Comparison of 3TIER Wind Simulations and Data:
Springerville – Tower 0506
Prepared for
National Renewable Energy Laboratory
Prepared by
Carson Pete, MS Candidate
Matt Acker, BS Candidate
Dr. Tom Acker, Department of Mechanical Engineering
November 20, 2008
Table of Contents
Table of Contents
1.Executive Summary
2.Background
2.1.NAU Data
2.2.NREL Data
2.3.MET Data
3.Description of Location
4.Wind Speed Characteristics
4.1.Yearly Characteristics
5.Conclusion
List of Figures:
Figure 1: NAU APA/APS wind speed simulation study area
Figure 2: Google Earth map of Springerville with data locations identified
Figure 3: TOPO map of Springerville with data locations identified.
Figure 4: Diurnal characteristics for average wind speed.
Figure 5: Monthly averaged wind speeds for reference period.
Figure 6: Daily averaged wind speeds for reference period.
Figure 7: 50m met tower Weibull distribution graph for hourly wind speeds.
Figure 8: NAU 50m Weibull distribution graph for hourly wind speeds.
Figure 9: NAU 100m Weibull distribution graph for hourly wind speeds.
Figure 10: NREL 100m Weibull distribution graph for hourly wind speeds.
Figure 11: NEW NREL 100m Weibull distribution graph for hourly wind speeds.
Figure 12: 50m met tower histogram for hourly wind speed changes.
Figure 13: NAU 50m histogram for hourly wind speed changes.
Figure 14: NAU 100m histogram for hourly wind speed changes.
Figure 15: NREL 100m histogram for hourly wind speed changes.
Figure 16: New NREL 100m histogram for hourly wind speed changes.
List of Tables:
Table 1: Data site locations and available data dates.
Table 2: Mean wind speed for July 2005 through March 2006.
Table 3: Year statistics for reference tower
Table 4: Yearly diurnal correlation matrix
Table 5: Monthly statistics for July 05 through March 06.
Table 6: Correlation matrix for monthly averaged data.
Table 7: Daily statistics data for reference period.
Table 8: Correlation matrix for daily data.
Table 9: Weibull parameters for distribution of hourly wind speeds.
Table 10: Statistical data for reference period.
Table 11: Correlation matrix for hourly wind speed changes.
Table 12: Mean wind speed for July 2005 through March 2006.
1.Executive Summary
NorthernArizonaUniversity (NAU) conducted a comparison and validation study oftwo different wind speed simulations and two different wind power simulations. The wind speed simulations were compared to measured anemometer data. The two wind power simulations were compared against each other.
NAUMetTower 0505 near Springerville, AZ was used in the comparison with available data from July 2005 through March 2006. Data point 1659 from the National Renewable Energy Lab (NREL) 3TIER wind data set was used for the wind speed and power comparison. In addition, new NREL data was reran for 1/3 of the time blocks, each time block consisting of 3 days, which had become corrupted or gone unstable while running the model. APS data point 1178 from the NAU 3TIER wind speed data set was used for the wind speed comparison. Table 1 below lists the available data and site locations.
Table 1: Data site locations and available data dates.
Latitude(°) / Longitude (°) / Location / Site ID # / Available data
MET / 34.195 / 109.503 / Springerville / 505 / July 05 - March 06
NAU / 34.181 / 109.52 / Springerville / APS 1178 / 1996- 2006
NREL / 34.208 / 109.508 / Springerville / 1659 / 2004 - 06
NEW NREL / 34.208 / 109.508 / Springerville / 1659 / 2004 - 06
Table 2 below shows the range of the mean monthly wind speeds as well as the mean yearly wind speeds for Springerville from July 2005 thru March 2006. The yearly wind speed comparisons show a strong positive correlation between the measured data and the NREL simulation. On average, the NREL data set had a higher correlation coefficient than the NAU data set. Although the NREL data set is strongly correlated, it some cases, it over predicts wind speeds by nearly 20%.
Table 2: Mean wind speed for July 2005 through March 2006.
MET 50m / NAU 50m / NAU 100m / NREL 100m / NEW NREL 100mMean Yearly Wind Speed (m/s) / 6.105 / 6.455 / 7.107 / 7.699 / 7.734
Although the NREL data set is strongly correlated, it over-predicts wind speeds on a yearly mean basis when compared to the met tower data.
2.Background
Northern Arizona University (NAU) conducted a comparison and validation study oftwo different wind speed simulations and two different wind power simulations for the state of Arizona. Both of the wind speed and wind power simulations were conducted by 3TIER Environmental Forecast Group. One set of simulation data was conducted for the northern portion of the state of Arizona for NAU and one for the Western United States for the National Renewable Energy Lab (NREL). Wind anemometer data was used to compare against the two different wind speed simulations. Wind data was collected from 7/1/05 through 3/5/06. Meteorological (“met”) tower data was measured at several locations in Arizona using 30m, 40m, and 50m met towers. One of these towers, a 50m met tower manufactured by NRG, was erected in Eastern Arizona near Springerville. The data set from the Springerville tower will be referred to throughout the report as the MET data.
2.1.NAU Data
The data produced for NAU by 3TIER consists of three data sets: a 5km, 1-minute resolution wind speed simulation, a 1km, 10-minute resolution wind speed simulation at two specific locations, and a modeled power simulation using the 3TIER SCORE methodology. The 5km simulation was conducted from 1996-2006 for the two broad regions of Arizona seen in Figure 1,and are referred to as the APA zones. The 1km simulation was conducted from 2003-2005 for Aubrey Cliffs and Grey Mountain, two regions of interest for further development. Both the 1km and 5km simulations include wind speed and direction data at heights of 10m, 20m, 30m, 50m, 80m, 100m, and 200m, temperature at 0m, 20m, and 50m, pressure, precipitation, solar insolation at 0m, and specific humidity at 2m.
Figure 1: NAU APA/APS wind speed simulation study area
2.2.NREL Data
The data provided by NREL consists of wind speed data and power data at 100m from 2004-2006 for the Western US using 2km grid spacing. The modeled power data uses 3TIER’s SCORE-Lite methodology for predicting power. The complete sets of data can be found at . In addition, new NREL data was obtained for the site location. The model was rerun for 1/3rd of the time blocks with that had become unstable or corrupted. The model is run for 3 day time period blocks. This data will be referred as NEW NREL data. Future data will be provided for wind speeds and directions at various elevations.
2.3.MET Data
NAU and Tucson Electric Power (TEP) have approximately 20 meteorological towers located throughout Arizona, consisting of 30m, 40m, and 50m height towers. Each tower’s data logger samples data every 1 second. This raw data is 10 minute averaged with standard deviation, max and min. The towers measure wind speed and direction at the max height (30m, 40m, or 50m) of the tower as well as wind speed and temperature at 10m. Data that will be used for the comparisons has been collected from 7/1/05 through 3/5/06.
3.Description of Location
A primary criterion for choosing a location for the comparison is that data from all three sources are readily available at a common location. The three data sources are from a meteorological tower (MET), 3TIER NAU simulation, and 3TIER NREL simulation. Having quality data that is complete for an acceptable time frame for the MET towers is another important criterion for choosing a location. Using Google Earth and TOPO! to map the data locations as seen in Figure 2 and Figure 3, respectively.It was found that the Springerville tower 505 in eastern Arizona had the most complete MET data. Additionally, NAU and NREL wind speed and wind power data was available in an acceptable range.
Springerville MET tower 505 has complete data from July 2005 to March 2006 which will be the reference period for the wind speed comparisons. The data location for NAU was APS location 1178 and the NREL location used was data point 1659. The NREL location is roughly a mile away from the MET tower and has similar terrain characteristics as the MET tower.
Figure 2: Google Earth map of Springerville with data locations identified
Figure 3: TOPO map of Springerville with data locations identified.
4.Wind Speed Characteristics
Each comparison plot is followed directly by a table of statistics and correlation matrix, respectively. Each table of statistics includes the mean, standard deviation, and variance for each plot as seen in Table 3 below. Each correlation matrix identifies the correlation coefficient, r, associated with each data set as seen in Table 4 below. The format of each comparison data set will follow: plot, table of statistics, and correlation matrix.
4.1.Yearly Characteristics
The comparison for the yearly wind speed characteristics consists of 5 comparisons: diurnal wind speed characteristics as seen in Figure 4, monthly averaged wind speeds as seen below in Figure 5, daily averaged wind speeds as seen below in Figure 6, and histogram of hourly wind speeds, the Weibull probability density function (PDF) overlaid on the histogram of hourly wind speeds shown below in Figure 7 through Figure 11, and lastly histograms of hourly changes in wind speed as shown below in Figure 12 through Figure 16.
Figure 4: Diurnal characteristics for average wind speed.
Table 3: Year statistics for reference tower
Reference Period (July 05 – March 06) / MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100Mean for Reference Period (m/s) / 6.1048 / 6.455 / 7.1071 / 7.6989 / 7.7338
Standard Deviation for Reference Period / 0.62028 / 0.35075 / 0.4628 / 0.73277 / 0.69791
Variance for Reference Period / 0.38474 / 0.12303 / 0.21419 / 0.53695 / 0.48708
As seen inTable 3,the highest yearly mean diurnal wind speed was recorded for the NEW NREL 100 data at 7.734 m/s while the lowest was recorded from the MET 50 data at 6.1048 m/s. It is also shown the highest standard deviation for the year was recorded from the NREL 100 data set at 0.73277, while the lowest was recorded for the NAU 50 data set at 0.35075. The highest variance was recorded from NREL 100 at 0.53695 while the lowest was recorded from NAU 50 at 0.12303.
Table 4: Yearly diurnal correlation matrix
MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100MET 50 / 1 / -0.138 / -0.23152 / 0.6600 / 0.6352
NAU 50 / 1 / 0.96047 / 0.24973 / 0.29617
NAU 100 / 1 / 0.33633 / 0.38361
NREL 100 / 1 / 0.99179
NEW NREL 100 / 1
Table 4 indicates the strongest yearly diurnal correlation was recorded between the NREL 100 and NEW NREL 100 data set at 0.99179, while the weakest correlation was recorded between the NAU 50 and MET 50 at -0.138.
Figure 5: Monthly averaged wind speeds for reference period.
Table 5: Monthly statistics for July 05 through March 06.
Reference Period (July 05 – March 06) / MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100Mean for Reference Period (m/s) / 6.1943 / 6.913 / 7.6486 / 7.8352 / 7.8795
Standard Deviation for Reference Period / 1.0723 / 2.0485 / 2.4203 / 1.5572 / 1.6361
Variance for Reference Period / 1.1497 / 4.1965 / 5.8578 / 2.4248 / 2.6
As seen inTable 5, the highest monthly mean wind speed for the year was recorded for the NEW NREL 100 data at 7.8795 m/s while the lowest was recorded from the MET 50 data at 6.1943 m/s. It is also shown the standard deviation for the year was recorded from the NAU 100 data set at 2.4203, while the lowest was recorded for the MET 50 data set at 1.0723. The highest variance was recorded from NAU 100 at 5.8578 while the lowest was recorded from MET 50 at 1.1497.
Table 6: Correlation matrix for monthly averaged data.
MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100MET 50 / 1 / 0.73285 / 0.72418 / 0.90069 / 0.90313
NAU 50 / 1 / 0.9995 / 0.78379 / .78786
NAU 100 / 1 / 0.78246 / 0.78603
NREL 100 / 1 / 0.99757
NEW NREL 100 / 1
Table 6 indicates the strongest yearly monthly averaged wind speed correlation was recorded between the NAU 100 and NAU 50 data set at 0.9995, while the weakest correlation was recorded between the NAU 100 and MET 50 at 0.724.
Figure 6: Daily averaged wind speeds for reference period.
Table 7: Daily statistics data for reference period.
Reference Period (July 05 – March 06) / MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100Mean for Reference Period (m/s) / 6.105 / 5.886 / 7.107 / 7.699 / 7.734
Standard Deviation for Reference Period / 3.507 / 3.259 / 4.194 / 4.643 / 4.800
Variance for Reference Period / 12.30 / 10.62 / 17.59 / 21.56 / 23.04
As seen inTable 7, the highest daily mean wind speed for the year was recorded for the NEW NREL 100 data at 7.734 m/s while the lowest was recorded from the NAU 50 data at 5.886 m/s. It is also shown the highest standard deviation for the year was recorded from the NEW NREL 100 data set at 4.8, while the lowest was recorded for the MET 50 data set at 3.507. The highest variance was recorded from NEW NREL 100 at 23.04 while the lowest was recorded from NAU 50 at 10.62.
Table 8: Correlation matrix for daily data.
MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100MET 50 / 1 / 0.772 / 0.769 / 0.815 / 0.820
NAU 50 / 1 / 0.997 / 0.797 / 0.820
NAU 100 / 1 / 0.798 / 0.819
NREL 100 / 1 / 0.987
NEW NREL 100 / 1
Table 8 indicates the strongest yearly monthly averaged wind speed correlation was recorded between the NAU 100 and NAU 50 data set at 0.997, while the weakest correlation was recorded between the NAU 100 and MET 50 at 0.769.
For visual clarity, below in Figure 7, the Weibull PDF for the reference MET tower is plotted in red whereas each specific Weibull PDF is plotted in black. A table of Weibull parameters can be found below in Table 9 below.
Figure 7: 50m met tower Weibull distribution graph for hourly wind speeds.
Figure 8: NAU 50m Weibull distribution graph for hourly wind speeds.
Figure 9: NAU 100m Weibull distribution graph for hourly wind speeds.
Figure 10:NREL 100m Weibull distribution graph for hourly wind speeds.
Figure 11: NEW NREL 100m Weibull distribution graph for hourly wind speeds.
Table 9: Weibull parameters for distribution of hourly wind speeds.
MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL100
Weibull C (m/s) / 6.7767 / 7.2169 / 7.9098 / 8.5126 / 8.5444
Weibull K / 1.4642 / 1.5798 / 1.5086 / 1.4233 / 1.4156
As seen inTable 9, the highest Weibull C and K parameters for the distribution of hourly wind speeds is 8.54 and 1.58 for NEW NREL 100 and NAU 50, respectively. The lowest Weibull C and K parameters were 6.78 and 1.42 for MET 50 and NEW NREL 100, respectively.
Figure 12: 50m met tower histogram for hourly wind speed changes.
Figure 13: NAU 50m histogram for hourly wind speed changes.
Figure 14: NAU 100m histogram for hourly wind speed changes.
Figure 15: NREL 100m histogram for hourly wind speed changes.
Figure 16: New NREL 100m histogram for hourly wind speed changes.
Table 10: Statistical data for reference period.
Reference Period (July 05 – March 06) / MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100Mean for Reference Period / -5.465e-4 / 9.234-4 / 1.161e-3 / 4.307e-4 / 4.307e-4
Standard Deviation for Reference Period / 1.7189 / 1.051 / 1.1506 / 1.8594 / 1.6462
Variance for Reference Period / 2.9546 / 1.1046 / 1.324 / 3.4575 / 2.7101
As seen inTable 10, the highest mean for the hourly change in wind speed for the year was recorded for the NAU 100 data at 1.161e-3 while the lowest was recorded from both the NREL 100 and NEW NREL 100 at 4.307e-4. It is also shown the highest standard deviation for the hourly change in wind speed data was recorded from the NREL 100 data set at 1.86, while the lowest was recorded for the NAU 50 data set at 1.051. The highest variance was recorded from NREL 100 at 3.46 while the lowest was recorded from NAU 50 at 1.11.
Table 11: Correlation matrix for hourly wind speed changes.
MET 50 / NAU 50 / NAU 100 / NREL 100 / NEW NREL 100MET 50 / 1 / 0.080559 / 0.07158 / 0.093993 / 0.10893
NAU 50 / 1 / 0.98056 / 0.11195 / 0.12084
NAU 100 / 1 / 0.11549 / 0.12239
NREL 100 / 1 / 0.84312
NEW NREL / 1
Table 11 indicates the strongest hourly wind speed change correlation was recorded between the NAU 100 and NAU 50 data set at 0.98056, while the weakest correlation was recorded between the NAU 100 and MET 50 at 0.07158.
5.Conclusion
Northern Arizona University conducted a comparison and validation study oftwo different wind speed simulations and two different wind power simulations. The wind speed simulations were compared to measured anemometer data taken in the field.
The Table 12 below shows the range of mean yearly wind speed for Springerville from July 2005 thru March 2006.
Table 12: Mean wind speed for July 2005 through March 2006.
MET 50m / NAU 50m / NAU 100m / NREL 100m / NEW NREL 100mMean Yearly Wind Speed (m/s) / 6.1048 / 6.455 / 7.1071 / 7.6989 / 7.7338
The yearly wind speed comparisons show a strong positive correlation between the NREL data and the NEW NREL data simulation, followed by NAU 100 data and NAU 50 data. The weakest correlation was noted by NAU 100 and MET 50 the majority of the time.
In many cases the monthly Weibull parameters did not match up when plotted against the “yearly characteristics”. This could be due to days in the month, time of the year, location sites, and amount of data collected, such as the month of March 2006 have only partial of the month data collected.
Although the NREL data set is strongly correlated, it over predicts wind speeds by nearly 20% on a yearly mean basis. The range of wind speeds found at Springerville varies from class 2 to class 8.
.
1