Minimizing the Cost of Wind Energy for Vashon Island – a Low...
36
Minimizing the Cost of Wind Energy for Vashon Island – a Low Wind Speed Site Eron Jacobson
Transcript of Minimizing the Cost of Wind Energy for Vashon Island – a Low...
Minimizing the Cost of Wind Energy for Vashon Island – a Low Wind Speed
Site
Eron Jacobson
Outline
IntroductionWind Resource AssessmentOptimum Turbine Design for a Low Wind Speed RegimeGrid Integration and Energy StorageConclusion
Introduction
Vashon Island is located in the Puget Sound – a predominantly low wind resource region.
Vashon IslandClass 1
Class 2
Class 3
Class 4
Class 5
Class 6
Class 7
0 - 5.6m/s
5.6 - 6.4m/s
6.4 – 7.0m/s
7.0 - 7.5m/s
7.5 - 8.0m/s
8.0 - 8.8m/s
8.8 - 11.9m/s
Average Wind Speed at 50m
Wind Power Class
Introduction
Most wind farms are developed in Class 4 wind resources or greater to be cost competitive with other energy generation.– Class 6: cost competitive – Class 4: cost competitive with 1.8¢/kWh PTC
A wind farm on Vashon Island could not compete with Washington State’s energy mix.The Vashon Island community may support the higher cost of energy from a wind farm on the island.
Goal of the Study
Determine the wind turbine and wind farm design to provide the least cost of energy (COE) for generating 26GWh of electricity annually on the island.Predict the reduction in COE over the next 10 years to facilitate deciding the best time to build the wind farm.Assess the cost of integrating the wind farm into the existing electrical grid and potential benefits of an energy storage system.
Wind Resource Assessment
Characterizing the wind resource is the critical task for a wind farm feasibility assessment: – 5% error in velocity equates to a ~15% error in
power.Typically, wind data is collected with an anemometer tower at the proposed turbine sites for at least one year.A comparison is then made with long term data from a nearby anemometer.
Data Collection and Turbine Sites
SeaTac
Beall
Vashon Island
Maury Island
Wind Data Analysis
SeaTac airport anemometer (1996-2004)10 meter measurement height
Beall anemometer (12/04 - 5/05)26, 35, 49 meter measurement heights
The data is grouped into 30º directional sectors and analyzed for:– Sector frequency– Average wind speed– Distribution of wind speeds (and Weibull fit)
Directional Sector AnalysisSector Distribution Frequency Sector Average Wind Speeds [m/s]
0.200.15
0.100.05
531
Distribution of Wind Speeds (Weibull), 205-235 Direction
0
0.05
0.1
0.15
0.2
0.25
0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 10.5
11.5
12.5
13.5
14.5
15.5
16.5
17.5
18.5
19.5
20.5
21.5
22.5
23.5
24.5
Wind Speed [m/s]
Freq
uenc
y Weibull Fit
Seatac Raw Data
Bucket 8
Bucket 9Bucket 10
, k=2.6
Terrain Effects on Wind Speed
Wind Farm Development Tool (WFDT) software program normally used to extrapolate data to turbine sites.These programs use analytic solutions for adiabatic turbulent boundary layer flow.In this study, simple guidelines for boundary layer flow were used to account for:– Surface coverage changes– Topographic changes– Height changes
Application of Guidelines
SeaTacMeasurement
UpstreamVelocity
Turbine Site
1. Speed-up at runway
2. Speed-up over water
3. Speed-up over ridgeline
4. Speed increase with height
Guideline Applications:
The guidelines are used to project the speed of each wind speed buckets for each directional sector from the collection sites to the speed expected at each of the ten turbine sites.
Wind Resource Results
4.8550
5.0160
5.1470
5.2480
5.3390
5.41100
Average Wind Speed [m/s]Height [m]
Velocity Shear Profile at Maury Island Turbine Sites
0
20
40
60
80
100
120
0 1 2 3 4 5 6 7Velocity [m/s]
Hei
ght [
m]
Distribution of Projected Wind Speeds for Turbine Site 8 at 50m
0
0.05
0.1
0.15
0.2
0.25
0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.510.511.512.513.514.515.516.517.518.519.520.521.522.523.524.5
W ind Speed [m/s]
Freq
uenc
y
ProjectedDistribution
Seatac OriginalDistribution
Average Distribution of Projected Wind Speeds at 50m
0
0.05
0.1
0.15
0.2
0.25
0.3
0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.510.5
11.5
12.5
13.5
14.5
15.5
16.5
17.5
18.5
19.5
20.5
21.5
22.5
23.5
24.5
Wind Speed [m/s]
Freq
uenc
y
Weibull Fit, k = 1.6
ProjectedDistribution
Wind Resource Results
The Guidelines are only estimates – more accurate results are required for wind farm development.The wind resource needs to be characterized better by:– Installing an anemometer at the south end of
Maury Island for one year– Using WFDT to extrapolate this data to the
turbine sites
i = the turbine sitej = the directional sectork = wind speed bucketPi,j,k = power for the wind speed bucketηt = the gearbox efficiencyηg = the generator efficiencyηpe = the power electronic efficiency ρa = 1.225 kg/m3, the density of airCp = 0.50, the aerodynamic rotor power coefficient for state-of-the-art turbines U’i,j,k = the projected wind speed for the bucketD = the diameter of the turbine rotor
Probabilistic Energy Production Model
4'
21 2
3,,,,
DUCP kjipapegtkji πρηηη=Power
Probabilistic Energy Production Model
∑∑= =
=12
1,,
25
1,, ***8760
jkjij
kkjii fbfsPE
Ei = the annual energy for turbine iPi,j,k = the power for the wind speed bucketfsj = the frequency of the sectorfbi,j,k = the frequency of the wind speed bucket
Energy
Turbine Energy by Sector 0.39 SR at 50m Height
0
200400
600
800
10001200
1400
355-2
5
25-55
55-85
85-11
511
5-145
145-1
7517
5-205
205-2
3523
5-265
265-2
9529
5-325
325-3
55
Directional Sector
Ann
ual E
nerg
y [M
Wh]
Turbine 1Turbine 2Turbine 3Turbine 4Turbine 5Turbine 6Turbine 7Turbine 8Turbine 9Turbine 10
88Turbine 8
76Turbine 10
72Turbine 6
67Turbine 5
100Turbine 3
97Turbine 2
97Turbine 1
96Turbine 9
92Turbine 4
90Turbine 7
Energy [%]Height [m]
Probabilistic Energy Production Model
Optimizing Turbine Design
Wind turbines are not designed to be optimized for such low wind speed regimes.To minimize the cost of energy, turbine designs are often tailored for an application including:
– Rotor diameter– Generator rating– Tower height
Often, turbines are tailored by specific power:
AreaSweptRotorRatingGeneratorPowerSpecific =
Turbine Model
A turbine design and cost model tool was developed that is scalable with rotor diameter, generator rating and tower height.The most important structural loads are:– Rated torque:
– Extreme thrust on the rotor:
33
161 D
VVCQ
T
RpaR πρ=
22)*85.0(81 DSCVT DEXaR πρ=
Turbine Components and Systems
16. Hydraulic and Lubrication System
15. Tower
14. Electrical Connections
13. Yaw Drive and Bearing
12. High-Speed Shaft, Coupler
11. Nacelle Cover and Bedplate
10. Variable Speed Electronics
5. Low-speed Shaft
9. Main Bearings
8. Controls and Safety System
7. Generator
6. Gearbox
4. Mechanical Brake
3. Pitch Mechanism
2. Hub
1. Rotor Blades
9
10
4
5
3
61
2
Roads
Overhead Lines
Recloser
Underground Cables
Balance-of-Station Model
12. Manufacturer Mark-up
11. Inspection Costs
10. Surveying Costs
9. Engineering Costs
8. Permitting Costs
7. Assembly and Installation
6. Crane Cost
5. Transportation
4. Foundation
3. Crane Pad
2. Roads and Civil Works
1. Electrical Interface
COE Model
COE = Levelized cost of energy ($/kWh)FCR = Fixed charge rate (0.106/year)ICC = Initial capital cost ($)AEPnet = Net annual energy production (kWh/yr)O&M = Operating and maintenance cost ($/kWh)
MOAEP
ICCFCRCOEnet
&)*(+=
COE and Optimum Turbine
Optimum Turbine Design
10¼COE [¢/kWh]
0.325Specific Power [kW/m2]
70Tower Height [m]
1.5Generator Rating [MW]
83.5Rotor Diameter [m]
Specific Power Study
10.0
10.5
11.0
11.5
12.0
12.5
0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Specific Power [kW/m^2]
CO
E [c
ents
/kW
h]
1.5MW, 70mHH
2.0MW, 70-80m HH
1.0MW, 60-70m HH
Ten 1.5MW turbines would will provide 25GWh at the least COE.
Increasing or decreasing the turbine rating results in an unbeneficial compromise between energy production to costs.
Future Component Cost Reductions
Improvements in components and systems are likely to decrease the overall COE including:– Drivetrain– Power Electronics System– Rotor Blades and Controls– Tower and Nacelle Installations
A COE of 8¢/kWh or a 25% reduction of COE is calculated from reductions in component and system costs determined by previous studies.
Grid Integration and Energy Storage
Energy generation must be balanced with grid load by the grid operator in real time.Balancing generation and load occurs over a significant range of timescales:– Grid operators plan which generators should be on-
line the following day based on capacity and economic considerations
– Automatic generation control (AGC) software ramps generators up or down to correct quick fluctuations
The stochastic nature of wind energy makes balancing of the grid more difficult.
Three Timescales for Grid Operation
Day-ahead (unit commitment): – One day, with one-hour time
increments – Generators scheduled 12 hours in
advance of the subject day. Load following:
– One hour, with five to 10-minute increments
– Economic-dispatch model every five to 10 minutes
Regulation: – One minute, with one to five-second
increments– AGC are used to match generation
to load
Seconds to minutes
Minutes to hours
0 6 12 18Time (Hour)
Sys
tem
Loa
d
Regulation Load Following
Sys
tem
Loa
d
1 2 3Days
Day Ahead
Integration of Wind Power
The grid load is not entirely predictable.Load forecasts errors dictate that traditional generating sources must be online that may be ramped up down to balance load and generation (ancillary services).Similarly, wind farms use forecasting techniques to predict power output for the three timescales, and have associated forecast errors.
Integration of Wind Power
Time
Load
Err
or
Load
Load + Wind
Overall, wind power tends to increase forecast errors slightly depending on the wind penetration level.
Since wind power mitigates load forecast errors about 50% of the time, the grid operator must balance aggregate load and generation.
Costs of Wind Integration
The necessary increase of ancillary services with the integration of wind energy is passed onto the wind farm.These costs are estimated by using:– wind forecasting techniques – statistical methods – and yearly data
A table correlating wind data at SeaTac airport for 1997 to power output for the Maury Island wind farm is developed.
Forecast Error Calculations
Day-ahead forecast error:
ave
nnhourave
aheadday P
PPE
24
24
1,∑
=−
−=
Pave: – daily power (lower bound) – monthly power (upper bound)
Load following error:
1,,ˆ
−= nhournhour PP
nhournhourfollowload PPE ,,ˆ−=−
Persistence forecast model
Assume a normal distribution.
Costs of Ancillary Services
A probabilistic model estimates the increase in ancillary service costs based on reserve generation costs.
6,6300.260.0430.0520.16Maury Island Study
96,2003.700.842.700.16PSE Study
Total Costs for 26GWh
[$]Total
[$/MWh]
Day-ahead Costs
[$/MWh]
Load-Following
Costs [$/MWh]
Regulation [$/MWh]Results
Example PSE Forecast Error Distributions
00.020.040.060.08
0.10.120.140.160.18
-10 -5 0 5 10Percent Error
Prob
abili
ty o
f Per
cent
Erro
r
PSE Load
PSE Load +Wind
Energy Storage
An energy storage system may be economically beneficial to the Maury Island wind farm by:– Reducing ancillary services for the three
timescales– Shifting generation from off-peak to on-peak times
Load following support and load shifting capabilities are the most beneficial. – Load following ancillary services are the most
expensive and the number of cycles are manageable
– Load shifting revenues have a large potential
Vanadium Redox Flow Batteries
VRBs applicable characteristics:
– Scalable– Fast response times– Able to sustain many
cyclese-e-
V5+
V4+
pump
V2+
V3+
H+
V2+/V3+
Electro-lyte
Tank
pump
V5+/V4+
Electro-lyte
Tank
Membrane
Positive Electrode
Negative Electrode
Cell
Operation is similar to a fuel cell, and is a reaction of Vanadium Ions only.VRBs are about 75-80% efficient.
Energy Storage Results
A VRB energy storage system has no economic benefit.The benefits of avoided ancillary costs and value of shifted energy are small compared to the system ICC and value of lost energy.
Present Value and ICC for Energy System vs. Capacity
-505
1015202530354045
0 10 20 30 40
Storage System Capacity [MWh]
Val
ue a
nd C
ost
[Mill
ion
$]
PresentValueICC
Conclusions
The average wind speed at 50 meters for Maury island turbines is 4.9m/s.10 turbines could produce 25GWh of electricity at a COE of about 10¢/kWh provided the turbines have:
– 1.5MW ratings– 83.5 meter rotors– 70 meter towers
The COE may drop by 25% over the next 10 years. Integrating the wind farm to the PSE grid would cost an additional 0.03 – 0.4 ¢/kWh.An energy storage system would not be economically beneficial.
Questions?
Comparison with Other Studies
Average wind speed Weibull shape parameter
Comparison of COE vs. W indspeed
2468
1012141618
3 4 5 6 7 8 9 10 11
Wind Speed [m/s]
CO
E [c
ents
/kW
h] W indPACT
Maury Island
Comparison of Previous Specific Power Studies for COE Minimization
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
3 4 5 6 7 8 9 10 11
Wind Speed [m/s]
Opt
imum
Spe
cific
Po
wer
[kW
/m^2
]
WindPACTBurtonMaury Island
COE results are slightly high:
Balance-of-Station costs are slightly higher because of turbine locations.Conservative assumptions were generally used.
Specific power results are dependent on both:
Webiull Shape Parameter Effects on Distribution
00.020.040.060.080.1
0.120.140.16
0 5 10 15 20 25
Wind Speed [m/s]
Freq
uenc
y k=1.7, Maury
k=2, WindPACT & Burton
