Application of LiDARs in

Annual Energy Production

Assessment of Wind Turbines

Andrew Marina
Masoud Asgarpour

August 2016
ECN-E--16-049
Acknowledgement

This report is written within “Lidar Application for WINdfarm Efficiency” or LAWINE
project partially funded by Dutch government under TKI Wind op Zee framework. The
LAWINE project is coordinated by ECN, in cooperation with Delft University of
Technology, Avent Lidar Technology and XEMC Darwind. This body of work is carried
out under subtask A of the LAWINE project (internal project number 52241)

‘Although the information contained in this report is derived from reliable sources and reasonable care
has been taken in the compiling of this report, ECN cannot be held responsible by the user for any
errors, inaccuracies and/or omissions contained therein, regardless of the cause, nor can ECN be held
responsible for any damages that may result therefrom. Any use that is made of the information
contained in this report and decisions made by the user on the basis of this information are for the
account and risk of the user. In no event shall ECN, its managers, directors and/or employees have any
liability for indirect, non-material or consequential damages, including loss of profit or revenue and loss
of contracts or orders.’
 Contents
Summary 4

1 Introduction 5

2 Wind Resource Assessment 6

2.1 EWTW Test Site 6
2.2 EWTW MM3 8
2.3 EWTW LiDARs 13
2.4 Wind Resource Selection 15
2.5 Filtering of Data 15

3 Wind Power Assessment 19

3.1 Turbine Data 19
3.2 Methodology 20
3.3 Estimated Production using Hub Height Wind Speed 23
3.4 Estimated Production using REWS 28

4 Discussion & Conclusion 34

References 36

ECN-E--16-0 3
 Summary
The first work package of the LAWINE project is allocated to the application of LiDARs in
wind resource assessments. In this report, energy production of a prototype 2.5 MW
wind turbine at “ECN Wind turbine Test site Wieringermeer” or EWTW are estimated
using a meteorological mast, as well as a ground-based LiDAR and these are compared
to the actual power production of the turbine.

The comparison of predicted and actual energy yield utilising hub height wind speed
shows that using a meteorological mast, the energy yield prediction was 2.9% lower
than the actual production. The ground based LiDAR showed a smaller deficit, being
0.7% lower than the actual energy yield. There was evidence in the results to suggest
that the difference in the values is not due to limitations in either device, however it is
more likely due to a small offset (under prediction) of the local wind speed from the
meteorological mast measurement device.

The main benefits of ground based LiDAR devices are their better economics as well as
the ability to measure easily at multiple heights covering the rotor area. Part of this
report was also dedicated to investigating if the calculation of the rotor equivalent wind
speed (REWS) resulted in better prediction of the energy yield compared to the
calculations using the hub height wind speed. The analysis using the rotor equivalent
wind speed at this site was inconclusive and made it difficult to draw conclusions.
Further work should be completed utilising a larger data set and more turbines with
REWS verified power curves in order to gain stronger conclusions on this subject.

4
 1
Introduction
Uncertainty in annual energy production (AEP) forecast is one of the main barriers in
wind energy financial decisions. The MEASNET “evaluation of site-specific wind
conditions” [1] outlines the requirements for a bankable AEP report.

Based on this guideline there should be minimum one year on-site measurement data
and preferable 10 to 30 years long-term measurement data for a bankable AEP report.
The measurement height should be minimum 2/3 of the hub height of the turbines and
the meteorological mast should be maximum 10 km far for simple terrains and
maximum 2 km far for complex terrains. Due to the high installation, calibration and
maintenance costs of meteorological masts, on and offshore, the MEASNET
requirements are not always fulfilled.

Moreover, rotor diameters of state of the art wind turbines are greater than 100 meters
which means that considering only wind speed at the hub height in AEP forecasts no
longer provides a sufficiently accurate estimate. Therefore, measurements at different
heights along the rotor are required (e.g. from 100 to 250 meters height). It is not
always possible to have such a tall meteorological mast and even if possible, it directly
increases the costs. In addition to this, proper AEP forecast models are required to
make use of all available measurement heights.

A potential solution to these challenges is LiDARs. LiDARs can be easily placed in any
location (ground based and nacelle based for onshore wind farms and floating LiDARs
for offshore wind farms) and they can provide accurate measurements in several
heights, without any additional costs [2].

In this report a case study is presented, where power production of a prototype 2.5 MW
turbine located at the EWTW test site is estimated using a ground based LiDAR as well
as a meteorological mast measurement instrument. Afterwards, the estimated results
are compared to the actual wind turbine production and conclusions are drawn as to
their usage in providing accurate wind resource assessments.

ECN-E--16-0 Introduction 5
 2
Wind Resource Assessment

2.1 EWTW Test Site

At the end of 2002, the Energy research Centre of the Netherlands (ECN) initiated a
wind turbine test site in Wieringermeer located on the north-east of the province
Noord-Holland. The test site is named EWTW which stands for ECN Wind turbine Test
site Wieringermeer. The EWTW is located about 60 km North of Amsterdam and 35 km
east of the ECN headquarters in Petten. The EWTW is located in the vicinity of the
artificial lake, the Ijsselmeer [3].

The EWTW is equipped with state of the art infrastructure for site assessment and wind
turbine testing and certification. Currently the test site consists of:

 Five research turbines in locations WT5 to WT9 of Figure 1. The research turbines
are used for experimental research conducted by ECN wind unit and other European
research institutes.
 Six prototype locations in locations WT11, WT10 and WT1 to WT4 of Figure 1. The
wind turbine manufactures use prototype locations for testing and certification of
their new wind turbines.
 Five meteorological masts (hereafter referred to as MET mast or MM) MM1 to MM5
in Figure 1.
 10kV cables to connect turbines to a 50/100 kV transformer station and then, direct
connection to a high voltage substation.
 Glass fibre network to transmit the measurement data.
 The EWTW office as measurement pavilion of the site.

6
Figure 1: The EWTW infrastructure: research turbines in green, prototype locations in red, neighbour
turbines in orange and meteorological masts in purple [Google]

ECN-E--16-0 7
 2.2 EWTW MM3

Measurements taken from MM3 will form the basis of the work conducted in this
report. This section is intended to give a short description of the MET mast and the
wind measurements taken from the mast. On MM3, meteorological instrumentations
are installed on booms at two heights and on the top of the mast. In Figure 2 an
overview of the MM3 instrumentation is illustrated. More details are given in [3].

Figure 2: Overview of the MM3 instrumentations [3]

109.1m: Gill sonic anemometer

80m: Gill sonic anemometer
80 m: two Risø anemometer
79.2m: two Thies wind vanes
78.4m: Vaisala air temperature,
pressure and relative humidity

52m: Gill sonic anemometer

52 m: two Risø anemometer
51.2m: two Thies wind vanes

Air temperature difference

(10m-37m)

The complete list of sensors installed on the MM3 is given in Table 1. The sensors
mounted at different height level are separated with a black line.

Table 1: Sensor list of the MM3 [3]

Sensor
Signal Signal Name Unit Height Brand
Type
(m)

Wind speed 108 m MM3_S108 m/s 109.1 Gill 1086 M

Wind direction 108 m MM3_S108_VDIR deg 109.1 Gill 1086 M
Wind speed 80 m MM3_S80N m/s 80 Gill 1086 M
Wind direction 80 m MM3_S80N_VDIR deg 80 Gill 1086 M
Wind speed SE 80 m MM3_WS80_120 m/s 80 Risø P2456A
Wind speed SW 80 m MM3_WS80_240 m/s 80 Risø P2456A

8
Wind direction SE 80 m MM3_WD80_120 deg 79.2 Thies 4.3150
Wind direction SW 80 m MM3_WD80_240 deg 79.2 Thies 4.3150
Air temperature 80 m MM3_Tair80 °C 78.4 Vaisala HMP 155
Air humidity 80 m MM3_RH80 % 78.4 Vaisala HMP45A
Air pressure 80 m MM3_Pair80 hPa 78.4 Vaisala PTB 210
Wind speed 52 m MM3_S52N m/s 52 Gill 1086 M
Wind direction 52 m MM3_S52N_VDIR deg 52 Gill 1086 M
Wind speed SE 52 m MM3_WS52_120 m/s 52 Risø P2456A
Wind speed SW 52 m MM3_WS52_240 m/s 52 Risø P2456A
Wind direction SE 52 m MM3_WD52_120 deg 51.2 Thies 4.3150
Wind direction SW 52 m MM3_WD52_240 deg 51.2 Thies 4.3150
Temperature difference MM3_dT_37min10 °C -- Rense --

In Figure 3 yearly average of wake-free wind speed measurements on MM1 and MM3
for the period of 2005 to 2014 are depicted.

Figure 3: Wake-free wind speed measurement on MM1 and MM3 for the period 2005 to 2014

ECN-E--16-0 9
 In Figure 4 and Figure 5 monthly average, probability distribution and fitted Weibull
distribution wake-free wind speed measurement on MM3 at 80 m height for the period
2005 to 2014 are shown.

Figure 4: Monthly average wake-free wind speed measurement on MM3 at 80 m height for the period
2005 to 2014

Figure 5: Probability distribution and fitted Weibull distribution of wake-free wind speed measurement
on MM3 at 80 m height for the period 2005 to 2014

Fitted Weibull parameters:

λ = 8.50901 [8.49714, 8.5209]
k = 2.26463 [2.25953, 2.26974]

10
In Figure 6 and Figure 7 monthly average and probability distribution of wake-free
turbulence intensity on MM3 at 80 m height for the period 2005 to 2014 are illustrated.

Figure 6: Monthly average wake-free turbulence intensity on MM3 at 80 m height for the period 2005
to 2014

Figure 7: Probability distribution of wake-free turbulence intensity on MM3 at 80 m height for the
period 2005 to 2014

ECN-E--16-0 11
 In Figure 8 and Figure 9 monthly average and probability distribution of wake-free flow
inclination on MM3 at 80 m height for the period 2005 to 2014 are illustrated.

Figure 8: Wake-free flow inclination measurement on MM3 per wind speed bins for the period 2005 to
2014

Figure 9: Probability distribution of wake-free flow inclination on MM3 for the period 2005 to 2014

12
2.3 EWTW LiDARs

Within the LAWINE project and the Norwegian Centre for Offshore Wind Energy
(NORCOWE) campaign, several ground-based and nacelle-based LiDARs were installed
at the EWTW test site. In [4], [5] and [6] an overview of the instrumentation, signals and
availability of these LiDARs is given. In Figure 10 the approximate location of these
LiDARs at the test site is illustrated.

Figure 10: Available LiDARs within the LAWINE project and the NORCOWE campaign

LiDARs within the LAWINE project and the NORCOWE campaign located at the EWTW
test site are:
 Ground-based LiDARs
o Wind Cube (WLS-100S-2) – in the wake of research turbines in 12.2D distance
o Wind Cube V1 (WLS7-065) – in the wake of research turbines in 3.5D distance
o Wind Cube V1 (WLS7-037) – in the wake of research turbines in 1.75D distance
o Wind Cube V1 (WLS7-067) – located next to the MM3
o Wind Cube V2 (WLS7-127) – located next to the MM3
o Wind Cube V2 (WLS-258) – located next to the MM3
o Zephir 300 (308) – located next to the MM4

 Nacelle-based
o Avent Wind IRIS – located on the prototype WT6 turbine, forward looking
o Avent Prototype 5 beam – located on WT10, forward looking
o Zephir 300 (328) – located on the prototype WT6 turbine, backward looking

In [4] the instrumentation of these LiDARs and their signals are discussed in detail. In
[5], the daily availability of measurements of these LiDARs is given. In [6] a public

ECN-E--16-0 13
 summary of the LAWINE LiDAR campaign is given. In Figure 11 and Figure 12, a
summary of availability of these LiDARs is given.

Figure 11: Availability of LAWINE/NORCOWE LiDARs at EWTW 2012/2013 [4]

2012/2013 2012-11 2012-12 2013-01 2013-02 2013-03 2013-04 2013-05 2013-06 2013-07 2013-08 2013-09 2013-10 2013-11 2013-12
mm31 20-nov
MM4 13-sep
2 2
WT1-WT5
Nordex 3-mei
XEMC
WT10 Darwind 12-sep
WLS7-127 17-mrt 15-sep
WLS7-2584 20-nov
Zephir 3085 12-sep 10-nov
WindIris (nac) 6 1-jun
WindIris REV (nac)
5-beam demonstrator7 12-sep
WLS100s-28 31-okt
WLS7-0379 16-nov
WLS7-0659 16-nov
WLS7-0679 1-nov
Zephir 328 (nac) 14-dec

Figure 12: Availability of LAWINE/NORCOWE LiDARs at EWTW 2014 [4]

2014 2014-01 2014-02 2014-03 2014-04 2014-05 2014-06 2014-07 2014-08 2014-09 2014-10 2014-11 2014-12
MM31,3
MM4 3

Nordex
WT1-WT522

XEMC
WT10 Darwind 31-mei
WLS7-127
WLS7-258 30-jan
Zephir 308
WindIris (nac) 25-apr
WindIris REV (nac) 26-jun 27-nov
5-beam demonstrator 31-mei
WLS100s-2 11-mei
WLS7-037 21-jul
WLS7-065 21-jul
WLS7-067 16-jun
Zephir 328 (nac) 26-nov
1
additional precipitation signals of other meteorological masts at the EWTW are added from 03-05-2013
2
from 25-02-2014 the main shaft measurement has been removed because of replacement of the main
bearings of the turbine and not reinstalled because of low priority
3
the prototype turbines measurements continued in 2014 but with lower priority, validated and data
acquisition in calibration, but no calibration checks on the loads signals.
4
height configuration changed at 15-08-2013
5
height configuration changed at 04-10-2013
6
Distance configuration changed at 05-06-2013 & 17-06-2013, scan rate changed at 25-06-2013
7
height configuration changed at 26-09-2013
8
relocated at 10-12-2013, after relocation the settings changed and as result the scan rate changed as well
9
relocated at 29-11-2013 and height configuration changed at 10-12-2013

Meteorological measurements
Turbine measurements
ECN ground based LiDAR measurements
ECN ground based LiDAR measurements, before change of configuration
Avent nacelle LiDAR measurements
Avent nacelle LiDAR measurements before change of configuration
Norcowe LiDAR measurements
Norcowe LiDAR measurements before change of configuration

14
2.4 Wind Resource Selection

In order to demonstrate the application of LiDARs in AEP estimation, in this case study
energy yield of a 2.5 MW prototype wind turbine is estimated using:
o MET mast 3 (MM3)
o Wind Cube V2 (WC258)

The Wind Cube V2 is located in the vicinity of MM3, the location of which is outlined
previously in Figure 3.

The measurement data of WT6 and MM3 are available from 2005 up to now. However,
WC258 data is only available for a limited time period. Based on the information given
in Figure 11 and Figure 12, it can be seen that the period whereby measurement data
from the ground based LiDAR is available is in the period 20/11/2012 - 30/01/2014 and
as such, this will be used of the analysis and comparison of ground based LiDAR to the
MM3 data.

It is noted that the remaining LiDARs can’t be used in this study, since they are not
located in the vicinity of the WT6 wind turbine, they are in the wake of turbines, or they
are not ground based types. Furthermore, other turbines cannot not be investigated
due to lack of available data, running in reduced power mode (especially the case for
WT5 in the period in question), or large vicinity from the measurement devices.

2.5 Filtering of Data

In order to ensure the quality of measured data used in this study, a filtering process is
undertaken. This filtering concerns availability, quality, turbine operation modes and
wake-free status of measured data. This analysis of ground based LiDAR for energy yield
assessments is a two part study. The first part of the study compares the power
estimations using data at a single height from both MM3 and WC258 (the hub height,
80 m). The second part of the study involves the calculation of a rotor equivalent wind
speed (REWS) whereby measurements at multiple heights are used to estimate the
turbine power and energy production. Both parts of the study require separate filtering
of the data as outlined below.

Filtering of the data is conducted based on four primary criteria:

1. Availability of the LiDAR measurement
2. Quality of the recorded data
3. Power production of the turbine
4. Wake disturbance of the flow

For filtering the sectors which may be disturbed by wake effects, an overview of wind
farm was investigated as seen in Figure 13. The disturbed wind sectors around WT6
and MM3 can be seen from this overview of the wind farm and are accounted for in the
filtering process. In order to ensure as many data points as possible for the analysis, the

ECN-E--16-0 15
 prototype turbines at the southern side of the wind farm are negated during filtering.
Although it would ideally be better to include only undisturbed wind sectors, this is
necessary in order to have enough data points for the analysis. Therefore, in this study
some of the prototype turbines, although relatively far away, may still disturb the
measurements.

Figure 13: Disturbed wind sectors around WT6 and MM3

2.5.1 Hub Height Wind Speed Study Filtering

The filtering criteria for this part of the study are as follows:
 Availability
o Mm3_WC258_D80_Avail_Q5 > 90 (%)
 Quality (only wind speed in [0,100] interval and only wind direction in [0-360]
interval)
o MM3_H80_Ws_Q1_avg > 0 & MM3_H80_Ws_Q1_avg < 100
o MM3_H80_Wd_Q1_avg ≥ 0 & MM3_H80_Wd_Q1_avg ≤ 360
o Mm3_WC258_D80_Ws_Q5 > 0 & Mm3_WC258_D80_Ws_Q5 < 100
o Mm3_WC258_D80_Wd_Q5 ≥ 0 & Mm3_WC258_D80_Wd_Q5 ≤ 360
 Power production (only normal power production and no reduced-power mode)
o Wt06_Ct_Popmode_Q5_avg>10.9 & Wt06_Ct_Popmode_Q5_avg<12.1
Wt06_Ct_Pred_pow_Q5_avg>-0.1 & Wt06_Ct_Pred_pow_Q5_avg<80
 Wake-free (no data be used when the measurement device is in the wake of
operating turbines around)
o Not 244.6° - 345.9° sector, only when WT5 is operating

16
  Wt05_Ct_Pgspd_Q5_avg > 1 || Wt05_Ct_Pgspd_Q5_avg <-1
o Not 353.7° - 67.3° sector, for the wake of WT6 turbine
o Not 44.9° - 125.8° sector, only when WT7 is operating
 Wt07_Ct_Pgspd_Q5_avg > 1 || Wt07_Ct_Pgspd_Q5_avg <-1
o 174.2° - 210.9° sector, for the wake of the neighbour turbine located in middle of
EWTW (operational data of this turbine is not available)

2.5.2 REWS Study Filtering

The filtering criteria for this part of the study are as follows:
 Availability
o Mm3_WC258_D80_Avail_Q5 > 90 (%)
o Mm3_WC258_D40_Avail_Q5 > 90 (%)
o Mm3_WC258_D52_Avail_Q5 > 90 (%)
o Mm3_WC258_D100_Avail_Q5 > 90 (%)
o Mm3_WC258_D108_Avail_Q5 > 90 (%)
o Mm3_WC258_D140_Avail_Q5 > 90 (%)
 Quality (only wind speed in [0,100] interval and only wind direction in [0-360]
interval)
o MM3_H80_Ws_Q1_avg > 0 & MM3_H80_Ws_Q1_avg < 100
o MM3_H80_Wd_Q1_avg ≥ 0 & MM3_H80_Wd_Q1_avg ≤ 360
o MM3_H52_Ws_Q1_avg > 0 & MM3_H52_Ws_Q1_avg < 100
o MM3_H52_Wd_Q1_avg ≥ 0 & MM3_H52_Wd_Q1_avg ≤ 360
o MM3_H108_Ws_Q1_avg > 0 & MM3_H108_Ws_Q1_avg < 100
o MM3_H108_Wd_Q1_avg ≥ 0 & MM3_H108_Wd_Q1_avg ≤ 360

o Mm3_WC258_D80_Ws_Q5 > 0 & Mm3_WC258_D80_Ws_Q5 < 100

o Mm3_WC258_D80_Wd_Q5 ≥ 0 & Mm3_WC258_D80_Wd_Q5 ≤ 360
o Mm3_WC258_D40_Ws_Q5 > 0 & Mm3_WC258_D40_Ws_Q5 < 100
o Mm3_WC258_D40_Wd_Q5 ≥ 0 & Mm3_WC258_D40_Wd_Q5 ≤ 360
o Mm3_WC258_D52_Ws_Q5 > 0 & Mm3_WC258_D52_Ws_Q5 < 100
o Mm3_WC258_D52_Wd_Q5 ≥ 0 & Mm3_WC258_D52_Wd_Q5 ≤ 360
o Mm3_WC258_D100_Ws_Q5 > 0 & Mm3_WC258_D100_Ws_Q5 < 100
o Mm3_WC258_D100_Wd_Q5 ≥ 0 & Mm3_WC258_D100_Wd_Q5 ≤ 360
o Mm3_WC258_D108_Ws_Q5 > 0 & Mm3_WC258_D108_Ws_Q5 < 100
o Mm3_WC258_D108_Wd_Q5 ≥ 0 & Mm3_WC258_D108_Wd_Q5 ≤ 360
o Mm3_WC258_D140_Ws_Q5 > 0 & Mm3_WC258_D140_Ws_Q5 < 100
o Mm3_WC258_D140_Wd_Q5 ≥ 0 & Mm3_WC258_D140_Wd_Q5 ≤ 360

 Power production (only normal power production and no reduced-power mode)

o Wt06_Ct_Popmode_Q5_avg>10.9 & Wt06_Ct_Popmode_Q5_avg<12.1
Wt06_Ct_Pred_pow_Q5_avg>-0.1 & Wt06_Ct_Pred_pow_Q5_avg<80
 Wake-free (no data be used when the measurement device is in the wake of
operating turbines around)
o Not 244.6° - 345.9° sector, only when WT5 is operating
 Wt05_Ct_Pgspd_Q5_avg > 1 || Wt05_Ct_Pgspd_Q5_avg <-1
o Not 353.7° - 67.3° sector, for the wake of WT6 turbine

ECN-E--16-0 17
 o Not 44.9° - 125.8° sector, only when WT7 is operating
 Wt07_Ct_Pgspd_Q5_avg > 1 || Wt07_Ct_Pgspd_Q5_avg <-1
o 174.2° - 210.9° sector, for the wake of the neighbour turbine located in middle of
EWTW (operational data of this turbine is not available)

18
 3
Wind Power Assessment
In this two-part study, the power production from the prototype turbine WT6 located
on the EWTW test farm is predicted utilising wind speed data from a MET mast and
ground based LiDAR located a short distance from the turbine. In the first part, the
power production is estimated using the hub height wind speed, and in the second part,
utilising the rotor equivalent wind speed (REWS). The methodology and results are
described in detail.

3.1 Turbine Data

In order to calculate the expected power production the turbine, power output of the
prototype 2.5 MW turbine is required through having a validated power curve. In this
study the verified turbine data, which is derived from measurements performed in [7],
is used. This data are given in Table 2.

Table 2: Power and thrust data for the prototype 2.5 MW turbines at EWTW site [7].

Wind speed Power Thrust coefficient Power coefficient Rotor speed

[m/s] [kW] [-] [-] [rpm]
4 26 0.8447 0.132 11.5
5 117 0.7982 0.304 12.1
6 251 0.7956 0.377 13.0
7 433 0.7939 0.410 14.1
8 669 0.7888 0.424 15.4
9 948 0.7438 0.422 16.5
10 1267 0.6878 0.412 16.4
11 1618 0.6512 0.395 16.8
12 1956 0.5766 0.368 17.0
13 2228 0.5083 0.329 17.0

ECN-E--16-0 19
 14 2398 0.4239 0.284 17.0
15 2472 0.3297 0.238 17.0
16 2494 0.2654 0.198 17.0
17 2499 0.2184 0.165 17.0
18 2500 0.1828 0.139 17.0
19 2500 0.1550 0.118 17.0
20 2500 0.1330 0.102 17.0
21 2500 0.1152 0.088 17.0
22 2500 0.1007 0.076 17.0
23 2500 0.0886 0.067 17.0
24 2500 0.0786 0.059 17.0
25 2500 0.0701 0.052 17.0

3.2 Methodology

3.2.1 Hub Height Wind Speed

To predict the power production of turbine WT6, the wind speed as given by MM3, as
well as WC258 was used as an input into the power curve given in Table 2. As the
filtered signals are 10 minute averages, the power can be multiplied by a value of (1/6)
to give the energy in kWh. These are summated over the entirety of the data set to give
the estimated energy yield of the turbine. The power output of the turbine is available
in the signal data at all time periods. For the data points in question, the power output
value is multiplied by 1/6 and then these are summated to give the actual deviation of
in the energy yield between the actual and the estimated values.

3.2.2 Rotor Equivalent Wind Speed

For the second part of this study, the rotor equivalent wind speed is utilised to estimate
the energy yield. The same method outlined above is utilised, except the REWS is used
as opposed to the hub height wind speed. Ideally the power curve should be validated
utilising REWS, however in the present study, this has been only conducted utilising the
hub height wind speed power curve.

In [8] this method is described and a formula for calculating the Rotor Equivalent Wind
Speed is presented:

20
There are two formulas presented above for the hub height wind speed: the integral
equation on the left, and the summation on the right side. We will refer to these from
now on as the continuous and discreet methodologies respectively.

The continuous methodology is utilised if the wind speed profile as a function of the
height is known, whereas the discreet method is used if there are wind speeds at
discreet heights within the rotor area. The discreet methodology is depicted in Figure
14. For this study, despite only having the wind speeds at discreet heights, both
methods will be utilised. In order to create a wind speed profile as a function of height,
th
the discreet points are interpolated utilising Matlab with the ‘pchip’ method and a 7
order polynomial then fitted the data to give a close fit with the points at all heights. It
should be noted that the continuous methodology is not described in the draft standard
FDIS IEC 61400-12-1, ‘Power Performance Testing’ and as such is limited to a research
tool at this time.

Figure 14: Concept of Rotor Equivalent Wind Speed method [8].

It should also be noted that within this study, there are far greater measurement
heights available for the LiDAR device than are available for MM3. The heights available
for MM3 are as follows and depicted in Figure 15.

For the period 20/11/2012 to 15/08/2013, the heights measured by the WC258
relevant to this study are (Period 1):
 40 m, 52 m, 80 m, 100 m, and 108 m

From the period 15/08/2013 to 20/01/2014, the measured heights by the WC 258 are
(Period 2):
 40 m, 52 m, 60 m, 80 m, 100 m, and 108 m

Therefore, we need to calculate different REWS for different periods of the data.

For Period 1, we need to use the following segments and associated areas for
calculation of the REWS:
 Segment 1 – 40 m – 40 to 46 m
 Segment 2 – 52 m – 46 to 66 m
 Segment 3 – 80 m – 66 to 90 m

ECN-E--16-0 21
  Segment 4 – 100 m – 90 to 104 m
 Segment 5 – 108 m – 104 to 120 m

For Period 2, we need to use the following segments and associated areas for
calculation of the REWS:
 Segment 1 – 40 m – 40 to 46 m
 Segment 2 – 52 m – 46 to 56 m
 Segment 3 – 80 m – 56 to 70 m
 Segment 4 – 80 m – 70to 90 m
 Segment 5 – 100 m – 90 to 104 m
 Segment 6 – 108 m – 104 to 120 m

For calculating the continuous REWS, the wind speed at the height of 140 m is also used
such that extrapolation of the curve is not necessary

The heights for which the areas are calculated to be used in the discreet REWS
calculation when using MM3 are also indicated:
 Segment 1 - 52 m – 40 to 66m
 Segment 2 - 80 m – 66 to 94m
 Segment 3 – 108 m – 94 to 120 m

Figure 15: Rotor Equivalent Wind Speed method based on Prototype 6 turbine and Wind Cube V2 data

22
3.3 Estimated Production using Hub Height

Wind Speed

To predict the power production of turbine WT6, the wind speed as given by MM3, as
well as WC258 was used as an input into the power curve given in Table 2. As the
filtered signals are 10 minute averages, the power can be multiplied by a value of (1/6)
to give the energy in kWh. These are summated over the entirety of the data set to give
the power of the turbine for the periods which pass the filtering process.

Prior to the analysis, the wind speed ratio defined below was tested to determine any
further sectors which may need to be filtered if the value deviated from unity.

It was determined that there were irregularities in the sector 305-330 degrees and as
such, this sector was also filtered from the result. These irregularities are likely an
effect induced due to the close proximity of the WT5 structure a short distance
upstream which is effecting the local wind speed measurements despite not rotating at
that time. The wind speed ratio of the filtered data plotted against the direction sector
can be seen in Figure 16 which follows. In total, there were 8845 data points,
equivalent to 1475 hours or 61 days for prediction of the energy yield of the turbine.
Additionally, Figure 17 shows a comparison of the wind speed measurements from each
source. It can be seen that the slope is equal to unity. There is very little scatter in the
data points, however there are a few outliers which may contribute to variances,
although these are minimal. The small y axis offset in the best fit line may suggest a
small offset or calibration error and higher measured wind speeds from the ground
based LiDAR (WC258) compared to the MM3 measurement. This will be further
investigated as part of the analysis.

ECN-E--16-0 23
 Figure 16: Wind Speed Ratio of sectors used for study

Figure 17: Comparison of wind speed measurements

24
Figure 18: Power curve prediction using MM3 (left) and WC258 (right)

Figure 19: Wind Speed deviation (<2300 kW)

Figure 20: Energy yield deviation, MM3 vs Wind Cube WC258

ECN-E--16-0 Wind Power Assessment 25

 For the comparison of the wind speed measurement methodologies for predicting
energy yield, the wind speed of both devices was plotted against the actual power
production of the WT6 turbine. The verified power curve was overlaid on these data
sets. The results of this can be seen in Figure 18. It is generally difficult to make
conclusions on these figures although in general it is noted that both measurement
methodologies are able to quite accurately reproduce the power curve. In general
there appears to be slightly more scatter or outliers in the MM3 measurement set.

To investigate the deviations in the expected wind speed against the measured wind
speed, histograms were compiled for the MM3 and WC258 devices. The expected wind
speed was calculated by using the inverse of the power curve and using the power
output measured from the turbine. Obviously, as the power curve reaches rated
power, the power is constant over a number of wind speeds. As such this method
cannot be used for all data points. Therefore, only data where the power was less than
2300 kW was used in compiling these figures.

It appears as though deviations in wind speed predictions are in general lower for the
MM3 measurements. That is, the peak of the curve as seen in the left side of Figure 19
is higher than that of the right side. Despite this, the tails of the data generated from
the MM3 deviations also appear a bit longer, which coincides to what is expected based
on the observations in Figure 18. Also, the centre of the histogram appears to be
slightly less than zero for the MM3 measurements which may indicate a small under
prediction of the actual wind speeds.

The primary goal of the study is to determine the applicability of LiDAR as a

measurement instrument for energy yield when compared to the conventional MET
mast measurement instrument. The method for predicting the energy yield has been
described previously. The deviation of the actual energy yield compared to the
predicted using the LiDAR and MM3 is illustrated in Figure 20. The deviation is less for
the LiDAR (10.7 MWhr, 0.7%) compared to the MM3 anemometer (44.3 MWhr, 2.9%)
and in both cases, the energy yield was under predicted. For the case of the WC258,
this is equivalent to a constant deviation of 7.25 kW (0.29% of P rated) over the data set.
For the MM3 anemometer this is slightly higher at 30.06 kW (1.20% of Prated), however
still a small overall error.

Finally, in Figure 21, the progression of the energy yield deficit is investigated to
determine its behaviour as the number of data points increases, and also if the values
using MM3 as well as the WC258 eventually converge to a certain value. Obviously
relative values are only investigated to determine convergence.

From Figure 21, there are two primary observations that can be made in relation to the
data. The first of these is that the relative difference in the expected vs actual energy
yield has not fully converged. It is reasonable to say however that after approximately
4000 data points, the relative error is less than 1% for both measurement devices. The
second observation is that the difference in the energy yield for both MM3 and the
Wind Cube WC258 remains relatively constant from approximately 1500 data points.
The trends in the relative error are observed for the energy yield prediction using both
devices. This indicates that there is likely a positive offset error in the Wind Cube device
(compared to MM3) which leads to a higher prediction in the energy yield compared to

26
the MM3 device. This theory is further supported by Figure 17 which showed this
offset in the MM3 device.

Figure 21: Progression of the energy yield deficit

ECN-E--16-0 27
 3.4 Estimated Production using REWS

The second part of this study involves redoing the analysis, except this time utilising the
REWS as opposed to the hub height wind speed. One of the quoted benefits of LiDAR
systems is their ability to accurately measure at a number of heights extending the
whole rotor area. This method is examined to determine if there is a significant benefit
in utilising the REWS as opposed to the hub height wind speed.

For the calculation, six wind speeds are calculated and compared. There are as follows:

1. HHWS – MM3
2. HHWS – WC258
3. Discreet REWS – MM3
4. Discreet REWS – WC258
5. Continuous REWS – MM3
6. Continuous REWS – WC258

The results obtained from each method are described in the work which follows. From
the filtered data there are 8514 data points equivalent to approximately 59 days. The
hub height wind speed comparison is included in some calculations for reference.

Once again, the wind speed of the two devices are compared. This was again done at
80 m height as seen in Figure 22. For the new filtered data set, the comparison was
also compared at 108 m as seen in Figure 23. The comparison at 108 m is chosen as a
sonic anemometer is used as opposed to a cup anemometer as used at the other
measurement heights. Devices show good correlation at both measurement heights,
with once again there appearing to be a slight over prediction of the wind speed in the
case of the WC258 compared to the MM3 equivalent measurement. The good
correlation suggests there should be no issue in the prediction of REWS using the MM3
devices.

28
Figure 22: Wind Speed comparison – 80 m

Figure 23: Wind speed comparison – 108 m

The calculated wind speed based on the measurement data is plotted against the
measured power output as is illustrated in Figure 24. All methods appear to follow the
verified power curve with small variances between each wind speed calculation
method. The MM3 measurements appear once again to contain more extreme outliers.

ECN-E--16-0 29
 Figure 24: Power curve prediction using MM3 (top) and WC258 (bottom) for discreet (left) and continuous (right) methods

30
Figure 25: Wind Speed deviation (<2300 kW) MM3(top) and WC258 (bottom) using discreet REWS (left) and continuous REWS
(right) methods

Figure 24 shows the deviation in the expected wind speed based on the inverse of the
power curve for power output less than 2300 kW. As was seen with the hub height
wind speed measurements, the centre of the MM3 anemometer REWS values are
slightly less than zero, again suggesting the power will be slightly under predicted.

The deficit in the actual energy yield vs the expected energy yield based on the wind
speeds measured or calculated, combined with the verified power curve is given in
Figure 26. Immediately evident is the fact that the hub height wind speed using the
WC258 gives the most accurate representation of the energy yield. The relative error is
0.7%. The least accurate is the MM3 hub height wind speed which gives a relative error
of 2.8%. For the MM3 measurement, the continuous REWS calculations showed better
correlation with energy yield than the discreet REWS ones, however this was not the
case with the WC258 results. This makes it difficult to draw conclusions on the accuracy
of continuous and discreet calculation of REWS. It is noted that small errors in the
verified power curve could lead to the offset seen below. The average power deviation
from left to right on Figure 26 is:

ECN-E--16-0 31
  MM3 hub height wind speed – 29.0 kW (1.2% of Prated)
 WC258 hub height wind speed – 7.4 kW (0.3% of Prated)
 MM3 discreet REWS – 26.2 kW (1.0% of Prated)
 WC258 discreet REWS – 15.1 kW (0.6% of Prated)
 MM3 continuous REWS – 23.9 kW (1.0% of Prated)
 WC258 continuous REWS – 14.1 kW (0.6% of Prated)

From this it is evident that small deviations in the verified power curve, or small
errors in the wind speed can lead to the quite significant differences in the energy
yield predictions. It is perhaps logical to revisit the analysis after re-verifying the
power curve using the REWS as opposed to the hub height wind speed as has been
used in this analysis.

Figure 26: Energy yield deviation absolute (top) and relative (bottom)

32
The progression of the relative error in the energy yield can be seen in Figure 27.
Similarly to the hub height wind speed results, there are two noticeable trends in the
data. All the methods of calculation for the wind speed do not seem to have converged
at the conclusion of the measurement period. The deficit in the actual energy yield to
the predicted yield appears to be increasing suggesting a longer measurement period
should be used. Furthermore, the trends in each prediction appear to be the same for
increasing data points after approximately 3000 points. This suggests that each method
gives similar trends, however gives slightly varying predictions of wind speeds.

Figure 27: Progression of relative energy yield deviation

ECN-E--16-0 33
 4
Discussion & Conclusion
Within this report, energy yield predictions were carried out utilising a MET mast and a
ground based LiDAR device at the EWTW test field at Wieringermeer. Whilst strong
conclusions could not be drawn as to the accuracy of energy yield estimations using
ground based LiDAR compared to MET masts, a number of conclusions can be drawn
over about their usage in general as well as further steps that could be taken with this
research. These main conclusions are as follows.

 In general the LiDAR and MET mast devices showed similar predictions in the
energy yield. The predictions were of high accuracy and likely within the
uncertainty range induced from the measured data and the verified power
curve.
 There was no evidence to suggest that one wind speed measurement
technique used in this study was inferior to another.
 The LiDAR device showed slightly better prediction capability in this study
when using both the hub height wind speed and the rotor equivalent wind
speed.
 The maximum relative difference in the expected and calculated energy yield
was found to be 2.9% for the hub height wind speed using MM3. A 1.2% error
in the power curve would account for this deficit.
 The minimum relative difference in the expected and calculated energy yield
was found to be 0.7% for the hub height wind speed using MM3. A 0.29%
error in the power curve would account for this deficit.
 The use of the REWS to calculate the energy yield showed better prediction
capabilities when using MM3 measurements, however the deficit was
increased when using the ground based LiDAR, making it difficult to draw
conclusions.
 When investigating the relative error progression with increasing data points, it
was evident that the difference in the expected and actual energy yield
showed the same trends for each of the measured and derived wind speed
methods. This suggests that deficits in the energy yield may be due to a small
offset in the wind speed measurement (calibration error) rather than one
measurement method having increased accuracy over another.

34
  The previous point is supported by the histograms showing the deficit between
actual wind speed and wind speed based on the inverse of the power curve
which showed a slightly left centre for measurements taken with the MET mast
 ECN’s experience with AEP predictions has shown overall estimations are
typically within an error band of approximately ±10%. The results using both
the hub height as well as the REWS for MM3 as well as the LiDAR device were
will within this typical uncertainty band indicating high quality of the
measurements as well as the validated power curve.

The following steps should be conducted in order to gain further insight into the use of
LiDAR devices for better energy yield assessments:
 A longer time period should be utilised for the study. Despite the large data
set used in this study, it was evident that the results had still not converged.
Further to this, better insight could be gained if the analysis was conducted on
more turbines, as well as an area where there were fewer disturbed wind
speed sectors.
 To give insight into the error in the power curve due to the use of the hub
height wind speed, the power curve should be verified using the REWS (both
continuous and discreet) and the analysis re-performed
 A better error analysis should be accounted for within the calculations,
indicating how uncertainties in measurements or the verified power curve
could lead to deviations in the relative and actual energy yield.

ECN-E--16-0 35
 References
[1] MEASNET Procedure: Evaluation of Site Specific Wind Conditions. Version 1.
November 2009.
[2] Boquet, M., Callard, P., Deve, N., Osler, E.G.. Return on Investment of a Lidar
Remote Sensing Device. DEWI Magazine, No. 37. August 2010
[3] Asgarpour, M; Verhoef, H. Wind Regime Analysis of ECN Wind Test Site EWTW
Period 2004-2014. ECN-X--14-178. March 2015. Petten, The Netherlands.
[4] Bergman, G; Wagenaar JW; Boorsma, K. LAWINE instrumentation report. ECN-
X--14-085. July 2015. Petten, The Netherlands.
[5] Bergman, G. Measurement overview LAWINE experiments, Completion report.
ECN-X--15-082. August 2015. Petten, The Netherlands.
[6] Asgarpour, M; Wagenaar JW. Overview of ECN/NORCOWE LIDAR campaign at
EWTW wind turbine test site. December 2015. Petten, The Netherlands.
[7] Bot, ETG. FarmFlow validation against full scale wind farms. ECN-E—15-045.
August 2015. Petten, The Netherlands.
[8] Wagner R. Equivalent Wind Speed for AEP. VindKraftNet. Risø. November
2011. Denmark

36
ECN
Westerduinweg 3 P.O. Box 1
1755 LE Petten 1755 ZG Petten
The Netherlands The Netherlands

T +31 88 515 4949

F +31 88 515 8338
info@ ecn.nl
www.ecn.nl

ECN-E--16-0 37