WIND AND WAVE CONDITIONS

Dutch Offshore Wind Energy Converter project

Wim Bierbooms (ed.) Section Wind Energy Delft University of Technology

DOWEC 47 rev. 2

Note In order to meet the confidential requirements in obtaining the NEXT data most tables in this public report are intentional left blank. The information from the tables has been used in other Work Packages of the DOWEC project.

DOWEC

Table of Contents

1.

INTRODUCTION..................................................................................................................................... 1

2.

GENERAL ANALYSIS NEXT DATABASE......................................................................................... 4

3.

SCATTER DIAGRAMS .......................................................................................................................... 9

4.

EXTREMES ............................................................................................................................................ 12

5.

WIND CLIMATE ................................................................................................................................... 22

6.

WEATHER WINDOWS ........................................................................................................................ 31

REFERENCES ................................................................................................................................................. 38 ANNEX I: NL7 DATA ..................................................................................................................................... 39

CONTRIBUTIONS BY ECN ANNEX II: TURBULENCE AND WIND SHEAR ....................................................................................... 64 ANNEX III: DESIGN WIND CLIMATE ...................................................................................................... 87

ii

Wind and wave conditions

1. Introduction This report is part of the DOWEC (Dutch Offshore Wind Energy Converter) study; it concerns task 5 of work package 1. The objective of this task is to generate sufficiently detailed and accurate wind, wave, current, soil and other site data, necessary for design of a large scale wind farm at the two selected locations as specified in pre-EET DOWEC concept study (indicated by “NL3” and “NL7”). The mean wind, wave and current characteristics are dealt with in this report. Contributions by ECN to this report concern the wind shear and turbulence as well as the design wind climate; they are given in annex II (the integral text from [4]) and III. The site data are specified in a separate report: [3]. For the determination of the wind, wave and current characteristics extensive use is made of the NEXT database (NEXT is the successor of NESS). For a detailed description of this database the reader is referred to the appendices 1 and 2 of [1]. The NEXT data are given on a 30 km by 30 km grid. The closest NEXT grid points to the requested “NL3” and “NL7” location are (see fig. 1.1): • “NL3” location: 52 gr 51 min 53 sec N, 4 gr 24 min 46 sec E. • “NL7” location: 52 gr 14 min 56 sec N, 3 gr 35 min 33 sec E For convenience these NEXT grid points will further on be indicated by NL3 and NL7.

Fig. 1.1: The locations of the NEXT gridpoints and the NL3 and NL7 site.

1

DOWEC The NEXT database consist of hindcast data and are given in a 3-hour interval covering the winter months of the period October 1964 to March 1995. For 9 years also the data of the summer months is included; thus for these years the complete data is available. In this study the following variables from the NEXT database are used: • (1-hour) mean wind speed (m/s) at 10 m height • wind direction (degrees) • significant wave height (m) • mean zero crossing period (s) • wave direction (degrees) • current speed with tide (m/s); depth averaged value • current speed without tide (m/s); depth averaged value • current direction with tide (degrees) • current direction without tide (degrees) • tide plus storm surge (m) with respect to MSL • storm surge (m) with respect to MSL Unless specified otherwise, the variables concern 3-hour values. Notes: in the NEXT database the ‘current speed with tide’ is indicated by ‘total current speed’ and ‘current speed without tide’ by ‘residual current speed’. Something similar holds for the water level: ‘tide plus storm surge’ is indicated in NEXT by ‘total height’ and ‘storm surge’ by ‘residual height’. The NEXT database also includes the significant wave height, wave direction and peak period from wind sea state and swell separately. Furthermore some directional spreading variables are available. In NEXT no information is given on turbulence and wind shear; as mentioned before these site specifications can be found in annex II. The given mean wind speed can be regarded as the actual wind speed at 10 m height; thus not a potential wind speed based on some reference roughness. The applied sign convention in NEXT is: • wind and waves: coming from a certain direction (with respect to true North) • current: going to a certain direction (with respect to true North) In the analysis, 12 direction sectors are used according to the usual convention: the 1st sector includes the directions between 345 (inclusive) and 15 degrees (with respect to true North); the 2nd sector 15 (inclusive) to 45 degrees, etc. The results for the NL3 site are given in the main text; for clarity all results for the NL7 site are given separately in the final chapter. Acknowledgement We would like to express our thanks to the NESS Users Group and to Shell in particular for their co-operation in making NESS data available to us for this research project.

2

Wind and wave conditions Furthermore the Dutch KNMI (Royal Netherlands Meteorological Institute) is acknowledged for the use of the online database of wind speeds.

3

DOWEC

2. General analysis NEXT database In this section the results are shown of some general analysis work of the NEXT database in order to give an indication of the environmental conditions. For this purpose, data is selected from the database covering full years. In fig. 2.1 the frequency of wave direction is compared to the distribution of wind direction. The distribution of the waves is even more pronounced, due to the fact that wind coming from land will result in small waves only (limited fetch). The ‘wave rose’ is given in fig. 2.2.

Fig. 2.1: Frequency of occurrence of wave direction (solid line) and wind direction (dashed line) for the NL3 site.

4

Wind and wave conditions

Fig. 2.2: Polar plot of the mean significant wave height (left) and mean zero crossing period (right) for the NL3 site. The spread in current direction is very limited due to the tidal effect, see fig. 2.3. It can be assumed that there is little correlation between the waves (or wind) and the current. Comparing the current strength with (fig. 2.3) and without tide (fig. 2.4; notice the difference in scale) it may be concluded that the current is dominated by the tide.

Fig. 2.3: Frequency of occurrence of current with tide (left) and polar plot of current speed (right) for the NL3 site.

Fig. 2.4: Frequency of occurrence of current without tide (left) and polar plot of current strength (right) for the NL3 site.

5

DOWEC Histograms for several variables are depicted in figure 2.5.

Fig. 2.5: Histograms of significant wave height, mean zero crossing period, current (with and without tide), storm surge (with and without tide) for the NL3 site.

6

Wind and wave conditions For offshore wind turbines the misalignment of wind and waves is of importance. This misalignment is given in the figures 2.6 and 2.7. For higher wind speeds the local wave field will be dominated by the local wind field, which will lead to a higher correlation between the wind and wave direction. This can be observed in figure 2.7.

Fig. 2.6: Histogram of the misalignment of wind and waves for the NL3 site.

Fig. 2.7: Histogram of the misalignment of wind and waves for the NL3 site; conditioned on wind speeds below 5 m/s (left) and above 15 m/s (right).

7

DOWEC

8

Wind and wave conditions

3. Scatter diagrams In table 3.1 the (relative) wave scatter diagram (significant wave height Hs versus mean zero crossing period Tz) for the NL3 location is given; the applied bin size for Hs is 0.5 m and 0.5 s for Tz. In order to include the seasonal variation only NEXT data covering full years are taken into account: 1977 through 1979 and 1989 through 1994 (26288 data points in total). Tz (s) 0.0 - 0.5 - 1.0 - 1.5 - 2.0 - 2.5 - 3.0 - 3.5 - 4.0 - 4.5 Hs (m) 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 4.5 5.0 4.0 4.5 3.5 4.0 3.0 3.5 2.5 3.0 2.0 2.5 1.5 2.0 1.0 1.5 0.5 1.0 0.0 0.5 CONFIDENTIAL

- 5.0 - 5.5 - 6.0 - 6.5 - 7.0 5.5 6.0 6.5 7.0 7.5

Sum

Sum

Table 3.1: Wave scatter diagram for NL3; values in parts per thousand (ppt); due to the effect of round off, the totals of columns and/or rows may be a little bit different from the values given under ‘Sum’. For a correct fatigue analysis of the support structure of an offshore wind turbine it is important to apply a 3D scatter diagram, with variables Hs, Tz and the (1-hour) mean wind speed V (see for more information on fatigue analysis of offshore wind turbines [6] and [7]). The 3D scatter diagram for the NL3 site is given in table 3.2; for the width of the wind speed bins 2 m/s has been taken. This results in total in about 80 different “sea states” (given by a particular combination of Hs, Tz and V). This number of 80 “sea states” will probably lead to a too time consuming fatigue analysis of the support structure (in the final design phase). In order to limit the involved computational time the number of load cases can be reduced by lumping of sea states. See e.g. [7] for some ideas how to do so. In [2] scatter diagrams are given based on data of the RIKZ (Rijksinstituut voor Kust en Zee) for several locations from the Measuring Network North Sea. The location ‘IJmuiden depot’ is closest to the NL3 site. Concerning the distribution of the significant wave height, table 3.1 above correspond well with the data shown in table 2 from [2]. However, there is some deviation in the distribution of the mean zero crossing period: the variation of Tz as specified in table 3.1 (for each Hs) is smaller than the variation given in the wave scatter diagram in [2]. Further analysis is needed for a more thorough comparison. 9

DOWEC The scatter diagrams of the NL3 and the NL7 site (see Appendix 1) appear to be very similar. This may be explained by the fact that the dominating wave directions are the 9th and the 12th sector (fig. 2.1). Roughly speaking waves from the 9th sector travel parallel to the Dutch coast, which is also about parallel to the line from NL7 to NL3. Since the water depth does not change so much between NL7 and NL3 one may anticipate that the waves travelling parallel to the coast will not change significantly between NL7 and NL3. Waves from the 11th / 12th sector travel about perpendicular to the coast, so will arrive at about the same time at the NL3 and NL7 site, leading to a similar wave climate. It is stressed that one should be cautious in generalizing the above reasoning. In general, scatter diagrams will differ from site to site, depending on amongst others the local bathymetry (e.g. a nearby sand bank will result in lower and shorter waves).

10

Wind and wave conditions Tz (s) 2.5 - 3.0

3.0 - 3.5

3.5 - 4.0

4.0 - 4.5

4.5 - 5.0

5.0 - 5.5

5.5 - 6.0

6.0 - 6.5

6.5 - 7.0

Hs (m) 4.0 - 4.5 3.5 - 4.0 3.0 - 3.5 2.5 - 3.0 2.0 - 2.5 1.5 - 2.0 1.0 - 1.5 0.5 - 1.0 0.0 - 0.5

CONFIDENTIAL

Table 3.2: Three dimensional scatter diagram for NL3; values in parts per thousand (ppt). The upper value of the mean wind speed bins is given in each box on the left (and is succeeded by a colon); e.g. “4:” means that the (1-hour) mean wind speed (at 10 m) is in between 2 m/s (inclusive) and 4 m/s. Due to round off errors the totals of each Hs-Tz bin may be a little bit different from the corresponding value given in table 3.1.

11

DOWEC

4. Extremes Considering estimation of extremes there is no general method what can be considered as the best method for all situations. Instead, depending on the actual application suitable method(s) from literature should be selected, based on engineering intuition and experience. Afterwards the suitability of the selected method can be evaluated, e.g. on basis of a Bayesian analysis; see [8] for a more profound description on these matters. Concerning the data, to be taken into account for the analysis, several options are available. One could use all available data from the NEXT database; this will however result into too high estimates of the extremes since for most years the (calm) summer months are not included. Another option is to select all 3-hour values covering complete years. This has the disadvantage that the estimation of extremes will be based on data which consist mostly of modest values (which probably have the largest relative errors due to e.g. the starting behavior of measuring equipment like anemometers). Furthermore, for the extrapolation (to the desired return period) of the probability function it is assumed that the data are independent. This is certainly not the case for 3-hour data. To overcome this problem, it is opt to use the annual extremes for the 30 years (1965-1994) what are available in the NEXT database. This is justified since the annual extremes can be expected to occur during the winter months. For the fitting procedure the Maximum Likelihood Estimation method is applied. According to this method, the parameters of the Extreme Value Distribution are chosen such that the chance of occurrence of the actual data (i.e. the 30 annual extremes) is as large as possible; under the assumption that the chosen Extreme Value Distribution is correct. As Extreme Value Distribution the 3-parameter Weibull distribution is chosen. The results given below all have passed the KolmogoroffSmirnov test on goodness of fit. Other choices in the method for the fit procedure and/or Extreme Value Distribution may lead to difference in the 100-year estimate of the order of 8%. In the tables 4.1 to 4.8 and figures 4.1 to 4.4 the extreme values for the significant wave height, wind speed and current (with and without tide) as well as minimum and maximum storm surge (with and without tide). The considered return periods are 1, 5, 10, 25, 50 and 100 year. For directional data, the extremes are determined for each direction sector. Notes: Just for completeness the extreme value for a return value of 1 year is given. This is, by definition, the value for which the Extreme Value Distribution equals 0. On basis of the annual extremes it is difficult to give an accurate estimate for this 1-year value (and therefore it is not recommended to use this 1-year value). The current as well as the height contain a deterministic part, i.e. the tide. For a proper estimation of the extremes this part should in fact be treated separately. For simplicity this has not been done here.

12

Wind and wave conditions Significant wave height (m) RP=1 year

NL3 5

10

25

50

100

sector

1 2 3 4 5 6 7 8 9 10 11 CONFIDENTIAL 12 all directions Table 4.1: The extreme value of the significant wave height for the NL3 site; for 12 wave direction sectors and several return periods. 1000

0.99 0.98 0.96

100 50 25

0.90

10

0.80

5

0.50

2

0.25

1.3

0.10

1.1

0.01

1.01

4 5 significant wave height (m) − NL3

6

Return period (year)

Probability

WEIBULL GRAPH 0.999

7

Figure 4.1: A Weibull graph of the significant wave height for the NL3 site (all wave direction sectors included). The annual extremes are indicated by ‘+’; first the N (=30) extremes are ordered to magnitude and then assigned a probability F=i/(N+1) with i=1 to N. 13

DOWEC

Mean wind speed (m/s) RP=1 year

NL3 5

10

25

50

100

sector

1 2 3 4 5 6 7 8 9 10 11 CONFIDENTIAL 12 all directions Table 4.2: The extreme value of the (1-hour) mean wind speed (at 10 m) for the NL3 site; for 12 wind direction sectors and several return periods. 1000

0.99 0.98 0.96

100 50 25

0.90

10

0.80

5

0.50

2

0.25

1.3

0.10

1.1

0.01

1.01

18

Return period (year)

Probability

WEIBULL GRAPH 0.999

20 22 24 26 28 mean wind speed (m/s) − NL3

Figure 4.2: A Weibull graph of the mean wind speed (at 10 m) for the NL3 site (all wind direction sectors included); the annual extremes are indicated by ‘+’. 14

Wind and wave conditions

Current (m/s) with tide

NL3 RP=1 year

5

10

25

50

100

sector

1 2 3 4 5 6 7 8 9 10 11 CONFIDENTIAL 12 all directions Table 4.3: The extreme value of the current (with tide) for the NL3 site; for 12 current direction sectors and several return periods. 1000

0.99 0.98 0.96

100 50 25

0.90

10

0.80

5

0.50

2

0.25

1.3

0.10

1.1

0.01

1.01

0.6 0.7 current speed (m/s)− with tide − NL3

Return period (year)

Probability

WEIBULL GRAPH 0.999

0.8 0.9 1

Figure 4.3: A Weibull graph of the current speed (with tide) for the NL3 site (all current direction sectors included); the annual extremes are indicated by ‘+’. 15

DOWEC

Current (m/s) without tide

NL3 RP=1 year

5

10

25

50

100

sector

1 2 3 4 5 6 7 8 9 10 11 CONFIDENTIAL 12 all directions Table 4.4: The extreme value of the current (without tide) for the NL3 site; for 12 current direction sectors and several return periods. 1000

0.99 0.98 0.96

100 50 25

0.90

10

0.80

5

0.50

2

0.25

1.3

0.10

1.1

0.01

1.01

0.3 0.4 0.5 current speed (m/s)− without tide − NL3

Return period (year)

Probability

WEIBULL GRAPH 0.999

0.6 0.7 0.80.9 1

Figure 4.4: A Weibull graph of the current speed (without tide) for the NL3 site (all current direction sectors included); the annual extremes are indicated by ‘+’. 16

Wind and wave conditions

Height - max (m) Tide plus storm surge RP=1 year

NL3 5

10

25

50

100

CONFIDENTIAL

Table 4.5: The extreme value of the maximum height (tide plus storm surge) for the NL3 site; for several return periods. Storm surge - max (m) RP=1 year

NL3 5

10

25

50

100

CONFIDENTIAL

Table 4.6: The extreme value of the maximum storm surge for the NL3 site; for several return periods. Height - min (m) Tide plus storm surge RP=1 year

NL3 5

10

25

50

100

CONFIDENTIAL

Table 4.7: The extreme value of the minimum height (tide plus storm surge) for the NL3 site; for several return periods. Storm surge - min (m) RP=1 year

NL3 5

10

25

50

100

CONFIDENTIAL

Table 4.8: The extreme value of the minimum height (storm surge) for the NL3 site; for several return periods.

17

DOWEC Above, in figure 4.1, the extreme values for the significant wave height are shown. In order to give an impression of the correlation, scatter diagrams are determined for the mean zero crossing period, fig. 4.5, and the storm surge, fig. 4.6. The values of the mean zero crossing period and storm surge are taken at exactly the same time instant at which the annual maximum of the significant wave height (regardless the wave direction) occur. scatter diagram extremes 9

mean zero crossing period (s)

8 7 6 5 4 3 2 1 0

0

1

2

3 4 5 significant wave height (m)

6

7

Fig. 4.5: The scatter diagram of the annual maxima of the significant wave height versus the mean zero crossing period (at the same moments) for the NL3 site. The lines indicate the deep water relations between Tz and Hs; steepness 14 (solid) and 16 (dashed).

scatter diagram extremes 2.2 2

storm surge (m)

1.8 1.6 1.4 1.2 1 0.8 0.6 3.5

4

4.5 5 5.5 significant wave height (m)

6

6.5

Fig. 4.6: The scatter diagram of the annual maxima of the significant wave height versus the storm surge (at the same moments) for the NL3 site.

18

Wind and wave conditions

Similar graphs for the maxima of the significant wave height versus the mean wind speed and vice versa, are shown in fig. 4.7 and 4.8.

Fig. 4.7: The scatter diagram of the annual maxima of the significant wave height versus the mean wind speed (at the same moments) for the NL3 site.

Fig. 4.8: The scatter diagram of the annual maxima of the mean wind speed versus the significant wave height (at the same moments) for the NL3 site.

19

DOWEC In the preceding the extremes of the mean wind speed and significant wave height have been treated separately. From amongst others, the scatter diagrams of chapter 3 it follows that a certain correlation exists between these two variables. To obtain a continuous joint occurrence probability of the sea states, described by V and Hs, a statistical procedure has to be applied to the NEXT data. Here the same procedure is chosen as used in [10]. The basic idea is that the joint probability density function can be written as a product of the marginal and conditional probability density functions: f(V,Hs) = f(V)*f(Hs|V) where f(V) is the marginal distribution of the mean wind speed V and f(Hs|V) is the conditional distribution of the significant wave height Hs for a given mean wind speed V. The marginal distribution of the mean wind speed is obtained by fitting all mean wind speeds (of the 9 year of continuous data) to a 2 parameter Weibull distribution function (analogous as what will be done in the next chapter concerning the determination of the wind climate). Next, wind speed bins of width of 2 m/s are made. The significant wave heights that correspond to each wind speed class is fitted to a cumulative distribution function; here again the 2 parameter Weibull distribution is applied. The fitted distribution parameters (i.e. scale and shape parameter) can be written as a function of the mean wind speed. By doing so, a continuous description of the conditional distribution of Hs is obtained. It turns out that the fitted Weibull parameters can be satisfactorily described by a polynomial in V of degree 2. It is now possible to calculate the joint probability density for any given value of the mean wind speed and significant wave height. Also contour plots of the joint pdf can be made. However it is not straightforward to ascribe a certain probability (or return period) to such a contour plot. In order to do so another statistical procedure is required. The joint probability density of the wind speed and significant wave height can be transformed in a standard normal space of uncorrelated variables using a Rosenblatt transformation (see some textbook on statistics). The e.g. 100 year contour line is described by a circle in this normalised space. The radius r of the circle is determined by r=Rinv(1-1/N100) with Rinv the inverse cumulative Rayleigh distribution and N100=8*365*100, the number of sea states in 100 years. The contour in the physical space can be determined by transforming the standard normal variables back to the physical variables, see fig. 4.9. Each contour line of this figure can be interpret as follows: all the points on a line have the same probability density; events outside the contour line will occur, on average, once in the specified return period (i.e. 10, 50 or 100 years).

20

Wind and wave conditions

Fig. 4.9: Contour plots of the joint density function of the mean wind speed and significant wave height for the NL3 site; for a return period of 10, 50 and 100 years.

21

DOWEC

5. Wind climate For a correct estimation of the wind climate, i.e. taking the seasonal variation into account, data must be used which covers full years. Concerning the NEXT database, for nine years the complete data is available: 1977 through 1979 and 1989 through 1994 (26288 data points in total). The wind speed given in the NEXT database concerns the wind speed at 10 m height. For the estimation of the yearly energy yield the wind speed given here should be scaled first to rotor axis height, applying the wind shear formula, see annex II and III. For wind energy application it is most common to apply the European Wind Atlas method [5] for the Weibull fitting. Two requirements have to be fulfilled: • The total wind energy in the fitted Weibull distribution and in the observed data must be the same • The frequenc y of occurrence of wind speeds higher than the observed average wind speed must be the same This leads to two (non-linear) equations for the Weibull shape and scale parameter, which can be solved by a standard numerical routine. The results, for all wind directions, for the NL3 site are given in fig. 5.1.

Fig 5.1: The fitted Weibull distribution for the NL3-site compared to the original data, given by a histogram. Left: probability density function; right: cumulative distribution function The results for each wind direction sector are given in fig. 5.2 and table 5.1. The usual sector definition is applied: the 1st sector are the wind directions between 345o and 15o; the 2nd sector the directions between 15o and 45o; etc. In the figure the wind rose is given of the: • Frequency of occurrence • Mean wind speed • Weibull scale parameter A • Weibull shape parameter k As expected the dominating wind direction is South-West.

22

Wind and wave conditions

Fig. 5.2: The wind rose of the frequency of occurrence, (1-hour) mean wind speed (at 10 m) and the calculated Weibull parameters for the NL3 site. Sector 1 2 3 4 5 6 7 8 9 10 11 12 Total

Frequency [-] 6.4 6.4 6.6 6.2 5.9 5.5 6.7 13.2 15.4 11.9 8.2 7.6 100.0

A [m/s] 7.6 7.4 7.9 7.8 7.2 7.4 8.3 9.7 10.1 9.5 8.8 8.6 8.6

k [-] 2.2 2.3 2.2 2.2 2.3 2.3 2.2 2.3 2.6 2.1 2.1 2.2 2.1

Table 5.1: The calculated Weibull parameters (at 10 m) for the NL 3 site.

23

Vm [m/s] 6.8 6.6 7.1 7.0 6.5 6.6 7.5 8.8 9.0 8.5 7.9 7.5 7.8

DOWEC The results per season are shown in table 5.1a to 5.1d. The mean wind speed Vm, as given in the tables, are determined directly from the corresponding wind speeds from the data base; therefore small deviations (~ 0.1 m/s) may occur compared to the mean values based on the corresponding, fitted Weibull parameters: Vm=A*Γ(1+1/k). By the division in season and wind direction, the number of wind speed values in some particular bin may be as small as 300. This may imply that rare large wind speed values can be absent in this subset, which brings about large values for the shape factor k. However, this will not be critical for an estimate of the energy yield per season. Sector Frequency [-] A [m/s] k [-] Vm [m/s] 1 2 3 4 5 6 7 8 9 10 11 12 CONFIDENTIAL Total Table 5.1a: The calculated Weibull parameters (at 10 m) for the NL 3 site – winter (Dec/Jan/Feb).

Sector Frequency [-] A [m/s] k [-] Vm [m/s] 1 2 3 4 5 6 7 8 9 10 11 12 CONFIDENTIAL Total Table 5.1b: The calculated Weibull parameters (at 10 m) for the NL 3 site – spring (Mar/Apr/May).

24

Wind and wave conditions

Sector Frequency [-] A [m/s] k [-] Vm [m/s] 1 2 3 4 5 6 7 8 9 10 11 12 CONFIDENTIAL Total Table 5.1c: The calculated Weibull parameters (at 10 m) for the NL 3 site – summer (Jun/Jul/Aug).

Sector Frequency [-] A [m/s] k [-] Vm [m/s] 1 2 3 4 5 6 7 8 9 10 11 12 CONFIDENTIAL Total Table 5.1d: The calculated Weibull parameters (at 10 m) for the NL 3 site – autumn (Sep/Oct/Nov).

25

DOWEC Although the period of 9 years is considerable it is short compared to a period of 30 years what is used in the meteorology for climate studies. In order to arrive at a long term (30 year) estimate of the wind climate the NEXT results can be compared to some long term data record. For the latter the available online database of the Dutch KNMI (Royal Netherlands Meteorological Institute) is used: http://www.knmi.nl/samenw/hydra/index.html. An overview of the KNMI stations is given in fig. 5.3.

Fig. 5.3: Overview of the KNMI wind speed measuring network. The station Light Isle Goeree has been selected since it is representative for the Dutch offshore wind climate and it covers the most years, namely 20 (Note: only years with at least 90% of valid data are taken into account). The correlation between the annual mean wind speeds for the Light Isle Goeree and the NL 3 site, see fig. 5.4, is 94 %, which justifies application of a correlation method. For completeness also data of the Measuring post Noordwijk is shown in fig. 5.4, which also exhibits a strong correlation with the Light Isle Goeree data. On basis of the corresponding years, a long term correction factor of 0.98 is determined for the calculated Weibull parameters of the NL3 site. The corrected wind climate for NL3 is tabulated in table 5.2.

26

Wind and wave conditions

Fig. 5.4: The variation of the annual mean wind speed. o: NL3 site; NEXT database *: Light Isle Goeree; KNMI database +: Measuring post Noordwijk; KNMI database Sector Frequency [-] A [m/s] k [-] Vm [m/s] 1 6.4 7.5 2.2 6.6 2 6.4 7.3 2.3 6.5 3 6.6 7.7 2.2 7.0 4 6.2 7.6 2.2 6.8 5 5.9 7.1 2.3 6.4 6 5.5 7.2 2.3 6.5 7 6.7 8.1 2.2 7.3 8 13.2 9.5 2.3 8.6 9 15.4 9.9 2.6 8.8 10 11.9 9.3 2.1 8.3 11 8.2 8.6 2.1 7.7 12 7.6 8.4 2.2 7.4 Total 100.0 8.5 2.1 7.6 Table 5.2: The calculated Weibull parameters (at 10 m) for the NL 3 site; after application of a long term correction factor. Finally the variation of the mean wind speed for an averaged day and averaged year is given in the figures 5.5 and 5.6 respectively. As expected there is hardly a daily variation, the figure just shows some scatter, by absence of a daily cycle of warming of the surface. In contrast the annual variation is large relative to an onshore site. These finding are in good agreement with the results given in [2].

27

DOWEC

Fig. 5.5: The hourly variation of the mean wind speed (at 10 m) for the NL3 site.

Fig. 5.6: The monthly variation of the mean wind speed (at 10 m) for the NL3 site.

28

Wind and wave conditions The figures 5.4-5.6 shows the annual variation of the mean wind speed; below the annual variation of the Weibull scale and shape parameters are given. Please note that the parameters are now determined from a limited data set (i.e. per year) so, as noted before, the uncertainty in the determined values is also larger.

Fig. 5.4a: The annual variation of the Weibull scale parameter.

Fig. 5.5a: The hourly variation of the Weibull scale parameter.

Fig. 5.6a: The monthly variation of the Weibull scale parameter. 29

DOWEC

Fig. 5.4b: The annual variation of the Weibull shape parameter.

Fig. 5.5b: The hourly variation of the Weibull shape parameter.

Fig. 5.6b: The monthly variation of the Weibull shape parameter. 30

Wind and wave conditions

6. Weather windows Specification of weather windows is important information for installation and O&M operations. Again on basis of the 9 full years of the NEXT database the weather windows are determined. Please note these weather windows are based on the significant wave height and/or mean wind speed and not on visibility, since the latter is not included in the NEXT database. In table 6.1 the percentage of time is specified for which the significant wave height Hs is below several values during uninterrupted time intervals (no restriction on the mean wind speed). Since the NEXT data concerns 3 hour data, time intervals are taken of 3, 6, 9, 12 and 15 hours (i.e. 1, 2, 3, 4 and 5 succeeding values of Hs lower than the specified levels). The windows are given for each season separately: winter (Dec, Jan, Feb), spring (Mar, Apr, May), summer (Jun, Jul, Aug) and autumn (Sep, Oct, Nov). In order to obtain the percentage of time an operation of, say, 6 hour is possible, the values for 6, 9, 12, 15 and > 15 should be added. At first one may think that these tables should be monotonically increasing; i.e. the percentage of time should increase for higher values of the threshold for the significant wave height. In fact that is in general only true considering the summation over all time intervals, and not for some particular time interval. As example, from table 6.1 it can be noticed that, above a value for Hs of 1.5 m, the percentage of time for a 3 hour interval decreases for increasing Hs. Apparently, for higher thresholds the intervals become longer and are counted in the intervals of 6 hours (or longer).

31

DOWEC

Winter Uninterrupted time interval (h) 3 6 9 12 15 > 15 Spring Uninterrupted time interval (h) 3 6 9 12 15 > 15 Summer Uninterrupted time interval (h) 3 6 9 12 15 > 15 Autumn Uninterrupted time interval (h) 3 6 9 12 15 > 15

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

NL3 < 1.25

< 1.5

< 2.0

< 2.5

CONFIDENTIAL

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

NL3 < 1.25

< 1.5

< 2.0

< 2.5

CONFIDENTIAL

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

NL3 < 1.25

< 1.5

< 2.0

< 2.5

CONFIDENTIAL

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

NL3 < 1.25

< 1.5

< 2.0

< 2.5

CONFIDENTIAL

Table 6.1: The weather windows concerning the significant wave height for the NL3 site.

32

Wind and wave conditions The above is an appropriate way to represent weather windows with respect to analysis of scour protection. Concerning O&M operations it makes more sense to consider the cumulative values as given in table 6.2. Weather window (percentage of time) NL3 Means of transport / operation 1 2 3 4 5 Significant wave height limit (m) 0.75 1.50 2.00 3.00 NA Wind speed limit (m/s) - 1 hour average NA 10.00 11.50 15.00 20.00 Uninterrupted time interval (h) >= 3 >= 6 >= 12 >= 24 >= 33 >= 42 CONFIDENTIAL >= 54 Weather window (percentage of time) – SUMMER (Apr.- Sept.) Uninterrupted time interval (h) >= 3 >= 6 >= 12 >= 24 >= 33 >= 42 CONFIDENTIAL >= 54 Weather window (percentage of time) - WINTER (Oct. - March) Uninterrupted time interval (h) >= 3 >= 6 >= 12 >= 24 >= 33 >= 42 CONFIDENTIAL >= 54 Table 6.2: The weather windows concerning the significant wave height and mean wind speed for the NL3 site.

33

DOWEC

A distinction is made between winter (Oct-Mar) and summer (Apr-Sep) period. Here it is assumed that the maximum required time for any particular O&M action (including travel time) is 54 hours. Note that the overall accessibility of a given access system is given by windows larger or equal 3 hours, since the NEXT data are given in a 3-hour interval. As stated earlier, the weather windows for the helicopter are based on the wind speed only; i.e. fog is not included. The operational limits used for table 6.2 correspond to the following access systems: No 1 2 3 4 5

Access system

Significant wave height (m) Fictitious 0.75 Rubber boat, jump onto 1.5 ladder Offshore Access System 2 (OAS) Offshore Access System + 3 (optimistic assumption) Helicopter NA

Mean (1-hour) wind speed (m/s) NA 10 12 15 20

Table 6.3: Preliminary wind speed and wave height criteria for different access systems. Please note that the operation limits are still subject to discussion, so the here presented statistics of weather windows are merely examples. In order to study the influence of these operational limits on the availability of an offshore wind farm a fictitious access system is added as well as an (very) optimistic assumption for an OAS. For the final values of the operational limits of the considered means of transport / lifting equipment as well as the corresponding weather windows, reference is made to [9]. By the way, the limits of the special purpose vessel Svanen coincide with the limits for the rubber boat. The weather windows from table 6.2 are related to the percentage of time of the considered time period (here 9 years). Another way to present weather windows is to consider the distribution of the length of the windows (corresponding to the operational limits of some access system). An example of such a distribution function of length of weather windows (concerning a total year) for the OAS is given in figure 6.4 in Annex I (i.e. for the NL7 site). One may also consider the distribution of the length of the weather windows for which the operational limits are exceeded (dashed line in fig. 6.4 of Annex I). These 2 contributions (per means of transport / access system; and if desired per season) are further processed by ECN in other to determine the waiting times, [9]. The weather windows as expressed in table 6.2 make sense in case multiple crews will work in shifts of 24 hours a day to conduct maintenance operations. In case the

34

Wind and wave conditions maintenance activities are interrupted for going to shore to stay overnight another analysis is required. For such a situation it is possible to extract the number of days certain limits are not exceeded during daytime. At the NL3 site there are 258 ± 31 days in a year for which the significant wave height is less than 2.5 m and a wind speed below 10 m/s during daytime (weather window of at least 9 hours). In table 6.4 the annual variation is shown. Number of days Significant wave height limit (m) Wind speed limit (m/s) 1 hour average Year 1977 1978 1979 1989 1990 1991 1992 1993 1994 mean standard deviation

NL3 1.5 2.5 10

10

CONFIDENTIA L

Table 6.4: The number of days for which the significant wave height and mean wind speed are (for at least 9 hours) below (or equal) the specified levels during daytime; for the NL3 site. From the table it follows that the number of days for the limits Hs 15 Summer Uninterrupted time interval (h) 3 6 9 12 15 > 15 Autumn

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

NL7 < 1.25

< 1.5

< 2.0

< 2.5

CONFIDENTIAL

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

NL7 < 1.25

< 1.5

< 2.0

< 2.5

CONFIDENTIAL

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

NL7 < 1.25

< 1.5

< 2.0

< 2.5

CONFIDENTIAL

Weather windows - % of time Significant wave height Hs (m) < 0.5 < 0.75 < 1.0

Uninterrupted < 1.25 < 1.5 < 2.0 time interval (h) 3 6 9 12 CONFIDENTIAL 15 > 15 Table 6.1: The weather windows concerning the significant wave height for the NL7 site.

57

NL7 < 2.5

DOWEC

Weather window (percentage of time) NL7 Means of transport / operation 1 2 3 4 5 Significant wave height limit (m) 0.75 1.50 2.00 3.00 NA Wind speed limit (m/s) 1 hour average NA 10.00 11.50 15.00 20.00 Uninterrupted time interval (h) >= 3 >= 6 >= 12 >= 24 >= 33 >= 42 CONFIDENTIAL >= 54 Weather window (percentage of time) - SUMMER (Apr.- Sept.) Uninterrupted time interval (h) >= 3 >= 6 >= 12 >= 24 >= 33 >= 42 CONFIDENTIAL >= 54 Weather window (percentage of time) - WINTER (Oct. - March) Uninterrupted time interval (h) >= 3 >= 6 >= 12 >= 24 >= 33 >= 42 CONFIDENTIAL >= 54 Table 6.2: The weather windows concerning the significant wave height and mean wind speed for the NL7 site.

58

Wind and wave conditions

Fig. 6.1: Scatter diagram of wind and waves for the NL7 location. Number of days Significant wave height limit (m) Wind speed limit (m/s) 1 hour average Year 1977 1978 1979 1989 1990 1991 1992 1993 1994 mean standard deviation

1.5

NL7 2.5

1.5

2.5

10

10

NA

NA

CONFIDENTIAL

Table 6.4: The number of days for which the significant wave height and mean wind speed are (for at least 9 hours) below (or equal) the specified levels during daytime; for the NL7 site.

59

CONFIDENTIAL

Figure 6.2 The distribution of the significant wave height of the NL7 site.

61

Wind and wave conditions

DOWEC

120

100

fict / sum RB / sum

percentage of time

80

OAS / sum fict / win

60

RB / win OAS / win

40

OAS+ / win 20

heli / win

0 0

10

20

30

40

50

60

weather window (hour)

Fig. 6.3: Percentage of time that the operational limits (maximum significant wave height and wind speed) are not exceeded for several access systems versus the (minimum) length of the (uninterrupted) time interval – graphical representation of table 6.2.

Fig. 6.4: Cumulative distribution functions of the weather windows for the OAS (total year) at the NL7 location. Solid line: CDF of weather windows which do not exceed the operational limits; dashed line: CDF of windows which exceed limits.

62

Wind and wave conditions

63

DOWEC

Annex II: TURBULENCE AND WIND SHEAR A LITERATURE STUDY AND MEASUREMENTS R.R. Nino and P.J. Eecen, ECN

64

Wind and wave conditions 1. INTRODUCTION The goal of the research described in this report is to find all available data regarding turbulence and wind speed profile for the two DOWEC locations in the North Sea near the Dutch coast rather than developing a theoretical model describing turbulence and wind shear. Therefore literature is searched for measured data and in addition the only available offshore turbulence data of the Dutch coast has been investigated. 2. TURBULENCE INTENSITY 2.1 QUOTES FROM LITERATURE West Sole The measurements at West Sole show that the turbulence intensity offshore is considerable lower than over land. Above a wind speed of approximately 10 m/s mechanical turbulence dominates over convective turbulence. In that range the turbulence intensity is described well by a simple expression when using the correct roughness length according to the Charnock relation. Typically, a turbulence intensity of 8-10% is found at all heights. The lowest levels show higher turbulence levels, but these levels are disturbed by the West Sole main deck. [1]

Figure 1. Overview of meteorological stations used in the Measuring Network North Sea. West Sole station is clearly indicated.

65

DOWEC

Figure 2. Description of the meteorological mast at the West Sole Platform.

66

Wind and wave conditions

Figure 3. 3-min average (I3) and 60-minute average(I60) turbulence intensities versus wind speed at the measuring heights at West Sole station. Data from June 1983 to May 1984.

Vindeby

67

DOWEC Relationships between wind speed, turbulence and fetch are highly complex. Minimum turbulence intensity offshore is associated with wind speeds of about 12m/s. At lower wind speeds, stability effects are important while at higher wind speeds wind and wave interactions appear to dominate. On average, turbulence intensity offshore at 50m height is approximately 0.08 if no coastal effects are present. However, the effect of the coastal discontinuity persists in wind speed and turbulence characteristics for considerable distances offshore. This distance is estimated here as approximately 40 km although the majority of the adjustment of turbulence appears to occur within 20 km of the coast. In contrast, coastal effects persist in wind speed measurements to 50 m height over distances of at least 40 km and further in stable conditions. [2]

Figure 4. Location of monitoring and reference sites in Denmark.

68

Wind and wave conditions

Figure 5. Variation of turbulence intensity with wind speed at Vindeby SMW.

69

DOWEC Figure 6. Vertical profile of turbulence intensity offshore.

Figure 7. Mean turbulence intensity by direction at SMW.

Figure 8. Location of the masts (triangles) and wind farm (crosses) at Vindeby.

70

Wind and wave conditions

Figure 9. Relationship between turbulence intensity and fetch. Horns Rev The turbulence intensity decreases with increasing wind speed to about 12 m/s and then begins to increase slightly at higher wind speeds. The increase in turbulence intensity is more pronounced at level 15 m, probably due to the response to increasing wave height. The observed increase in turbulence level with declining height is as expected, however, there is a clear tendency towards higher turbulence intensities at 15m than might be expected assuming a linear relationship as in [2]. In agreement with [2] the profile indicates that turbulence intensity continues to decrease with increasing height. [3]

Figure 10 Geographical layout of the wind farm and the measurement systems at Horns Rev. 71

DOWEC

Figure 11. the meteorological mast at Horns Rev is a square cross section lattice raised on a 1700mm diameter monopile. All heights refer to DNN.

Figure 12. Observed wind speeds averaged by directions at four heights. Data from Horns Rev May 1999 to January 2000.

72

Wind and wave conditions

Figure 13. Variation of turbulence intensity with wind speed. Turbulence intensity is the weighed average for all direction sectors at each level. Data from Horns Rev May 1999 to January 2000.

Figure 14. Variation of turbulence intensity with direction sector. The turbulence intensity is the weighed average for all wind speeds bins at each level. Data from Horns Rev May 1999 to January 2000. * Based on a huge amount of 10-minute shallow water offshore wind statistics, originated from Vindeby and the Gedser experiments, the variability of the standard deviation of the horizontal wind speed is quantified in terms of a set of probabilistic models conditioned on the mean wind speed. Measured distributions are parameterised by fitting to a threeparameter Weibull distribution. Combining a simple heuristic load model with the parameterised probability density functions of the turbulence standard deviations, an empirical offshore design turbulence intensity is determined:

TI

D

= 0.0032 U10 + 0.0234 +

0.3807

U

10

73

[4]

DOWEC * The turbulence standard deviation depends on mean wind speed, upstream conditions, measuring height and thermal convection. Defining a population of turbulence standard deviations, at a given measuring position, uniquely by the mean wind speed, variations in the upstream conditions and the atmospheric stability will appear as variability in the turbulence standard deviation. Distributions of such turbulence standard deviations, conditioned on the mean wind speed, are quantified by fitting the measured data to logarithmic Gaussian distributions (instead of 3-parameter Weibull distributions as in [4]). Combining a simple heuristic load model with the parameterised probability density functions of the turbulence standard deviations, an empirical offshore design turbulence intensity is determined:

TI

2

D

= −0.0002 U10 + 0.0123 U10 − 0.1423 +

1.5389

U

[5]

10

The expressions from references [4] and [5] were derived based on the same data sets (Vindeby and Gedser). The recommendations from these two proposals are compared in Figure15 for mean wind speeds corresponding to a conventional operational mean wind speed regime. In addition the reccomendation for on-shore sites originated from IEA-61400 is also shown for the lowest possible turbulence specification.

- - - - IEA-61400 onshore recommendation ___equation from reference [4] - - - equation from reference [5] Figure 15. Comparison of design turbulence intensities.

As seen, the proposal from [4] under estimates the design turbulence intensity for mean wind speeds below 10 m/s where the turbulence contribution from thermal convection becomes important. However, in this mean wind speed regime usually a minor part of the total fatigue loading is experienced due to the relative low wind

74

Wind and wave conditions speed. For mean wind speeds above 10 m/s the predictions from the two proposals are rather similar. Comparing with the recommendations given by IEA 61400 for on-shore siting it appears that the present proposed design turbulence intensity for (shallow waters) offshore applications qualitatively exhibits the same functional behaviour with the mean wind speed. However, quantitatively it presents a considerable reduction in the fatigue design load.

75

DOWEC 2.2 DATA 2.2.1 ECN DATA KNMI is not able to provide the 10-minute averaged wind speed together with its standard deviation for the concerned time intervals. ECN has a measurement mast of 50 m at the coast, located in the dunes. From the upper anemometer an indication of the turbulence at the North Sea is obtained by considering the western sector (between 240 and 300 degrees) for the available data from June, 1 to September, 30. Figure 16 and 17 show that the turbulence in that specific sector is quite different from the turbulence of all sectors together.

Figure 16. Turbulence intensity at 50 m measured at the ECN metmast considering all sectors.

Figure 17. Turbulence intensity at 50 m measured at the ECN metmast considering sectors from 240 to 300 degrees. 76

Wind and wave conditions

Table 1. Main statistical values for the measuring period June 2001 to September 2001 at ECN metmast at 50 m considering sectors between 240 and 300 degrees. Bin with width of 1 m/s around 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Mean Wind speed (m/s)

Turbulence Intensity (%)

Standard deviation (%)

5.00 6.05 7.01 7.99 8.94 10.01 10.98 11.97 13.02 13.93 15.09 15.98 17.01 17.90 18.89 19.92 20.86

10.1 9.7 9.6 9.0 8.2 8.4 7.9 8.4 8.3 8.1 8.6 8.4 8.3 8.5 9.1 9.7 10.5

3.0 2.5 2.5 3.0 2.2 2.8 1.8 2.4 1.7 1.4 1.6 1.4 1.3 1.4 1.6 1.5 1.6

Characteristic turbulence intensity (%) 13.2 12.2 12.2 12.0 10.4 11.2 9.7 10.7 10.0 9.6 10.2 9.8 9.6 9.9 10.7 11.2 12.1

ECN mast at 50m (240-300)

Turbulence intensity (%)

14.0

Characteristic turbulence intensity

12.0

Turbulence intensity

10.0 8.0 6.0 4.0 2.0 0.0 0

5

10

15

20

25

Wind speed (m/s)

Figure 18. The turbulence intensity and the characteristic (mean plus standard deviation) turbulence intensity versus wind speed for the ECN metmast data considering sectors between 240 to 300 degrees. 77

DOWEC

The measured turbulence intensity resembles the characteristic shape of other offshore turbulence measurements. There is a minimum between 10m/s and 17 m/sand at lower and higher wind speeds the turbulence increases as it was expected.

78

Wind and wave conditions 2.2.2 HORNS REV DATA The Horns Rev site is located at a reef approximately 14 km off Jutland in the North Sea.The water depths at the erection site for the wind farm vary from 6 to 12m.The site is characterised by a large undisturbed over water fetch in the western sectors towards the North Sea. Figure 19 shows the turbulence intensity measurements at Horns Rev at 62 m height. It is important to mention that raw data was not available. I 0.3

Turbulence

0.2

0.1

0 0

5

10

15

20

25

30

35

40

45

Mean wind speed

Figure 19. Turbulence intensity measurements at Horns Rev, h=62m, No. of 10min values 7500.

Table 2. Main statistical values for the measuring period May 1999 to January 2000 at Horns Rev. Height Mean wind speed (m/s) Max. wind speed (m/s) Mean turbulence intensity (%)

62 m

45 m

30 m

15 m

9.7

9.2

8.8

8.2

45.4

43.1

40.7

39.5

8.8

9.7

10.1

11.9

79

DOWEC

Turbulence Intensity at Horns Rev

Height (m)

100

10 8

9

10

11

12

13

Turbulence Intensity (%)

Figure 20. Turbulence intensity as a function of height at Horns Rev.

Windshear at Horns Rev

Height (m)

100

10 8

8,5

9

9,5

Wind speed (m/s)

Figure 21. Win shear at Horns Rev.

80

10

Wind and wave conditions

3. WIND SHEAR 3.1 QUOTES FROM LITERATURE * Wind speed profiles offshore should conform to logarithmic profiles and hence be easy to predict. The fact that they do not always do so can be ascribed to four factors: -

flow distortion in the measurements caused by the mast structure the presence of internal boundary layers non-neutral stability conditions effects of wind/wave/roughness on the wind speed profile

It is worth noting that changes in the state of the sea surface should not lead to distortions of the profile away from logarithmic predictions but to changes in roughness which can not be predicted using the Charnock equation relating roughness to friction velocity. At all the offshore sites, the observed wind speeds at 48 m are higher than those predicted using 10 m wind speeds and the logarithmic profile. This could be due to atmospheric conditions being slightly stable offshore – however further analysis is required to clarify this point. Since offshore wind speed profiles are affected by both stability and roughness it appears that a model which takes account both effects is required for their accurate prediction. Internal boundary layers develop in response to discontinuities in surface conditions (roughness, temperature and humidity changes). Figure 22 shows averaged IBL plotted by fetch for different stability conditions. As expected the IBL grows most slowly in stable conditions when its development can be expected to have the most impact on near-surface wind speed profiles. Mean IBL heights by direction range between 300 and 800 m suggesting that on average the IBL is not a major influence on the wind speed profile at the site. However, minimum IBL heights in strongly stable conditions can be as low as 20 m.

81

DOWEC

Figure 22. Average Internal Boundary Layer heights (m) by fetch distance (km) for different stability classes. Fetch and stability have a strong influence on the wind speed profile which can be identified by calculating the ratios of wind speed at 50 m to those at 10 m. Figure 23 illustrates these relationships by fetch distance. In unstable conditions, the wind speed profile adjusts more rapidly to the new surface conditions and the increase of the ratio with fetch reaches an equilibrium value much quicker. Conversely in stable conditions the ratio continues to increase with fetch to well beyond 50 km. Hence the near-surface wind speed profile is continuing adjusting even at considerable distances from the coast. [6]

82

Wind and wave conditions

Figure 23. Ratios of wind speeds at 50 m to those at 10 m averaged by fetch distance and different stability classes.

* 83

DOWEC

Measurements of the wind speed error resulting from the influence of a lattice tower on the flow were analysed showing the expected decrease in wind speed upstream of the tower and speed-up effects on both sides of the tower. The magnitude of the error on the wind speed was from less than one percent to about five percent depending on the size of the boom relative to the side length of the tower. One important result was that the error averaged over 360 degrees was always positive and not zero as could be assumed for a first approximation. The vertical extrapolation of corrected and uncorrected wind profiles from 7-30m to 60 m leads to approximately the same results. Using one wind speed (at 7 m) and different estimates for the roughness length for the extrapolation leads to differences up to 7% at 60 m. If we were to take into account the deviations from the logarithmic profile above 30 m also, we could encounter even larger differences. The corrections were used to correct measured wind speeds from the RASEX experiment at Vindeby and the results show that the influence of stability on the corrected profiles is consistent with the expressions that were developed over land, except for very stable conditions. For neutral conditions deviations were seen for heights above 25-30 m. One main conclusion was that even with flow distortion corrected wind profiles, extrapolation of measured wind speeds to higher levels must be done very carefully. See Figure 24.[7]

Figure 24. Nondimensional velocity profiles for long (solid) and short (dashed) fetch situations at Vindeby (RASEX experiment).

84

Wind and wave conditions

4. CONCLUSIONS For the long fetch offshore turbulence the ECN metmast data is used for the sector in which the wind is coming directly from the sea (240-300 degrees). Using the measurements at 50 m ensures that the data is only little corrupted by the sea-land transition and the sector is chosen such that is not influenced by the dunes or structures around. Note: the ECN data does not cover a complete year. The turbulence intensity at 50 m height is around 8% for a large wind speed interval from 9 m/s to 18 m/s, below and above this interval the turbulence increases. The dependency of the turbulence on the distance to the shore is determined from Vindeby data at 50 m height, (see Figure 9). Summarised the results approximately indicate the following:

Distance to shore 5 km 10 km 20 km 40 km

the Turbulence intensity 9.8% 9.0% 8.3% 8.0%

When the location is known offshore, for all sectors the distance to the shore is known and therefore the turbulence intensity can be determined at 50m height. From Figure 14 the variability of the turbulence intensity over the sectors is shown for Horns Rev, which is revealed to vary considerably. The dependency of the turbulence intensity with the height can be determined from the Horns Rev data (Figure 20). Also at Vindeby the data was extrapolated (Figure 6), however a linear fit seems unrealistic. A logarithmic fit with height seems more appropriate. For the description of the wind shear there is little data available. Mostly the logarithmic profile for the wind shear is applied based on neutral atmospheric conditions. Comparing observed and predicted data from the Rodsand station [2] it is claimed that the wind speed at higher altitudes tends to be larger than is expected using a logarithmic profile. At the same time is verified that integrating the stability corrections improves the prediction. It must be noted that these measurements are available only in the surface layer. Extrapolation to higher altitudes is therefore based on theory and is unreliable.

85

DOWEC

5. REFERENCES [1] DESCRIPTION OF THE NORTH SEA WIND CLIMATE FOR OFFSHORE WIND ENERGY APPLICATIONS. TNO Institute of Environmental and Energy Technology. 1991 [2] MONITORING OFFSHORE WIND AND TURBULENCE CHARACTERISTICS IN DENMARK R. J. Barthelmie. BWEA 1999 [3] EVALUATION OF THE STAND-ALONE WIND AND WAVE MEASUREMENT SYSTEMS FOR THE HORNS REV 150 MW OFFSHORE WIND FARM IN DENMARK Soren Neckelmann. OWEMES 2000 [4] DESIGN OFFSHORE WIND CLIMATE Gunner C. Larsen, Hans E. Jorgensen. EWEA 1999 [5] OFFSHORE FATIGUE DESIGN TURBULENCE Gunner C. Larsen. 2000 [6] OFFSHORE WIND RESOURCES AT DANISH MEASUREMENTS SITES R.J. Barthelmie, M.S. Courtney. EWEA 1999 [7] VERTICAL EXTRAPOLATION OF OFFSHORE WIND PROFILES J. Hojstrup. EWEA 1999

86

Wind and wave conditions

Annex III: DESIGN WIND CLIMATE J.M. Peeringa, ECN The design wind climate given below is for the NL3 site and based on the main text of this report [1]. The data is for a 10 m reference height and 1 hour mean wind speeds. Only the total data is used for the design climate, so effects of the wind direction are not included. However the transformation methods described below are also applicable for the data of a certain wind sector. For the design load calculations information about the wind climate at hub height is necessary. This section describes the way the wind climate at hub height is determined using the wind data at 10 m height. For the design load calculations the following is of importance: - the vertical wind shear - the turbulence intensity at hub height - the Weibull distribution at hub height (annual average and shape factor k) - the extremes at hub height Vertical Shear For (neutral) offshore conditions, the vertical wind shear can be modelled by a logarithmic wind profile. For this wind profile the roughness z0 should be known. At the Hydra workshop on November 8 2001 Verkaik [2] presented a table with wind speed depended roughness z0 for the North Sea. See column 2 in the table below. Using this table and the logarithmic wind profile, the wind speed at hub height can be determined. Windspeed [m/s] North Sea z0 [m] at 10 m 5 0.0001 10 0.0006 15 0.0016 20 0.0034 25 0.0059 30 0.0095

z0 [m] Charnock β=0.032 0.0001 0.0005 0.0016 0.0033 0.0059 0.0097

z0 [m] Charnock β=0.0185 0.00006 0.0003 0.0009 0.0019 0.0034 0.0056

Turbulence intensity Offshore the turbulence intensity is on average smaller than on land and decreases with height. An increase of the turbulence intensity for higher wind speeds is observed offshore. Nino and Eecen [3] discuss the turbulence intensity offshore. Based on reference [3] a characteristic turbulence intensity I15 = 9 % at 15 m/s wind speed is selected. The characteristic turbulence intensity I15 consists of 7 % turbulence intensity and a 2 % standard deviation. For a range of wind speeds the characteristic turbulence intensity is modelled as follows: - According to IEC with I15 = 9 % and a = 5 for wind speeds ≤ 15 m/s - Using I= 1/ln(z/z0) for wind speeds > 15 m/s with z = hub height and z0 = roughness 87

DOWEC It should be noted, that I= 1/ln(z/z0) is determined for a roughness z0 corresponding with the wind speeds at hub height. The turbulence intensity given here is the ambient turbulence. Wake effects of the wind farm are not included. Weibul distribution The wind speed distribution at hub height is described by a Weibull distribution. Reference [1] presents the Weibull distribution at 10 m height. The shape factor k and the annual average wind speed define the Weibull distribution. The annual average wind speed at hub height is determined according to reference [4] by transforming the annual average wind speed at 10 m. to hub height for a roughness z0 = 0.0002 m. using:

 ln( z hub / z 0 )   a = a ref   ln( z / z )  ref 0   For the shape factor k no height correction is applied. See reference [4]. Extremes For the calculation of the extreme loads the extreme wind speeds should be known. In the code IEC 61400-1 second edition the reference wind speed Vref is defined as the 10 minute mean wind speed at hub height with a return period of 50 years. Using the Ness data base, reference [1] estimates the one hour mean extremes at 10 m height for the selected sites. The 50 year one hour mean wind speed at 10 m height (26.7 m/s) is transformed to hub height using a roughness z0 = 0.008. The one hour mean (extreme) wind speed is transformed to a 10 minutes mean using reference [5] (1 hour mean = 0.916*10 minute mean). The relation between the (extreme) reference wind speed and the annual average wind speed can be expressed by the ratio of the two wind speeds. In the IEC 614001 code the ratio Vref/Vave is 5. Reference [6] page 1sub C.14 figure 2.1 shows that for shape factor k=2.1 the ratio is 4. In reference [7] the annual average and the extreme mean wind speed are given for several sites. The table below shows the wind speeds and the ratio Vref/Vave for these sites and the DOWEC site NL 3. Site UK-1 the Wash NL-1 IJmuiden NL-5 IJmuiden DK-1 Hornsrev DE-1 Rostock S-1 Blekinge NL-3 Bierbooms NL-3 IEC 61400

Reference speed 41.5 41.5 41.5 43 40.5 43 38.2 46

wind Annual average 8.2 9 9.5 9.2 7.8 8.4 9.2 9.2 88

Ratio Vref/Vave 5.1 4.6 4.4 4.7 5.2 5.1 4.2 5

NL-3 [6]

reference 36.8

9.2

Wind and wave conditions 4

It can be seen that the sites at the eastern part of North Sea (NL-1, NL-3, NL-5 and Horns Rev) have a ratio smaller than 5. Based on this information it was decided to select a ration of 4.5 for NL-3. Using the annual average wind speed at hub height the reference wind speed is determined using this ratio. The 10 minute wind speeds are converted to a 3 s gust according to reference [5]. The 10 minute wind speed is increased by 25%. Design Wind Climate Using the transformation methods as described above, the design wind climate at hub height is given in the tables below.

Parameter at hub height Average wind speed [m/s] Weibull shape factor Ambient characteristic turbulence intensity Air density [kg/m3] Hill slope Wind shear exponent Extreme 1 year wind speed (10 min. average) Ve1_10min= Ve1/1.25 Extreme 1 year wind speed (3 sec. average) Ve1=0.75 Ve50 Extreme 50 year wind speed (10 min. average) Vref Extreme 50 year wind speed (3 sec. average) Ve50= 1.25 Vref Ambient characteristic turbulence intensity Wind speed at hub Ambient characteristic height [m/s] turbulence intensity 4 0.131 6 0.123 8 0.107 10 0.098 12 0.094 14 0.091 16 0.090 18 0.092 20 0.094 22 0.096 24 0.098 Reference 89

Value 9.2 2.1 See table 1.225 0o 0.11 31.1 38.9 41.5 51.9

DOWEC [1] Main text of this report [2] J. Verkaik, Geografische interpolatie van windvelden (KNMI) Presentation Hydra Workshop november 8 2001 http://www.knmi.nl/onderzk/index-nl.html [3] Annex II [4] J.P. Coelingh, Wind and wave data compiled for the DOWEC concepts study DOWEC-T56-JC-00-R01_F-P.doc or IW-00162R [5] DNV/Risoe, Guidelines for design of wind turbines [6] Pierik, J.T.G, Dekker, J.W.M, European Windturbine Standards II [7] Opti Owecs, report A typical design solution for an OWECS

90