ENERGY
What Your Anemometer Calibration
Really Means
Carl Ostridge and Taylor Geer
2 June 2015
1
DNV GL ©
2 June 2015
SAFER, SMARTER, GREENER
Overview
Why are we talking about
anemometers again?!
Why do anemometer
measurements vary?
Why does this matter?
2
DNV GL ©
2 June 2015
Why are we talking about this again?!
3
DNV GL ©
2 June 2015
Why Are We Talking About This Again?!
 Do all cup anemometers measure the same thing?
No, probably not.
 Implications for pre-construction P50 estimates and power
performance testing
– Anemometer Model + Calibration Facility could cause >2%
deviation on wind speed, 3-4% on energy.
– Not all uncertainty – some biases
– Disconnect between anemometer models and calibration
facilities used in pre-construction assessments and power
performance testing
We are talking about this because if we
aren’t careful, significant biases can enter
our work.
4
DNV GL ©
2 June 2015
Why Do Anemometer Measurements
Vary?
5
DNV GL ©
2 June 2015
Why Do Anemometer Measurements Vary?
1. Anemometer Design & Atmospheric Conditions
 Anemometer Design
– Cup shape, size
– Aerodynamics
– Bearing
– Signal generation
 Atmospheric Conditions
– Inflow angles
– Turbulence Intensity
– Temperature
– Air Density
6
DNV GL ©
2 June 2015
Wind Speed and Turbulence Intensity Measurement Example
Wind speed
Turbulence Intensity
 Two different anemometer models calibrated in different facilities, mounted in parallel
 Strong divergence and trend at lower wind speeds, better agreement at mid to high wind
speeds, still some bias in both wind speed and turbulence
 Deviation in correlation may seem small, but can result in big impacts
– Long-term mean wind speed and energy – up to 3-4% energy
– Smaller impacts on loss factor calculations and turbine suitability due to turbulence
intensity
7
DNV GL ©
2 June 2015
Why Do Anemometer Measurements Vary?
2. Wind Tunnel Calibrations
 Wind tunnels vary in size and design
– Large variation in test cross section area between MEASNET tunnels
– A mix of open and closed test sections
– Boundary and blockage effects have been shown to influence anemometer calibrations
 T. Blodau, A. Janzen, K. Neumann: “Anemometer Calibration Variability”. DEWEK 2012.
 O. Frost Hansen, S.O. Hansen and L. Kristensen: “Wind tunnel calibration of cup anemometers“. AWEA
WindPower 2012.
 S. Clark: “SOH Wind Engineering Qualification, Calibration, and Accreditation”, Renewable NRG Systems.
8
DNV GL ©
2 June 2015
Wind Tunnel Calibrations – Working Group Round Robin
 Working Group formed to investigate
variability of calibrations seen at
different MEASNET facilities
 All common anemometer types and
MEASNET facilities included
Working Group Members
AWS Truewind
DNV GL
NREL
Pattern Energy
Renewable NRG Systems
 1 or 2 anemometers of each type sent
to each facility and calibrated
Siemens
SOH Wind Engineering
WSP
 Finally, each instrument is returned to
the first facility and recalibrated
9
DNV GL ©
2 June 2015
Wind Tunnel Calibrations – Results
2%
Deviation relative to average at 8 m/s
Each anemometer’s results are normalized across all facilities
1%
Facility A
Facility C
Facility D
-1%
-2%
10
Facility B
0%
DNV GL ©
Anem 1
2 June 2015
Anem 2
Anem 3
Anem 4
Anem 5
Anem 6
Deviation relative to average [%]
Wind Tunnel Calibrations – Results
1.50%
Anem 1
1.00%
0.50%
Facility A
Facility B
0.00%
Facility C
Facility D
-0.50%
Facility A
-1.00%
-1.50%
 Slope and offset variations
between tunnels cause
wind speed-dependent
variation in results
Consensus
0
4
8
12
16
20
 Deviations can exceed 1%
threshold at different wind
speeds
Deviation relative to average [%]
Wind Speed [m/s]
1.50%
Anem 5
1.00%
0.50%
Facility A
Facility B
0.00%
Facility C
Facility D
-0.50%
Facility A
-1.00%
-1.50%
Consensus
0
4
8
12
Wind Speed [m/s]
11
 Repeatability of test
results may also be
anemometer-dependent
DNV GL ©
2 June 2015
16
20
 Caution required due to
small sample size
Wind Tunnel Calibrations – Results
 Variation in calibration results between
wind tunnels can approach or exceed
the 1% threshold
 Impact on measurements is wind
speed dependent and therefore sitespecific
 No strong bias towards any particular
anemometer or facility
 Analysis is limited by small sample size
and single snapshot of facility
performance – time variance of
calibration results not captured
12
DNV GL ©
2 June 2015
Why Does This Matter?
13
DNV GL ©
2 June 2015
Summary
 A variety of factors influence the wind speeds recorded by anemometers in the
field
– Response to atmospheric conditions
– Calibration
 Wind tunnel calibrations can vary by +/- 1% for a given anemometer type
– Variation seems to be dependent on anemometer model
– Unclear impact of wind tunnel size and design
 Pre-construction assessments and power performance tests impacted by
anemometer type and calibration facility
– Impact on P50 can be up to 3-4%, meaning €Millions in NPV of projects
– AEP can vary by ~2% for power curve tests
 Reduced bias risk with mixed anemometer models and calibration facilities
14
DNV GL ©
2 June 2015
Thank you
Carl Ostridge
[email protected]
Taylor Geer
[email protected]
www.dnvgl.com
SAFER, SMARTER, GREENER
15
DNV GL ©
2 June 2015