Equivalent wind speed for AEP Rozenn Wagner
VindKraftNet 03-11-2011 Risø
Experimental setup & Data processing N
Filters: • wind direction; • no rain; • lidar signal availability 100% at all heights; • turbine status=1.
Risø DTU, Danmarks Tekniske Universitet
Profiles classification
u fit ( z ) = RSS = i
m uhub
(u
Risø DTU, Danmarks Tekniske Universitet
fit
z zhub
m fit ( zi ) ui
RSS0.1
Standard power curve
2 groups of profiles result in 2 different power curves Risø DTU, Danmarks Tekniske Universitet
Equivalent wind speed U1
Concept:
U2
One wind speed representative of the whole wind speed profile in front of the wind turbine rotor in term of power production
U3
Ueq
U4 U5
A1 A2
Definition:
U KE
1 = A
1/3
R 3
u ( z ) c( z )dz R
Risø DTU, Danmarks Tekniske Universitet
Ai u A i =1 N
3 i
1/3
A3 A4 A5
Power curve with equivalent wind speed
Similar power curves are obtained for both groups of profiles Risø DTU, Danmarks Tekniske Universitet
Comparison of the power curves Difference due to the shear distribution during the power curve measurement.
How can the equivalent wind speed power curve be used for AEP estimation?
Risø DTU, Danmarks Tekniske Universitet
Annual Energy Production
AEP
X
=
Power curve at wind farm site
Risø DTU, Danmarks Tekniske Universitet
Wind speed distribution at wind farm site
AEP estimation
Predicted = AEP
X
Reference power curve: measured at a reference site
Wind speed distribution at wind farm site
Uhub power curve
Uhub distribution
Ueq power curve
Ueq distribution
Risø DTU, Danmarks Tekniske Universitet
Illustration of the 2 cases with Høvsøre data Data Group 1 Reference site
Reference power curve
Data Group 2 Estimated site
Measured wind speed
Total power estimation case1: with uhub case 2: with Ueq Risø DTU, Danmarks Tekniske Universitet
Illustration of the 2 cases with Høvsøre data Case 1
prediction: + 1.76%
Risø DTU, Danmarks Tekniske Universitet
Illustration of the 2 cases with Høvsøre data Case 1
Case 2
prediction: + 1.76%
prediction: 0.005%
Improved AEP estimation by using the equivalent wind speed both in the power curve and the wind speed distribution. Risø DTU, Danmarks Tekniske Universitet
More realistic application
Power curve and wind distribution from 2 separate sites
Østerild
Høvsøre
Risø DTU’s Test Site for Large Turbines Høvøsre Risø DTU, Danmarks Tekniske Universitet
More realistic application
Power curve and wind distribution from 2 separate sites
Power curve measured at Høvsøre in Feb-March 2009
Wind speed distribution measured at Østerild May 2010- May 2011 All
nb hours
400 300 200 100 0
4
6
8
10 m s
4 possible combinations, but no turbine yet. Risø DTU, Danmarks Tekniske Universitet
12
14
Combination 1: • Equivalent power curve
Account for the shear during the power curve measurement; expected to be the same power curve at any site
Reference AEP
• Equivalent wind speed distribution
Risø DTU, Danmarks Tekniske Universitet
Account for the shear at Østerild
Combination 2: • Hub height power curve
Underestimates the power produced because of the shear during the power curve measurement.
-2.3%
• Hub height wind speed distribution
Risø DTU, Danmarks Tekniske Universitet
Slightly underestimates the energy available because does not account for the shear at Østerild (assumes flat wind speed profiles)
Combination 3: • Equivalent power curve
Account for the shear during the power curve measurement; expected to be the same power curve at any site
-0.5%
• Hub height wind speed distribution
Risø DTU, Danmarks Tekniske Universitet
Slightly underestimates the energy available because does not account for the shear at Østerild (assumes flat wind speed profiles)
Summary
U_hub power curve Ueq power curve U_hub distribution
(-2.3%)
U_eq distribution
(-0.5%) (ref)
The error depends both on: • the Ueq/Uhub distribution during the power curve measurement (1.8%) • and the Ueq/Uhub distribution at the assessed site (-0.5%)
Risø DTU, Danmarks Tekniske Universitet
More examples
Ueq/Uhub< 1
Risø DTU, Danmarks Tekniske Universitet
Ueq/Uhub> 1
Case1: Ueq/Uhub> 1 Profiles with larger kinetic energy than flat profiles
U_hub distribution
U_hub power curve
Ueq power curve
(-3.8%)
(-2.1%)
U_eq distribution
(ref)
Part of the the error due to Ueq/Uhub distribution at the assessed site larger than before (-2.1%); Overall error larger than previous case.
Risø DTU, Danmarks Tekniske Universitet
Case2: Ueq/Uhub< 1 Profiles with smaller kinetic energy than flat profiles
U_hub distribution
U_hub power curve
Ueq power curve
(0.00%)
(+1.8%)
U_eq distribution
(ref)
Specific case: The Ueq/Uhub distribution are very similar for both datasets.
Risø DTU, Danmarks Tekniske Universitet
Conclusions 1 The shear influences the AEP estimation in 2 ways: 1)Error in power curve due to the shear during the power curve measurement 2)Error in available energy at the assessed site. Missing uncertainty terms in the standard AEP estimation Equivalent wind speed results in a repeatable power curve. Improved AEP estimation with equivalent wind speed It requires to measure the wind speed profiles for site assessment Risø DTU, Danmarks Tekniske Universitet
Conclusions 2 What to do if the equivalent wind speed distribution at the assessed site is not available? If the Ueq/Uhub distributions at the two sites are similar: use the standard AEP calculation (wind speed at hub height). If the Ueq/Uhub distributions are different: combine the hub height speed distribution with the equivalent power curve. But to know the Ueq/Uhub distribution… … you need to measure the shear!
Acknowledgement: EU SafeWind Risø DTU, Danmarks Tekniske Universitet
Wind lidar calibration • Comparison to cup anemometers at 5 heights
16
y 0.005 F
• Horizontal wind speed, sensing height error, uncertainty, direction
1.005 x
R2 0.998542
12
y 1.006 x R2 0.998541
• Synchronised lidar/mast data
10 8 0.6
6 4 4
6
8 cup
lidar error 40m m s
lidar
40m m s
14
• Accrediation for Windcubes calibration DANAK
0.4 0.2
0.0 12 0.2 40m m s 0.4
10
0.6
14
4
16
6
8
10
12
ref wind speed 40m m s
Risø DTU, Danmarks Tekniske Universitet
14
16
Wind lidar measurement •Rent out a Windcube •Set up, data transfert •Data analysis
Risø DTU, Danmarks Tekniske Universitet
Wind Lidar – A practical course 7-10 May 2012 •Lectures about the working principles and limitations of lidar systems. •Interpreting data •Hands-on experience with a lidar on-site
More information and registration http://www.risoe.dtu.dk/da/conferences/vea_lidar_course_2012.aspx?sc_lang=en Risø DTU, Danmarks Tekniske Universitet