W I ND E NG I NE E R I NG V O L UM E 25, NO. 5, 20 0 1 P P 30 1 30 6 3 0 1 Sodar Wind Velocity Measurements of Offshore Turbine Wakes L. Folkerts 1, R. Barthelmie 2, P. Sanderhoff 2, F. Ormel 3, P. Eecen 3 and O. Stobbe 4 1 Ecofys bv, P.O. Box 840 8, NL-350 3 RK Utrecht, the Netherlands, tel. +31 30 280 830 0, email <L.Folkerts@ecofys.nl>, 2 Risø National Laboratory, Roskilde, Denmark. 3 Energy Research Centre of the Netherlands (ECN), Petten, the Netherlands. 4 Ecofys Energieberatung und Handelsgesellschaft mbh, Köln, Germany. ABSTRACT In A pril 2001 the first offshore wake measurements with a SODA R (sonic detection and ranging) were conducted at Vindeby. Wake effects were measured at distances from the turbine varying from 1.4 to 7.1 rotor diameters. The corresponding calculated relative velocity deficits cover a range from 12% up to 56%. The recorded wake profiles provide excellent reference cases for the development and evaluation of offshore wake models. 1. INT RODUCT ION The ENDOW project (Barthelmie et al., 2001) aims to reduce uncertainties in estimating power production introduced due to wake effects in large offshore wind farms. The m ajor objectives are to evaluate wake m odels in offshore environments and to develop and enhance existing wak e and boundary-layer m odels accounting for complex stability variations to produce a design tool to assist planners and developers in optimising offshore wind farms. To provide additional data on offshore wak es and to form part of the evaluation process for wak e models within the ENDOW project a large-scale experiment was conducted at the 5 MW offshore wind farm at V indeby in D enmark. Especially in near-w ake situations, very few data sets are available, even for land sites. F or the first time, SOD A R instrumentation (sonic detection and ranging) was m ounted on a boat and used to measure wind profiles in the turbine wake in an offshore wind farm. Selective operation of turbines in carefully m onitored conditions recording wind speed, wind direction and atmospheric stability allow the direct impact of turbine operation on wak e effects to be measured at varying distances from the turbine. Use of a SODA R sy stem provided vertical wind speed profiles to hub and rotor heights, w hich supplements ongoing m easurem ents on three meteorological m asts. 2. EXPERIMENT The measurement campaign was conducted at the Vindeby wind farm in Denmark during the period 21-28 A pril 2001. The Vindeby farm consists of 11 BONUS 450 kw turbines in two row s oriented towards the southwest. The hub-height is 38m and the rotor diameter is 35m. The distance between two adjacent turbines is 300 m. This site was chosen because it is one of very few operating offshore wind farms and has three monitoring masts (two offshore and one at the coast) providing detailed meteorological measurements to 50 m height. The site has the advantage of relatively low water depth (2-5 m) offering the possibility of relatively low wave heights and swell.
3 0 2 SO DA R W I ND V E L OC I T Y M E A SUR E M E NT S OF O F F SH OR E T UR B I NE W A K E S The SODA R used was an Aerovironment mini-soda R model 3000 equipped with a model 4000 speaker array. The SOD A R was operating at 4500 Hz with a pulse length of 50 ms. The SOD A R was mounted on a ship, 5.5 m wide and 18.6 m long ( Seaw orker, Figure 1). The orientation of the SODA R with respect to the ship was such that one of the tilted beams of the SOD A R (tilt angle 16 degrees off vertical) was oriented towards the port side and the other towards the bow of the ship. Figure 1. SODA R mounted on the ship Seaworker at the Vindeby wind farm. Positioning of the boat in the direct wake of a wind turbine, using 3 anchors, took about 40 minutes. The ship was approximately oriented with its stern in the wind. A ccurate position and orientation of the SOD A R were determined using G PS and compass. Using position logging, the accuracy of the GPS readings improved to 4 m eter. Profile measurements were tak en over periods of approximately half an hour. Free stream reference m easurem ents were conducted with the wind turbine in question switched off. The ship engines were off during the m easurements. There is a noticeable change in the noise level between on and off operation of the nearest wind turbine. The noise level from the turbine was acceptable even at the closest distance (50 m). The movement of the SOD A R was monitored with an inclinometer, recording both the longitudinal and transversal tilt angles (see figure 2). Considering the opera tion of the SOD A R, only data taken with limited swing (< ±2 degree) was accepted in the analysis. F or an estimate on the error induced by the swing it is important to realise that the ship was oriented more or less in line with the wind direction. Therefor, the longitudinal movement is of m ore consequence than the transversal. The swing in the longitudinal direction in the accepted data is < ±0.5 degree, resulting in a maximum error of ± 3% for an individual m easurement. Over several m easurements the effect of the swing will average out.
W I ND E NG I NE E R I NG V O L UM E 25, NO. 5, 20 0 1 30 3 2.5 2 1.5 1 0.5 0-0.5-1 0 15 30 45 60 Figure 2. Recorded longitudinal (upper line) and transversal (lower line) tilt angles. Variation in distance from the turbine was an important goal of the ex periment. We succeeded in getting useful wake profiles at a minimum distance of 1.4 rotor diam eters, and up to 7.1 rotordiameters. Especially at the larg er distance, small variations in the wind direction mean that the Sodar moves in and out of the centre of the wake. Using the wind direction of the mast SMW, the off-center distance (or angle) can be calculated. Only data near the wak e center was accepted in the analysis. Figure 3 show s the layout of the wind farm at Vindeby and the positioning of the SODA R at 3.5 rotor diam eters behind turbine 1W. ~ 34 0 o 3.5 D Figure 3. Layout of the wind farm at Vindeby (d shows each wind turbine and m the two sea masts) and the positioning of the SODA R (j ) in the wake of turbine 1W. On the left the positioning of the SODA R is shown in more detail with distance in units of turbine diameter. 3. RESULTS The first ex periment was taken at approximately 3.5D (rotor diameter D = 35m) in the direct wak e of turbine 1W and the wake measured for approximately 40 minutes. D uring this period the wind direction was 343 ± 2 (single wake only) and the stability at the land m ast was nearneutral. The average wind profiles for wak e measurem ent and the corresponding free stream m easurem ents are show n in figure 4. R esulting wind profiles were composed by averaging over the recorded one-minute profiles. The resulting standard deviation on the avera ged wind speeds corresponds m ainly to the actual variations in the wind speed, but also includes some deviations resulting from the measuring technique (e.g. the movement of the SOD A R, see section 2).
3 0 4 SO DA R W I ND V E L OC I T Y M E A SUR E M E NT S OF O F F SH OR E T UR B I NE W A K E S Average wind speeds measured at 50m (at the land mast) decreased by 15% in the time lapse between the free stream and wak e m easurement. This change in wind speed is derived from the free stream and wake velocities well above blade tip maximum height. A correction is applied to increase the SODA R wake profile accordingly. 100-0.2 0.0 0.2 0.4 0.6 0.8 1.0 80 60 40 20 0 0 5 10 1 5 20 Figure 4. Profile of the free stream wind (d ) and the wake profile (m corrected for the change in free wind speed) measured at 3.5D behind turbine 1W. A lso shown is the calculated relative velocity deficit (u, top axis). Height is above sea level. The shown standard deviations result mainly from the actual variations in the wind speed (see text). 1988: The figure also show s the calculated relative velocity deficit, defined as in Hogstrom et al., D U/U freestream = 1 2 U wake /U freestream A s expected, the velocity deficit is approximately zero well above the tip height of the rotor. The m aximum velocity deficit in this case is 37% around hub-height. D uring the wak e measurements (turbine on), a variation in the wind direction m eant that the Sodar profile was taken at various distances from the centre of the wake. This was especially critical at the larger distances. In the analysis of the wake profile tak en at 7.1D a gate was set to only include profiles near the centre of the wake. The maximum displacement from the wake centre allowed is 0.5D. Figure 5 show s the measured wake profile at 7.1D behind turbine 1E, the corresponding free stream profile and the calculated relative velocity deficit. The velocity deficit at hub height at this distance am ounts to about 12% Wake profiles were measured at 13 different positions. The corresponding calculated maximum relative velocity deficits are show n in figure 6 as a function of the distance. The line in figure 6 is a meant to guide the eye to the obvious trend. There is considerable scattering around 3D dow nstream distance. One important aspect is the variation in wind speed between the various measurements from 6 to 12 m/ s at hub height. Other atmospheric conditions (turbulence intensity, stability) could play an important role as well. Futher work will include detailed comparison with m odel calculations.
W I ND E NG I NE E R I NG V O L UM E 25, NO. 5, 20 0 1 30 5 100-0.2 0.0 0.2 0.4 0.6 0.8 1.0 80 60 40 20 0 0 5 10 1 5 20 Figure 5. Same as figure 4, but for 7.1D behind turbine 1E. During this wake measurement the wind direction was 206 ±2. 4. FUTURE WORK The ENDOW project focuses on the evaluation and development of wak e models for use in an offshore environment. The six partners performing wake modelling within the project and a brief description of the models are given in Barthelmie et al., 2001. So far, comparison of the model output with experimental results was limited to the Vindeby mast measurement at 8.6D. The results from the ex periment described here provide near-w ake wind speed profiles for further direct comparison with the various wake models in the ENDOW project. 1. 0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 1 2 3 4 5 6 7 8 9 10 Figure 6. Relative velocity deficit by distance (shown here as number of rotor diameter). The trendline is a fit to the points to guide the eye. 5. ACKNOWLEDGEMENTS Financial support for this research was given in part by the European Commission s Fifth F ra mework Progra mm e under the Energy, E nvironment and Sustainable D evelopment Progra m m e. Project R eference: E R K 6-1999-00001 END OW ( E fficient D evelopment of Offshore W indfarms). A dditional funding was supplied by NOV EM for Ecofys and ECN.
3 0 6 SO DA R W I ND V E L OC I T Y M E A SUR E M E NT S OF O F F SH OR E T UR B I NE W A K E S 6. REFERENCES Bar thel mi e R., et al., (2001) Efficient developm ent of offshore windfarms: w ak e and bou ndar y-l ay er interactions, E W EA Offshore Conference, D ecem ber 2001, W ind Engineering, vol. 25, no. 5, pp. 263-269. F air all C.W., W hite A.B., E dson J.B. and Hare J.E., (1997) Integrated Shipboar d Measu rements of the Marine Bou ndery Lay er, Journal of A tmospheric and Oceanic Technology, vol. 14, no. 3, pp. 338-359. Hogstr om U., A simak opoulos D.N., K ambezidis H., Helmis C. and Smedman A., (1988) A field study of the wake behind a 2 MW wind turbine, A tmospheric Environment, vol. 22, no. 4, pp. 803-820.