserve for wind uncertainty in one step, then move forward in time and solve again with an updated forecast. This work is being considered by A. Tuohy at University College Dublin (Tuohy, 2006). Such a model should also be tested to quantify the effect of using the reserve for wind.

Here the work has calculated wind reserve based on accounting for 90% of the errors in wind speed forecasts. However, further work should be conducted to determine how this value might change with increasing levels of wind penetration to maintain the current level of system security.

It has not been possible to accurately represent the effect a turbine will have on the flow around it and the separation distance required for the wake effects to dissipate behind it. Some preliminary model work based against scale models has been done by Thomson (2004), however measurements must be taken following the installation of the first prototype with two blades mounted side by side to enable full model development and verification. This work is essential before any large-scale tidal farms are developed to understand the spacing between rows and the exact effect on the local tidal regime.

The effect of an impeller turbine on seals remains uncertain. It is therefore necessary to conduct a monitoring program to determine how seals interact with the turbine. MCT are intending to place an underwater camera on the monopole to monitor this.

In areas of high current flow eddies can be formed as a result of turbulence, and these may cause some fluctuations in power output. Following the installation of the two-bladed prototype, power fluctuations should be measured to assess the effects of tower shadow, velocity gradient and turbulent flow.

187

To develop a site such as those presented here it would be necessary to conduct a detailed site survey and to make accurate current velocity measurements over at least one springneep cycle using an instrument such as an Acoustic Doppler Current Profiler (ADCP). Accurate models should be developed using an understanding of the effect of a turbine on the current flow to determine the effect on the local tidal regime, sediment transport and population dispersal (such as sewage discharges).

188

References
ABB (2004) Technical Status of Power Electronics Technologies Today A manufacturers perspective. GridWorks R&D Planning Workshop, Chicago. (Available at www.abb.com).

All Island Project (2005) Capacity Payment Mechanism and Reserve Charging High Level Decision Paper. All Island Project, Ireland. (Available at www. allislandproject.org).

All Island Project (website) http://www.allislandproject.org

ANEMOS (website) http://anemos.cma.fr

Arakawa, A. & Lamb V.R. (1977) Computational design of the basic dynamical processes of the UCLA general circulation model. Methods Comput. Phys., 17, 174-265. ISSN: 00766860.

Beckers, J.M. (1991) Application of a 3D model to the western Mediterranean. Journal of Marine System, 1, 315 332. ISSN: 0924-7963.

Binnie Black & Veatch (2001) The commercial prospects for tidal stream power. Departement of Trade and Industry, London. DTI Pub URN 1/1011.

Blue Energy Ltd. (website) http://www.bluenergy.com

Bryan, K. (1997) A numerical method for the study of the circulation of the world ocean. Journal of Computational Physics, Vol. 135, Pp. 154169. ISSN: 0021-9991.

Bryans, A.G. (2003) A Two Dimensional Model of Tidal Flow in the Menai Strait. MSc Thesis, University of Wales, Bangor.

189

Bryans, A.G. (2004a) Capacity for addition generation on the NI transmission system. System Operator Northern Ireland, Belfast.

Bryans, A.G. (2004b) Tidal Energy Feasibility Study: Phase I of III Report. Sustainable Energy Ireland, Dublin.

Bryans, A.G., Denny, E., Fox, B., Crossley, P. & OMalley, M. (2005) Study of the effect of tidal generation on the Irish grid system: resource and emissions. International Council on Large Electric Systems (CIGRE) Symposium, Athens.

Bryans, L. (2005) Personal communication, System Operator Northern Ireland, Belfast.

Bryans, L. (2006) Grid integration of large-scale wind power. Queens University Belfast, PhD thesis.

Bryans, L., Bryans, G.R. & Fox, B. (2005b) Studies and progress towards intermittence management on a small system. CIGRE, Athens symposium.

Bryden, A. (2003) Energy Extraction from Tidal Flows. IMechE. (Available at www.sbe.hw.ac.uk/staff/arthur/shsg/library.htm).

Callaghan, D (2003) Personal communication.

Commission for Energy Regulation (2004) A regulatory approach to ESB dominance. Commission for Energy Regulation, Dublin. CER Pub 04053.

Connolly, D. & Canning, T. (2005) Desulphurisation of flue gas in a large peat-fired CFB boiler. Universities Power Engineering Conference, Cork, Ireland.

Danish Wind Industry Association (website) http://www.windpower.org/

190

Delabroy, O., Haile, E., Lacas, F., Candel, S., Pollard, A., Sobiesiak, A. & Becker H.A. (1998) Passive and active control of NOx in industrial burners. Experimental Thermal and Fluid Science, Vol. 16, pp. 64-75. ISSN: 0894-1777.

Denny, E. and O'Malley, M. (2006) Wind Generation, Power System Operation and Emissions Reduction. IEEE Transactions on Power Systems, vol. 21, pp. 341-347. ISSN: 0885-8950.

Denny, E., and O'Malley, M. (2006b) A Quantitative Analysis of the Net Benefits of Grid Integrated Wind. IEEE Power Engineering Society General Meeting, Montreal.

Department of Communications, Marine and Natural Resources (2005) All-Island Energy Market: Renewable Electricity A 2020 Vision Preliminary Consultation Document. Department of Communications, Marine and Natural Resources, Dublin. (Available at www.dcmnr.gov.ie).

Department of Communications, Marine and Natural Resources (2006) Renewable Energy Feed in Tariff (RE-FIT - 2006). Department of Communications, Marine and Natural Resources, Dublin. (Available at www.dcmnr.gov.ie).

Department of Energy (1990) Offshore Installations: Guidance on design, construction and certification. HMSO, London. ISBN: 011413328.

Department of Enterprise, Trade and Investment (2004) Energy, A strategic framework for Northern Ireland. Department of Enterprise, Trade and Investment, Belfast. (Available at www.detini.gov.uk)

Department of Enterprise, Trade and Investment (website) www.detini.gov.uk.

Department of Environment, Heritage and Local Government (Ireland) (2003) Strategy to reduce emissions of transboundary pollution by 2010 to comply with National Emissions

191

Ceilings. Department of Environment, Heritage and Local Government (Ireland). (Available at www.environ.ie).

Department of Trade & Industry (2004) The World Offshore Renewable Energy Report 2004-2008. Department of Trade & Industry, London. (Available at www.dti.gov.uk).

Dillon, D.J. (1998) Scheduling of a Hydrothermal Power System. University College Dublin, MEng Thesis.

Dippner, J.W. (1993) A frontal-resolving model for the German Bight. Continental Shelf Research, 13, 49-66. ISSN: 02784343.

Doodson, A.T. and Warburg, H.D. (1941) Admiralty Manual of Tides. HMSO, London, UK.

Ekanayake, J. & Jenkins, N. (2004) Comparison of the Response of Doubly Fed and FixedSpeed Induction Generator Wind Turbines to Changes in Network Frequency. IEEE Transactions on Energy Conversion, VOL. 19, NO. 4, Pp. 800 - 802. ISSN: 08858969.

Electricity Supply Board (website) http://www.esb.ie/

Electricity Supply Board National Grid (2005) Grid Code, version 1.2. Eirgrid, Dublin.

Energy Information Administration (2006) International Energy Outlook 2006. Energy Information Administration, Department of Energy, Washington. Report #DOE/EIA-0484.

Engineering Business Ltd. (website) http://www.engb.com/

Environmental Protection Agency (2004) National Allocation Methodology 2005 - 2007. The Environmental Protection Agency, Wexford. (Available at www.epa.ie).

192

ESRI (website) http://www.esri.ie/

EU (2001a ) Directive of the European Parliament and of the Council of 23 October 2001 on National Emission Ceilings for Certain Atmospheric Pollutants. Official Journal of the European Communities (2001/81/EC).

EU (2001b) Directive of the European Parliament and of the Council of 27 September 2001 on the promotion of electricity produced from renewable energy sources in the internal electricity market. Official Journal of the European Communities (2001/77/EC).

EU (2002) Council Decision of 25 April 2002 concerning the approval, on behalf of the European Community, of the Kyoto Protocol to the United Nations Framework Convention on Climate Change and the joint fulfilment of commitments there-under. Official Journal of the European Communities (2002/358/CE).

EU (2003) Directive of the European Parliament and of the Council of 13 October 2003 on Establishing a scheme for greenhouse gas emission allowance trading within the Community and amending Council Directive 96/61/EC. Official Journal of the European Communities (2003/81/EC).

European Commission (2005) Seaflow, Pilot project for the exploitation of marine currents. Office for Official Publications of the European Communities, Luxembourg. ISBN 92-8944593-9.

Falconer, R.A. & Owens, P.H. (1987) Numerical simulation of flooding and drying in a depth-averaged tidal flow model. Proceeds of Insinuaton of Civil Engineering. 83, 161180.

Fitzgerald, J. (2004) Generation adequacy in an island electricity system. The Economic and Social Research Institute, Dublin. (Available at www.esri.ie).

193

Flather, R.A. & Heaps, N.S. (1975) Tidal Computations for Morecambe Bay. The Geophysical journal of the Royal Astronomical Society, 42, 489-517. ISSN: 00168009.

Flather, R.A. (1993) A Storm Surge Prediction Model for the Northern Bay of Bengal with Application to the Cyclone Disaster in April 1991. Journal of Physical Oceanography, 24, 172-190. ISSN: 00223670.

Fraenkel, P.L. (2002) Power from marine currents. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy, Vol. 216, Pp. 1-14. ISSN: 09576509.

Frau, J.P. (1993): Tidal energy: promising projects - La Rance, a successful industrial-scale experiment. IEEE Transactions on Energy Conversion, Vol. 8, 3, pp. 552-558. ISSN: 08858969.

Garrad Hassan (2003) The Impacts of Increase Levels of Wind Penetration on the Electricity Systems of the Republic of Ireland and Northern Ireland. Commission for Energy Regulation, Dublin. (Available at www.cer.ie).

Giebel G. (2003) The State-Of-The-Art in Short-Term Prediction of Wind Power, A Literature Overview, Version 1.1. ANEMOS Project, (Available at www.anemos.cma.fr)

Giebel, G., Landberg, L., Kariniotakis, G. & Brownsword, R. (2003) State-of-the-Art on Methods and Software Tools for Short-Term Prediction of Wind Energy Production. European Wind Energy Conference & Exhibition EWEC, Madrid, Spain.

Haidvogel, D.B., & Beckmann, A. (1998) Numerical Models of the Coastal Ocean. The Sea, 10, 457 482. ISBN: 0471115444.

194

Haidvogel, D.B., Beckmann, A. and Hedstrom, K.S. (1991) Dynamical simulations of filament formation and evolution in the Costal Transition Zone. Journal of Geophysical Research, 96, 15017 15040. ISSN: 01480227.

Hammerfest Stromas (2002a) Tidal power plant in Kvalsunde, The installation of the prototype has started. Press release, Hammerfest Stromas, Norway. (Available at www.etidevannsenergi.com).

Hammerfest Stromas (2002b) The tidal power plant in Kvalsundet, The shore terminal is completed and the umbilical in position. Press release, Hammerfest Stromas, Norway. (Available at www.e-tidevannsenergi.com).

Hammerfest Stromas (website) http://www.e-tidevannsenergi.com

Howarth, M.J. (1990) Atlas of Tidal Elevation and Currents around the British Isles. Her Majesty's Stationery Office, London.

Hunter, S. (1982) Formation of SO3 in Gas Turbines. Journal of Engineering for Power, vol. 104, pp. 4451.

Hydraulic Current Turbines (2003) Hydraulic Current Turbines Ltd. Data Sheet, Hydraulic Current Turbines, UK.

Hydrohelix _stand=418

Energies

(website)

http://cci-entreprises.icomme.fr/public/stand.php?id

Hydroventuri (website) http://www.hydroventuri.com.

Idaho National Laboratory (website) http://www.inl.gov/wind/software/

195

Intergovernmental Panel on Climate Change (2001) Climatechange 2001: the scientific basis. Cambridge University Press. ISBN: 0521014956.

J.A. Consultants (2004) www.tidalstream.co.uk

Kazarlis, S.A., Bakirtzis, A.G., & Petridis, V. (1996) A genetic algorithm solution to the unit commitment problem. IEEE Transactions on Power Systems, Vol. 11, P. 83-92. ISSN: 0888950.

Kirk McClure Morton, Queens University Belfast & Natural Power Consultants (2004) Tidal & Current Energy Resources in Ireland. Sustainable Energy Ireland, Dublin. (Available at www.sei.ie).

Lalor, G., Mullane, A., & OMalley, M. (2005) Frequency Control and Wind Turbine Technologies. IEEE Transactions on Power Systems, Vol. 20, No. 4, P. 1905-1913. ISSN: 08858950.

Lalor, G.R. (2005) Frequency Control on an Island Power System with Evolving Plant Mix. PhD Thesis, University College Dublin.

Lambeck, K. (1980) The Earths Variable Rotation: Geophysical causes and Consequences. Cambridge University Press, Cambridge, UK. ISBN: 0521673305.

Larsson, A. (2000) The Power Quality of Wind Turbines. Chalmers University of Technology, Goteborg, Sweden.

Leonhard, W. & Grobe, E.M. (2004) Sustainable Electrical Energy Supply with Wind and Pumped Storage - a Realistic Long-Term Strategy or Utopia? IEEE Power Engineering Society General Meeting, Vol. 2, Pp. 1221 - 1225. (Available at http://ieeexplore.ieee.org).

Lunar Energy (website) http://www.lunarenergy.co.uk

196

Marine Current Turbines (Website) http://www.marineturbines.com

Marine Current Turbines Ltd. (2003) Seaflow Project, Press Pack 16th June. Marine Current Turbines Ltd., Bristol.

Marine Institute (Website) http://www.marine.ie

McErlean, T., McConkey, R., & Forsythe, W. (2002) Strangford Lough: An Archaeological Survey of the Maritime. The Blackstaff Press Ltd, Belfast. ISBN: 0856407232.

MOSEK (2002)The MOSEK optimization tools version 3.2 (Revision 8) Users manual and reference. MOSEK ApS, Denmark. (Available at www.mosek.com/).

Mullane, A. & O'Malley, M. (2005) The inertial-response of induction-machine based wind-turbines. IEEE Transactions on Power Systems, Vol. 20, P. 1496 - 1503. ISSN: 08858950.

Mullane, A.P. (2003) Advanced control of wind energy conversion systems. University College Cork, PhD thesis.

Munk, W. & C. Wunsch (1998) Abyssal recipes II: energtics of tidal and wind mixing. Deep Sea Research part1, Vol. 45, pp.1977-2010. ISSN: 09670637.

Munk, W. (1997) Once Again Tidal Friction. Progress in Oceanography, vol. 40, pp. 735. ISSN: 00796611.

OceanTecs (2004) Oeantecs, Renweable Energy. Data Sheet, Oeantecs, Hartlepool, UK.

Ofgem (2005) Renewables Obligation - an overview of the second year. Ofgem, factsheet 49 14.02.05, Belfast. (Available at www.ofgem.gov.uk).

197

Oilnergy (website) http://www.oilnergy.com

Ongsakul, W. & Petcharaks, N. (2003) Transmission Constrained Unit Commitment in Competitive Environment by Enhanced Adaptive Lagrangian Relaxation. IEEE Bologna PowerTech Conference, Italy.

OpenHydro (website) http://www.openhydro.com

Pond S.

& Pickard G.L. (1983) Introductory Dynamical Oceanography. 2nd edition.

Butterworth-Heinemann, Oxford. ISBN: 0750624965.

Pugh, T.G. (1981) Tidal amphidrome movement and energy dissipation in the Irish Sea. The Geophysical journal of the Royal Astronomical Society, Vol. 67, pp 515-527. ISSN: 00168009.

Pugh, T.G. (1987) Tides, Surges and Mean Sea-Level. John Wiley & Sons Ltd., Chichester.1-472. ISBN: 047191505.

Richtmyer, R.D. (1967) Difference methods for initial value-problems. 2nd ed. Interscience Publishers, New York. ISBN: 0894647636.

Rourke, S. (2003) Locational Marginal Pricing of Electricity. University College Dublin, MEng Thesis.

Sasaki, H., Watanabe, M., Kubokawa, J., Yorino, N., & Yokoama, R. (1992) A solution method of unit commitment by artificial neural networks. Transactions on Power systems, Vol. 7, P. 974-981. ISSN: 08858950.

Schellnhuber, H. J., Cramer, W., Nakicenovic, N., Wigley, T. & Yohe, G. (2006) Avoiding Dangerous Climate Change. Cambridge University Press. ISBN: 0521864712.

198

Simpson, J.H. (1998) Tidal Processes in Shelf Seas. The Sea, Chap 5, Vol. 10. K.H Brink & A.R. Robinson, John Wiley & Sons Inc. UK. ISBN: 0851153135.

SMD Hydrovision (2004) TidEl. Data Sheet, SMD Hydrovision, Newcastle upon Tyne, UK.

Song, Y. & Haidvogel, D.B. (1994) A semi-implicit ocean circulation model using a generalized topography-following coordinate. Journal Of Computational Physics, 115, 228 244. ISSN: 00219991.

Streiffert, D., Philbrick, R. & Ott, A. (2005) A Mixed Integer Programming Solution for Market Clearing and Reliability Analysis. IEEE, Power Engineering Society General Meeting, P. 195-202. (Available at http://ieeexplore.ieee.org).

Sustainable Energy Ireland (2004) Operating Reserve Requirements as Wind Power Penetration Increases in the Irish Electricity System. Sustainable Energy Ireland, Dublin. (Available at www.sei.ie).

Sustainable Energy Ireland (2006) Renewable Energy in Ireland, 2005 Update. Sustainable Energy Ireland, Dublin. (Available at www.sei.ie).

System Operator for Northern Ireland (2003) Seven year transmission statement 2003/04 2009/10. System Operator for Northern Ireland, Belfast. (Available at www.soni.ltd.uk/).

System Operator for Northern Ireland (2004) Personal communication.

Takriti, S. & Birge, J.R. (2000) Using Integer Programming to Refine Lagrangian-Based Unit Commitment Solutions. IEEE Transactions on Power Systems, Vol. 15, P. 151-156. ISSN: 08858950.

199

Taylor, G.I. (1919) Tidal Friction in the Irish Sea. Philosophical Transactions, Vol. 220,193. ISSN: 03702316.

The Engineering Business (Website) http://www.engb.com/

Thomson, M. (2004) The Flow Around Marine Current Turbines. MSc Thesis, University of Bristol.

Transmission System Operator Ireland (2003) Generation Adequacy Report 2004 2010. Transmission System Operator Ireland, Dublin, Ireland. (Available at www.eirgrid.com).

Transmission System Operator Ireland (2005) Transmission Forecast Statement 2005 2011. Transmission System Operator Ireland, Dublin, Ireland. (Available at

www.eirgrid.com).

Trapp, T. (2004) Stingray tidal stream generator. WATTS Conference, London , 16th March.

Tuohy, A. (2006) Personal communication, Electricity Research Centre, Dublin.

Twidell, J.W. & A.D. Weir (1986) Renewable Energy Resources. The University Press, Cambridge, UK. ISBN: 0419253300.

United Nations (1997) Kyoto Protocol to the United Nations Framework Convention on Climate Change. United Nations. FCCC/CP/1997/L.7.

Van den Noort Innovations bv (website) http://www.noort-innovations.nl/

Verdant Power (website) http://www.verdantpower.com

200

Walsh, M.P. & O'Malley (1997) Augmented Hopfield Network for Unit Commitment and Economic Dispatch. IEEE Transactions on Power Systems, Vol. 12, Pp. 1765-1774. ISSN: 08858950.

Walsh, M.P. (1998) A Novel Neural Network for Power System Scheduling. Michael Patrick Walsh. University College Dublin, PhD thesis.

Webb, D.K. (2002) Current and Tidal Flow in the Menai Striat. BSc Thesis, University of Wales, Bangor.1-99.

Whittaker, T., P L Fraenkel, A Bell & L Lugg (2003) The Potential for the use of Marine Current Energy in Northern Ireland. DTI. (Available at www.dti.gov.uk).

Wikipedia (website) http://en.wikipedia.org/

Wood, A.J. & Wollenberg, B.F. (1996) Power Generation Operation and Control. John Wiley & Sons inc. New York. ISBN: 0471586994.

Wright, M. (2004) SeaFlow tidal current turbine. WATTS Conference, London , 16th March.

201

1.0 Appendix
The model of the All-Island system was constructed in PLEXOS using a combination of data sources as indicated below. Node Data Name

- Number looked up in the PSSE file - The abbreviation was provided in the All Island project data set Percentage of total system load - ROI - provided in the All Island project data set - NIE - looked up in the PSSE file Voltage - provided in the All Island project data set Line Data Name Rating Resistance Reactance From Node To Node Transformer Name Rating Resistance Reactance From Node To Node

- generated from the bus numbers and names - provided in the All Island project data set - provided in the All Island project data set - provided in the All Island project data set - looked up in the PSSE file - looked up in the PSSE file

- generated from the bus numbers and names - provided in the All Island project data set - provided in the All Island project data set - provided in the All Island project data set - looked up in the PSSE file - looked up in the PSSE file

Generators Name - provided in the All Island project data set Node - looked up in the PSSE file Maximum capable generation - provided in the All Island project data set Minimum stable generation - provided in the All Island project data set Start times - provided in the All Island project data set Start costs associated with the start times - calculated from start energy and fuel price Maximum ramp rate - provided in the All Island project data set Minimum ramp rate - provided in the All Island project data set 202

Thermal Plant Hydro Plant

Probability of forced outage Committed - set to 0 on the first run Fuel type - provided in the All Island project data set No load heat rate - provided in the All Island project data set Incremental heat rate - provided in the All Island project data set Head storage o Max capacity - provided by ESB o Inflow rate - provided by the ESRI Foot storage - no limitations set

Load All island load profile - provided in the All Island project data set

Fuel

Price per GJ

- provided by the ESRI

203

2.0 Appendix grid connection options


The distribution system re-enforcements considered to facilitate the grid connection of tidal generation, and the options considered for connecting areas of tidal resource to the 33 / 38 kV distribution system.

1. 2. 2.1 Grid connection for the North Coast and Rathlin Island resource.

Figure A-2-1. Scenario A, no change to the existing 33 kV distribution system.

204

Figure A-2-2. Scenario B, upgrade of line BC1 from Coleraine to Ballycastle.

Figure A-2-3. Scenario C, upgrade of line BC2 from Coleraine to Ballycastle.

205

Figure A-2-4. Scenario D, upgrade of line BC1 and DC2 from Coleraine to Ballycastle.

Figure A-2-5. Scenario E, upgrade of line BM1 from Coleraine to Bushmills.

206

Figure A-2-6. Scenario F, upgrade of line BM2from Coleraine to Bushmills.

Figure A-2-7. Scenario E, upgrade of line BM1 and BM2 from Coleraine to Bushmills.

207

Figure A-2-8. Scenario F, upgrade of line BC2 from Coleraine to Ballycastle to 110 kV.

Figure A-2-9. Scenario T-A, operation of the planned line to Rathlin Island at 33 kV.

208

Figure A-2-10. Scenario T-B, upgrade of line R6 and construction of line R2.

Figure A-2-11. Scenario T-C, operation of the planned line to Rathlin Island at 33 kV and construction of N1 a 33 kV line to the North Coast from Ballycastle.

209

Figure A-2-12. Scenario T-D, operation of the planned line to Rathlin Island at 33 kV, construction of N1 a 33 kV double circuit to the North Coast from Ballycastle and construction of the 33 kV cable R5 from the North Coast to Rathlin Island.

210

Figure A-2-15. Scenario T-E, operation of the planned line to Rathlin Island at 33 kV, construction of N1 a 110 kV line to the North Coast from Ballycastle and construction of the 110 kV cable R5 from the North Coast to Rathlin Island.

211

2.2 Grid connection for the Maiden Islands resource.

Figure A-2-16. Scenario A, no change to the existing 33 kV distribution system.

Figure A-2-17. Scenario B, upgrade of one line from Larne main to Larne North 33 kV sub station.

212

Figure A-2-18. Scenario C, upgrade of one line from Larne main to Larne North 33 kV sub station.

Figure A-2-19. Scenario D, upgrade of both lines from Larne main to Larne North 33 kV sub station.

213

2.3 Carnsore Point

Figure A-2-20. Scenario A, no change to the existing 38 kV distribution system.

Figure A-2-21. Scenario B, upgrade of the 38 kV line from Killinick to Wexford.

214

Figure A-2-22. Scenario C, upgrade of the 38 kV line from Killinick to Wexford to a double circuit.

215

2.4 Malin Head

Figure A-2-23. Scenario A, no change to the existing 38 kV distribution system.

216

Figure A-2-24. Scenario B, upgrade of the 38 kV line from Buncrana to Carndonagh.

Figure A-2-25. Scenario C, upgrade of the 38 kV line from Buncrana to Carndonagh to a double circuit.

217

3.0 Appendix Strangford model parameters


The parameters used to setup the model of the 11 kV system to Strangford in DigSilent (see table A-3-1 and Figure 6-21) were based on simplifications made from each of the line sections listed in NIEs database.
Positive Sequence Line section Sec. 1 Sec. 2 Sec. 3 Sec. 4 Sec. 5 Sec. 6 Sec. 7 Sec. 8 Sec. 9 Sec. 10 Sec. 11 Sec. 12 Sec. 13 Sec. 14 Sec. 15 Sec. 16 Sec. 17 Sec. 18 Sec. 19 Resistance (ohms) 0.1363 0.2243 0.2715 0.4421 0.1559 0.2061 0.2407 0.1359 0.4894 0.5060 0.0404 0.0459 0.0760 0.4142 0.1464 0.3232 0.3209 0.7410 0.9867 Reactance (ohms) 0.0973 0.1606 0.1943 0.2066 0.1116 0.1475 0.1723 0.0973 0.3503 0.3621 0.0289 0.0329 0.0544 0.2965 0.1048 0.2314 0.2297 0.3111 0.3754 Suscep. (S) 0.0009 0.0006 0.0007 0.0502 0.0004 0.0005 0.0006 0.0004 0.0013 0.0013 0.0001 0.0001 0.0002 0.0011 0.0004 0.0008 0.0009 0.0003 0.0181 Negative Sequence Resistance Reactance Suscep. (ohms) (ohms) (S) 0.1776 0.3974 0.0002 0.2840 0.6580 0.0003 0.3438 0.7964 0.0003 1.2617 0.6518 0.0002 0.1974 0.4573 0.0002 0.2609 0.6044 0.0002 0.3049 0.7062 0.0003 0.1721 0.3987 0.0002 0.6198 1.4357 0.0006 0.6408 1.4842 0.0006 0.0512 0.1186 0.0000 0.0581 0.1346 0.0001 0.0962 0.2229 0.0001 0.5245 1.2149 0.0005 0.1854 0.4295 0.0002 0.4093 0.9481 0.0004 0.4063 0.9412 0.0003 0.8536 1.2492 0.0001 1.3004 1.4389 0.0000 Load (at end of sec.) MW 0.0034 0.0460 0.0155 0.0284 0.0011 0.0080 0.0032 0.0069 0.1376 0.0333 0.0344 0.0538 0.1002 0.0215 0.0103 0.0088 0.0080 0.0088 0.2419 MVAR 0.0011 0.0151 0.0051 0.0093 0.0004 0.0026 0.0011 0.0023 0.0452 0.0110 0.0113 0.0177 0.0329 0.0071 0.0034 0.0029 0.0026 0.0029 0.0795

Table A-3-1. The line properties used to represent the 11 kV system to Strangford.

218

Anda mungkin juga menyukai