. . . . . . . . . . .

Centre for Renewable Energy Sources 19th km Marathon Avenue GR-190 09 Pikermi, Greece Τel: +30 210 66 033 00 e-mail: dfousek@cres.gr
http://www.cres.gr/windrose

A wind data analysis tool

.

.

WindRose
. . .
User’s Guide

.

.

.

.

.

N NNW NW NNE NE

WNW

ENE

W

E

WSW

ESE

SW SSW S SSE

SE

Document version: 8.1 Excel COM+ Add-In version: 3.88 Analysis DLL version: 5.89

Table of Contents
Introduction ............................................................................. 1 Installation ............................................................................... 3 Running the program

Running WindRose for the 1st time ..................................... 4 Registration / Demo version ............................................... 5
The Input worksheets

Input ................................................................................... 6 Multiple anemometers - vanes (within the same mast) ........... 9 PowerCurve ........................................................................ 13
Quick Check of data ............................................................. 15 The Output worksheets

Results ................................................................................ 16 Windrose ............................................................................ 20 Shear................................................................................... 21 AnnexK .............................................................................. 25 TimeCharts ......................................................................... 26 Tables ................................................................................. 27 3D ....................................................................................... 28 12pie, 12diurnal.................................................................. 29 BarCharts, Weibull ............................................................. 30 UhourT, UhourG, DIRhourT, DIRhourG ........................... 32 WTprodT, WTprodG.......................................................... 33 TempT, TempG, SRadT, SRadG........................................ 34 TempData ........................................................................... 35
Correlation –Prediction of missing data – MCP method

WindCorr worksheet........................................................... 36 Methodology ...................................................................... 38
Appendix

Advanced Options .............................................................. 45 Air-density variation with Height....................................... 47 Weibull distribution methods ............................................. 49 Power Curve correction ...................................................... 52
Installation issues

WindRose.XLS Security warning....................................... 53
References ............................................................................... 55

WINDROSE : Wind Data Analysis software

INTRODUCTION
WindRose is a software tool dedicated to the analysis of wind characteristics (speed, direction, turbulence, temperature). It is not a standalone program, but an Add-In to the Microsoft Excel® 2000/XP/2003/2007, for the Windows 9x/ME/NT4/2000/XP/Vista/7 operating systems. The analysis results are stored graphically and numerically into spreadsheets, which can be further used as ordinary Excel files. The program is designed to provide all the results of the data analysis, in a customisable form to meet any particular needs. Thus, the user can rearrange all the graphs, resize them, change their colours, copy or link them to other sheets or programs (i.e.: embedded links to Microsoft Word® document), create new tables using the numerical results, etc. Data analysis complies with the requirements imposed by the IEC and MEASNET standards.
Main features

It performs complete statistical analysis of the wind data, including Weibull distribution constants (per direction sector and global), turbulence intensity evaluation and polar plots (wind roses) of the time and energy distribution of the wind. It correlates data from two sites calculating correlation coefficients globally and per ranges of wind speed and direction. As a result, it provides the predicted time-series for the missing data of a site, based on the other site’s complete set of data (MCP method). Where multiple anemometers and/or wind vanes are present, the vertical wind shear is calculated (per wind speed and per direction sector). Thus, the extrapolation of the wind speed at higher heights (ie: hub height of a wind turbine) is performed more accurately. A dozen of power-curves from a variety of wind turbines are included, providing a good estimate of the expected energy production.

PAGE 1

WINDROSE : Wind Data Analysis software

It has a user configurable time step (10 minutes, 1 hour, etc) and it is able to analyse huge data sets (limited only by the available computer memory). In case of measurements with systematic errors, a linear correction can be applied for all the measured quantities and separately for each file. It includes 3 methods for the air-density correction (due to the site elevation height and to the temperature), which is critical for the correct calculation of the wind energy. It produces monthly charts and tables with the per hour variation of wind speed, wind direction, expected wind turbine’s power, temperature and solar radiation.

How it works

WindRose reads ASCII files containing columns of data, an output format supported by the majority of data-loggers. Five 1 columns of data are necessary for the program to run: wind speed, wind direction, standard deviation of the wind speed, time and date. Several formats are supported when dealing with date and time. Whenever temperature atm. pressure, humidity, solar radiation, flow inclination are recorded, the appropriate analysis is performed.

1 If the wind speed’s standard deviation is not recorded, the program can still run, using the following trick: provide the same column numbers for both U (wind speed) and Usdv and then ignore the Turbulence Intensity results (which will be constant to: 100%).

PAGE 2

WINDROSE : Wind Data Analysis software

INSTALLATION
Assuming that Microsoft Excel is properly installed, WindRose can be installed using the SETUP.EXE program. Administration privileges are required in Windows NT4/2000/XP. At the end of the installation, the WindRose option is added in the main menu of the Excel (Figure 1). By opening the WindRose.XLS file (default location: c:\Program Files\WindRose), you are ready to run the program for the first time (see below RUNNING THE PROGRAM).

Running WindRose within Excel.
Digital Signature

Microsoft Excel has security levels that allow users to run macros based on whether or not they are digitally signed by a trusted macro developer. The WindRose.XLS file includes some macros and has a digital signature named WindRose. The first time that the WindRose.XLS file is opened, the user is asked whether or not the macros should be enabled. Answer yes so that WindRose runs properly and never asked in the future the same question. If you are not allowed to open the WindRose.XLS file with the included macros enabled (might happen if the Security Level of Excel is set to: High), then consult the Appendix: INSTALLATION ISSUES.

PAGE 3

8. “Run” 2 . 3D 12.TXT and GenASCII. WtprodT 17. The content of all the worksheets is explained in detail in the following paragraphs.TXT) containing wind data are included in the package to facilitate the first run of the program. 23.WINDROSE : Wind Data Analysis software RUNNING THE PROGRAM Running the program for the first time Three sample data files (ST-FLASH.XLS. by saving it at the end of the analysis (using the Save As… option) with another name. 2 3 During the program execution and under certain circumstances. 6. UhourT 10. all the required information to run the wind data analysis for the specific data files. WindRose has to be run twice. the results of the analysis are stored in forms of graphs and/or tables into the different worksheets. 5. once per each site. DIRhourG DIRhourT TempG TempT SradG SradT TempData Air-Density Two worksheets (Input and PowerCurve) are used for defining all the necessary parameters to run the program. Input PowerCurve Results WindCorr Tables Weibull Upolar UhourG 9. 24. Moreover. All the user has to do is. 18. PAGE 4 . The status bar of the Excel should then display the different steps of the program’s execution3 . 2. NOMAD. 22. “WindRose” and then. 4. to select from the Excel Main Menu. The remaining worksheets are used for presenting graphically and numerically the results of the analysis. the PC may not respond for a few seconds. 7. 3. the Input worksheet contains already. At the end. keep it unchanged for future use. BarCharts 15. Before calculating the correlations/predictions between 2 sites. 12diurnal 13. 12pie 14.TXT. 19. 21. Note • When working with the original WindRose. then modify accordingly the paths of the 3 provided sample files in the Input sheet. WtprodG 16. If the default program path: c:\Program Files\WindRose was changed during the installation. TimeCharts 11. 20. The WindRose Excel workbook is composed by the following 24 worksheets: 1.

but is limited to analysing measured data that do not exceed 31 days. Registered users are given a code. Security Center) and turn it back on afterwards. therefore some graphs and tables may not display correctly. which is entered using the WindRose menu within Excel. If the PC’s OS is Windows Vista. then before entering the code. can be imported using the “Import…” option of the WindRose menu (see figure below). Import OLD WindRose. PAGE 5 .XLS has some new input parameters and graphs. this version of WINDROSE. turn off the User Account Control (Control Panel.XLS files Excel files created with older versions of WindRose. Regarding older files.WINDROSE : Wind Data Analysis software Demo version Registration The demo version of the WindRose (for unregistered users) incorporates all the features of the full version.

direction etc). etc. Note that wildcards (* and ?) are accepted to specify multiple files File names and column numbers (sheet Input). the signal quality of Lidars (CNR for Windcube. Some of the columns of the ASCII data file can be assigned as control signals. followed by at least 5 columns of data. enabling thus data filtering depending on e. Delimiting characters (within the ASCII data files) It is expected that ASCII files contain some comment lines in the beginning. The acceptable range of each control signal can be assigned in WindRose’s Advanced Options… menu. The accepted characters delimiting (separating) the columns are the following: • The space • The comma (.g. additional memory is needed for that.WINDROSE : Wind Data Analysis software THE INPUT WORKSHEETS Input Four tables compose this worksheet. If the order of the files is not the chronological one (from older to newer data).) • The 2 parentheses • The tab • The double quotation marks (“) PAGE 6 . The number of the lines to skip in the beginning of each file.) • The semicolon (. ?) in their names. However. points in fit for ZephIR). the rain. containing blank lines or comments is also given here. The first table contains the names of the ASCII files to be read together with the column numbers of the requested quantities (wind speed. which deals with the input data and the necessary parameters required by the program. Multiple data files can be specified by using wildcards (*. then sorting of data is performed automatically without any user action.: the data logger’s battery level.

the last two represent the seconds. If the American format is used (mm/dd/yy. when reading ASCII files. ddmmyyyy . Date format WindRose defaults to the European date format. dd-mm-yyyy . etc) then. yyyy.e. I. then depending on the time step they are considered either as minutes. there are cases where time comes in the form of incremental steps from the beginning of the day. Measurements data coming from the NOMAD data logger use another date format.m.e.e. If the “international” date format is used (yymmdd.: for the 10th column we write 10u. the following forms: hh:mm:ss .mm.: for the 26th September 1998 it is written: PAGE 7 .e. In this case we add the character s right after the column number. If 3-4 digits are found then. hh.mm. if just 1-2 digits are found.e. the middle two the minutes and the first one(s) the hour. hh.: 13h50) can be supported on demand.dd. hhmm Note that.: 1:50pm) and the French-style h (i. For example if the time column was the 9th we put 9s. d. etc) then after the column number of the date. are present and 5-6 digits are found then. ddmmyy . right after the column number of the date.: the 12-hour form using am and pm (i. If the date has one of the following (European) formats. if none of the delimiting characters: : . hh:mm .WINDROSE : Wind Data Analysis software Time format WindRose accepts as time format. d/m/yy . Other custom formats i.ss . the time column is filled with integers increasing from 1 to 144 (time step=10min) or from 1 to 1440 (time step=1min) etc. Finally. d-m-yy . mmddyyyy.: 5 is taken as 05:00:00).yy .yyyy .. we put the character i i. .: 5 is taken as 00:00:05). The acceptable delimiting characters for the date format are: / . or as the hour of the day. the last two represent the minutes and the first one(s) the hour of the day. then it is directly read. hhmmss . dd/mm/yyyy dd. but also accepts other date formats.e. if the time step ≥ 1h (i. we put the character u i.: for the 10th column we write 10i. Thus. if the time step < 1h (i.e.e.mm . Finally.mm.

3 12.0 3.2000 12.7 21.1.3 14.4 20.9 21.4 21.4 12.4 19.1 20.8 2.4 13.1 13.37.8 150.9 2.2 3. the above sequence should be considered as one column (instead of 3).3 14.: for the 10th column we write 10c However.9 21.0 3. Additionally.27.9 2.9 2.9 Gust ----22.4 13.1 22.3 142.: for the 10th column we write 10n.5 Gust ----22.8 2.9 2. Below.5 144.2 13.27.7 21. PAGE 8 Lines to skip Lines to skip Time Date Lines to skip Time Date 2 2 2 .1 140.3 148.1 20.6 2.6 146.4 13. WindSpeed --------14.0 146.: 11th in the previous example) ii) should not be the file’s last one and iii) is normally counted when numbering the columns of the ASCII file.1 12.12.8 150.0 2.00 015700 020700 Date -------01/12/00 1/12/00 1/12/0 1-12-2000 1.6 146. Some models of the CAMPBELL data loggers use another particular date format: An integer number is given indicating the number of days elapsed from the beginning of the year.9 14.WINDROSE : Wind Data Analysis software Sep 26.6 146.9 21.6 144.1 22.3 14.9 2.8 12.3 13.5 3.2 13.1 147. the first lines of the date and time columns contain various format types.3 12.3 13.6 21.e.9 2.8 Dir ----137.6 12.1 140.9 21.4 19.3 142.00 12-1-2000 12/1/2000 12012000 12012000 12012000 12012000 12012000 Gust of Wind Speed [m/s] Solar Radiation (W/m2) Wind Direction (°) Wind Speed (m/s) Temperature (°C) Wind Speed SDV 1 2 3 4 5 6u Gust of Wind Speed [m/s] Solar Radiation (W/m2) Wind Direction (°) Wind Speed (m/s) Temperature (°C) Wind Speed SDV WindSpeed --------14.4 20.7 21.00 01.6 144. Thus.9 21.5 Usdv ---2.00 1.8 Dir ----137.6 21.8 2.12.e.6 146.7 2.8 2.5 144.0 3.5 3.3 13.0 3.2000 1.9 Dir ----137.1 20.e.47.3 Usdv ---2.47.1 13.6 12.00 1.00 1-12-2000 1/12/2000 01122000 01122000 01122000 01122000 01122000 Gust of Wind Speed [m/s] Solar Radiation (W/m2) Wind Direction (°) Wind Speed (m/s) Temperature (°C) Wind Speed SDV 1 2 3 4 5 6 WindSpeed --------14. the daycolumn: i) has to be next to the year column (i.8 12.4 12.2 3.4 20.0 146.1 Time -----0:7 0:17:00 0:27:00 Date -------00/12/01 00/12/1 0/12/1 4 Time 1 2 3 4 5 6i Date Intentionally (for better understanding).0 2.9 2.4 20.6 2.6 144. some examples are given showing how the Input sheet should be filled4 .1. 1998 When reading such files. right after the column name we put the character n i.1 Time -----0:7 0:17:00 0:27:00 00:37:00 00:47:0 0057 0107 0117 1.4 13.1 Time -----0:7 0:17:00 0:27:00 00:37:00 00:47:0 0057 0107 0117 01.7 21.7 21.9 14.1 12.1 147.37.3 Usdv ---2.6 146.7 2.1 20.00 015700 020700 Date -------12/01/00 12/1/00 12/1/0 12-1-2000 12. for the 26th September 1998 it is written: 1998.5 Gust ----22.4 21.00 1.6 144.5 144.8 2.6 144. 269 This case is treated by adding the letter c right after the date column number i.5 144.3 148.

WINDROSE : Wind Data Analysis software 14.---------------.1 11.2 222.2000 Mar 1.4 2 1 33 Multiple anemometers and vanes: Example form of Input sheet’s 1st table.4 277 20:04 13.9 21.9 236 18:59 11.3 2.4 20.4 227.18.47.6 37 20.9 2.0 8.1 12.2000 Mar 1. the heights of the anemometers should be set accordingly (in ascending order) in the 2nd table of the Input sheet.00 015700 020700 2000-12-1 2000.6 21.2000 Mar 1.7 10.6 144.5 22.3 3 minute(s) Vane A Wind Van Average 10 minut ° -------222.8 9.---------------1.0 225 19:22 13.4 236. Weibull distributions per height.13.8 250 20:21 14.4 219 19:30 0.0 2.00 1.3 246 20:25 Gust of Wind Speed [m/s] Solar Radiation (W/m2) Temperature (°C) 8 2 1n 13 Multiple anemometers . Furthermore.6 12.2 222.8 9. should be separated by a semicolon .2 226.2 9.5 12.2 250.3 142.2000 ----18:57 19:00 19:10 19:20 19:30 19:40 19:50 20:00 20:10 20:20 Anem A Anem A Anem A Other An Other An Other An Avg Dev Minimum Maximum 10 minut 10 minut 10 minut m/s m/s dir time m/s dir time -------.6 146.2 10.5 144.12.1 251.14.1 243 19:54 12.8 226.4 12.1 225 19:44 12.0 3.1 00.6 211 18:59 0.10.0 3.7 21.4 13.vanes (within the same mast) When multiple anemometers and wind vanes are present. PAGE 9 Lines to skip Time Date Lines to skip Time Date .1 11.3 12.2000 Mar 1.12.8 245 20:18 1.4 20.4 12.2000 Mar 1.8 10.8.e.1 13.2000 Mar 1.000 21..4 11.5 33.7 9.7 2.7 3 11 4 Wind Direction (°) Wind Speed (m/s) Wind Speed SDV TIME ----------Mar 1.0 146. Case with 5 anemometers and 3 wind vanes.7 245 19:59 1.2000 Mar 1.1 140. In that case.2 12.2 3.3 148.4 19.1 20.00 1.2 13.8 144.4 224 19:19 0.9.25 22.6 226 19:17 13.8 150.4 221 19:05 13.16.3 253.6 9.29.27.37.8 2.4 13.2 219 19:04 0.2000 Mar 1.8 12.2000 Mar 1.1 2000-12-01 2000/12/01 20001201 20001201 20001201 20001201 20001201 Site Name: Site Number: Start Time: Finish Time: Total Time: DATE SOMEWHERE 1 18:57 03/01/2000 15:00 03/27/2000 25 day(s) 20 hour(s) Anem A Other An Average 10 minut m/s -------10.17.6 2.9 10.4 21.6 228 19:31 12.2 8.1 20.: wind shear. Gust of Wind Speed [m/ Column Numbers Solar Radiation (W/m2) Wind Direction (°) Wind Speed (m/s) Wind Speed SDV Temperature (°C) File Names c:\data\site1\st010604.12.2 7. WindRose provides some additional results (i.0 250 20:06 1.5 3.1 147. the column numbers of each device.1 228 19:25 0.0 233 19:49 0.2 246 20:15 15.7 10. etc).1 00:37:00 00:47:0 0057 0107 0117 1.

In this example the 21st column contains the wind speed at 10m height.50 Setting the corresponding anemometers heights (Input sheet’s 2nd table).30.20.08 "paper" method 4 16 Wind Speed Uncertainty Calibr. 12 or 8) that the program will use (recommended value: 16) Title1 Title2 Start from End at Minutes between data Limit for calms (m/s) Number of Direction Sectors Measurements Height : above Ground level (m) above Sea level (m) Calcul. The second table of the Input sheet contains all the necessary parameters characterising the analysis to be performed. which is the value of the wind speed below which the response of the wind direction measuring device is not reliable (recommended value: 2m/s) • The number of the direction sectors (16. of mean turbulence at wind speed (m/s) ± bin width (m/s) Wind Turbine selected Exponent coefficient (α) Weibull method low limit (m/s) high limit (m/s) ( Site's name ) ( comment ) 18/12/1999 29/5/2000 10 2 16 10 300 10 1 NEG Micon Tacke 600 750/48 0.WINDROSE : Wind Data Analysis software Measurements Height : above Ground level (m) 10. These parameters are: • The measuring period (start and end dates) • The time interval of the measurements (10 minutes. error at Umean (m/s) 0. etc) • The limit for the calm.06 Anemometer's max speed m/s 40 Data Logger's precision (bits) 8 Site Correlation file name Export processed data file name The necessary parameters for the wind data analysis (sheet Input) PAGE 10 . 1 hour.40.

5 The second method is recommended. more details in § SHEAR). This is particularly useful when delivering long term data (i.02.e.: yearly) that are stored originally in many files and/or to which correction coefficients were applied7 . PAGE 11 7 . Thus. then at 10m/s is 0. 5 The 2 values below the selected Weibull calculation method are taken into account only if the 2nd method is selected.: 4-16m/s). The given value refers to 5m/s. etc. If two or more anemometers are present.e. • The exponent coefficient of the Power Law. a simple model is used to “expand” it: every 5m/s this value is divided by two. then this coefficient is calculated analytically (per wind speed bin and direction sector. • The name of an ASCII file into which all the row data are exported. 2 methods are proposed for the calculation of the (k. The created export file includes all the corrections. A linear regression is used for the intermediate values and the double of it is considered at 0m/s (0. if 0. In brief. C) coefficients: a) using only the mean and the standard deviation of the wind speed and b) using the data distribution.08 is set. Since it is not realistic to consider a constant value for the entire wind speed range. restricted or not to a specific range (i. so no corrections should be applied if it is further processed. at 15m/s is 0. This coefficient is used only when one anemometer is present on the Mast. More details can be found in the Appendix § WEIBULL DISTRIBUTION.: the output signal of the anemometer is analogue voltage). 6 Uncertainty calculation for “pulsed-ouput” anemometers will be added in the next program release.e.04.6 • The name of the (intermediate) file that must be created when site correlation will take place (details in the CORRELATIONS chapter). which is used to extrapolate the Mast’s wind speed at the Hub height of the wind turbine. • The wind speed uncertainty in case of analogue anemometers (i. These parameters are not used in case of anemometers producing pulses/revolution.16 in the above example). • The Weibull distribution calculation method.WINDROSE : Wind Data Analysis software • The wind turbine for which energy calculations will take place.

The correction applied has the form: α·x+β and is intended for the processing of measured quantities. The third table of the Input sheet includes correction coefficients per each file. when: 1. If data contain time-series of atmospheric pressure and temperature. wind direction in radians. etc) 2. then the multipliers α and the offsets β are set to 1. additional offset for the “zero” of the wind vane. PAGE 12 Time OFFSET (sec) Temperature MULT Wind Speed MULT Time MULT .0 respectively. Although. The following important issues have to be pointed out when applying correction coefficients: • If the wind speed has to be adjusted by means of the correction coefficients (y=α·x+β). if the resultant direction angle falls out of the [0°. wind speed in miles/h. • Applying correction coefficients on the time/date should be done with caution.0). can seriously affect the reliability of the measurements.0 and 0.WINDROSE : Wind Data Analysis software Solar Radiation OFFSET (W/m2 Correction Coefficients Wind Speed OFFESET (m/s) Wind Direction OFFSET (°) Temprature OFFSET (°C) Solar Radiation MULT Wind Direction MULT -38 600 A sample of the correction coefficients table (sheet Input). the multiplier α (if ≠ 1. which seriously affects the wind energy calculations. Recorded data have different units (temperature in Fahrenheit degrees. 360°] interval then. is automatically adjusted. as it modifies the sampling rate (time step) of the measurement campaign. • When adjusting wind directions. there is no problem using the offset β (in seconds). The fourth table of the Input sheet deals with the calculation method of the air-density.e. then its standard deviation is automatically multiplied by the coefficient (α). etc…) If no values are given. A systematic error occurred during the measurements campaign (i. wrong calibration factors. in order to simulate a wrong day or time setting.

compose this worksheet. but the user can add more or delete the existing ones. etc.e. • PAGE 13 . In the opposite case.). Density correction.5ºC/km) and the estimated mean ground temperature (for the measurements period). then the recorded mean value is used instead. three are the possible choices: • The “first method” uses an empirical formula relating air-density only to the anemometer elevation (site height + anemometer height). One restriction applies here: The user has to respect the form of the existing power curves (i. These values can be retrieved either from a nearby meteorological station or set intuitively. put in the Row1 cell the model name of the wind turbine. choose the power control strategy. Two additional values are required: the vertical temperature gradient Γ (recommended value: 6. Power Curve Pairs of columns. one per each month of the year. The mean value of the air density used during the calculations is given in the Results sheet. Some representative power curves are included by default. More details on the air-density calculation methods and on the power curve corrections are given in the Selection of the method for the Air Appendix. • The “second method” calculates air-density using both the height and the temperature. the power is considered as zero and the wind turbine as out of operation.WINDROSE : Wind Data Analysis software then this table is not used. between the given values of the power curve. If temperature data exist into the data files. in the Row2 cell of the next column the tower height. since the air density is calculated from real data complying with IEC-61400-12. containing power curves of various wind turbines. • The “third method” uses 12 preset values. Whenever the measured wind speed is lower than the minimum value or greater than the maximum value of the power curve’s wind speed then. Note • A linear interpolation is performed in order to calculate the exact electrical power of the wind turbine.

3 23 692.3 14 729.5 13 696.5 16 750 17 744.8 24 695.3 8 244.1 6 97.5 12 640.6 9 349.4 7 155.WINDROSE : Wind Data Analysis software NEG Micon 750/48 tower (m): 45 diameter(m): 48 control: stall U (m/s) Power (kW) 3 0 4 19.8 19 723 20 711.2 11 564. Right: Its graphical representation including the correction due to the site elevation height (sheet Input) PAGE 14 .4 22 694.9 21 701.6 18 734. Power Curv e Correction applied 400 300 200 100 0 0 5 10 U (m /s) 15 20 25 30 Left: A sample of a power curve table (sheet PowerCurve).2 25 700.2 10 462. control:stall 800 700 600 500 Power (kW) Ref .8 15 745.6 NEG Micon 750/48 tow er=45m.5 5 53.

Note that: a) zoom operation is possible by selecting a region with the mouse b) scrolling (panning) is achieved by holding the mouse down and dragging it to the desired direction. since Excel graphs cannot contain unlimited amount of data and the user interaction (zooming. When a smaller data “window” is displayed. as he can quickly deduce the quality of the data (spikes. PAGE 15 . Execution time depends on the amount of data. normally a Pentium-4 PC will take ½ minute to run 1-year data. When the execution is finished the figure below will appear displaying all the processed valid data.). An experienced user will find this screen particularly useful. the user can launch the WindRose by selecting Run in English or Run in Greek depending on the language in which the results will be displayed.WINDROSE : Wind Data Analysis software QUICK-CHECK OF THE DATA Once the input sheets are filled correctly. Displaying a smaller data “window” is also possible when the View Mode is set to: Zoom X & Y (or Zoom X if zooming is done only in the time axis). Detailed view of all the processed data. the user can smoothly move forward or backward. scrolling) is limited. Select part of the data for detailed view. fault operation of a device. etc. by selecting Pan X & Y (or Pan X just for the time-axis) and holding down the mouse while dragging it to the desired direction. This is an important feature of the program.

3 kWh 23.WINDROSE : Wind Data Analysis software THE OUTPUT WORKSHEETS Results This one page sheet presents the summary of the performed analysis.64 % 29. PAGE 16 .2 27.7 m/s 48280 5593 4280 (8.8 kWh NNW SSE NNW SSE 41. Annual Mean Wind Speed at 67m height WT Annual Energy Production & cap.8 2477.T.148kg/m3) (general mean= 6.1% ) 3 740 376. The displayed quantities are: Measurements Period : from 4/2/2002 to 3/2/2003 Mean Annual Wind Speed (at 40m height) Mean Turbulence Intensity (at 10m/s) Max. energy production W.1 m/s % m/s m/s m/s Watt/m² kWh/m² (general mean= 6.59 % 23.2 % 6.5 m/s) 4 157 412.62 6. Capacity Factor Calc.51 % Mean Wind Speed variation NNW 10 8 6 4 2 N NNE NE ENE W WSW SW SSW S SSE SE E ESE W 0 E WSW ESE SW SSW S SSE SE % % % % % % % % N NNW NW WNW W WSW SW Time Distribution >15m/s 10-15m/s 5-10m/s 2-5m/s SSW S SSE SE ESE E ENE NE NNE The main results of the WindRose program (worksheet: Results).0 m/s) (6/1/2003 07:50) (6/1/2003 07:40) 1.2 34.0MW (ρave=1.71 % 23. 10min Average Wind Speed Maximum Gust Uncertainty of Wind Speed measurement Mean Wind Power Total Wind Energy Weibull Distribution constants shape factor (k) scale factor (C) Total number of valid data Included number of calms (<2m/s) Missing data Expected W.6 m/s (Gamesa G80-2.22 % 20.T.3 kWh 4 121 376. factor calculated from data distribution calculated from Weibull distribution Best Sector in Energy contain 2nd best Sector in Energy contain Best Sector in Time distribution 2nd best Sector in Time distribution Main Directions NNW N NW WNW NNE NE ENE WNW NW 6.1 0.2 9.2 307.95 % 23.

• The maximum wind speed and the date of its occurrence • The maximum gust (1sec value) and the date of its occurrence (if a gust column was set in the Input sheet). • The capacity factor of the wind turbine (percentage of the nominal power of the wind turbine. then the AEP is calculated from the 12 distributions. the “annual mean” is also provided. the AEP is calculated from the general average distribution. apart from the mean value of all the wind speed data. • The total number of valid data within the specified time period (by the Start-Stop dates given at the Input sheet) • The percentage of the missing data. • The number of calms (wind speed below a threshold) • The expected electrical energy production (kWh) of the selected wind turbine during the given period of measurements. Both are calculated by two methods: a) from the data distribution and b) from the Weibull distribution. Similarly with the measured mean wind speed. PAGE 17 . • The turbulence intensity (at the specified wind speed range). are not taken into account. Note that. In the opposite case (less than 1-year data). • The two best direction sectors in wind energy contain. If 12-month data exist. • The mean power of the wind per area (in units of kW/m2). such as seasonality. if more than 1-year data exist then. • The expected Annual Energy Production (AEP) in kWh and the corresponding capacity factor. • The coefficients of the Weibull distribution that fits all the data. for all the valid data. • The two best direction sectors in terms of time.WINDROSE : Wind Data Analysis software • The mean wind speed. Each monthly average value is weighted with the completeness of the given month for all the years. The calculation method of this value (see § below) assures that nonstationary effects. then the given value is the average of 12 monthly values. • The total wind energy per area (kWh/ m2). at which the machine should operate continuously to produce the expected electrical energy) • The estimated mean wind speed at the nacelle height of the wind turbine. if more than 1-year data exist.

6+7. Below. 2003 May 2003 Jun .1+7. Jul.6 100 % 7.3+6.54)/1. as it is calculated only by the valid data.45 91 % 7.5 100 % 6. Dec.0+6. 2003 Feb.1 100 % 7.45+ +7.64 99 % 6.0 100 % 6. 2003 Aug. an arithmetic example is given.4 100 % 8. In fact. while the other 2 refer to the annual production.36+6.7 99 % 8.7 +8.86 = (9.5 90 % 6.4 100 % 8. -------. assures that the result a) represents the 1-year reference period and b) is not biased by data that fail to cover exactly 12.0 100 % 6. If the given time-period (defined by the Start.9*0.4 +8. 2003 Mar. 2004 Mar.0 100 % 7. Nov. within the given time-period. 2004 Mean Data Wind Speed Complet.---------9. 2004 Feb. 2003 Jan. Energy production? • The 1st value refers to the energy production of the wind turbine for the given time period.86+7. Thus. Apr. 24.3*1.5 +6. May Jun.3 100 % 5.2003 Jul. 2003 Apr. the 1st and the 2nd values should be practically the same.3 100 % 6. 2004 May 2004 Jun. the 3 values should converge. 2004 Apr.3 100 % 8.9 100 % 6. The other 2 values are the outcome of the projection of the 2 distributions (data and Weibull) to 1-year. 2003 Dec. 2003 Nov.7 99 % 8. 2003 Oct. Aug. 36. Sep.86(*) 77 % 7. etc months.36 80 % 6.4 = ( 9. The 1st value depends on the missing data.T. Mean Data Wind Speed Completeness ----------------------9. Feb.0 100 % 6. Oct.3 98 % 6.9 54 % 8.3 100 % 7.3 60 % 5. Mar.9 )/12 (*)9. during all the years of measurements.54 Why 3 values for the expected W.9 75 % STEP 3 Mean Annual Wind Speed = 7. it means that the given Weibull distribution coefficients (k.9 82 % STEP 2 Month ----Jan. If the 3rd value is close enough to the other two. if there is no missing data then.WINDROSE : Wind Data Analysis software Note that.3 100 % 6. C) fit • • PAGE 18 .64+6. Averaging the 12 months.0+10.9 75 % 10. Stop dates) is one year then. the “annual mean” wind speed calculation takes into account the various data completeness of each month.9 100 % 7. STEP 1 Month --------Jan. representing the described procedure.1 100 % 7. 2003 Sep. the notion of missing data is not applicable to them.

this local air density and the type of the power control of the wind turbine (i.: stall or pitch). How the W. Each bar is composed by 4 parts. Energy production is calculated? • The wind turbine energy production is based on the wind speed estimation at hub height. If measurements are not taken at the WT’s hub height. then the wind speed is extrapolated to that height. Finally. the air density at the specific site elevation needs to be evaluated (using 3 different methods. depending on the percentage of time that PAGE 19 . etc). When the wind speed is measured at different heights. then the detailed wind shear (per direction and per wind speed) is calculated.e. the energy production is evaluated using the “corrected” power curve. then a uniform shear is assumed. the power curves given by the WT manufacturers refer to the sea level. • For the same reasons there are also 3 wind turbine capacitor factors. the “power law” is used in order to calculate the wind shear. based on the exponent coefficient α of the power law. including or not temperature timeseries. • • • Two charts are shown below the results table: A pie chart presenting schematically the most prevailed directions of the site and a polar chart showing the variation of the mean wind speed as a function of the direction. Therefore. For that.225kg/m3. provide the “corrected” power curve. according to the IEC-61400-12 recommendation. Usually. Finally. where ρ=1. If the wind speed is measured at only one height. Having calculated the wind speed time-series at hub height.T.WINDROSE : Wind Data Analysis software well the data and can be used safely to characterize the site. the data distribution in time per direction sector is given.

he has to set the calm limit to 0m/s and re-run the program. in most of the cases the two distributions are identical. They present the wind speed distribution (%) per direction. [10-15m/s). where the wind turbine is supposed to shut down. May N NNW NW 7 6 5 4 3 2 1 0 NNE NNW % energy N % time NNE NE 30 25 20 15 10 5 0 NE NW WNW ENE WNW ENE W E W E WSW SW SSW S SSE SE ESE WSW ESE SW SSW S SSE SE Samples of a site’s wind roses (total and monthly) Note that. Finally. i. It is reminded here that. WindRose This sheet is composed by the total and the monthly wind roses (distribution chart in polar co-ordinates). each 5. in order to calculate both time and energy distributions per direction sector. note that the energy refers to the energy of the wind. [5-10m/s). the value of the distribution at 0° is the mean value of all the data found within the sector [-2.WINDROSE : Wind Data Analysis software the wind speed at the specific sector falls in the 4 preset ranges8 : [2m/s-5m/s).e. Although. only data with wind speeds greater than the calm limit (set in the Input worksheet) are taken into account.8125). PAGE 20 . represents the mean value of all the data within this sector. differences might exist in case of very high wind speeds (greater than the cut-out wind speed). for maximum precision. [>15m/s). in the time and energy domain. If the user wants to change this and take into account the total data. the polar plots are divided in 64 sectors. +2. 8 The low limit is actually the calm threshold value (as set in the Input sheet). below of which the response of the wind direction measuring device is not reliable. not the estimated electrical energy of the wind turbine.625° (=360°/64) wide.8125. Each point of the drawn line.

a) this value is user selectable from “Advanced Options…” of the WindRose menu and b) it is displayed on the top of that table.4 m/s 20 6.7 m/s 0 0 2 4 6 8 Wind Speed [m/s] Mean wind speed per height – measurements & mean vertical shear (sheet Shear) The next table shows the measured mean and max wind speeds per direction and per height. Values of α are considered valid. Note that. Mean vertical wind profile 60 α 50 6.95. only if the goodness of fit R2> 0. along PAGE 21 .6 m/s 30 6.7 m/s 40 Height [m] 6. the calculated values are given for the hub height of the selected wind turbine. Both tables provide results at hub height. The first graph shows the mean wind speed values (as measured) along with the average vertical wind shear.2 m/s 10 5. The last table of this worksheet presents in detail the average value of the exponent coefficient α of the “power law” per wind speed and per direction. Wind shear calculations are based on the “power law”: ⎛ z ⎞ ⎟ u ( z ) = u ( z ref ) ⎜ ⎜z ⎟ ⎝ ref ⎠ where the zref denotes the height at which the wind speed u(Zref) is measured and α is the wind shear exponent coefficient. For each time step one value of α is calculated. The variation of the Weibull coefficients per height and per direction is given in another separate table.WINDROSE : Wind Data Analysis software Shear This Excel worksheet is created automatically only if multiple anemometers are present. At the end. based on the extrapolated wind speed time-series at that height (from the higher anemometer) using the wind shear results.

10 5.5 2.65 15.02 7.34 5.51 16.32 19.57 16.9 All Dirs 2.83 5.45 6.25 4.85 ENE 5.24 6.01 6.75 4.48 6.6 1.94 6.58 7.42 16.08 6.60 6.01 16.92 6.02 5.76 7.37 15.62 19.29 15.33 8.08 18.46 4.35 9.40 8.51 7.4 2.9 SSE 1.4 2.2 1.76 10.6 2.98 18.84 4.62 9.6 2.30 4.5 2.88 4.57 13.53 7.25 N 6.66 5.24 5.2 2.1 S 2.82 7.0 WNW 2.7 N 2.01 4.1 2.5 NNW 1.97 3.96 17.7 2.57 4.68 6.34 8.54 6.95 rejected.5 2.70 5.15 6.9 1.54 4.1 2.8 2.33 7.70 13.07 6.20 3.80 6.39 6.49 14.8 E 2.17 3.46 6.77 13.55 9.1 2.8 2.5 ENE 2.10 7.75 19.0 1.48 6.66 13.0 2.90 20.4 2.71 7.06 5.49 6.7 2.37 4.2 2.85 6.1 2.11 6.04 15.3 2.78 8. per height and per direction (sheet Shear) PAGE 22 .06 S 5.11 5.50 7.88 10.60 9.70 13.7 2.3 WSW 2.46 6.85 R2 = 0.0 1.8 2.01 3.72 6.06 16.45 7.78 6.94 7.24 9.37 16.3 2.84 5.WINDROSE : Wind Data Analysis software with the data percentage satisfying that condition.31 8.67 20.8 2.3 2.64 19.80 4.42 3.06 6.07 7.78 7.66 5.96 3.67 6.7 2.03 7.88 6.46 5.59 16.83 6.975 Z Z U U Measured wind speed (black dots) and calculated wind shear: Goodness of fit criterion: Left: R2<0.07 14.05 6.63 5.04 4.7 2.5 1.64 5.4 2.65 4.78 7.39 13.04 WNW 3.03 7.2 2.00 18.67 13.8 2.84 5.64 16.0 2.45 19.9 1.18 4.46 16.07 14.55 7.63 16.49 6.43 4.87 WSW 5.41 18.62 14.66 16.17 19.24 SSW 5.07 11.62 E 3.85 6.27 4.49 6.57 4.29 17.4 2.84 16.2 2.1 1.03 16.86 9.84 3.64 8.52 15.52 3.15 5.3 2.41 ESE 3.00 14.1 2.66 5.69 20.73 3.74 16.94 3.63 7.27 8.65 16.33 (calculated) (measured) Mean wind speed per height and per direction (sheet Shear) Weibull parameters per Height Height 10 20 30 40 50 80 Const k C k C k C k C k C k C NNE 2.72 18.60 7.10 14.4 1.52 5.18 3.62 8.9 Variation of the Weibull distribution parameters.39 9.92 5.79 7.03 6.65 8.49 7.0 2.5 1.03 16.0 1.4 2. Right: R2>0.4 2.04 12.50 6. R2 = 0. providing thus an indication of the validity of the Power Law at the specific site.9 2.33 All Dirs 5.3 2.2 2.19 8.49 7.09 7.04 7.46 6.50 6.76 4.2 2.47 6.86 W 4.85 7.92 3.37 8.3 2.8 2.3 1.8 2.55 6.61 13.59 13.3 ESE 1.3 2.27 13.98 19.77 14.59 9.48 6.60 SSE 4.35 8.91 5.95 accepted Average and Max (10min) wind speed per direction Height 10 20 30 40 50 80 U Ave Max Ave Max Ave Max Ave Max Ave Max Ave Max NNE 6.08 5.25 NNW 6.35 8.67 7.9 2.35 13.24 4.48 4.51 15.4 2.86 4.7 1.03 6.79 3.2 2.85 7.52 6.5 2.44 19.60 7.7 1.56 6.7 1.5 SSW 2.50 9.04 7.63 5.60 10.45 15.7 W 2.08 7.07 9.47 9.65 15.45 6.72 14.62 7.02 20.41 7.

475 0.363 0.049 0.093 0.130 0.167 0.104 0.079 0.069 0.097 0.070 0.371 0.95 : 90.218 0.166 0. A typical graph showing the diurnal variation of the exponent coefficient α with the hour of the day.221 SSW 0.118 0.058 0.075 0. This is achieved by plotting the wind speed ratios above a threshold (4m/s).175 0.391 0.164 0. if such values of α exist (per hour.099 0.170 0. no matter the wind direction.090 0.139 0.525 0.097 0.250 S 0.070 0.073 0.186 0.474 0.122 0.068 0.223 0. (sheet Shear) Here it is underlined that.146 0.078 0. then they are preferably used.129 0.088 0.081 0.045 0.080 0.103 0.099 0.058 0.095 0.396 0.398 0.079 0. assuming that the reference anemometer is installed on the top of the mast.120 0.109 0.249 ESE 0.310 0.066 N average 0.086 0.178 0.124 0.116 2 0.091 0.401 0.086 0.102 0.388 0.099 0.482 0.109 0. a series of polar graphs are created showing the influence of the mast tower “shadow” on the (lower) anemometers.168 0.365 0.188 0.059 0.074 0.126 0.110 0.106 0.WINDROSE : Wind Data Analysis software Wind Shear exp. At the end of the Shear sheet.404 0.092 0.077 0.159 0. The reason is the significant variability of the exp.166 0.100 0.082 0.172 0.369 0.465 0. coef.088 0.101 0.160 0.176 0.087 0. Thus.090 0.084 0.165 0.062 0.374 0. per wind speed and per direction).185 0.108 0.175 0.080 0.181 0.119 0.134 0.227 0.207 0.113 0.180 0. Result from Lavrio site (worksheet: Shear).104 0.086 0.357 0.070 0.344 0.162 0.179 E 0.147 0.144 0.344 0. the above table is calculated also for each month and for each hour of the day.156 Data percentage with R >0.079 0.050 0.152 0.078 0.070 0.113 0.351 0.062 0.093 0.119 0. coefficient α with these two parameters. when calculating the wind speed at a higher height (ie: hub height).369 SSE 0.242 0.107 0.120 0.211 0. α m/s 5-6 6-7 7-8 8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 20-21 21-22 22-23 23-24 24-25 25average NNE 0.099 0.100 0.355 0.144 0.135 0.059 0.110 0.424 0.100 0. rather than the (summarized) ones of the above table.411 0.404 NNW 0.078 0. PAGE 23 .100 0. internally. where the wind flow is undisturbed.017 WSW 0.076 0.25% W WNW 0.138 0. per month.278 0.124 0.083 0.124 Wind shear exponent results per wind speed and direction.077 0.129 0.095 0.165 0.391 0.186 0.081 ENE 0.090 0.075 0.161 0.035 0.107 0.075 0.

However. PAGE 24 .5. without any user interaction. note that at ~300° the anemometers ratio falls around 0. diametrically at ~120°. Comparison of anemometers and vanes mounted on the same Mast (worksheet: Shear). as expected. These graphs are automatically created for all the low anemometers in respect to the reference one. Polar plot of the wind speed ratio of a lower anemometer versus the top one (worksheet: Shear).WINDROSE : Wind Data Analysis software Influence of the mast tower “shadow” on the (lower) anemometers. In the above example. A less pronounced drop of the ratio is also observed. providing valuable information about a) the exact orientation of the boom and b) the boom length sufficiency according to the standards.

Finally. a linear regression is performed between the two anemometers.3 0 5 6 7 8 9 10 11 12 13 14 15 16 16 15 14 y = 0.0 5 6 5 0 .0 0 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 4 Wind speed bin [m/s] The main result of the Annex K sheet PAGE 25 Anemometer at 30m [m/s] 13 . cup anemometers should be calibrated in a wind tunnel after the measurement period.2 5 Squared sum of deviations [m/s] 0 . in order to assure the accuracy of the measurements.9992 0 .2 0 12 11 0 .1 1 .WINDROSE : Wind Data Analysis software since Excel cannot handle more that 32000 data per graph. using data from the initial 1-2 months.2 5 deg ) E valuation period (regres s ion) from: 1 2 /1 /2 0 0 8 to: 8 /3 /2 0 0 8 A pplic ation period (deviation res ults ) from: 1 6 /1 1 /2 0 0 8 to: 1 1 /1 /2 0 0 9 Anemometer at 28. during the entire measurement campaign.1 5 Squared Sum of Dev iations Comparison of anemometers 0 .9618x + 0. Briefly. This procedure is defined in the Annex K of the IEC-61400-12-1 standard.5m [m/s] 4 0 . Consequently. Another (inferior) possibility is to perform an in-situ comparison between the top and the control anemometer (usually 1. the procedure can be summarized as follows: Data are filtered for a narrow direction sector and classified in bins of 1m/s.1 0 10 9 8 7 0 .XLS.5m and at 30m ( wind direc tions : 3 4 8 . as other Excel graphs in WindRose.7 5 deg . an external graph library is used to create them in a form of a PNG picture. these graphs cannot further customized.5m below the top one). In-Situ comparison of anemometers at 28. Ideally.3264 R 2 = 0. the regression results are applied to the last 1-2 months of the measurements and compared to the real data. AnnexK This Sheet is created only if multiple anemometers are present and sufficient data exist (ie: several months). Then.

for the given time-step (i. All the data are taken into account here.e.WINDROSE : Wind Data Analysis software The main result of the AnnexK sheet is a (complex) graph with 4 axes. y-right) the linear regression between the two anemometers is presented. Actually. More detailed “views” of the results are presented in the next worksheets. the maximum value is of course. • The 2nd figure presents the Diurnal distribution. but the maximum 10minutes wind speed occurred during this day). if each bar remains lower than 0. However. Time Charts Three important figures are presented in this worksheet.1m / s ⎝ N⎠ 2 2 where: ΔU = Utop – (α*Ucontrol+β) with (α. not the instantaneous 1sec gust. values) • The 1st one shows the evolution of the daily mean wind speed. In the blue axes (x-top.1m/s. these parameters can be modified within the “Advanced Options…” of the WindRose menu. displaying the results of the wind data analysis in a “global” form. the maximum wind speed occurred per day. as well as. y-left) the square sum of the systematic and statistical deviations per wind speed bin are plotted. independently of the wind PAGE 26 . if 10minutes series are recorded. The examined condition can be written as follows: ⎛σ ⎞ ΔU + ⎜ ΔU ⎟ ≤ 0. In the green axes (x-bottom. Mean Daily Wind speed variation 25 Mean 20 Wind speed (m/s) Max 15 10 5 0 20/2/93 12/3/93 1/4/93 21/4/93 Date 11/5/93 31/5/93 20/6/93 10/7/93 Wind speed variation per day (mean & max. β) being the regression coefficients for the initial period and Ν the number of data per each wind speed bin (Ν≥3 10min data). The default and recommended parameters for this method are: i) select data from 6m/s to 12m/s ii) use intervals of 8 weeks for both the first and last part of the measurements and iii) apply the method to the main wind direction sector. The in-situ comparison is considered successful (and a recalibration of the anemometers is not necessary). this is the distribution of the wind speed as a function of the hour of the day.

The following points should be noted here: The “binning” of the wind speed is done in 25 steps of 1m/s. The 1st one shows the variation of the % turbulence intensity (σu/U*100) over the wind direction and the wind speed. Tables Two very essential tables are presented in this worksheet. 1m/s). Mean Monthly Wind Speed Variation over a year 12 10 Wind Speed (m/s) 8 6 4 2 0 Feb Mar Nov Jan Jun Apr Jul May Month Dec Aug Sep Oct Mean wind speed variation per month. PAGE 27 .9 7.8 7. Diurnal Distribution 8. The table of the turbulence intensities does not show the first five velocity bins (the Excel’s “Hide cells” option is enabled). dominating the rest ones in the graphical representation of the table (see worksheet:3D).6 7. the ratio σu/U takes very high values.4 7. The 2nd one shows the total data distribution over both wind speed and direction. starting from the [0m/s.3 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour of the Day Distribution of the wind speed per hour of the day.7 7.2 8. If more than a year period is examined. In that case.1 Wind speed (m/s) 8 7. the 3rd figure shows the mean wind speed per month. as no physical meaning can be extracted from turbulence intensities. The Diurnal distribution per month is presented in the 12diurnal worksheet. when the mean wind speed approaches to zero. with the 26th one gathering all data higher than 25m/s.5 7. • Finally. each monthly mean wind speed is calculated from all the corresponding months.WINDROSE : Wind Data Analysis software direction.

then the width of each sector is 22.WINDROSE : Wind Data Analysis software Data distribution vs Wind Speed & Wind Direction m/s NNE 32 0-1 41 1-2 31 2-3 16 3-4 9 4-5 18 5-6 7 6-7 7-8 1 8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 20-21 21-22 22-23 23-24 24-25 25totals 155 NE ENE 61 50 55 80 49 69 27 61 5 18 8 11 2 1 E 78 142 187 139 60 33 14 15 10 7 5 4 10 4 5 4 8 10 12 4 1 ESE 124 141 239 206 167 139 144 115 69 46 31 46 23 16 14 6 6 6 1 1 SE 100 175 255 215 209 243 218 265 168 184 218 205 211 191 122 86 93 58 37 18 18 8 1 SSE 92 134 140 127 158 220 218 204 122 78 77 71 81 37 34 27 26 12 7 11 8 6 4 2 3 1 1900 S SSW 154 110 158 158 141 139 153 161 189 142 148 105 130 91 54 91 31 93 28 52 26 42 19 23 26 26 23 20 13 15 4 2 6 3 2 1 SW WSW 104 141 224 315 304 483 361 720 284 899 236 609 208 430 177 246 138 151 135 122 83 76 71 72 61 30 33 22 51 6 63 3 22 8 2 W WNW 189 229 436 493 782 784 1338 797 2165 862 2351 782 1988 1004 1463 1131 1272 1298 995 1413 1163 1439 1159 1338 1004 1245 938 1238 813 1100 430 933 162 684 29 524 6 470 1 308 189 89 25 5 3 NW NNW 166 38 337 47 428 64 288 18 106 10 65 22 88 10 106 5 120 137 1 176 250 1 177 1 57 2 44 1 40 1 58 2 30 11 11 3 N 36 42 22 7 7 13 3 1 2 totals 207 290 752 1540 3298 1309 1270 2565 4325 18684 18383 2698 223 1704 2978 4117 4634 5290 5003 4556 3873 3475 3198 3336 3259 2895 2581 2218 1599 1067 680 548 354 220 103 30 7 6 1 133 57732 Tables worksheet: Data distribution per wind speed and direction. the total data distribution again per wind speed and wind direction. The number of successive rotations of the wind vane (clockwise and counter clockwise) is given in the 2nd table. the wind turbine usually rotates for an optimum orientation. which could be calculated. Obviously. These revolution numbers. 3D This worksheet contains the 3D graphical representation of the Tables worksheet. provide a good estimate of the number of times that the electrical power cables are twisted. the accuracy of the two numbers depends on the data completeness (low number of missing data). The number of the wind direction sectors is the one selected in the Input worksheet (8. per wind speed and direction bin. showing the turbulence intensity distribution per wind speed and direction. For example: suppose that 16 sectors are selected (the default and recommended value).5° (=360°/16) and the North winds are all whose angle falls into the [348. only if the detailed wind speed time-series were available. as well as. is calculated using the classical averaging methods. this simple method provides a satisfactory approximation of the “true” turbulence intensity. 12 or 16). Although not precise. no correction takes place.25°) sector. For this calculation. As mentioned before. indicating the number of times that a wind turbine would have been rotated around itself. +11. each direction refers to the mean angle of the corresponding sector. all the data (including calms) have been taken into account.75°. The mean turbulence intensity. considering that even during low wind speeds. PAGE 28 .

it is the hourly variation of the wind speed per month. the calms are not included here too. PAGE 29 . At the end of the 2nd page a table is also given with the numerical representation of the figures. In case that.6 9.18 21.22 NNE NE ENE E ESE SE SSE S SSW SW WSW W WNW NW NNW N 10 5 0 Turbulence intensity per wind speed and direction (3D worksheet).14 17. a prevailed sector (referring to time) is not among the best two sectors (referring to energy) then a blue colour is used (or a green colour in the opposite case). Note that. 12diurnal This one-page sheet is composed by 12 bar-plots presenting the Diurnal distribution of the data. As in the WindRose sheet. per month of the year (all years included).10 13.WINDROSE : Wind Data Analysis software Turbulence Intensity (%) vs Wind Speed & Wind Direction 35 30 25 20 15 5. showing the two dominant wind directions. January February NNW NW NW N NNE NE ENE E NW NW NNW N NNE NE ENE E W WSW SW SSW S SSE SE ESE W WSW SW SSW S SSE ESE SE Main wind directions (blue: based on the time. one per month of the year. In fact. green: based on the energy content). it can be found in the TimeChart sheet. all the directions taken into account. 12pie In this 2-page worksheet 12 pie charts are plotted. if the total diurnal distribution is needed.

Energy Distribution vs Wind Speed 700 600 Energy (kWh/m²) 500 400 300 200 100 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Wind speed (m/s) Wind energy distribution versus wind speed.7m/s Example of the Diurnal distribution per month (12diurnal worksheet).6m/s Mean Wind Speed: 6.WINDROSE : Wind Data Analysis software Important Note • If data exceed 1-year period the (monthly) graphs cumulate data from different years. All the wind directions have taken into account when calculating the shape (k) and the scale (C) coefficients of the Weibull distribution in this graph. the graph refers to the reference anemometer (usually the highest one). Bar Charts The 4 most common representations of the wind distribution are plotted in the present worksheet. (BarCharts worksheet). The last two show the time-based distribution of the wind as a function of the wind direction (3rd one) and the wind speed (4th one). March 12 Wind speed (m/s) 10 8 6 4 2 0 0 2 4 6 8 10 12 Hour 14 16 18 20 22 April 12 Wind speed (m/s) 10 8 6 4 2 0 0 2 4 6 8 10 12 Hour 14 16 18 20 22 Mean Wind Speed: 8. The first two show the distribution of the wind. When it is necessary to see the diurnal distribution of only one specific month. based on the energy (not the expected energy production of the wind turbine) as a function of the wind direction (1st one) and the wind speed (2nd one). If multiple anemometers are present. once per year (specifying accordingly the start and end dates in the Input worksheet). then the program should be run. PAGE 30 . Weibull The main figure of this worksheet shows the Weibull distribution that fits the wind data.

25 .8 8.2 14. future versions will include a graphical representation.53 1.7 3.75 258.75 .5 1.8 10. the site’s topography.07 8.e.40 5.44% 7. 9. Some important information is included in the table below the graph.56.02 0. PAGE 31 .I.78.98 11.25 56.0 11.03 0.213.36% 31.23% Mean Wind Speed (m/s) 2.0 8.29% 2.67% 5.0) Wind speed (m/s) Data distribution together with the calculated Weibull distribution that fits the data.71% 3.7 5.: WindFarm® by ReSoft).75 .75 .07 0.27 2.87 1.25 . the calculated Weibull coefficients per month are given only as values at: TempData!M109:O122 however.75 .75 213.05 0. mean wind speed and turbulence intensity.25 236. the mean wind speed.37 3.67% 0.7 8. Direction NNE NE ENE E ESE SE SSE S SSW SW WSW W WNW NW NNW N Angles (deg) 11.97 5.01 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Data Distribution Weibull Distribution (1.04 0. In this version of WindRose.25 .71 1.25 Weibull shape 1.30% 2.75 303.27% 0.1 0.32 7.6 12.84% 4.11.9 2.65 1.02 Data Distrib 0.25 .9 4.25 4. data percentage. not on the Weibull distribution).25 .3 14. Here.75 . The 2nd graph of the worksheet is the wind data accumulated probability (based on the data distribution.75 168.236.1 5.3 T.281.WINDROSE : Wind Data Analysis software 0. require as an input this table’s data.79 1.74 1. (%) at 10m/s 11.75 .11 7.25 191.0 5.25 .20% 4.2 4.42 1. i.3 6.70 1.90. per direction sector.191.72 1.36 2.36% 0.326.5 9.e.25 146.25 101. dealing with optimum wind turbine arrangement within a wind farm (i.16 2.49% 32.25 .27% 2.0 9. how much of the time the wind speed exceeds the cut-in or the cut-out speed of the wind turbine.88 6.25 326.75 78.09 0.58 1.123.2 15.9 7.1 Weibull distribution coefficients.50% 1.85 2.4 15.146.73 2.08 0.303. Some thirdparty programs.75 .0 2.3 10. It is a useful tool to estimate the percentage of the time that the wind speed exceeds a specific value.25 .75 33. the turbulence intensity and the starting and ending values of the direction sector.168. the coefficients of the Weibull distribution are displayed per wind direction sector together with the corresponding percentage of the total data found in it.75 348.258.28 9.37 8.101.348.55 Weibull scale 3.39% 0.69 6.33.2 6.23 6.06 0.11 1.75 123.75 .9 4. as well as.65 1.25 281.

(24 * 31=744 points plotted). (i. Obviously.WINDROSE : Wind Data Analysis software Cumulative Probability (%) 100 90 Cumulative Probability (%) 80 70 60 50 40 30 20 10 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Wind speed (m/s) Cumulative data probability.9% Variation of the mean wind speed per hour. a zero is displayed.e. 2000 Mean Wind Speed: 8. The 3rd graph shows the variation of the turbulence intensity over the wind direction. Safe conclusions can be obtained only when significant amount of data are found in each direction sector. PAGE 32 . The calculation is done using only wind speeds within the specified range in the Input worksheet (usually 10m/s ± 1m. The discussion held in the description of the Tables worksheet about averaging turbulence intensities.s). applies here too.UhourG These 2 worksheets contain monthly tables (UhourT) and graphs (UhourG). representing the evolution of the mean hourly wind speed per day. the wind speed is higher than 10m/s).6m/s Data completeness:99.DIRhourG Similarly to the previous 2 worksheets.: 33% of the time. DIRhourT . DIRhourT and DIRhourG. show the variation of the wind direction per day (mean value per hour). 24 Wind Speed [m/s] 21 18 15 12 9 6 3 0 1 3 5 7 9 11 13 15 17 D ay 19 21 23 25 27 29 31 February. UhourT . if a direction sector has no data in the specified wind speed bin. during the month of January. for each month.

Note that the energy production results depend on the data completeness.6m/s 4 188 7 739 23 707 0 1 31 463 30 219 307 76 60 217 171 146 731 719 740 715 278 613 637 711 722 334 663 729 5 132 2 735 2 709 1 11 19 511 37 226 619 33 60 168 176 376 736 740 734 633 315 636 667 706 739 395 735 720 6 95 0 731 3 723 2 3 45 500 2 59 734 168 116 189 160 411 715 603 731 697 348 615 705 709 729 469 728 739 7 133 35 739 68 727 0 5 34 558 284 131 700 120 158 210 144 536 613 572 735 675 191 552 704 702 716 500 727 744 8 151 129 739 178 743 0 28 10 517 588 119 555 30 297 325 294 565 633 518 734 590 40 560 721 694 712 480 738 741 9 25 256 740 395 724 0 1 6 591 676 307 631 10 291 314 184 446 541 610 716 668 198 650 715 711 734 551 727 737 10 107 454 738 525 721 0 0 32 511 702 243 574 1 486 198 123 404 580 644 642 664 34 665 691 722 736 504 739 667 11 58 546 723 684 651 0 0 161 566 715 169 290 39 376 354 97 198 467 700 590 711 0 681 675 683 667 590 742 596 12 21 613 701 735 506 1 2 301 702 729 280 262 68 322 518 85 366 608 731 639 691 0 672 665 565 208 567 714 213 13 18 677 697 722 172 2 0 328 734 219 102 123 106 323 504 22 378 672 725 568 735 5 684 595 490 369 500 731 90 14 87 714 709 718 16 4 8 435 726 514 285 169 52 277 614 31 525 611 695 606 732 10 661 548 244 436 432 725 73 15 75 730 713 664 6 2 38 526 517 287 489 35 2 256 670 66 586 667 680 639 732 2 687 554 156 389 299 732 21 16 99 715 730 656 58 2 73 530 530 40 603 12 110 166 550 41 500 718 696 692 691 29 686 668 36 298 414 740 2 17 63 739 673 654 193 0 102 615 524 72 658 51 42 208 622 51 438 722 725 689 732 143 586 635 3 135 359 713 0 18 8 740 473 652 106 0 61 614 401 189 555 44 6 102 546 62 424 698 727 647 719 108 639 682 0 39 372 722 0 19 5 720 351 660 83 0 39 565 312 294 571 28 0 84 442 70 681 698 616 659 583 308 695 685 0 0 148 739 3 20 20 721 242 435 113 0 5 299 137 143 636 51 3 70 89 45 612 683 465 656 630 262 646 708 2 47 260 717 15 21 38 666 241 544 70 7 0 228 134 130 575 17 3 38 259 43 224 713 405 718 677 262 677 719 21 92 389 707 35 22 24 725 228 659 8 35 0 199 78 196 520 8 0 4 285 98 210 668 543 738 686 543 688 735 83 66 559 722 197 23 Sum 3 2851 736 9992 151 14685 656 9698 2 9741 17 137 0 384 252 5239 61 10785 313 6495 659 8175 0 7029 12 919 0 3880 350 7556 83 3170 622 8828 572 15901 741 15314 722 16558 593 16441 446 5625 651 14890 740 16100 278 10420 106 10813 503 9591 734 16730 79 9303 WTprodT: Table showing the hourly sum of the wind turbine energy production (kWh). PAGE 33 .Energy [kWh] 15000 10000 5000 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 9 9 . per day of a month. 2000 360 315 270 225 180 135 90 45 0 1 3 5 7 9 11 13 15 17 D ay 19 21 23 25 27 29 31 Dir. per month of the year. February 2000 T otal: 2 6 7 2 5 2 kWh M ean Wind Speed: 8 . [deg] U<5m/s 5<U<10m/s U>10m/s Monthly Variation of the mean wind direction per hour WTprodT .WINDROSE : Wind Data Analysis software February. production by Nordex N50/800: 267252kWh Mean Wind Speed: 8. February 2000 Hour 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 0 313 4 735 36 658 63 1 0 350 102 113 507 1 16 1 383 58 722 681 742 709 657 584 697 742 690 252 642 709 1 276 23 718 5 678 0 0 0 451 75 126 422 2 8 1 253 46 696 554 740 732 444 586 670 735 721 175 444 737 2 410 27 711 8 680 0 1 1 467 88 295 411 0 82 53 249 12 732 544 738 719 446 374 636 719 737 243 607 727 3 504 15 727 11 689 0 4 9 443 68 236 477 37 81 79 239 63 706 678 743 729 557 403 650 710 724 297 545 729 Total exp.6 m/s D ata c ompletenes s : 20000 El.9 % 16 12 8 4 0 U [m/s] E U D ay WTprodG: Variation of the WT’s daily produced energy together with the mean wind speed.WTprodG These 12 tables (WTprodT) and 12 graphs (WTprodT) represent the sum of the hourly wind turbine energy production per day.

for each month of the year. so that the correct hourly/daily sums are calculated. then in these 2 worksheets (similarly to the previous ones).WINDROSE : Wind Data Analysis software Variation of the Wind Turbine Energy production per month 70000 60000 WT Production (kWh) 50000 40000 6 30000 20000 10000 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month 4 2 0 WT Production (kWh) Wind Speed (m/s) 12 10 8 Wind Speed (m/s) 31 WT’s energy production and mean wind speed per month TempT . provided that in the Input worksheet. Some pyranometers (solar radiation measuring devices). SRadT . Note some missing data between 3 and 5 April.10 1 3 5 7 9 11 13 15 Da y 17 19 21 23 25 27 29 April Mean Temperature: 16. record during the night. small negative values instead of zeros. one can find for each month the mean hourly temperature per day. 35 30 25 20 15 10 5 0 -5 . column numbers for the temperature were supplied. these 2 worksheets contain the sum of the hourly solar radiation per day.SRadG Again. column numbers were specified for the solar radiation then.TempG If in the Input worksheet. due to a limited precision of their A/D converter. PAGE 34 . WindRose takes this into account by considering these data as zero. during 1-month period. Similarly to the above the tables and the graphs are in the SRadT and SRadG worksheets. respectively.2°C Variation of the mean temperature per hour.

TempData Finally. when selecting “XValues” and “Names and Values”. In order to improve the readability of the sheet.WINDROSE : Wind Data Analysis software December 2003 T otal: 5 4 8 7 7 W/m2 D ata c ompletenes s : 9 9 .8 % 3500 S. new graphs can be created using all the listed data. [Wh/m2] 3000 2500 2000 1500 1000 500 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 D ay Solar radiation daily sum during one-month period. the user can see the corresponding data columns. the data are grouped in sets of columns of different colours.Rad. in this worksheet the detailed information. PAGE 35 . Moreover. It is reminded here that. which is indispensable for the all graphs appearing to all the above-mentioned worksheets is presented. (worksheet: SRadG). which will be updated automatically each time the program runs. by double-clicking on the every Excel graph.

Based on the common data. with mean values μ1 and μ2 and standard deviations σ1 and σ2. Predict) and deals with the correlation between the measured data at two different sites. Site Correlation file name Creating the necessary file for the Correlations (sheet Input) Important Note • The two measurement campaigns must contain data from a common time period. the program runs by selecting the Calculate Correlations option.0 show similar (in-phase) variations. Otherwise. correlation coefficient values close to 1. When done. by running WindRose twice (setting two different file names. in Input sheet’s appropriate cell).N are the σ 1 ⋅σ 2 common data of the two time-series x1 και x2. where x11.2.2. while values close to 0.0 show 9 Goodness of fit : N R2 = 1− 2 SSE SSM with: N ∧ ⎛ ⎞ SSE = ∑ ⎜ yi − yi ⎟ ⎠ i =1 ⎝ and SSM = ∑ yi − y i =1 ( ) 2 where: y = ax + b ∧ (α. correlations cannot be calculated.N and x21.WINDROSE : Wind Data Analysis software CORRELATION – MISSING DATA PREDICTION WindCorr This worksheet uses the MCP method (Measure.…. but for all the wind speeds (above calms) for all wind speeds. two files (one per site) have to be created. All the above are calculated: per wind speed range and per direction sector per wind speed range but for all directions per direction. Before using it. linear regression coefficients (a. PAGE 36 . b of the best-line fit) and the goodness of fit (R2) 9 . b are the slope and the offset of the best-line fit) Correlation coefficient: ρ= ∑ (x1 − μ )(x2 i =1 i 1 N i − μ2 ) . no matter the direction It is reminded here that. calculations are performed to estimate correlation coefficients (ρ).…. Correlate. in the toolbar of the Excel’s main menu.

• A Boolean value (Yes/No) indicating whether or not an hourly averaging will occur. wider wind speed bins and hourly averaging. goodness of fit values close to 1. Optionally. these files are generated by running twice WindRose before this step. Input parameters The first column of the WindCorr sheet contains the necessary parameters for the program to run. two files can be also created containing i) the concurrent pair of data and ii) the predicted time-series (see § below: Output Files). in order to perform reliable calculations (recommended: 2hours.WINDROSE : Wind Data Analysis software irrelevant variations (uncorrelated phenomena). 10 11 . before processing the data. Recommendation When correlation results are not satisfactory. • The width of each wind speed bin. After running the “Calculate Correlations” command. As pointed out earlier. The remaining input parameters are: • The initial time-shift of the 2 time-series. Often. The main input is 2 files containing the data from the Reference and the Target site. • The number of the wind speed bins.0 show that the best-fit line represents very well the data “cloud” (using the linear regression coefficients). try less direction sectors. 12 data) • The number of the direction sectors (8. the WindCorr sheet holds all the results in form of tables and graphs. PAGE 37 . • The minimum number of common hours that must exist (per wind speed bin and direction sector). i. if the time step is 10min. provided that the clocks are synchronized. Note that this number has nothing to do with the one set during the WindRose analysis. Furthermore. 12 or 16)11 . used to calculate the best possible correlation coefficient (see below: Methodology)10 .e. the hourly averaging provides better correlations. For nearby sites 2 hours is sufficient.

-2h. +2h) of the Target time-series. hours of data per bin 2 Number of Dir. Input File with the Reference timeseries c:\data\ref. Time-shift values close to zero show that wind phenomena occur simultaneously to both sites.e.bin Input File with the Target (incomplete) time-series c:\data\target. 0. by shifting Target series up to : [hours] 2 Transform time-series to : 1-hour data Number of wind speed bins 20 Width of wind speed bin [m/s] 2 Min.bin Output File for the concurrent pairs of data (optional) c:\data\sync. … . sectors 8 Correlation Method y=ax 2 The input parameters of the WindCorr Sheet. The time-shift.txt Locate best correlation. at which the maximum value of the correlation coefficient is found. -1h 50min.txt Output File for of the predicted timeseries (optional) c:\data\pred. is then inserted into the Target time-series (by adding it to its time/date array)12 and displayed at the Table 2 of the WindCorr sheet.WINDROSE : Wind Data Analysis software Methodology Consider correlating measured data from a Reference site (long-term data) and a Target site (short-term or incomplete data). +1h 50min. in respect to the Reference ones. …. This time lag can be determined by examining the correlation coefficient for various time-shifts (i. Sometimes nearby sites log wind data with a time lag. depending on their distance (and their data-loggers clock settings). Global correlation coefficients close to 12 This step assures that time-series from data-loggers with unsynchronized clocks will be correctly processed !! PAGE 38 .

Therefore.2m/s and the wind direction is φ=330° then. for the wind speed. if WindRose is asked to predict missing data. then for the wind speed it uses the regression coefficients and for the wind direction the difference of the mean direction values. linear regression coefficients.141 and for the wind direction.5° During the prediction phase.75°) and their wind speed falls into the [6m/s-8m/s) bin.e. Hence the concurrent pairs of data are 9-month long.5° = 320.WINDROSE : Wind Data Analysis software zero show uncorrelated phenomena and should alarm the user about the exploitation of the results.4m/s (varying from 6. The predicted gust value is taken as: Umax(predicted)=3*USDV(predicted).25° to 348. Now. The corresponding concurrent (simultaneous) 100 data from the Uncompleted time-series (not necessarily within the [6m/s-8m/s) bin and the NNW sector) have an average wind speed of 8.022 and β=1. that the Reference site holds 1-year long data and the Target site holds 9-month data (missing months: January. and their average wind speed is 7. Arithmetic example: Assume 16 direction sectors and 2m/s wide wind speed bins. the same procedure is followed for the correlation – prediction of the turbulence intensities. Therefore. For both data sets. the predicted values (u’. Suppose that 100 Reference time-series data have directions from the NNW sector (326. The next step is to select the data of the Reference time-series that “belong” to a specific wind speed bin and direction sector.2m/s. the difference of the means is: Δφ = 321°-330. etc) are calculated and stored.5° = -9. In the same time.8m/s) and average direction 321° (varying from 315° to 327°).141 = 7.48m/s and: φ’ = φ+Δφ = 330°-9. we investigate their concurrent pairs from the Target time-series. Let’s now assume. Notes on the methodology example • The correlation results depend on the common data period. if the Reference time series were 5- PAGE 39 .5m/s to 9. the slope and the offset of the linear regression are calculated as: α=1.5°. φ’) for the other site are: u’=αu+β = 1. if in February the wind speed of the Reference site is u= 6. for simplicity reasons.5° Similarly.: mean values. February and October). all the statistical quantities (i. The mean value of their directions is 330.022*6.2 + 1.

91 Max. Again.site General results of the Correlations (WindCorr sheet). from that sector).: a) per wind speed bin.WINDROSE : Wind Data Analysis software or 15 years long. no matter the direction. assume that only during October (another missing data period) extremely high wind speeds (>25m/s) occurred from the NNW direction.162 0. Assume that SSE wind directions appeared only during January (when no data exist for the Target time-series).e. However. the correlation results would be the same and based to the common 9-month period.856 (**) of Target site relative to Ref. correlation results exist for the NNW direction but only up to 18m/s. Therefore.21 (*) of the 4176 concurrent data 0.11 0. • The mean wind speed. • Apart from the partial correlation results (per wind speed bin and per direction).876 0. no matter the direction). the predicted values of these high speeds will be based on the global NNW correlation results (i. Then. In the beginning are given: • The total number of the data in file.926 0 0. in the next section are written: • The number of the common data PAGE 40 . no correlation results exist and consequently the prediction will be based on the global correlation results (per wind speed bin. global correlations are also calculated i.: all the wind speeds above calms. • The dates of the first and last data points. Therefore. Correlation Uncertainty for the Target site [m/s] • • Number of data Data start at Data end at Mean(*) Wind Speed [m/s] Number of concurrent data 4464 1/2/2000 2/3/2000 23:50 6. Wind Speed Correlation coefficient at a time(**) shift [min] General regression coefficients of wind speeds slope offset [m/s] R2 4176 4176 1/2/2000 0:00 29/2/2000 23:50 6.e. Output Tables The 2nd table of the WindCorr sheet contains some statistics of the two input files and the global correlations results. b) per wind direction for all wind speeds above calm.

Note that this is not a static table and its contents can change depending on the selected direction in the drop-down button. Correlation coefficient occurred. based on the data wind speed and direction distribution. correlation uncertainty = σu PAGE 41 . The displayed value is an average weighted value. as follows: • N where: σu is the standard deviation of the (concurrent) wind speeds of the Target site and Ν the number of data per each interval. in the last section are written: • The correlation uncertainty of the wind speed • The global coefficients of the best-line fit (all wind speeds.WINDROSE : Wind Data Analysis software The maximum correlation coefficient after the timeshifting of the two-series (see above) • The time-shift for which the max. Finally. no matter the direction) • The goodness of fit R2 The calculation of the correlation uncertainty is performed analytically. The 3rd table of the WindCorr sheet presents the detailed correlation results. per each wind speed bin and direction sector of the Reference site. per direction and wind speed bin.

29 13.198 Slope Offset 0.088 0.7379 0.8 19.15 2.057 1.74 20.78 5.8789 0. these are the 12-month long data). Site [m/s] y = 0.20 20 . Note that the contents change according to the direction selection of the drop-down button.910 0.18 18 .5 40.609 3.056 -6.077 2.5 15. Selected Direction : NNE 16 Wind Speed of Target Site [m/s] 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 Wind Speed of Ref.325 1.98 12.675 8.0 13.778 0. which were used for the correlation/prediction.09 10.98 5.631 -3.3 12.3194 0.Ref. Select direction of the Reference Site: NNE Wind Speeds Unc.26 51 91 127 147 144 137 101 58 59 26 18 1.14 14 .207 1.453 3. site Mean value 30. site Mean value 6 2 11 15 17 15 23 22 29 27 27 Ref.563 -53.269 -1.0480 0.91 18.97 9. there are not included.027 0.94 9.01391 0.007 The table with the detailed results of the correlations (WindCorr sheet).204 -1.8 Unc. site Wind speed bin Number of Mean Mean .154 6.90 3.498 0.205 0.21052 0.251 0.070 0.5 14.6 15. Output Files Apart from the results presented in the worksheet’s tables and graphs.5 20.1 11.77246 0.97 14.435 0.98 16.6 14. site Ref.78 0.895 4.034 1.70 14.54443 0.Intensities [%] Corr.810 0.9 15.207 5.WINDROSE : Wind Data Analysis software Correlations Table.954 0.008 5.10204 0. Note.62122 0.24 24 .58 7.2229 Slope Offset [m/s] -0.440 R2 Directions Ref.17 1.1 15.9923 Another result of the WindCorr sheet: Mean wind speeds correlation per direction sector.03 6.430 0.3 10.2 13.08434 0.336 0.4588 0.152 -7.004 1.464 0.84 13.10 10 .7882 0.1179 0.118 6.79 11.8 13.6512 0.856 -1. The 2nd one includes time-series created by applying the calculated coefficients to the Reference time-series (in the given example.113 0.025 0.16 16 .22 13.165 0. site Mean value 22 22 22 20 20 19 19 17 19 18 18 Unc.515 10. in the above example 9-month data).840 0. The 1st one contains the concurrent pair of data (i.5 15. three ASCII files are created containing the detailed results.377 -1.097 1.501 0.8 14.6 Turb.e.3883 0.15078 0.8785x R 2 = 0.051 12.42402 0.626 0. Coef.2 14. 0. PAGE 42 .6 18.46 Correlation Coefficient 0.22 22 .412 0.2904 0.047 0.946 -56. site Mean value 14.773 0. that even if data existed in the timeseries.00230 0.870 3.12 12 . site [m/s] data value value [m/s] [m/s] 0-2 2-4 4-6 6-8 8 .04968 0.350 1.663 9.9 12.

39 12. February and October). 0. Site 552 65 47 31 30 63 83 123 104 49 118 58 44 476 1305 19 40 67 86 138 159 181 203 223 248 271 290 318 342 0 19 38 72 92 136 149 172 196 189 243 266 298 333 330 356 Corr.22 11.09 8.: in the above example the predicted data for January. it is checked whether data exist in all the 4 quarters of the circle.25 78. the 4th table of the WindCorr sheet contains the correlation results per wind direction.07 10.75 146.25 258.25 348.75 281.2877 0. for all wind speeds above 5m/s.25 168.75 236.25 123.69 9. circle.25 33.25 303.3424 0.88 10.25 303.2810 0.86 9.29 8.25 348.57 10.07 Unc.1365 0.75 101.32 10. Correlation of directions for wind speeds > 5m/s Direction of Ref.64 10.25 213. Site of Unc.75 326.1057 0.2443 -0.75 To 33. site Mean wind speed 11.2049 0.4160 Ref.25 213.69 9.75 236.75 146.79 11.75 101.95 8. PAGE 43 Mean Wind Speed of Ref. site Mean wind speed 12.37 9.25 258.75 326.34 9.4145 0.81 8. Correlation of directions for U>5m/s 40 30 20 14 12 10 8 0 6 -10 -20 -30 -40 NNE NE ENE E ESE SE SSE S SSW SW WSW W WNW NW NNW N 10 4 2 0 A graphical representation of the Directions Correlation together with the mean wind speed (WindCorr sheet). In that case. Finally.58 7.55 8.3108 0.27 12. Difference [deg] .25 168. Coeff.75 281. ) Wind Direction correlations (WindCorr sheet).4507 0.37 9. When averaging wind direction values.75 11.88 ( Red color: Mean values result from the 4 quarters of the trig.75 56.68 10.75 191.44 9.14 8.25 78.75 56.e. Particular notice should be given to the angular shift of the wind direction vector. results appear in red to remind that no reliable conclusions can be made.2704 0.39 8.25 Number Mean Direction Mean Direction of data of Ref.98 9.5551 0.75 191.WINDROSE : Wind Data Analysis software The 3rd one is again the predicted data but only for the missing data period (i.04 10.30 9. 360°) point.25 123.3074 0. site [m/s] Dir. Site NNE NE ENE E ESE SE SSE S SSW SW WSW W WNW NW NNW N From 11. due to the discontinuity of the (0°.0900 0.

Inc. y1 . x2 .k) φ'i = φi + Δφ +∆Τ time Shift 3 Locate the maximum corr. x2 . time Pred.m) wind speed & direction bin that belongs to. yN 12-14 14-16 16-18 18-20 k th DIR bin Table: m/s 0-2 2-4 4-6 6-8 8-10 10-12 12-14 14-16 16-18 18-20 NE U(Inc)mean & DIR(Inc)mean E SE S SW W NW N time l th U bin ? m th DIR bin 1... . coef. .. yN 4. Wind Direction: Use the differences of the mean bin angles as a displacement angle.∆Τ 2 Ref. x1 . Thus: ∀ (ui . Corr. xN that belong to the (n. time Shift time series by ΔΤ Ref. + ∆Τ Prediction method Wind Speed: Use the regression coefficients of the wind speed bin to obtain the predicted value. k) bin. U (t) dt Ref. data x1 . . coef.. Isolate the Ref.k) and (l. the predicted (u'i ..k) ui + b(n. Calculate the mean value ymean of the simultaneous data y1 . 4 Inc. t Inc. using the Tables with the results of the statistical analysis Table: Incomplete time series U (t) Ref.m) denote the Reference and Incomplete time-series bins. Apply the time shift dt time Table: U (t) Ref. y2 . Locate the (l.m) Mean ..k) Mean The (n.φ(n. Calculate their mean value: xmean 3.. PAGE 44 . -∆Τ dt where: Δφ = φ(l. respectively.. b: ymean = axmean+ b · goodness of fit : R2 · correlation coefficient : ρ Methodology of the correlation – prediction procedure.k) wind speed & direction bin..k) .. . . y2 . xN m/s NE E 5 Correlate the synchronized data U(Ref)mean & DIR(Ref)mean SE S SW W NW N dt n th U bin 0-2 2-4 4-6 6-8 8-10 10-12 Inc. φi ) ∈ (n. φ'i ) is: u'i = a(n.WINDROSE : Wind Data Analysis software 1 Locate common data U (t) Reference time series 6 Generate Predicted time series. 5. Calculate for each bin the following: · linear regression coefficients a.. dt U(Ref)mean & DIR(Ref)mean m/s 0-2 2-4 4-6 6-8 8-10 10-12 12-14 14-16 16-18 18-20 NE E SE S SW W NW N (n. 2.

WINDROSE : Wind Data Analysis software APPENDIX Advanced Options Spike detection / Warnings.5 If 5<u<10 then ε=ε * 2.8 Error: ε = 2. even if some of the advanced parameters were modified (they appear in red in TempData sheet). If “strange” values are noticed. The alarm conditions are summarized below. they are all stored into the TempData sheet for traceability reasons. When multiple anemometers are used. the way spikes are detected. After each run. (2-3) and (3-4). then a warning or an error is produced and an entry is added into the Log file. These parameters include among others: the acceptable limits of each quantity. They are based on the acquired experience and refer to each pair of successive anemometers and/or vanes. then 3 successive tests will be performed for the anemometer pairs (1-2). the tests to be performed to check “frozen” anemometers. Consequently. Most of them are straightforward and self-explanatory.5 ● Max Wind Speed Vertical Profile U2 / U1 < ε Warning: ε = 1. errors The “Advanced Options…” menu of the WindRose includes various parameters which affect the program’s behaviour but (most of the times) are not related to a specific site and consequently can be applied to several WindRose.2 Error: ε = 1. etc. ● Mean Wind Speed Vertical Profile U2 / U1 < ε Warning: ε = 1. it should be always possible to reproduce the same results. if a mast has 4 anemometers.xls files. a check is performed between the (simultaneous) wind speed values per height. Note that.0 PAGE 45 . Changing the preset parameters of the WindRose program.0 The final value of ε depends on the wind speed: If 0<u<5 then ε=ε * 2.

5 If 15<u<65 then ε=ε * 1. List of the Advanced Options of the WindRose program. PAGE 46 .0 ● Mean Wind Direction Vertical Profile Δφ (for U > 5m/s) Warning: Δφ = 25deg Error: Δφ = 45deg ● Spike detection (for consecutive time steps) Umean: 10m/s Umax: 13m/s ● SDV constant or invariant during consecutive time steps.WINDROSE : Wind Data Analysis software If 10<u<15 then ε=ε * 1.

41 Umax(30m)=12.25588 ] Then. 3 empirical methods are used to deduce the air density. The last depends on the temperature and is calculated as follows: Pw = 2.25577 ⋅10 −5 ⋅ H hub [( ) 5.05 ⋅ 10 −7 ⋅ e 0. then air-density estimation due PAGE 47 .76 Umax(30m)=12. assumed value: Hbaro=2m). 1013. as recommended by IEC (sea level. the air density can be calculated using the following equation: ρ= Bcorr φ ⋅ 0. T is the measured temperature (in °Κ).25588 − 1 − 2.25 ⋅ 1 − 2.25577 ⋅10 −5 ⋅ H baro ( ) 5.25mbar). while: Umean(45m)=7. according to the following formula: B Bcorr = Bmeas + 1013. Case B: Atm. for accurate air density measurements. temperature (and relative humidity) have to be measured simultaneously with the wind speed.378 ⋅ Pw − R ⋅T R ⋅T where: R=287 Jοule deg-1 kg-1 . then a constant value of 0. airdensity varies considerably and its calculation has to be done accurately.55. thus Bmeas data are “extrapolated” to hub height Hhub. pressure is not recorded.txt Line: 21742 01/07/2000 21:00 Dir Error: Values beyond the error limit for the vertical profile Dir1= 230. 15°C.3. atmospheric pressure is measured at a low height (ie: data logger level.txt Line: 5334 Air density variation with height WindRose expects power curves (given in the PowerCurve sheet) to be normalized.WINDROSE : Wind Data Analysis software Examples of the WindRose log file when an error occurs: 16/07/2004 14:30 Umax Error: Value beyond the error limit for the vertical profile Umax(45m)=39.64 Umean(30m)=9.0631846⋅T Finally.98 File: c:\data\site1.5 (50%) is assumed. At higher altitudes and different temperatures. atmospheric pressure. pressure time-series not available When atm.51 Umean(30m)=6. In the opposite case.93 Umax(45m)=13. Ideally. pressure and temperature data Usually.0 Dir2= 329. φ the relative humidity και Pw the vapor pressure (in hPa).87 File: c:\data\site1. while: Umean(45m)=10. since it influences significantly the energy calculations. Analytically: Case A: using atm. if relative humidity φ data are not available.

4th Table): 1st Method This method. conclude to the equation: ρ = ρ0 T0 ⎛ Γz ⎞ ⎜1 − ⎟ T ⎜ T0 ⎟ ⎝ ⎠ PAGE 48 .WINDROSE : Wind Data Analysis software to the height (site elevation + anemometer height) can be done as follows. employs an empirical formula in which air density is a function only of the height [m] from the sea level: ρ = 1. ρ the air-density and z the height. therefore: dp = − gρ dz where g=9. depending on the user’s selection (Input sheet. mainly used when temperature and pressure measurements are not available.Γz With Τ0 = 288°Κ (=273+15) and Γ is the vertical temperature gradient usually taken as: 6. we obtain: dp g =− dz p RT Assuming that the temperature Τ varies with height: Τ = Τ0 . Integrating the above equation: dp g dz ∫ p = − R ∫ T0 − Γz 0 p0 p z and taking the logarithm: ln p g ⎛ Γz ⎞ ln⎜1 − ⎟ = p0 RΓ ⎜ T0 ⎟ ⎠ ⎝ g ΓR We. Using the ideal gas equation: p = R ρT where R=287 Jοule deg-1 kg-1 is the gas constant. p the atmospheric pressure and T the temperature in Kelvin [°K].81m/s² is the gravity acceleration.226 ⋅ e −3.5°Κ/km. finally.1089⋅10 2nd Method −5 ⋅Height Another method assumes the atmosphere’s adiabatic variation.

WINDROSE : Wind Data Analysis software that relates the air density to the temperature Τ and the height z of a site. then a mean temperature value for the measurement period has to be set in the fourth table of the Input sheet. Note that. These values can be retrieved either from a nearby meteorological station or set intuitively. 3rd Method This method uses 12 preset values. Two methods exist for the calculation of these parameters: 1st Method: Using only the Mean and SDV values Solving {1} for k and C. requires that the mean value μ and the standard deviation σ must be expressed as a function of k and C. one per each month of the year. At the end of the document a comparison table is given showing the differences of the first two methods. From the mean-value definition: ⎛k ⎞ ⎛u⎞ μ = Ε(u ) = ∫ uf (u )du = ∫ u ⎜ ⎟ ⎜ ⎟ c⎠⎝c⎠ 0 0 ⎝ ⎛u⎞ Using the transformation t = ⎜ ⎟ we obtain: ⎝c⎠ k ∞ ∞ k −1 e ⎛u⎞ −⎜ ⎟ ⎝c⎠ k du u = ct and Therefore: 1 k ⎛c⎞ ⎛u⎞ ⎜ ⎟dt = ⎜ ⎟ ⎝k⎠ ⎝c⎠ k −1 ⎛u⎞ −⎜ ⎟ k ⎝c⎠ k −1 du μ = ∫ u ⎜ ⎟⎜ ⎟ 0 ∞ ⎛ k ⎞⎛ u ⎞ ⎝ c ⎠⎝ c ⎠ ∞ e du = c ∫ t k e −t dt 0 ∞ 1 Recalling the gamma function definition: Γ(q ) = ∫ t q −1e −t dt . if the ASCII data files do not contain temperature columns. q > 0 0 we deduce that: PAGE 49 . Weibull distribution The probability density function ƒ(u) of the Weibull distribution is given by the formula: ⎛k ⎞⎛u⎞ f (u ) = ⎜ ⎟ ⎜ ⎟ ⎝c⎠⎝c⎠ k −1 e ⎛u⎞ −⎜ ⎟ ⎝c⎠ k {1} where k and C are the characteristic parameters of the distribution.

5) e − x 2π A( x ) {5} where: A( x ) = 1 + with : a1 a 2 a3 a 4 + + + +L x x2 x3 x4 a1 = 1 1 139 571 . starting from an initial value. a2 = . For the (ν) iteration we can PAGE 50 . a3 = − . we have: ⎛ E (u 2 ) = c 2 Γ⎜1 + ⎝ So. for the standard deviation σ is valid to write: E (u − μ ) = σ 2 = E (u 2 ) − μ 2 2 ⎛ ⎝ 1⎞ k⎠ {2} { } Following the same transformation. finally we conclude: 2⎞ ⎟ k⎠ σ 2 = c 2 ⎨ Γ⎜ 1 + ⎟ − Γ 2 ⎜ 1 + ⎟ ⎬ ⎧ ⎛ ⎩ ⎝ 2⎞ k⎠ ⎛ ⎝ 1 ⎞⎫ k ⎠⎭ {3} The Weibull distribution in its cumulative form P(u≤u1) is written as: ⎛ k ⎞⎛ u ⎞ P (u ≤ u1 ) = ∫ ⎜ ⎟⎜ ⎟ c ⎠⎝ c ⎠ 0⎝ u1 k −1 e ⎛u⎞ −⎜ ⎟ ⎝c⎠ k du or: P (u ≤ u1 ) = 1 − e ⎛u ⎞ −⎜ 1 ⎟ ⎝ c ⎠ k {4} The following asymptotic series is used as good approximation of the gamma function Γ: Γ( x ) = x ( x −0. a4 = − L 51840 2488320 12 288 1 we finally obtain the k 2 Dividing {2} and {3} and setting x = following non-linear x -function: ⎛σ ⎞ Γ(1 + 2 x ) g( x) = 2 −1− ⎜ ⎟ = 0 ⎜μ⎟ Γ (1 + x ) ⎝ ⎠ Resolving g(x) = 0 is achieved by applying the Newton iterative method.WINDROSE : Wind Data Analysis software μ = cΓ ⎜ 1 + ⎟ Now.

if we plot a graph with the Χ and Υ axes to be ln(u) and PAGE 51 .WINDROSE : Wind Data Analysis software write: x (ν ) =x (ν −1) g (ν −1) − (ν −1) g′ Its 1st x derivative of g(x) is: g′ = ∂g ( x ) Γ′(1 + 2 x ) Γ(1 + x ) − 2Γ(1 + 2 x )Γ′(1 + x ) = Γ 3 (1 + x ) ∂x ∂Γ() ∂x where Γ′() = The gamma function’s derivative Γ′(x ) is calculated using the ′ relationship (x x ) = x x (ln x + 1) that finally gives: Γ′( x ) = B ( x ) − Γ( x ) x Using {5}. leads to the following form for the B (x ) function: ⎧ ⎤ ∂A( x ) ⎫ ⎡1 B( x ) = 2πx x x e − x ⎨ A( x ) ⎢ + ln x ⎥ + ⎬ ∂x ⎭ ⎦ ⎣ 2x ⎩ Keeping just the first 4 terms we obtain: 3a ⎫ ⎧ 2a ⎡1 ⎤ a B( x ) = 2πx x x e − x ⎨ A( x ) ⎢ + ln x ⎥ − 1 − 32 − 43 ⎬ 2 x x ⎭ ⎣ 2x ⎦ x ⎩ As a result. 2nd Method: Using the data distribution (Weibull paper) This method (also referred as “Weibull paper”) can be applied only if the wind speed distribution is already known. that is the asymptotic approximation of the Gamma function.k ln(C) Therefore.: the probability that the wind speed is greater than a given value) is written as: Q (u ≥ u1 ) = e ⎛u ⎞ −⎜ 1 ⎟ ⎝ c ⎠ k Taking twice the logarithm. The cumulative form of the Weibull distribution (i. B(x) is approximated permitting the calculation of the g’(x) th at the ν iteration. we obtain: ln[ .e.ln(Q(u) ] = k ln(u) .

The formulas used are the following: Α/Γ stall control: P = Pnorm ⋅ ρ ρ norm − Α/Γ pitch control: ⎛ ρ ⎞ ⎟ U = U norm ⋅ ⎜ ⎜ρ ⎟ ⎝ norm ⎠ 1 3 PAGE 52 . Note that. Power Curve correction Following the IEC-61400-12 recommendations. the power curve is corrected depending the power control (stall. so that extreme values (accentuated by the logarithms) will not affect the results. It is suggested to fit the data within the range: 4-16m/s.WINDROSE : Wind Data Analysis software ln[ . then the Weibull distribution becomes a straight line with slope k and intercept point to Y-axis the quantity: -k ln(C). in the Input sheet it is given the possibility to “restrict” the wind speed range in which the fit is performed.ln(Q(u) ] respectively. from which the C parameter can be calculated. pitch control) of the wind turbine.

XLS Security Warning Under certain configurations (i. Macro.e. you may not be able to open the WindRose. PAGE 53 . Choose Then.: Office XP and/or Windows XP professional). select View Certificate and afterwards Install Certificate. The purpose is to end up with the WindRose macro to be among the Microsoft Excel’s trusted sources. Set it to Medium. WindRose. A security warning appears. This will prevent WindRose to function properly. The following steps need to be taken to overcome this situation. open Details…. Now. open Excel (without Security.WINDROSE : Wind Data Analysis software INSTALLATION ISSUES WindRose.XLS). go to Tools. Therefore.XLS. WindRose.XLS file with the included macro enabled.

come back to the previous screen. PAGE 54 . you may check that WindRose is among Excel’s Trusted Macros. Then. check the box Always trust macros from this source. Click on Enable Macros and start working with WindRose. where the Enable Macros option is not anymore grayed (if not close Excel and re-open WindRose.XLS. you may re-set the Security Level to High. Finally.XLS). Moreover.WINDROSE : Wind Data Analysis software Now.

“Wind Energy Conversion Systems”. Z. 1989. Practice hall. J. May 1983. 6. Models and Methods for Engineering Applications”. Sorbjan. “Atmospheric Science. Inc. L. H. “Weibull coefficients estimation”. June 1999. E. Wallace. IEC-61400-12 International Electrotechnical Commission. John. 2. Academic Press. 1990 8. April 1994. Panofsky. “Wind data analysis ”.WINDROSE : Wind Data Analysis software REFERENCES 1. Lysen. “Atmospheric turbulence.H. “Introduction to Wind Energy”. PAGE 55 . D. “Wind Turbine Generator Systems. 5. The Netherlands. USA. CWD. Part 12: Wind turbine Power Performance Testing”. Foussekis. Peter V. 1984. Prentice Hall. L. 1997. 4. 3. Hobbs. 1997. 7. Dutton. An Introductory survey”. Centre for Reneable Energy Sources (CRES). John-Wiley & Sons. Freris. M. D. “Structure of Atmospheric Boundary layer”. Douvikas. CRES.

991 0.052 1.088 1.267 1.098 1.064 -5 1.056 1.130 1.071 1.087 1.038 1.232 1. Comparison of the 2 methods (Case B) Method 1 Height 1.244 1.164 1.316 1.154 1.043 0 1.964 0.129 1.139 1.162 1.068 1.084 1.147 1.159 1.925 0.143 1.096 1.225 1.055 1.163 1.025 1.202 1.067 1.957 0.131 1.988 0.034 1.021 1.107 1.121 1.256 1.974 0.069 1.099 1.060 1. .184 1.149 1.138 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 -10 1.220 1.118 1.008 0.919 0.945 0.188 1.304 1.935 0.167 1.003 0.167 1.175 1.118 1.050 1.048 1.977 0.141 1.316 1.164 1.951 0.110 1.198 1.976 0.151 1.017 1.177 1.255 1.024 1.106 1.107 1.097 1.129 1.043 1.170 1.971 0.120 1.006 0.928 0.161 1.112 1.001 0.280 1.053 1.141 1.140 1.223 1.086 1.200 1.125 1.158 1.174 1.101 1.058 1.190 1.045 1.077 1.938 0.942 0.215 1.232 1.140 1.966 20 1.994 1.073 1.122 1.232 1.181 1.304 1.162 1.145 1.174 1.152 1.031 1.177 1.123 1.994 0.192 1.107 1.910 0.151 1.101 1.114 1.204 1.267 1.054 1.211 1.208 1.268 1.998 0.255 1.152 1.341 1.174 1.996 0.085 1.153 1.900 Comparison of the first two methods for the air-density variation with height.222 1.142 1.136 1.959 0.200 1.075 1.109 1.257 1.015 1.984 0.978 0.011 1.046 1.059 1.209 1.047 1.196 1.132 1.004 0.075 1.076 1.117 1.208 1.128 1.986 0.108 1.118 1.235 1.174 1.156 1.984 15 1.197 1.090 1.961 0.035 1.064 1.181 1.949 25 1.Air-Density [kg/m³] as a function of Elevation.145 1.037 1.112 1.220 1.034 1.932 30 1.245 1.074 1.005 1.029 1.103 1.145 1.968 0.129 1.218 1.066 1.163 1.280 1.096 1.243 1.024 1.186 1.210 1.027 1.151 1.140 1.165 1.080 1.211 1.091 1.292 1.027 1.203 1.079 1.292 1.013 0.092 1.087 1.033 1.197 1.207 1.226 1.193 1.019 1.997 0.185 1.981 0.065 1.213 1.119 1.187 1.243 1.987 0.055 1.179 1.085 1.039 1.185 1.197 1.185 1.096 1.188 1.076 1.233 1.063 1.014 1.150 1.156 1.329 1.134 1.134 1.081 1.042 1.007 0.280 1.916 35 1.172 1.220 1.954 0.064 1.170 1.967 0.016 1.948 0.269 1.095 1.292 1.221 1.044 1.247 1.023 Method 2 Temperatures [°C] 5 10 1.