German Test Station for Remote Wind Sensing Devices

(1)

German Test Station for Remote Wind Sensing Devices

A. Albers, A.W. Janssen, J. Mander

Deutsche WindGuard Consulting GmbH, Oldenburger Straße 65, D-26316 Varel, Germany E-mail: [email protected], Tel: (++49) (0)4451/9515-15, Fax: : (++49) (0)4451/9515-29 Summary

Intensive testing of lidars is performed by Deutsche WindGuard since 2005. In this frame, the German Test Station for Remote Wind Sensing Devices has been put into operation in spring 2008. Two commer-cially available lidar systems, have been tested so far. The Windcube (offered by Leosphere) has reached an excellent accuracy, reproducibility and availability in flat terrain at least up to 135m measurement height above ground, which is the highest reference meas-urement height at the test station. The ZephIR (offered by Natural Power) has shown a high availability and accuracy in flat terrain, but lacking accuracy has been observed at large measurement heights (124m above ground), and the system has been found to be prob-lematic during fog below the target measurement height. Improvements of the ZephIR for large meas-urement heights are ongoing. It has been observed so far that the accuracy of series models of wind remote sensing devices can deviate from the accuracy of prototypes and that the origin of unacceptable meas-urement errors of individual units can often be identi-fied and resolved. Without testing of the individual units, the user would often not be aware of such prob-lems. The test station has been initiated for providing the wind industry the possibility to have tested lidars or sodars prior to the application in the field, similar to cup anemometer calibrations. Furthermore, the test station is intended to be used for the type specific classification of remote wind sensing devices.

1 Introduction

Remote wind sensing by the use lidar (LIDAR=LIght Detection And Ranging) gets more and more intention by the wind energy industry. Detailed testing of the first two commercially available lidars has been per-formed. In this frame, Deutsche WindGuard has set up the German Test Station for Wind Remote Sensing Devices.

2 Set-up of the Tests, German Test Station for Remote Wind Sensing Devises

2.1 Test of ZephIR

Detailed testing of the ZephIR has been performed against high quality cup and sonic anemometers at different masts with 65 m height and 124 m height from end of to 2005 to the beginning of 2006. Details of these tests are reported in reference [1] and [2]. All tests have been performed in flat terrain.

2.2 Test of Leosphere Prototype

Testing of the prototype of the Windcube has been performed by Deutsche WindGuard from February to March 2008 against high quality cup anemometers at a 99 m high mast in flat terrain. Details of this test are reported in reference [3].

2.3 Test Station for Remote Wind Sensing Devices So far, 5 series models of the Windcube have been tested at the German Test Station for Remote Wind Sensing Devices since its founding in May 2008. The test station basically consists of a 135 m high met mast, which is kindly made available by the wind turbine manufacturer Enercon. The mast is directly located at the German North Sea Coast in flat terrain. The mast is heavily equipped with MEASNET [4] calibrated sensors, which are mounted according to IEC 61400-12-1 [5] (see sensor list in Table 1). All data is measured and stored with a rate of 5 Hz. A test pad for lidars is available directly adjacent to the mast, allowing easy and fast installation of lidar systems, convenient data transfer and the measurement of certain specs of remote sensing devices, like power consumption, temperatures etc. The mast allows using a very wide free measurement sector of 154°-318°, thus ensuring a fast coverage of a large wind speed range. Quite a variation of roughness conditions from sea surface roughness to open farm land and industri-alised areas is present within the free measurement sector, what allows analysing remote sensing devices under different turbulence and wind shear conditions. Measurement

Height agl [m]

Sensor

135.1 2 cup anemometers Thies FC

132.6 cup anemometer Thies FC,

3D sonic Gill Windmaster

131.0 2 vanes Thies FC,

air temperature, air pressure, air humidity

vane Thies FC, air temperature

3D sonic Gill Windmaster vane Thies FC,

air temperature

35.0 2 cup anemometer Thies FC,

vane Thies FC, air temperature

Table 1: Specs of instrumentation of 135m-mast 3 The 2 Commercially Available Lidars

The two lidar systems commercially available today are briefly described in Table 2. Both lidar systems are ground based, small and compact and consists of standard telecommunications components (fiber la-sers). At both systems, measurements of all three wind

(2)

speed components are derived by rotating the laser beam on a cone.

The key difference is the determination of the meas-urement height. Zephir uses a continuous laser beam, while the measurement height is determined by focus-sing the beam on a certain height. After three rotations of the laser beam at a certain focus height (takes 3 seconds), the focus is shifted to the next height. By this, up to 5 measurement heights can be covered successively within about 16 seconds. At each focus height, the wind speed is measured at 50 azimuth angles.

In case of the Windcube, the measurement height is determined by range gates reserved for recording the backscatter (determination of elapsed time between sending and receiving the lidar beam). The rotation of the lidar beam is interrupted at each azimuth angle, and the measurements are taken at all measurement heights (up to 10 heights). Then, the laser beam is rotated to the next angle (4 angles per rotation). A full rotation takes about 6 seconds (4 second rotation was also tested).

Feature ZephIR Windcube

manufacturer Natural Power Leosphere

determination of measure-ment height

focussing of lidar

beam time window

laser beam continuous pulsed

wave lengths 1.575 m 1.54 m measurement heights up to 5, compro-mising output rate 10 number of azimuth angles 50 4 cone angle _30° 30° (15° optional) revolutions per measurement 3 1 output rate 1/3 Hz (1 height), 1/16 Hz (5 heights) 1/6 Hz (1/4Hz) total weight 134kg 50kg power

consumption about 100W, 24VDC about 130W, 24VDC

Table 2: Comparison of key technical specs of ZepiHR and Windcube

4 Testing Results

4.1 Availability of Lidar Measurements

The ZephIR has reached a rate of valid data of 99.7 % at 65 m measurement height and 96.1 % at 124 m measurement height in respect to the horizontal wind speed component (wind speed and wind direction). The Windcube series models have reached availability levels of about 98 % at 100 m measurement height, 95 % at 150 m measurement height and 90 % at 200 m measurement height. These are remarkable results for remote sensing devices, taking into account the partly

bad weather conditions during the measurement cam-paigns. Normally, even data periods with rain, snow or icing conditions had not to be filtered out.

Only the vertical wind speed component seems to be significantly disturbed by precipitation.

4.2 Horizontal Wind Speed Component

The ZephIR has reached a very good accuracy at 65 m measurement height when the so-called cloud correc-tion has been applied (see reference [1], Figure 1). At 124 m measurement height, the ZephIR had the ten-dency to systematically underestimate the wind speed with increasing vertical wind shear [1]. This is be-lieved to be due to the large focussing range of the ZephIR in large measurement heights. Different im-provements in this respect are awaited, e.g. an optimi-sation of the cloud correction as it is reported in refer-ence [6].

During events with heavy fog below the target meas-urement height, the ZephIR has shown the tendency to measure the wind speed in the fog layer rather than in the target measurement height, what must be awaited form the measurement principle.

The Windcube has shown a remarkably good and consistent accuracy in all tested measurement heights between 72 m and 135 m (Figure 2). It is noted that the Windcube does not provide measurement data below 40 m above ground. Furthermore, it is remark-able that all tested series models of the Windcube tested so far have shown a very consistent accuracy pattern (Figure 3). The S-shape observed in the per-centage deviation between the lidar and cup ane-mometer measurements has been found to be due to a problem of the spectrum analysis of the first software version of the Windcube. This originated from the low resolution of the frequency spectrum, which is a con-sequence of the finite length of the laser pulse. After implementation of an improved algorithm, the devia-tions of the Windcube to cup anemometer measure-ments is lower than the uncertainty of the cup ane-mometer measurements within the entire tested wind speed range (Figure 4).

The Windcube is offered with a 30° cone angle and optionally with a 15° cone angle. The smaller cone angle is intended to decrease the measurement circle in complex terrain applications in order to reach a more homogenous flow over the measurement vol-ume. A Windcube has been tested with both cone angles at the test station. Despite the flat terrain at the test station, the wind speed measured with the 15° cone angle has shown larger deviations to the cup anemometer measurements (Figure 5).

(3)

y = 0.981x + 0.110 R2_{= 0.991} y = 0.982x - 0.221 R2_{= 0.986} 0 2 4 6 8 10 12 14 0 5 10 15

wind speed cup [m/s]

w ind s p ee d Z eph IR [m/ s]

124m, with cloud correction 65m, with cloud correction

mean deviation 65m: -0.04m/s std of deviation 65m: 0.18m/s mean deviation 124m: -0.39m/s std of deviation 124m: 0.30m/s

Figure 1: Comparison of 10-minute averages of hori-zontal wind speed component measured with ZephIR and with cup anemometers at 65 m and 124 m height

with cloud correction applied.

y = 1.004x - 0.079 R2_{= 0.996} 0 5 10 15 20 25 0 5 10 15 20 25

w in d speed W ind cube [m /s ] 99m, no correction mean deviation: -0.03m/s std of deviation: 0.24m/s

Figure 2: Comparison of 10-minute averages of hori-zontal wind speed component measured with the Windcube prototype and with a cup anemometer at

99 m height. y = 0.988x + 0.143 R2_{= 0.999} y = 0.998x + 0.046 R2_{= 0.999} y = 1.003x + 0.025 R2 = 0.999 y = 0.995x + 0.087 R2_{= 1.000} 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 v-cup, 135m [m/s] v-W in d cu be [ m /s ] -5 -4 -3 -2 -1 0 1 2 3 4 5 v-W ind cube v -c up , u n ce rta inty m ast [% ]

lidar 1 lidar 2 lidar 3

lidar 4 v-lidar - v-cup, lidar 1 v-lidar - v-cup, lidar 2 v-lidar - v-cup, lidar 3 v-lidar - v-cup, lidar 4 uncertainty v-cup

Figure 3: Comparison of 10-minute averages of hori-zontal wind speed component measured with 4 Wind-cubes and with a cup anemometer at 135 m height.

y = 1.003x + 0.026 R2_{= 1.000} y = 1.003x - 0.004 R2_{= 1.000} 2 4 6 8 10 12 14 16 18 20 22 2 4 6 8 10 12 14 16 18 20 22 v-cup, 135m [m/s] v-W indc u be [ m /s ] -5 -4 -3 -2 -1 0 1 2 3 4 5 v-W in dc ub e v -cu p, un ce rt ai nt y v -cu p [ % ]

lidar before improvement lidar after inprovement

v-lidar - v-cup before improvement v-lidar - v-cup after inprovement uncertainty v-cup

Figure 4: Comparison of 10-minute averages of hori-zontal wind speed component measured with a Wind-cube and with a cup anemometer at 135 m height before and after improvement of the spectrum

analy-sis. y = 1.006x + 0.028 R2_{= 1.000} y = 1.032x - 0.274 R2_{= 0.999} 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 v-cup, 135m [m/s] v-W ind cu be [ m /s ] -14 -12 -10 -8 -6 -4 -2 0 2 4 v-W ind cube v -c up, un cert ai nt y m as t [ % ]

v-30° cone v-15° cone v-lidar - v-cup, 30° cone

v-lidar - v-cup, 15° cone uncertainty v-cup

Figure 5: Comparison of 10-minute averages of hori-zontal wind speed component measured with a Wind-cube and with a cup anemometer at 135 m height with

30° cone angle and 15° cone angle. 4.3 Wind Direction

The wind direction measurements performed by the ZephIR and by vanes correlate well in terms of 10-minute-averages (Figure 6). It has been found, that within 3-second averages sometimes the detected wind direction is switched around by 180°.

Also, the wind direction measurement performed by the Windcube works very well (Figure 7). Here, no disorientation of the wind direction at single 6(4)-second average has been found, what can be under-stood from the measurement principle.

y = 0.994x + 0.256 R2_{= 0.994} y = 1.016x + 0.355 R2_{= 0.982} 120 150 180 210 240 270 300 330 360 120 150 180 210 240 270 300 330 360 wind direction, vane on mast [°]

w ind di rect io n Ze phI R [°] 124m 65m

Figure 6: Comparison of wind direction measurement performed by ZephIR at 124 m and 65 m height above

ground and by a vane (10-minute averages).

y = 1.007x - 0.505 R2_{= 1.000} 150 170 190 210 230 250 270 290 310 330 150 200 250 300 350

wind direction, vane on mast [°]

w ind di recti on W in d cube [°] 135m

Figure 7: Comparison of wind direction measurement performed by Windcube at 135 m height above ground

and by a vane (10-minute averages). 4.4 Comparison of Vertical Wind Speed Component The vertical wind speed component as measured by the ZephIR and by the Windcube does not correlate well with the vertical wind speed component meas-ured by sonic anemometers (Figure 8, Figure 9). The

(4)

lidars need to be clearly improved in terms of the determination of the vertical wind speed component.

y = 0.520x + 0.089 R2_{= 0.548} 0.0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6

vertical wind speed component Ultra Sonic, 122m [m/s]

ve rt ic al w in d sp ee d c om pon en t Z ep hI R , 124m [ m /s ]

Figure 8: Comparison of vertical wind speed compo-nent measured at 124 m height above ground by ZephIR and measured at 122 m height above ground by an ultra sonic anemometer (10-minute averages).

y = 0.709x + 0.336 R2_{= 0.498} -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8

vertical wind speed component ultra sonic, 133m [m/s]

ve rti ca l w in d s peed co mp on en t W indc ube , 1 33 m [ m /s ]

Figure 9: Comparison of vertical wind speed compo-nent measured at 133 m height above ground by Windcube and by an ultra sonic anemometer

(10-minute averages).

4.5 Comparison of Standard Deviation of Horizontal Wind Speed Component

The standard deviation of the horizontal wind speed component as measured by the ZephIR is smaller than measured by cup anemometers (Figure 10), what can is expected due to the spatial averaging of the lidar and the longer pre-averaging period of 3 seconds. In contrast, the Windcube measures the standard deviation of the horizontal wind speed component astonishing well if the system is operated with the standard 30° cone angle (Figure 11). Here, the reduc-tion of the cone rotareduc-tion period from about 8 seconds at the prototype to about 6 seconds at the first series version to about 4 seconds in a testing mode has led to a small improvement. However, when the Windcube is operated with a 15° cone angle, the wind speed standard deviation is highly overestimated (Figure 12).

y = 0.741x + 0.056 R2_{= 0.776} y = 0.651x + 0.013 R2_{= 0.543} 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 standard deviation of horizontal wind speed cup [m/s]

st anda rd de ia ti on o f ho ri zo nt al w ind s pe ed Ze phI R [ m /s ] 124m 65m

Figure 10: Comparison of standard deviation of wind speed within 10-minute periods as measured by

ZephIR and cup anemometers at 124 m and 65 m height. y = 1.070x + 0.014 R2_{= 0.943} 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5

standard deviation of horizontal wind speed cup [m/s]

st anda rd dev ia ti on o f h or izo nt al w ind s p ee d W indC u b e [ m /s ] 135m

Figure 11: Comparison of standard deviation of wind speed within 10-minute periods as measured by Wind-cube with 30° cone angle and a cup anemometer at

135 m height. y = 1.362x + 0.096 R2_{= 0.883} 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5

standard deviation of horizontal wind speed, cup [m/s]

st an da rd d eviat ion h or iz on tal win d sp eed , W in dcu be [ m /s ]

Figure 12: Comparison of standard deviation of wind speed within 10-minute periods as measured by Wind-cube with 15° cone angle and a cup anemometer at

135 m height.

4.6 Comparison of Extreme Values of Horizontal Wind Speed Component

The maxima of the horizontal wind speed component within 10-minute periods measured by the ZephIR are underestimated compared to cup anemometer meas-urements, while the minima are overestimated (Figure 13). The origin can again be seen in the larger spatial averaging of the ZephIR and the larger pre-averaging period of 3 seconds.

The Windcube shows only a small overestima-tion/underestimation of the wind speed maxima/minima within 10-minute periods in the 30° cone angle mode (Figure 14).

y = 0.997x + 0.131 R2_{= 0.896} y = 0.920x + 0.246 R2_{= 0.960} y = 1.044x + 0.302 R2_{= 0.906} y = 0.909x - 0.051 R2_{= 0.974} 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20

w ind s p ee d Z eph IR [ m /s ]

Maxima, 65m Maxima, 124m Minima, 65m Minima, 124m

Figure 13: Comparison of extreme values of wind speed within 10-minute periods as measured by ZephIR and cup anemometer at 124 m and 65 m

(5)

y = 0.991x - 0.129 R2_{= 0.962} y = 1.025x + 0.004 R2_{= 0.982} 0 5 10 15 20 25 30 0 5 10 15 20 25 30

w ind spe ed W in dc ube [ m /s] Maxima 135m Minima 135m

Figure 14: Comparison of extreme values of wind speed within 10-minute periods as measured by Wind-cube and a cup anemometer at 135 m height (4 second

cone rotational mode). 5 Status of Remote Wind Sensing

From our view, remote wind sensing has reached the following status for applications in wind engineering: • The Windcube has reached a high level of

accu-racy and reproducibility of measurements in flat terrain conditions when it is applied with the stan-dard 30° cone angle. Measurements with this type of instrument are suitable for nearly all applica-tions in wind engineering, including power curve measurements and site assessments (at least in flat terrain). This level of confidence has so far not been reached by sodar instruments (see also refer-ence [7]). The Windcube’s performance observed with 15° cone angle is lacking, even in flat terrain. • The ZephIR has shown good performance in lower measurement heights, but problems were observed in large measurement heights (124 m) and during fog below the target measurement height. How-ever, the reported data has been gathered 3 years ago. Improvements have been announced in refer-ence [6].

• Individual testing of lidar and sodar units prior to the field application helps to identify system faults and to ensure the system accuracy. An example of an identification of a system fault of a lidar is shown in Figure 15. Furthermore, individual test-ing of remote wind senstest-ing devices helps to con-vince financing parties or investors of wind farms when using the instrument for site assessment pur-poses. This is in line with a conclusion drawn by an expert meeting of sodar and lidar experts or-ganised by IEA in the beginning of 2007 [7]. • In complex terrain sites, the influence of the

rela-tively large scanning volume of today’s lidars and sodars must be carefully considered in terms of its influence on the measurement accuracy [8].

0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 cup anemometer, 135m [m/s] Windc ube [ m /s ]

Outlier data: Defect of

Figure 15: Example of measurement results gained with a malfunctioning lidar. In the shown case, the origin of outlier data could be identified easily. 6 Acknowledgement

Thanks belong to Enercon to make available the 135 m met mast and a lot of technical support for the German Test Station Remote Wind Sensing Devices, as well as for the permit to use wind data from other masts for the reported work.

7 References

[1] A. Albers; Evaluation of ZephIR, proceedings of DEWEK 2006, Bremen

[2] Evaluation of ZephIR, Deutsche WindGuard Consulting GmbH, report PWG 06005, Version 4, 2006-06-02, www.windguard.de

[3] Evaluation of Windcube, Deutsche WindGuard Consulting GmbH, report PP 08007, 2008-03-16, www. windguard.de

[4] MEASNET, Cup Anemometer Calibration Proce-dure, 1997

[5] IEC 61400-12-1, Wind turbines, Part 12-1: Power performance measurements of electricity produc-ing wind turbines, 2005

[6] M. Courtney, R. Wagner, P. Lindelöw; Testing and comparison of lidars for profile and turbu-lence measurements in wind energy, 14th Interna-tional Symposium for the Advancement of Boundary Layer Remote Sensing, 2008, Roskilde [7] International Energy Agency (2007): Implement-ing Agreement for Cooperation in Research, De-velopment and Deployment of Wind Turbine Sys-tems Task 11 51st IEA Topical Expert Meeting State of the Art of Remote Wind Speed Sensing Techniques using Sodar, Lidar and Satellites. Risoe, Roskilde, Denmark, January 2007

[8] S. Bradley; Wind speed errors for LIDARs and SODARs in complex terrain; 14th International Symposium for the Advancement of Boundary Layer Remote Sensing, 2008, Roskilde