Loading...
HomeMy WebLinkAboutIndependent Wind Resource and Energy Assessment - Shovel Creek - Apr 2025 - REF Grant 7014029 UL Services Group LLC 23 British American Blvd. | Latham, NY 12110 | USA www.ul.com/renewables ©2025 UL Services Group LLC CLASSIFICATION CLIENT'S DISCRETION ISSUE A Energy Production Report Independent Wind Resource and Energy Assessment PREPARED FOR: Golden Valley Electric Association Ref. No.: PR-177068 SHOVEL CREEK Fairbanks North Star Borough, Alaska 16 April 2025 Shovel Creek Page ii Ref. No.: PR-177068 Issue: A Status: Final Golden Valley Electric Association 16 April 2025 NOTICE TO THIRD PARTIES1 This report was prepared by UL Services Group LLC (“UL Solutions”) and is based on information not within the control of UL Solutions. UL Solutions has assumed the information provided by others, both verbal and written, is complete and correct. While it is believed the information, data, and opinions contained herein will be reliable under the conditions and subject to the limitations set forth herein, UL Solutions does not guarantee the accuracy thereof. Use of this report or any information contained therein by any party other than the intended recipient or its affiliates, shall constitute a waiver and release by such third party of UL Solutions from and against all claims and liability, including, but not limited to, liability for special, incidental, indirect, or consequential damages in connection with such use. In addition, use of this report or any information contained herein by any party other than the intended recipient or its affiliates, shall constitute agreement by such third party to defend and indemnify UL Solutions from and against any claims and liability, including, but not limited to, liability for special, incidental, indirect, or consequential damages in connection with such use. GVEA may make the report publicly available for reference purposes only. GVEA should inform and seek written permission from ULS if this report is intended to be used for any other purposes. To the fullest extent permitted by law, such waiver and release and indemnification shall apply notwithstanding the negligence, strict liability, fault, breach of warranty, or breach of contract of UL Solutions. The benefit of such releases, waivers, or limitations of liability shall extend to the related companies and subcontractors of any tier of UL Solutions, and the directors, officers, partners, employees, and agents of all released or indemnified parties. KEY TO DOCUMENT CLASSIFICATION STRICTLY CONFIDENTIAL For recipients only CONFIDENTIAL May be shared within client’s organization UL SOLUTIONS INTERNAL ONLY Not to be distributed outside UL Solutions CLIENT’S DISCRETION Distribution at the client’s discretion FOR PUBLIC RELEASE No restriction 1 Notice to third parties Shovel Creek Page iii Ref. No.: PR-177068 Issue: A Status: Final Golden Valley Electric Association 16 April 2025 DOCUMENT CONTRIBUTORS AUTHOR SUPPORTING AUTHOR(S) REVIEWER(S) Arthanareeswaran Mayilsamy Senior Renewable Energy Analyst Anand Raj Renewable Energy Analyst Shannon Beebie Renewable Energy Analyst Katrina Mayba N.A. Wind Energy Lead Beanán O’Loughlin Global Technical Lead DOCUMENT HISTORY ISSUE DATE SUMMARY A 16 April 2025 Initial Report Shovel Creek Page iv Ref. No.: PR-177068 Issue: A Status: Final Golden Valley Electric Association 16 April 2025 TABLE OF CONTENTS 1. Introduction ............................................................................................................. 8 2. Site Description ...................................................................................................... 8 3. Wind Measurements ............................................................................................... 8 3.1 Meteorological Masts .......................................................................................... 8 3.2 Remote Sensing Devices .................................................................................... 9 3.3 Data Handling and Validation ........................................................................... 10 4. Wind Resource Characteristics ........................................................................... 11 4.1 Observed Statistics ........................................................................................... 11 5. Estimation of Long-Term Mean Wind Speed at Measurement Height .............. 13 6. Extrapolation to Hub Height ................................................................................ 15 7. Estimation of Long-term Energy production ...................................................... 15 7.1 The Sitewind System ......................................................................................... 16 7.2 Openwind ........................................................................................................... 17 7.3 Results ............................................................................................................... 18 8. Uncertainty Analysis ............................................................................................ 19 9. Summary ............................................................................................................... 21 Appendix A - Energy Production Losses .................................................................. 42 Appendix B – Monthly Diurnal Matrix ........................................................................ 46 Shovel Creek Page v Ref. No.: PR-177068 Issue: A Status: Final Golden Valley Electric Association 16 April 2025 LIST OF FIGURES Figure 1: Location of the Shovel Creek Wind Project in Alaska ................................................................ 22 Figure 2: Shovel Creek Monitoring Locations ............................................................................................ 23 Figure 3: Mast 101 Monitoring Configuration ............................................................................................. 24 Figure 4: Mast 9004 Monitoring Configuration ........................................................................................... 25 Figure 5: Lidar L1921 Monitoring Configuration ........................................................................................ 26 Figure 6: Views of Area Near Mast 101 towards East ............................................................................... 27 Figure 7: Views of Area Near Mast 102 towards Southeast ...................................................................... 27 Figure 8: Views of Area Near Mast 103 towards East ............................................................................... 28 Figure 9: Views of Area Near Mast 9004 towards East ............................................................................. 28 Figure 10: Mast 101 Observed Annual Wind Frequency Distribution and Fitted Weibull Curve ............... 29 Figure 11: Mast 101 and WRF Concurrent and Historical Monthly Mean Wind Speeds ........................... 29 Figure 12: Mast 101 and ERA5 Concurrent and Historical Monthly Mean Wind Speeds ......................... 30 Figure 13: Mast 101 Annual Diurnal Wind Speed and Shear Patterns ..................................................... 30 Figure 14: Monitoring Location Annual Wind Roses .................................................................................. 31 Figure 15: Reference Station Annual Mean Wind Speeds ........................................................................ 32 Figure 16: Scatterplot of Mast 101 and WRF Daily Mean Wind Speeds ................................................... 32 Figure 17: Scatterplot of Mast 101 and ERA5 Daily Mean Wind Speeds .................................................. 33 Figure 18: Shovel Creek Turbine Layout ................................................................................................... 34 Shovel Creek Page vi Ref. No.: PR-177068 Issue: A Status: Final Golden Valley Electric Association 16 April 2025 LIST OF TABLES Table 1: Monitoring Location Summary ..................................................................................................... 35 Table 2: Summary of Top-Level Anemometer Adjustments ...................................................................... 35 Table 3: Monitoring Location Monthly Wind Speeds and Data Recoveries (With Reconstructed data).... 35 Table 4: Remote Sensing Device Wind Speed Data Recovery with Height ............................................... 37 Table 5: Monitoring Location Observed Wind Resource Characteristics** ............................................... 37 Table 6: Mast 101 and Reference Coefficient of Determination Summary ............................................... 38 Table 7: Monitoring Location Long-Term Wind Speed Projection Summary ............................................. 38 Table 8: Lidar L1921 Shear Trend Table ................................................................................................... 38 Table 9: Lidar L1921 (Flow Corrected) Shear Trend Table ....................................................................... 38 Table 10: Extrapolation of Long-Term Wind Speeds to Hub Height .......................................................... 39 Table 11: Comparison of Observed and Predicted Speeds at 80 m ......................................................... 39 Table 12: Shovel Creek Wind Speed and Energy Production Detail ......................................................... 40 Table 13: Annual Production Estimates ..................................................................................................... 40 Table 14: Wind Speed and Energy Production Uncertainty Summary ...................................................... 41 Table 15: Estimated Energy Production and Net Capacity Factor at Five Confidence Levels .................. 41 Table A.1: Shovel Creek Detailed Energy Production Loss Accounting ................................................... 42 Table B.1: Shovel Creek P50 Monthly Diurnal Energy Matrix (MWh) ....................................................... 47 Shovel Creek Page 7 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 EXECUTIVE SUMMARY UL Services Group LLC was retained by Golden Valley Electric Association to evaluate the long-term wind resource and energy production potential of the proposed Shovel Creek wind project, located in central Alaska. The project, which will have a rated capacity of approximately 150 MW, is to consist of 18 Vestas V150-4.5 MW turbines with a rotor diameter of 150 m and a hub height of 105 m and 16 Vestas V136-4.3 MW (Low HH HWO) turbines with a rotor diameter of 136 m and a hub height of 82m. This report discusses the methods used to develop the wind resource, energy production and uncertainty estimates, and presents the results. The key aspects of the project and a summary of the analysis results are presented in the table below. While this project’s layout and turbine technology is still in development, the results can be considered preliminary in nature. Project Name Shovel Creek Project Location Alaska Rated Capacity 149.8 MW Turbine Model Vestas V150-4.5 MW Vestas V136-4.3 MW Rated Power 4.50 MW 4.30 MW Variant - Low HH HWO Rotor Diameter 150 m 136 m Weather/Climate Package Standard Weather Package Cold Weather Package Hub Height 105 m 82 m Number of Turbines 18 16 Total Number of Turbines 34 Array-Average Speed 6.81 m/s Gross Annual Production 486.2 GWh/yr Flow Effect Losses 3.4% Plant Losses 21.2% Total Losses 23.9% Net Annual Production (Capacity Factor) 370.2 GWh/yr (28.2%) P95 Production (Years 2-10) (Capacity Factor) 308.2 GWh/yr (23.5%) P99 Production (Annual) (Capacity Factor) 262.9 GWh/yr (20.0%) Shovel Creek Page 8 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 1. INTRODUCTION UL Services Group LLC (“UL Solutions”) was retained by Golden Valley Electric Association to evaluate the long-term wind resource and energy production potential of the proposed Shovel Creek Wind Project (the “Project”), located in central Alaska. The Project, which is proposed to have a rated capacity of approximately 150 MW, is to consist of 18 Vestas V150-4.5 MW turbines with a rotor diameter of 150 m and a hub height of 105 m and 16 Vestas V136-4.3 MW (Low HH HWO) turbines with a rotor diameter of 136 m and a hub height of 82 m. This report presents the preliminary results of UL Solutions’ analysis and briefly discusses the methods used to develop the wind resource, energy production, and uncertainty estimates. While the Project’s layout and turbine technology is still in development, the results can be considered preliminary in nature. 2. SITE DESCRIPTION The Shovel Creek wind project is approximately 40 km to the west of Fairbanks, Alaska and 50 km northeast of Nenana, Alaska. Its location is indicated on the regional map in Figure 1. The project area is situated in complex rolling hills, valleys and ridgelines. Figure 2 contains a topographic map of the project area. The mean turbine base elevation is about 699 m with an elevation range of approximately 305 m. UL Solutions has not conducted a site visit, but photographs provided by the Project indicate that the land cover primarily consists of shrubs and pine trees with a low surface roughness. 3. WIND MEASUREMENTS 3.1 Meteorological Masts Wind monitoring at the Shovel Creek project began in September 2022 with the installation of two monitoring masts, designated Mast 101 and Mast 102. Mast 103 was installed in October 2022. One additional mast, designated Mast 9004, was installed in October 2023 but raw data files are not available from this mast until the beginning of November 2023. All masts remain in operation and their locations are indicated in Figure 2. Table 1 presents basic information about the masts including their geographic coordinates, elevations, periods of record, and sensor types and heights. Since UL Solutions did not perform site visits, information about the masts contained in tower installation forms and site photos provided by the Project were used to inform the analysis. Raw binary files were received; each file contained 10-minute average wind speed, wind direction and temperature measurements, along with their standard deviations. 1-minute data was recorded at Mast 9004 for 10 months, though it was resampled to 10-minute intervals for appending to the other mast data. In addition, each file contained meteorological measurements from multiple sensor types. Calibrated NRG S1, Windsensor P2546C, Thies Clima First Class Advanced, Thies Clima First Class Advanced (Heated), NRG #40 anemometers have been used in the measurement program. Although the NRG #40 anemometers used in the measurement campaign were calibrated, the consensus slope (0.765 m/s/Hz) and offset (0.35 m/s) were applied to convert their raw logger counts to speed values based on UL Solutions research, which indicates that the results agree more closely with Class I anemometers, like the WindSensor P2546A, than when the calibrated (measured) coefficients are used.2 The sensor-specific calibration constants were employed to convert the raw logger counts to 2 Hale, E., “Memorandum: NRG #40 Transfer Function Validation and Recommendation”, AWS Truepower, 8 January 2010. Shovel Creek Page 9 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 speed values for the NRG S1, Windsensor P2546C, Thies Clima First Class Advanced, and Thies Clima First Class Advanced (Heated) anemometers. Mast 101 through 103 is a 60-m, guyed, tubular NRG Super 60m XHD TallTower that tapers from a diameter of 254 mm (10 in) at the base to a diameter of 203 mm (8 in) at approximately 30.9 m and then remains constant to the top. The wind speed has been measured at four heights, ranging from 30m to 60 m; the wind direction at two heights of 35.5 m and 55 m (Masts 102 and 103), and three heights at 35 m, 50 m and 55 m at Mast 101; and temperature at two heights of 3 m and 59 m. Two anemometers oriented roughly northeast and southeast are present at three levels (30 m, 45 m and 60m) except at 55 m; single anemometers, oriented toward the southeast, are present at the 55 m level. Mast 9004 is a 50-m, guyed, tubular NRG 50m XHD TallTower that tapers from a diameter of 254 mm (10 in) at the base to a diameter of 203 mm (8 in) at approximately 30.9 m and then remains constant to the top. The wind speed has been measured at four heights, ranging from 27.9 m to 49.9 m; the wind direction at three heights of 25.9 m, 41.9 m and 47.9 m; and temperature at two heights of 3 m and 49.9 m. Two anemometers oriented roughly northeast and southeast are present at three levels of 27.9 m,39.9 m and 49.9 m. A single anemometer, oriented towards the southeast, is present at the 47.9 m level. Each anemometer is mounted at the end of a 2.4-m horizontal boom, which provides a separation distance from the mast that exceeds the IEC-recommended minimum of seven mast diameters from a tubular tower.3 Figure 3 and Figure 4 present a view of the monitoring configuration on Mast 101 and Mast 9004 respectively. Views from the base of these masts, looking toward the east and southeast, the prevailing wind directions, are presented in Figure 6 through Figure 9. 3.2 Remote Sensing Devices The Project has conducted a remote sensing device (RSD) campaign utilizing a lidar (LIght Detection And Ranging) as part of the wind resource assessment program for the Shovel Creek wind project. A lidar instrument is a mobile device that uses light waves to remotely measure the wind speed and direction at heights that are well above typical mast heights and through the rotor plane of most commercial-scale turbines. A Lidar ZX300 (serial number 1921) was deployed at a location within the project area. This deployment – from 20 November 2023 until 16 September 2024 (Lidar L1921) – was adjacent to Mast 102. The monitoring location is indicated in Figure 2. Basic information about the lidar deployment, including geographic coordinates, elevation, period of record, and measurement heights, is included in Table 1. During the deployment, the lidar measured wind speed and direction at 11 heights ranging from 39 m to 200 m; external temperature measurement was collected. The lidar data were delivered in their raw format to UL Solutions where a series of quality control checks to screen and validate the data were employed. Since UL Solutions did not perform site visits, information about the remote sensing device was contained in commissioning form provided by the Project. Figure 5 shows the lidar configuration for the deployment collocated with Mast 102. UL Solutions was also provided complex flow corrected data for L1921 that was used to inform shear calculations. 3 Annex G, IEC 61400-12-1, “Power Performance Measurements of Electricity Producing Wind Turbines”, Geneva, Switzerland, 2005. Shovel Creek Page 10 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 3.3 Data Handling and Validation An experienced UL Solutions analyst inspected the data for completeness and reasonableness. The main issues addressed by the validation process included mast shadow effects, equipment failures, and icing. The following is a summary of the equipment failures and maintenance history of the monitoring devices. Mast 101 • 18 October 2022 – 05 January 2023: 50.0-m windvane failure. • 25 October 2022 – 16 September 2023: 45.0-m northeast-facing anemometer failure. • 11 March 2023 – 16 September 2023: 45.0-m southeast-facing anemometer failure. • 16 September 2023: Maintenance performed to replace all anemometers and wind vanes. Mast 102 • 26 October 2022 - 19 September 2023: 30.0 m northeast-facing anemometer failure. • 02 November 2022 - 19 September 2023: 35.5 m wind vane failure. • 16 December 2022 – 19 September 2023: 60.0 m southeast-facing anemometer failure. • 04 February 2023 – 19 September 2023: 60.0 m northeast-facing anemometer failure. • 31 March 2023 – 19 September 2023: 45.0 m southeast-facing anemometer failure. • 21 November 2023 – 22 October 2024: 45.0 m northeast-facing anemometer failure. • 26 April 2023 – 19 September 2023: 30.0 m southeast-facing anemometer failure. • 20 January 2024 – 22 October 2024: 59.0 m temperature sensor failure. • 19 September 2023: Maintenance performed to replace all anemometers (Except 55.0 m northeast-facing anemometer) and wind vanes. • 28 September 2024 – 22 October 2024: 60.0 m southeast-facing anemometer failure. Mast 103 • 12 March 2023 – 21 September 2023: 60.0-m southeast-facing anemometer failure. • 02 June 2023 – 21 September 2023: 30.0-m northeast-facing anemometer failure. • 02 June 2023 – 21 September 2023: 30.0-m southeast-facing anemometer failure. • 21 September 2023: Maintenance performed to replace all anemometers (Except 55.0 m northeast-facing anemometer) and wind vanes. Mast 9004 • 11 October 2023 – 09 November 2023: Missing Data. • 09 November 2023 – 25 July 2024: 47.9 m wind vane failure. • 25 July 2024: Maintenance performed to replace 47.9 m wind vane. • 11 May 2024 – 15 May 2024: Missing Data. Lidar L1921 • No issues The response of NRG #40 anemometers to turbulence differs from that of Class I anemometers used for turbine power performance testing and certification. High turbulence causes NRG #40 anemometers to overspeed (measure higher mean speeds) compared to Class I sensors, because they respond more quickly to gusts than to falling wind speeds. Conversely, at very low turbulence, speeds reported by NRG #40 anemometers tend to be below those reported by Class I sensors. UL Solutions has computed the following relationship to adjust the NRG #40 wind speed data to account for turbulence effects:4 𝑇𝐶𝑜𝑟𝑟𝑏𝑏𝑟𝑏𝑏=𝑇𝑂𝑏𝑟𝑏𝑟𝑣𝑏𝑏 / (0.095 ∗𝑇𝐼+0.992) 4 Filippelli, M.V., et al., “Adjustment of Anemometer Readings for Energy Production Estimates”, Proceedings of Windpower 2008, June 2008. Shovel Creek Page 11 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 The free-stream wind flow at the masts is characterized by low turbulence intensity. As a result, the adjustments, which range from, –0.0% (Mast 9004) to –0.2% (Mast 103), are small. All NRG #40 wind speed values in this report include this adjustment. Table 2 lists the mean adjustments applied to data from each top-level NRG #40 anemometer(s). For each mast, at measurement heights where two anemometers are present, a series of direction- based regression equations was developed using valid data from both sensors; the equations were used to reconstruct invalid readings at the same height whenever possible. UL Solutions has used a computational fluid dynamics (CFD) model to simulate the aerodynamic influences that a tubular tower has on the free stream wind flow5 and compared observed wind speeds with those of their modeled free-stream values. For masts having anemometer configurations that are consistent with IEC specifications, the average wind speeds derived from two concurrent independent measurements at the same monitoring level are consistent with the true free stream values. Therefore, UL Solutions has averaged all valid wind speed data samples at monitoring heights where two anemometers are employed; for direction sectors where one anemometer is shadowed by the tower, only valid wind speed observations from the unwaked anemometer are retained in the data stream. Anemometer data for the top measurement height at each mast were reconstructed with available data from the heated sensor measurements to ensure optimum data recovery for the analysis. The statistics presented in the report are indicative of this reconstruction. After data validation and adjustments, the wind speed data recovery rates ranged from 74.1% (60.0 m) at Mast 102 to 84.5% (60.0 m) at Mast 103. 4. WIND RESOURCE CHARACTERISTICS 4.1 Observed Statistics Table 3 presents the observed monthly mean wind speeds and data recovery rates for each monitoring location and Table 4 shows the data recovery rate by height of the RSD. Table 5 summarizes the wind resource characteristics observed over the periods of record at the onsite monitoring locations. The characteristics include the average and annualized average wind speeds, data recoveries, shear exponents, turbulence intensities, and Weibull parameters. The observed mean wind speeds ranged from 6.00 m/s (60.0 m) at Mast 103 to 7.06 m/s (60.0 m) at Mast 102. The annualized mean speeds, which take into account repeated months in the data record and weights each calendar month by its number of days, ranged from 6.06 m/s (60.0 m) at Mast 103 to 7.21 m/s (60.0 m) at Mast 102. The wind shear exponent represents the rate of increase of wind speed with height above ground according to the power law (described in Section 6). The annualized shear exponents, which ranged from 0.111 (Mast 9004) to 0.152 (Mast 103), were calculated from the mean wind speeds at the monitoring levels listed in Table 5 based on concurrent valid records at both heights. Only wind speeds greater than 4 m/s were used in the calculations. The turbulence intensity is estimated using fluctuations in the wind speed recorded by the anemometer in each 10-minute interval as a fraction of the average speed. The observed turbulence intensities at 15 m/s, which ranged from 0.066 (49.9 m) at Mast 9004 to 0.101 (60.0 m) at Mast 103, are low to moderate and consistent with the surface roughness and terrain complexity of the site. 5 Filippelli, Matthew and Pawel Mackiewicz, “Experimental and Computational Investigation of Flow Distortion Around a Tubular Meteorological Mast”, CanWEA Conference, Toronto, Ontario, October 2005. Shovel Creek Page 12 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 The Weibull function is an analytical curve that describes the wind speed frequency distribution, or number of observations in specific wind speed ranges. Its two adjustable parameters allow a reasonably good fit to a wide range of actual distributions. A is a scale parameter related to the mean wind speed while k controls the width of the distribution. Values of k typically range from 1 to 3.5, the higher values indicating a narrower distribution. The annual k values ranged from 1.64 (60.0 m) at Mast 101 to 1.79 (60.0 m) at Mast 102, indicate a somewhat variable wind resource. Figure 10 contains a chart showing the observed annual frequency distribution and fitted Weibull curve for Mast 101. Monthly patterns of variation are also useful indicators of the wind resource. The observed pattern of monthly mean wind speeds at Mast 101 is presented in Figure 11. Also plotted in this figure are the concurrent and historical monthly mean wind speeds from WRF.6 The concurrent wind speeds from the two datasets track each other reasonably well, though there is a tendency for the relationship to change between summer and winter. The historical record of WRF indicates that the strongest winds normally occur during the winter, while the weakest winds occur during the summer. The range of variation in the observed monthly average wind speeds at Mast 101 is about 2.21 m/s. Figure 13 depicts the variation in average wind speed with time of day at Mast 101 at 60.0 m, along with the variation in mean wind shear exponent. The mast indicates the average wind speed is highest during overnight while the average wind shear exponent varies from a minimum during the afternoon to a maximum during overnight. The directional distribution of the wind resource is an important factor to consider when designing the wind project to minimize the wake interference between turbines. The annual wind frequency and energy by direction plots (wind rose) for each monitoring location are shown in Figure 14. The wind roses indicate that the prevailing wind direction is from the east to south-east with some variation across the mast locations. The air density directly affects the energy production: the greater the density, the greater the power output of a wind turbine for the same speed distribution. The estimated energy-weighted air density at Mast 101, 1.179 kg/m3, was calculated from the following equation: 𝜌=𝑃𝑜𝑒[−𝑔𝑧(1.0397−0.000025𝑧) 𝑅𝑇] 𝑅𝑇 where ρ = Air density (kg/m³) P0 = Standard sea-level atmospheric pressure in Pascals (101325 Pa) R = Specific gas constant for dry air (287 J/kg·K) T = Air temperature (K) g = Acceleration due to gravity (9.8 m/sec2) z = Elevation of temperature sensor (m) This equation was applied to each 10-minute data record, and a weighted average was calculated in which the weight was proportional to the energy content of the wind. 6 See Section 5 for a complete description of the global dataset. Shovel Creek Page 13 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 5. ESTIMATION OF LONG-TERM MEAN WIND SPEED AT MEASUREMENT HEIGHT Since the wind climate can vary considerably over time scales of months to years, it is important to adjust the data collected at a site to represent historical wind conditions as closely as possible. The method used to make this adjustment is known as measure-correlate-predict, or MCP. In MCP, a linear regression or other relationship is established between two meteorological stations (or other sources of wind data, such as RSD or modeled data). One, the target site, spans a relatively short period and the other(s), the reference site(s), spans a much longer period. The complete record at the reference station would then be applied to this relationship to estimate the long-term historical wind climate at the target site. Normally, the most important factor determining the success of MCP is the choice of reference station(s), particularly the quality of its relationship with the target site (which should ideally be linear with a high correlation coefficient) and the consistency and length of the reference data record.7 In addition, when much less than a full year of data is available from the target site, it is necessary to consider the possibility that a seasonal change in the relationship between the target site and reference may bias the climatological adjustment. Historical wind speed data was obtained from the following monitoring stations that are part of the National Weather Service (NWS) Automated Surface Observing System (ASOS) / Federal Aviation Administration (FAA) Automated Weather Observing System (AWOS) / Western Regional Climate Center (WRCC) Remote Automated Weather Station (RAWS) monitoring networks and assessed their suitability as long-term references: • Caribou Peak (RAWS): January 2004 – November 2024 • Chatanika (RAWS): January 2004 – November 2024 • Fairbanks (ASOS): August 2006 – November 2024 • Livengood Alaska (RAWS): January 2004 – November 2024 • Nenana (ASOS): October 2005 – November 2024 • Wainright (AWOS): January 2006 – November 2024 The locations of the potential reference stations are indicated on the regional map in Figure 1. Hourly wind speed, direction, and temperature data for each ASOS station were obtained from the National Centers for Environmental Information (NCEI). Beginning in September 2002, a nation-wide initiative was conducted to replace the anemometers at the ASOS stations with standard, ultra-sonic, ice-free wind (IFW) sensors. Only data after the switch to the IFW sensors was utilized in this analysis. In addition to these measured data sources, data were also assessed from global datasets: • ERA5: ERA5, which was developed by the European Centre for Medium-Range Weather Forecasts (ECMWF), utilizes a variety of observing systems which have been assimilated into a global three-dimensional grid by their Integrated Forecast System (IFS) at a spectral resolution of T639 (31 km) and the ERA5 data is available on a 0.25 degree resolution on a regular lat-lon grid. • Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA- 2): MERRA-2, which was developed by the National Aeronautics and Space Administration (NASA), utilizes a variety of observing systems which have been assimilated into a global three- 7 Taylor, Mark, et al., “An Analysis of Wind Resource Uncertainty in Energy Production Estimates”, Proceedings of the European Wind Energy Conference, November 2004. Shovel Creek Page 14 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 dimensional grid by numerical atmospheric models at a horizontal resolution of 1/2° latitude and 2/3° longitude. • Weather Research and Forecasting (WRF)8: WRF is a state-of-the-art global or regional numerical weather prediction (NWP) model designed to simulate synoptic and mesoscale atmospheric circulations. WRF was developed by several organizations, including NCAR, NOAA (NCEP), AFWA, the Naval Research Lab, the University of Oklahoma, and the FAA in the early 2000’s. The WRF long-term dataset created by UL Solutions for this analysis spanned the period from January 2004 to present. The global datasets can be interpolated to the exact location of a meteorological mast. For this analysis, the model output for the four nearest grid cells were interpolated to the location of Mast 101 as it has the longest valid period of record and is located within the project’s layout. Although earlier data may exist for each surface station and global dataset, no data prior to January 2004 were utilized in this analysis. For stations with significantly longer data records, this mitigates the potential impact that longer-term trends could have on the climatological adjustment, while still providing the most significant benefit of the MCP process. Linear regression equations were established using concurrent daily mean wind speeds at Mast 101 and each potential reference. The strongest correlation was found with the WRF (r2 = 0.76); the remaining sites had r2 values between 0.14 (Wainwright) and 0.71 (ERA5). Table 6 contains a summary of the regression results. Figure 15 presents the respective time series of annual mean wind speeds from Caribou Peak, Chatanika, Nenana, Livengood Alaska, Fairbanks, Wainwright, MERRA-2 and the WRF and ERA5 datasets between 2004 and 2024. This plot was created to determine whether any abrupt changes or significant trends in mean wind speed occurred during the reference periods of record. A discontinuity in the surface-based data could indicate a problem with the measurement equipment or a relocation of the anemometry, whereas a statistically significant trend or abrupt changes in wind speed could be a sign of changing conditions around the station, such as tree growth/removal, or poor maintenance of the anemometer. A discontinuity or abnormal trend in modeled data could indicate a change in the source data or the analytical techniques used to estimate the wind speed. Any of these conditions would call into question the validity of the climatological adjustment. WRF and ERA5 exhibit the strongest correlations to the site data, track each other reasonably well, and show no inconsistent trends. They were therefore selected to estimate the long-term mean wind speed at Mast 101. The regression equations between Mast 101 and these references are as follows: Mast 101 = (WRF 6.74 m/s * 0.816) + 1.12 m/s = 6.62 m/s Mast 101 = (ERA5 4.26 m/s * 1.555) + -0.059 m/s = 6.57 m/s The results of these regression equations, when averaged, result in a long-term 60.0-m speed of 6.60m/s at Mast 101. This value is about 1.8% higher than the observed annualized mean wind speed, indicating that the wind speeds recorded during the measurement period were below the climatological average during the reference period of record. Scatterplots showing the relationship between the observed daily mean wind speeds at Mast 101 and WRF and ERA5 are contained in Figure 16 and Figure 17, respectively. The long-term wind speeds at the remaining monitoring locations were estimated using a similar technique, but with Mast 101 now serving as the reference. The regressions were performed using concurrent hourly wind speeds; the r2 values range from 0.75 (Mast 103) to 0.86 (Mast 102). Substitution of the estimated long-term speed at Mast 101 into the respective regression equations yields long-term 8 Skamarock, W. C. (2004). “Evaluating Mesoscale NWP Models Using Kinetic Energy Spectra”. Mon. Wea. Rev., vol. 132, pp. 3019-3032. Shovel Creek Page 15 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 mean wind speeds ranging from 6.23 m/s at Mast 103 (60.0 m) to 7.23 m/s at Mast 102 (60.0 m). The equations and speed projections are summarized in Table 7. MCP – with WRF serving as the reference – was also used to long-term adjust the temperature data collected at Mast 101. The long-term mean temperature at the mast is estimated to be -1.5⁰C at 59 m, which is slightly higher than the annualized mean of -1.4⁰C observed during the period of record. Assuming the standard atmospheric temperature lapse rate of 6.5⁰C per 1000 m, the corresponding long-term average at the mean turbine hub height elevation is -1.2⁰C. This value is an input into the calculation of the long-term site average air density as described in Section 7. 6. EXTRAPOLATION TO HUB HEIGHT The mean wind speed was extrapolated to the anticipated 82 -m and 105 -m hub-heights using the power law equation: 𝑇=𝑇𝑂(𝑍/𝑍𝑂)𝜌 where U = the unknown wind speed at height Z above ground; U0 = the known speed at a reference height Z0; and p = the shear exponent. This equation is an empirical relationship that is widely employed in wind resource assessment. Often, the main challenge is to determine the shear exponent between the top anemometer on the mast and the turbine hub height. A common assumption is that the shear exponent does not change with height. Lidar L1921 was used to inform shear assumptions at the project and suggested a decrease in shear with height. Therefore, reductions in annualized shear ranging from 10% to 35% were applied at each mast. The final adjusted shears can be found in Table 10. The shear values were then used to extrapolate the mast-top speeds to the requested hub heights. The resulting projected, 82 -m hub-height mean wind speeds ranged from 6.53 m/s at Mast 103 to 7.45 m/s at Mast 102 and 6.73 m/s (Mast 103) to 7.57 m/s (Mast 102) at the 105-m hub height. The hub-height wind speed projections are summarized in Table 10. Although constant shears are applied in these calculations, the shear exponents computed for each individual 10-minute record at each mast were used to create the hub-height wind speed frequency distributions for the wind flow modeling and energy production calculations. 7. ESTIMATION OF LONG-TERM ENERGY PRODUCTION The energy production of the Shovel Creek wind project was estimated using the Openwind software. Openwind was developed by UL Solutions as an aid for the design, optimization, and assessment of wind power projects.9 The primary input is a wind resource grid generated by a numerical wind flow model, in this case the Sitewind system. Other inputs include elements of the project design such as the turbine locations, hub height, power curve, and thrust coefficients, as well as the mast data. The Sitewind system and Openwind software and their applications in this project are briefly described below. 9 “Openwind – Theoretical Basis and Validation”, Version 1.3, AWS Truewind, LLC, April 2010. Shovel Creek Page 16 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 7.1 The Sitewind System Numerical wind flow models are used to calculate the wind resource variation across a project area due to changes in terrain and surface roughness. UL Solutions has developed the Sitewind system to perform these calculations.10, 11 Sitewind employs both mesoscale and microscale models to simulate the wind climate over a wide range of scales. The mesoscale model assesses regional climate conditions and simulates complex meteorological phenomena such as katabatic (downslope) mountain winds, channeling through mountain passes, lake and sea breezes, low-level jets, and temperature inversions. The microscale model accounts for the localized influences of topography and surface roughness changes and produces a detailed wind resource map and grid. As a final step, the predicted speed and direction are adjusted with onsite data from masts within the project area. This method has been found to be more accurate on the whole than microscale wind flow models alone.12 The mesoscale model used for this project was the open-source Weather Research and Forecasting (WRF13) model. WRF is a state-of-the-art numerical weather prediction (NWP) model designed to simulate synoptic and mesoscale atmospheric circulations. WRF was developed by several organizations, including NCAR, NOAA (NCEP), AFWA, the Naval Research Lab, the University of Oklahoma, and the FAA in the early 2000’s. The WRF model is updated frequently with new versions released twice annually and can use analysis or reanalysis datasets for initialization. The WRF simulations were initialized by the ERA514, 15 reanalysis dataset. Several studies by UL16 and others17 show that ERA5 has superior accuracy in terms of its correlation to meteorological mast data than other reanalysis datasets. UL Solutions uses a dynamical downscaling approach with nested grids of 9, 3 and 1 km resolution. The 1-km resolution WRF model outputs were then coupled to WindMap – a microscale mass-conserving model – which was run on a grid scale of 50 m.18 Finally, the output of WindMap was adjusted to the wind speed and direction distribution at the monitoring locations within the project area. This last step was performed within Openwind and the resulting wind resource map is shown in Figure 18. The Sitewind system performance can be independently validated by adjusting the wind map to the long-term hub-height wind speed at a single measurement location and comparing the map projections at other onsite locations with the long-term adjusted wind speed estimates. The standard deviation of the map biases at each measurement location – called the unbiased map error – is used to quantify the accuracy of the modeled wind map however, it may have limited significance with fewer measurement 10 Brower, M.C. (1999). “Validation of the WindMap Program and Development of MesoMap”. Proceeding from AWEA's WindPower conference. Washington, DC, USA. 11 Beaucage, P., M.C. Brower, J. Tensen (2014). “Evaluation of four numerical wind flow models for wind resource mapping”. Wind Energy, vol. 17, pp. 197-208. 12 Beaucage, Philippe and Brower, Michael C, “Wind Flow Model Performance – Do More Sophisticated Models Produce More Accurate Wind Resource Estimates?”, 6 February 2012. 13 Skamarock, W. C. (2004). “Evaluating Mesoscale NWP Models Using Kinetic Energy Spectra”. Mon. Wea. Rev., vol. 132, pp. 3019-3032. 14 Hersbach, H. and co-authors (2019). “Global reanalysis: goodbye ERA-Interim, hello ERA5”. ECMWF Newsletter, No. 159, p. 17-24 15 Beaucage, P., Gothandaraman A. et al (2021): “Validation of Sitewind Version 4”. Available at: https://www.awstruepower.com/knowledge-center/technical-papers/ 16 Beaucage, P. et al (2021): “A Global Comparison of Modern Reanalysis Dataset Wind Speeds”. Available on request 17 Jourdier, B. (2020). "Evaluation of ERA5, MERRA-2, COSMO-REA6, NEWA and AROME to simulate wind power production over France". Adv. Sci. Res., vol. 17, pp. 63-77 18 WindMap, developed by AWS Truepower, is a mass-conserving model that adjusts an initial wind field, here supplied by WRF, in response to local variations in topography and surface roughness. See, e.g., Michael Brower, “Validation of the WindMap Model”, Proceedings of WindPower 1999, American Wind Energy Association, June 1999. Shovel Creek Page 17 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 locations. The unbiased map error along with an evaluation of the terrain complexity, variation in predicted spatial wind resource, and the location of the onsite measurements with respect to the turbine locations, are all considered when assessing the modeling uncertainty for a given site. In this case, to compute the unbiased map error, the model output was adjusted only to data from Mast 101. The resulting unbiased map error was computed to be 0.3 m/s, or 4.5% of the average observed speed and is summarized in Table 11. The model comparisons were completed at a standard height of 80 m but are representative of the unbiased map errors at the project hub height. 7.2 Openwind Once the wind resource model has been run, the resource grid file is imported into Openwind to define the wind resource for the project area. The Weibull parameters in the file are converted to directional speed-up ratios relating the wind speed at each grid point to the speed at a reference mast. By associating the model data to a wind speed histogram file for the reference mast, the program is able to adjust the modeled speed distribution to the true speed distribution observed at a point. This method usually produces a more accurate estimate of the energy production than relying on the modeled distributions alone. A number of reference masts can be used to reduce errors in the predicted spatial variation of the wind resource across the project area. Conventionally, the project area is broken up into sub-regions, each of which is associated with a different mast using a distance-weighted interpolation between all masts. This avoids discontinuities in wind speeds across the boundaries of areas assigned to different masts and produces a more realistic picture of the spatial variation of the wind resource. Within Openwind, the adjusted wind resource grid is divided into sub-regions associated with different masts to capture variations in the observed speed frequency distribution, although the corresponding impact on energy production estimates is usually relatively small. UL Solutions uses the Openwind Deep Array Wake Model (DAWM) to calculate wake losses. This model actually contains two separate wake models operating independently. The first is the Eddy Viscosity model, which is based on the thin-shear-layer approximation of the Navier-Stokes equations assuming axisymmetric wakes of Gaussian cross-sectional form, as originally postulated by Ainslie.19 The model equations ensure that momentum and mass conservation are observed simultaneously. As inputs, the wake model requires the ambient turbulence intensity at hub height, which influences the initial wake deficit behind each turbine and the rate of wake dissipation; the speed and direction frequency distribution, based on a wind resource grid and associated mast files; the locations of the turbines; and the turbine thrust coefficient curves. Validation of the Openwind Eddy Viscosity model is described elsewhere.9 In response to evidence that conventional wake models like the Eddy Viscosity model underestimate wake losses in deep (multi-row) arrays of wind turbines, especially offshore, UL Solutions implemented a second model designed to handle such situations. This model is loosely based on a theory developed by Frandsen,20 who postulated that the effect of a deep array of wind turbines on the atmosphere could be represented as a region of increased surface drag, represented by a surface roughness length. Where the wind first impinges on the array, an internal boundary layer (IBL) is created, within which the wind profile is determined by the array roughness rather than by the ambient roughness. This IBL grows with downwind distance, and once its height exceeds the turbine hub height, the hub-height speed impinging upon turbines farther downwind is progressively reduced. According to the Frandsen theory, 19 Ainslie, J.F., “Calculating the flowfield in the wake of wind turbines”, Journal of Wind Engineering and Industrial Aerodynamics, 1988, Pages 213-224. 20 Sten Tronæs Frandsen, “Turbulence and turbulence-generated structural loading in wind turbine clusters”, Risø-R-1188(EN), Risø National Laboratory, January 2007. Shovel Creek Page 18 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 the effective array roughness is in the range of 1 m to 3 m, or typical of a forest, for mid-range speeds and typical turbine spacings. UL Solutions modified the Frandsen model to treat each turbine as an isolated island of roughness, a necessary change to permit rapid modifications to the turbine layout for array optimization. In addition, the IBL created by each turbine is assumed to be centered on the turbine’s hub height. In combining the two models, the DAWM implicitly defines “shallow” and “deep” zones within a turbine array. In the shallow zone, the direct wake effects of individual turbines dominate, and the unmodified Eddy Viscosity (EV) model is used to calculate wake deficits; in the deep zone, the deep-array effect is more prominent, and thus, the roughness model is employed. The DAWM has been validated at several offshore and onshore projects.21 In addition to the DAWM-EV combination, the Rankine Half-Body (RHB) induction model proposed by Gribben22 is incorporated to the total flow effect modeling. The Gribben RHB model combines uniform flow with a point source to simulate how the flow field is impacted by the presence of a turbine. While it does not take account of atmospheric stability, turbulence intensity or atmospheric boundary layer (ABL) height, and therefore does not attempt to model the global blockage effect, it is added to include some measure of the local induction effects. 7.3 Results The energy production was simulated for the 18 Vestas V150-4.5 MW turbines with a rotor diameter of 150 m and a hub height of 105 m and 16 Vestas V136-4.3 MW (Low HH HWO) turbines with a rotor diameter of 136 m and a hub height of 82m. The proposed turbine layout23 is shown in Figure 18. Each turbine in the layout was associated with the wind speed and direction distribution file from one of the onsite masts. The average air density was calculated from the wind speed and temperature data at Mast 101 and adjusted to the mean elevation of the turbines using a standard atmospheric lapse rate. The result was 1.193 kg/m3, with a range from 1.177 kg/m3 to 1.210 kg/m3 across the turbine array. The Vestas V150- 4.5 MW and Vestas V136-4.3 MW power curves used as input to Openwind were provided to UL Solutions on 15 November 2024. These power curves were found to be consistent with the version in UL Solutions’ in-house database; as such, the in-house power curves, including additional air densities, were used as input to Openwind to limit uncertainties related to interpolation of the power curve to the site specific conditions. Plant losses were estimated based on UL’s assessment of the actual performance of operating wind plants and an analysis of site-specific conditions.24 Loss estimates for six broad categories (along with an itemized summary of turbine production) are presented in Table 12; a detailed breakdown and explanation for each is contained in Appendix A. Table 13 includes the annual time varying losses and production estimates through the evaluation period. The turbine flow effects (wake and blocking) of 3.4% loss was calculated by Openwind. This value was estimated by taking into account the impact of the turbine wakes on each other. There are no existing or proposed wind farms within the vicinity of the Shovel Creek project. 21 Robinson, Nicholas M., “The Openwind Deep-Array Wake Model – Development and Validation Update”, July 2024. 22 Brian J Gribben and Graham S Hawkes (2019) Technical Paper: A potential flow model for wind turbine induction and wind farm blockage. Frazer-Nash Consultancy 23 The Project has indicated that the layout and turbine technology utilized is preliminary. 24 O’Loughlin, B., et al., “2024 Backcast Study and Methods Update”, UL Solutions, July 2024. Shovel Creek Page 19 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Total plant losses of 21.2% combined with turbine flow effect losses of 3.4% results in an overall project loss estimated to be 23.9%. Energy consumption of the maintenance building, auxiliary equipment within the substation, and site lighting are not included in this analysis. Typically, these are an insignificant fraction of the plant energy production. Since they are often metered separately from plant production, they are typically treated as an operational cost within the financial model. A matrix table summarizing energy output by month and hour of day (MDM) for the Shovel Creek wind project is presented in Table B.1 of Appendix B. The MDM shows that energy production will normally peak during overnight. 8. UNCERTAINTY ANALYSIS The following is a summary of the uncertainty elements associated with the wind speed and energy production estimates. For this purpose, the uncertainty is defined as the standard error for a normal probability distribution. All uncertainties are computed as weighted averages based on the number of turbines associated with each mast. 1. Site Documentation and Verification (0.3%): This uncertainty addresses the quality and independence of the available information describing the site characteristics and monitoring equipment, as well as the integrity of the meteorological data provided. Specific items considered include the quality, comprehensiveness and correctness of tower commissioning and verification and maintenance documents; the quality and number of photographs depicting each mast and its surroundings; and information regarding obstacles potentially affecting the wind flow at each mast; and the format of the provided data. Where UL Solutions has conducted an independent assessment of each mast, this uncertainty is reduced. 2. Wind Speed Measurements (1.2%): This is the uncertainty in anemometer readings of the free- stream wind speed. It reflects not just uncertainty in the sensitivity of the instruments when operating under wind-tunnel conditions, but also uncertainty in their performance in the field, where they may be subject to turbulent and off-horizontal winds, tower effects, and problems such as icing that may be missed in the validation. A component related to potential anemometer calibration degradation was assessed using In Situ testing techniques as described in Annex K of IEC standard 61400-12-1 2nd edition (2017). In addition, where applicable, the uncertainty in empirical adjustments applied to account for factors such as turbulence or the impact of wakes from existing turbines on observed wind speeds is considered. 3. Long-Term Average Speed (1.8%): This uncertainty addresses how accurately the site data, after the MCP adjustment, may represent the historical average wind resource. UL Solutions has undertaken a study of wind speed interannual variability and has produced an interannual variability map using the global ERA5 reanalysis dataset.16 The map suggests that the standard deviation of annual mean wind speeds for the Shovel Creek Wind Project is about 3.3%. It is assumed that the annual mean varies randomly according to the normal distribution, and thus the error margin varies inversely with the square root of the number of years. The estimated uncertainty accounts also for the degree of correlation between the target and reference stations, the length of the reference period of record, and the data recovery at each mast. If data reconstruction was performed to fill gaps or extend measurement periods, the impact on uncertainty has been accounted for here. 4. Evaluation Period Wind Resource (1.2%): This uncertainty is associated with how closely the wind resource over the evaluation period may match the long-term site average. The estimated value assumes a 10 -year evaluation period, 3.3% interannual variation in the mean speed, and 0.5% uncertainty associated with possible climate oscillations and trends. Shovel Creek Page 20 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 5. Wind Shear (1.3%): The wind shear uncertainty includes the uncertainty in the observed shear due to possible measurement errors and the uncertainty in the change in shear above mast height. The estimated value considers the site conditions, anemometer heights, hub heights, and measurement uncertainties at the mast. 6. Wind Flow Modeling (5.5%): The uncertainty in the array-average free-stream wind speed at the turbines, relative to the masts, depends on the wind climate, terrain complexity and vegetation height and variation, characteristics of the wind flow model, and number of masts used to adjust the resource grid and their placement relative to the turbine layout. 7. Wind Speed Frequency Distribution (1.2%): Like the mean speed, the wind speed frequency distribution varies over time. UL Solutions research indicates that the interannual variability of the energy production directly related to the wind speed frequency distribution is typically about 1.4%. The estimated uncertainty in the long-term energy production estimate considers this factor along with the onsite period of record and the length of the evaluation period. 8. Plant Losses (4.4%): UL Solutions has used operational data to quantify the uncertainties associated with some loss categories, while for others, UL Solutions applies uncertainty as a factor of the magnitude of estimated losses24 25. When these values are combined the plant operational loss uncertainty is estimated to be 4.3% over the 10-year evaluation period. (Uncertainties associated with grid curtailment losses are not considered here.) In addition, based on the DAWM validation findings21, the uncertainty in the flow effect loss calculations is estimated to be 20% of the total flow effect loss, or 0.7%. The operational and flow effect loss uncertainties are combined as the square root of the sum of their squares. The following steps were taken to determine the energy production at various desired confidence levels: • The uncertainty percentages in wind speed were combined as the square root of the sum of squares and multiplied by the predicted array-average mean speed to determine the uncertainty of the array-average mean speed. The result is 6.2%, or 0.42 m/s. • The sensitivity of the project energy output to changes in wind speed was determined to be approximately 9.1% for the given 6.2% uncertainty in mean wind speed. This ratio was calculated by comparing the energy output of a turbine at the predicted array-average wind speed of 6.81 m/s to the output of a turbine with an average speed of 6.39 m/s (predicted speed minus uncertainty). • The sensitivity of the project output to changes in wind speed was multiplied by the wind speed uncertainty to estimate the corresponding uncertainty of the project energy output. • The uncertainty in plant losses and wind speed frequency distribution were combined with the previous total using the square root of the sum of squares. • Assuming a normal distribution of errors, the energy production levels that would be exceeded by the project with 75%, 90%, 95%, and 99% confidence were calculated. The total and individual uncertainties for the project evaluation period (years 2-10) are shown in Table 14. The overall uncertainty margin in the energy production is 10.2%, or 37.7 GWh/yr. Table 15 presents the estimated net annual energy production and capacity factor at five confidence levels assuming a 9- year mature operation evaluation period and the same for the first year and for any single year thereafter. 25 Chris Ziesler, et al., “UL Harmonization Phase 2”, UL July 2021. Available at: https://www.awstruepower.com/knowledge- center/technical-papers/ Shovel Creek Page 21 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 9. SUMMARY The long-term wind resource at the proposed Shovel Creek wind project was estimated using data from four onsite masts, one onsite lidar, and correlation with ERA5 and WRF datasets. The energy production of the site was simulated using a wind resource grid developed using the Sitewind system, the Openwind software, a wind turbine layout and the 18 Vestas V150-4.5 MW turbines with a rotor diameter of 150 m and a hub height of 105 m and 16 Vestas V136-4.3 MW (Low HH HWO) turbines with a rotor diameter of 136 m and a hub height of 82m and site average air density of 1.193 kg/m3. The total wind plant loss is estimated to be 23.9%. Over the evaluation period, the net plant output is expected to average at least 308.2 GWh/yr, or 23.5% capacity factor, with 95% confidence. Following the first year of operation, the annual net plant output in any given year is expected to be at least 262.9 GWh, or 20.0% capacity factor, with 99% confidence. The expected average annual net production and capacity factor for the project are 370.2 GWh/yr and 28.2%, respectively, and the predicted array- average wind speed is 6.81 m/s. Shovel Creek Page 22 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 1: Location of the Shovel Creek Wind Project in Alaska Shovel Creek Page 23 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 2: Shovel Creek Monitoring Locations Shovel Creek Page 24 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 3: Mast 101 Monitoring Configuration Shovel Creek Page 25 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 4: Mast 9004 Monitoring Configuration Shovel Creek Page 26 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 5: Lidar L1921 Monitoring Configuration Shovel Creek Page 27 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 6: Views of Area Near Mast 101 towards East Figure 7: Views of Area Near Mast 102 towards Southeast Shovel Creek Page 28 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 8: Views of Area Near Mast 103 towards East Figure 9: Views of Area Near Mast 9004 towards East Shovel Creek Page 29 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 10: Mast 101 Observed Annual Wind Frequency Distribution and Fitted Weibull Curve Figure 11: Mast 101 and WRF Concurrent and Historical Monthly Mean Wind Speeds Shovel Creek Page 30 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 12: Mast 101 and ERA5 Concurrent and Historical Monthly Mean Wind Speeds Figure 13: Mast 101 Annual Diurnal Wind Speed and Shear Patterns Shovel Creek Page 31 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 14: Monitoring Location Annual Wind Roses Shovel Creek Page 32 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 15: Reference Station Annual Mean Wind Speeds Figure 16: Scatterplot of Mast 101 and WRF Daily Mean Wind Speeds Shovel Creek Page 33 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 17: Scatterplot of Mast 101 and ERA5 Daily Mean Wind Speeds Shovel Creek Page 34 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Figure 18: Shovel Creek Turbine Layout Shovel Creek Page 35 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Table 1: Monitoring Location Summary Name Site UTM Coordinates (WGS-84, Zone 6N) Elevation (m) Period of Record Monitoring Heights (m) Easting Northing Wind Speed Wind Direction Temp Pres RH Mast 101 431994 7203210 777 2022-09-26 to 2024-10-23 60, 55, 45, 30 55, 50, 35 59, 3 2 3 Mast 102 428056 7201132 852 2022-09-30 to 2024-10-22 60, 55, 45, 30 55, 35.5 59, 3 2 2 Mast 103 430366 7207843 664 2022-10-01 to 2024-10-23 60, 55, 45, 30 55, 35.5 59, 3 2 3 Mast 9004 423953 7196320 703 2023-11-09 to 2024-10-22 49.9, 47.9, 39.9, 27.9 47.9, 41.9, 25.9 49.9, 3 3 3 Lidar L1921 428015 7201062 854 2023-11-20 to 2024-09-16 200, 180, 160, 140, 120, 100, 80, 60, 55, 45, 39 2.5 Table 2: Summary of Top-Level Anemometer Adjustments Name TI Mast 101 -0.1% Mast 102 -0.1% Mast 103 -0.2% Mast 9004 -0.0% Table 3: Monitoring Location Monthly Wind Speeds and Data Recoveries (With Reconstructed data) Month-Year Mast 101 Mast 102 Mast 103 60.0-m Speed (m/s) Data Recovery (%) 60.0-m Speed (m/s) Data Recovery (%) 60.0-m Speed (m/s) Data Recovery (%) Sep-22 6.84 12.0 4.99 3.3 - - Oct-22 6.05 65.5 7.39 59.1 5.27 64.2 Nov-22 6.59 75.3 7.95 70.9 6.31 71.0 Dec-22 8.79 94.4 10.06 88.1 7.07 95.0 Jan-23 6.76 87.3 8.21 83.7 6.36 91.2 Feb-23 5.84 74.6 6.25 45.0 5.07 67.7 Mar-23 7.02 87.1 6.96 37.0 6.46 76.8 Apr-23 5.76 100.0 5.96 80.5 5.68 92.4 May-23 5.75 100.0 5.55 73.0 5.73 100.0 Jun-23 5.84 100.0 6.15 94.0 5.60 98.0 Jul-23 4.63 100.0 5.12 99.8 4.43 100.0 Aug-23 6.11 100.0 6.75 100.0 6.01 100.0 Sep-23 6.31 89.1 6.84 88.7 6.02 96.8 Oct-23 5.08 48.0 7.42 43.0 4.60 66.8 Shovel Creek Page 36 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Table 3: Monitoring Location Monthly Wind Speeds and Data Recoveries (With Reconstructed data) Month-Year Mast 101 Mast 102 Mast 103 60.0-m Speed (m/s) Data Recovery (%) 60.0-m Speed (m/s) Data Recovery (%) 60.0-m Speed (m/s) Data Recovery (%) Nov-23 7.54 58.9 9.37 44.5 7.58 53.2 Dec-23 5.71 89.9 7.17 62.5 5.45 82.8 Jan-24 7.58 67.2 8.91 81.0 7.16 69.2 Feb-24 7.23 77.4 9.00 64.3 7.11 92.0 Mar-24 6.08 84.1 6.88 62.7 5.60 94.0 Apr-24 5.04 92.6 5.19 75.3 5.09 98.4 May-24 4.99 100.0 5.33 100.0 4.90 99.1 Jun-24 5.00 100.0 5.28 100.0 4.72 100.0 Jul-24 6.99 100.0 7.51 100.0 7.06 100.0 Aug-24 7.22 100.0 7.88 99.3 7.38 100.0 Sep-24 5.60 91.7 6.08 87.0 5.37 94.4 Oct-24 6.43 46.0 8.07 28.8 6.98 30.0 Period of Record 6.38 82.9 7.06 74.1 6.00 84.5 Annualized Speed 6.48 7.21 6.06 Table 3 (Cont’d): Monitoring Location Monthly Wind Speeds and Data Recoveries (With Reconstructed data) Month-Year Mast 9004 49.9-m Speed (m/s) Data Recovery (%) Nov-23 9.75 31.7 Dec-23 6.75 84.5 Jan-24 8.32 73.6 Feb-24 8.24 88.2 Mar-24 5.72 92.9 Apr-24 5.16 95.0 May-24 4.99 87.1 Jun-24 4.96 100.0 Jul-24 7.10 99.5 Aug-24 7.33 89.4 Sep-24 5.95 92.7 Oct-24 6.10 44.2 Period of Record 6.63 83.2 Shovel Creek Page 37 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Table 3 (Cont’d): Monitoring Location Monthly Wind Speeds and Data Recoveries (With Reconstructed data) Month-Year Mast 9004 49.9-m Speed (m/s) Data Recovery (%) Annualized Speed 6.63 Table 4: Remote Sensing Device Wind Speed Data Recovery with Height Height (m) Lidar L1921 (%) 200 69.3 180 69.8 160 71.3 140 72.8 120 74.5 100 77.0 80 79.0 60 82.1 55 83.7 45 86.5 39 92.9 Table 5: Monitoring Location Observed Wind Resource Characteristics** Monitoring Location Height (m) Wind Speed Wind Shear* TI 15 m/s (-) Weibull Parameters Mean (m/s) Annualized (m/s) Data Recovery (%) Exponent (α) Heights (m) A (m/s) k (-) Mast 101 60.0 6.38 6.48 82.9 0.148 60.0 / 30.0 0.086 7.13 1.64 Mast 102 60.0 7.06 7.21 74.1 0.125 60.0 / 30.0 0.073 7.94 1.79 Mast 103 60.0 6.00 6.06 84.5 0.152 60.0 / 30.0 0.101 6.74 1.75 Mast 9004 49.9 6.63 6.63 83.2 0.111 49.9 / 27.9 0.066 7.42 1.68 *Only Speeds > 4 m/s used in calculation ** Statistics provided in table are indicative of reconstructed data Shovel Creek Page 38 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Table 6: Mast 101 and Reference Coefficient of Determination Summary Reference r² WRF 0.76 ERA5 0.71 Caribou Peak (RAWS) 0.66 MERRA-2 0.61 Chatanika (RAWS) 0.42 Nenana (ASOS) 0.35 Livengood Alaska (RAWS) 0.22 Fairbanks (ASOS) 0.17 Wainwright (AWOS) 0.14 Table 7: Monitoring Location Long-Term Wind Speed Projection Summary Name Monitoring Height (m) Reference Station Regression Equation R2 Long-Term Adjustment (%) Long-Term Wind Speed (m/s) Mast 101 60.0 WRF ERA5 y = 0.816x + 1.120 y = 1.555x + -0.059 0.76 0.71 1.8 6.60 Mast 102 60.0 Mast 101 y = 0.955x + 0.933 0.86 0.3 7.23 Mast 103 60.0 Mast 101 y = 0.790x + 1.017 0.75 2.7 6.23 Mast 9004 49.9 Mast 101 y = 0.920x + 0.821 0.81 3.9 6.89 Table 8: Lidar L1921 Shear Trend Table Calculated for Wind Speeds >4 m/s Concurrent With Monitoring Mast 102 Shear Layer Lidar L1921 Shear Layer Mast 102 120 m / 80 m 0.016 120 m / 60 m 0.027 100 m / 60 m 0.032 80 m / 60 m 0.042 60 m / 45 m 0.068 60 m / 45 m 0.117 60 m / 39 m 0.073 60 m / 30 m 0.115 Table 9: Lidar L1921 (Flow Corrected) Shear Trend Table Calculated for Wind Speeds >4 m/s Concurrent With Monitoring Mast 102 Shear Layer Lidar L1921 Shear Layer Mast 102 120 m / 80 m 0.033 120 m / 60 m 0.047 100 m / 60 m 0.054 Shovel Creek Page 39 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 80 m / 60 m 0.066 60 m / 45 m 0.094 60 m / 45 m 0.121 60 m / 39 m 0.098 60 m / 30 m 0.118 Table 10: Extrapolation of Long-Term Wind Speeds to Hub Height Name Monitoring Height (m) Long-Term Wind Speed (m/s) Effective Wind Shear (α) Projected 82-m Speed (m/s) Projected 105-m Speed (m/s) Mast 101 60.0 6.60 0.146 (82 m) 0.134 (105 m) 6.90 7.11 Mast 102 60.0 7.23 0.097 (82 m) 0.083 (105 m) 7.45 7.57 Mast 103 60.0 6.23 0.150 (82 m) 0.138 (105 m) 6.53 6.73 Mast 9004 49.9 6.89 0.109 (82 m) 0.096 (105 m) 7.27 7.40 Table 11: Comparison of Observed and Predicted Speeds at 80 m Name Projected 80-m Speed (m/s) Sitewind Adjusted to Mast 101 (m/s) Bias (m/s) Mast 101 6.88 6.88 0.00 Mast 102 7.44 7.07 -0.37 Mast 103 6.51 6.55 0.04 Mast 9004 7.26 6.64 -0.62 Average 7.02 6.79 -0.24 Std Deviation 0.3 (4.5%) Shovel Creek Page 40 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Table 12: Shovel Creek Wind Speed and Energy Production Detail Table 13: Annual Production Estimates Year Gross Plant Production (MWh/yr) Turbine Availability Loss BOP Availability Loss Blade Degradation Loss Environ. Curtailment Loss Fixed Annual Losses Net Plant Production (MWh/yr) Net Capacity Factor 1 486,159 6.2% 1.5% 0.6% 0.0% 18.8% 362,411 27.6% 2 486,159 5.2% 0.5% 0.7% 0.0% 18.8% 369,609 28.1% 3 486,159 4.6% 0.5% 0.8% 0.0% 18.8% 371,566 28.3% 4 486,159 4.6% 0.5% 0.9% 0.0% 18.8% 371,191 28.3% 5 486,159 4.6% 0.5% 1.0% 0.0% 18.8% 370,817 28.2% 6 486,159 4.6% 0.5% 1.1% 0.0% 18.8% 370,442 28.2% 7 486,159 4.6% 0.5% 1.2% 0.0% 18.8% 370,068 28.2% 8 486,159 4.6% 0.5% 1.3% 0.0% 18.8% 369,693 28.2% 9 486,159 4.6% 0.5% 1.4% 0.0% 18.8% 369,319 28.1% 10 486,159 4.6% 0.5% 1.5% 0.0% 18.8% 368,944 28.1% Project: Golden Valley Electric Association - Shovel Creek, AK Date: 11-Apr-2025 Comments: Client Layout Turbine Manufacturer/Model: Vestas V150- 4.5 MW Vestas V136- 4.3 MW (Low HH HWO) Turbine Rated Power (MW): 4.50 4.30 Hub Height (m): 105 82 Number of Turbines: 18 16 Total Number of Turbines: 34 Plant Capacity (MW): 149.8 Site Air Density (kg/m3): 1.192 1.194 Loss Accounting Overall Wind Plant Summary Turbine Flow Effect 3.4%Average Free Wind Speed (m/s)6.81 Availability 5.6%Gross Plant Production (MWh/yr)486,159 Electrical 2.3%Net Plant Production (MWh/yr)370,183 Turbine Performance 2.0%Net Capacity Factor 28.2% Environmental 12.8% Curtailments/Operational Strategies 0.0% Average Total Loss 23.9% Turbine Mast Coordinates (WGS84 UTM6N)Free Gross Array Array Total Net Turbine Net Capacity Total TI Turbine ID Association Easting (m)Northing (m)Speed (m/s)MWh/yr Eff. (%)Loss (%)Loss (%)MWh/yr Rank Factor (%)at 15m/s (%)Model A01 0103 430855 7209635 6.22 13,748 97.5 2.5 22.9 10,596 23 26.9 9.9 Vestas V150-4.5 MW A02 0103 430738 7209199 6.20 13,636 96.6 3.4 23.6 10,422 25 26.4 10.1 Vestas V150-4.5 MW A03 0103 430622 7208764 6.21 13,644 95.6 4.4 24.6 10,282 26 26.1 10.1 Vestas V150-4.5 MW A04 0103 430245 7208280 6.49 14,568 97.2 2.8 23.4 11,158 20 28.3 9.8 Vestas V150-4.5 MW A05 0103 430429 7207868 6.71 15,256 97.0 3.0 23.8 11,625 11 29.5 9.3 Vestas V150-4.5 MW A06 0103 430590 7207447 6.63 14,974 97.0 3.0 23.9 11,399 13 28.9 9.2 Vestas V150-4.5 MW A07 0103 430727 7207002 6.65 15,004 97.6 2.4 23.0 11,549 12 29.3 9.2 Vestas V150-4.5 MW A08 0103 430886 7206389 7.05 16,123 96.8 3.2 23.7 12,298 6 31.2 8.7 Vestas V150-4.5 MW A09 0103 430995 7205952 7.09 16,225 97.4 2.6 23.3 12,442 5 31.5 8.4 Vestas V150-4.5 MW A10 0101 432032 7204866 7.45 16,631 97.4 2.6 23.4 12,736 3 32.3 8.1 Vestas V150-4.5 MW A11 0101 432201 7204045 7.05 15,232 97.9 2.1 22.9 11,744 9 29.8 8.5 Vestas V150-4.5 MW A12 0101 432019 7203141 7.04 15,354 97.1 2.9 23.9 11,681 10 29.6 8.0 Vestas V150-4.5 MW A13 0101 431809 7202571 6.88 14,819 96.7 3.3 24.1 11,242 17 28.5 8.2 Vestas V150-4.5 MW A14 0101 431741 7202086 6.82 14,676 96.5 3.5 24.2 11,131 21 28.2 8.1 Vestas V150-4.5 MW A15 0101 431639 7201606 6.84 14,796 97.0 3.0 23.7 11,288 15 28.6 8.2 Vestas V150-4.5 MW B01 0102 428039 7202697 7.34 17,145 96.0 4.0 24.4 12,965 2 32.9 7.7 Vestas V150-4.5 MW B02 0102 428163 7202138 7.59 17,544 96.9 3.1 23.8 13,369 1 33.9 7.5 Vestas V150-4.5 MW B03 0102 428197 7201568 7.29 16,596 97.2 2.8 23.4 12,716 4 32.2 7.5 Vestas V150-4.5 MW B04 0102 427938 7200964 7.42 15,181 96.6 3.4 23.8 11,565 7 30.7 7.8 Vestas V136-4.3 MW (Low HH HWO) B05 0102 427841 7200442 6.69 12,621 95.7 4.3 24.9 9,475 29 25.1 8.9 Vestas V136-4.3 MW (Low HH HWO) B06 0102 427891 7200036 6.69 12,466 95.1 4.9 25.1 9,342 31 24.8 8.8 Vestas V136-4.3 MW (Low HH HWO) B07 0102 427941 7199630 6.46 11,861 94.2 5.8 25.7 8,809 33 23.4 9.0 Vestas V136-4.3 MW (Low HH HWO) B08 0102 427808 7199243 6.03 10,689 93.9 6.1 26.0 7,910 34 21.0 9.5 Vestas V136-4.3 MW (Low HH HWO) C01 9004 426473 7199744 6.58 12,764 94.3 5.7 25.7 9,489 28 25.2 7.6 Vestas V136-4.3 MW (Low HH HWO) C02 9004 426320 7199193 6.43 12,367 95.7 4.3 24.3 9,365 30 24.8 7.4 Vestas V136-4.3 MW (Low HH HWO) C03 9004 426064 7198775 6.22 11,676 96.7 3.3 23.4 8,946 32 23.7 8.0 Vestas V136-4.3 MW (Low HH HWO) C04 9004 425525 7198269 6.88 13,794 98.1 1.9 22.3 10,716 18 28.4 7.2 Vestas V136-4.3 MW (Low HH HWO) C05 9004 425310 7197865 7.01 14,063 97.6 2.4 22.9 10,847 14 28.8 7.4 Vestas V136-4.3 MW (Low HH HWO) C06 9004 424895 7197471 7.03 14,118 96.1 3.9 24.3 10,688 19 28.4 7.3 Vestas V136-4.3 MW (Low HH HWO) C07 9004 424253 7197213 7.06 14,151 96.2 3.8 24.0 10,749 16 28.5 7.5 Vestas V136-4.3 MW (Low HH HWO) C08 9004 424125 7196824 6.95 13,787 97.3 2.7 23.4 10,555 22 28.0 7.2 Vestas V136-4.3 MW (Low HH HWO) C09 9004 423955 7196311 7.27 14,803 97.5 2.5 23.0 11,400 8 30.2 6.5 Vestas V136-4.3 MW (Low HH HWO) C10 9004 424172 7195964 6.72 13,134 96.2 3.8 23.9 9,989 24 26.5 6.7 Vestas V136-4.3 MW (Low HH HWO) C11 9004 424471 7195619 6.57 12,713 96.5 3.5 23.7 9,694 27 25.7 7.1 Vestas V136-4.3 MW (Low HH HWO) Per Turbine Summary Shovel Creek Page 41 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Table 14: Wind Speed and Energy Production Uncertainty Summary (Evaluation Period [Years 2-10]) Uncertainty Source Wind Speed Energy Equivalent % m/s % GWh/yr Wind Resource Site Documentation and Verification 0.3 0.02 0.5 1.7 Wind Speed Measurements 1.2 0.08 1.8 6.5 Long-Term Average Speed 1.8 0.12 2.7 9.9 Evaluation Period Wind Resource 1.2 0.08 1.7 6.4 Wind Shear 1.3 0.09 1.9 7.0 Wind Flow Modeling 5.5 0.37 8.1 30.1 Total Wind Resource Uncertainty 6.2 0.42 9.1 33.8 Performance Wind Speed Frequency Distribution 1.2 4.5 Total Plant Losses 4.4 16.1 Total Energy Uncertainty 10.2 37.7 Table 15: Estimated Energy Production and Net Capacity Factor at Five Confidence Levels (Evaluation Period [Years 2-10], Annual, and First Year) Probability of Exceedance Evaluation Period Average Energy Production (GWh/yr) Evaluation Period Average Capacity Factor (%) Annual Energy Production (GWh/yr) Annual Capacity Factor (%) First Year Energy Production (GWh/yr) First Year Capacity Factor (%) P50 370.2 28.2 370.2 28.2 362.4 27.6 P75 344.8 26.3 339.1 25.8 328.0 25.0 P90 321.9 24.5 311.1 23.7 297.0 22.6 P95 308.2 23.5 294.3 22.4 278.4 21.2 P99 282.5 21.5 262.9 20.0 243.7 18.6 Shovel Creek Page 42 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 APPENDIX A - ENERGY PRODUCTION LOSSES Table A.1: Shovel Creek Detailed Energy Production Loss Accounting 10 Year Evaluation Period Turbine Flow Effects (Wakes and Blocking) First Year Long-Term Internal Effect of the Project 3.4% 3.4% External Effect of Existing or Planned Projects 0.0% 0.0% Turbine Flow Effects Total 3.4% 3.4% Availability Turbine Availability* 6.2% 4.7% Availability of Balance of Plant* 1.5% 0.5% Availability of Utility Grid 0.5% 0.5% Availability Total 8.1% 5.6% Electrical Electrical Efficiency 2.3% 2.3% Power Consumption of Extreme Weather Package 0.1% 0.1% Electrical Total 2.3% 2.3% Turbine Performance Sub-Optimal Operation 1.5% 1.5% Power Curve Adjustment 0.5% 0.5% High Wind Control Hysteresis 0.0% 0.0% Inclined Flow 0.0% 0.0% Turbine Performance Total 2.0% 2.0% Environmental Icing 11.0% 11.0% Blade Degradation* 0.6% 1.1% Low/High Temperature Derate and Shutdown 0.9% 0.9% Lightning 0.0% 0.0% Environmental Total 12.3% 12.8% Curtailments/Operational Strategies Directional Curtailment 0.0% 0.0% Production Limit Curtailment 0.0% 0.0% Environmental Curtailment 0.0% 0.0% Operational Strategies 0.0% 0.0% Curtailments/Operational Strategies Total 0.0% 0.0% Total Losses 25.5% 23.9% *Evaluation period average values presented here. See Table 13 for year-by-year values. Shovel Creek Page 43 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 The summarized loss categories presented in the main report are explained in detail below. Turbine Flow Effects (Wakes and Blocking) Wind turbines alter the free stream wind flow which may reduce the energy production of a wind project. Losses due to the flow effects are divided into the following categories: • Internal Flow Effect of the Project: This loss accounts for the flow effects, wakes and blocking, of the turbines within the project being analyzed. • External Flow Effect of Existing or Planned Projects: This loss accounts for the flow effects, wake and blocking, of existing or planned projects located adjacent to the project being analyzed for which sufficient information was available to make a precise estimate of their impact on the project being studied. Availability A plant or turbine is said to be available when it is capable of generating power, given sufficient wind. Availability losses occur when some turbines in a project, or an entire project, are inoperative for some reason. UL Solutions has studied operational data to estimate the energy-based availability losses presented in the categories below. These values are presented for the evaluation period specified in this report and incorporate the impacts of the time variability of the turbine and BoP availability described in Table 13. • Turbine Availability: This loss is an energy-based approximation related to any wind turbine downtime consistent with IEC 61400-26-2 definitions, though excluding potential effects of partial production, lightning, high wind hysteresis or temperature-related derating and shutdown and prescribed curtailments for acoustic, avian or wind sector management curtailments. o Site Access Component: Severe weather can limit access to some sites, which can reduce energy production because response times for repairs are increased. This situation often occurs in areas prone to heavy snow. However, offshore projects may also be strongly affected. This loss is estimated based on weather data and other site-specific information and is included as a component of turbine availability. This component of the loss equates to 0.3% at this site. • Balance of Plant Availability: This loss accounts for outages of the collection system and substation. It is typically assigned a value of 1.5% in the first year and 0.5% every year thereafter. • Grid Availability: This loss accounts for outages of the utility grid. It is typically assigned a value of 0.5%, which includes losses during the periods when turbine components are brought within operating specifications after a grid outage. Electrical • Electrical Efficiency: Losses are experienced in all electrical components of the wind project, including the padmount transformer, electrical collection system, and substation transformer. These losses are established in the electrical system design. The typical 2.3% value assumed here is intended to account for losses between the low-voltage terminals of the turbine (where the output is measured in a power curve test) and the point of interconnection. Even if the revenue meter is not placed at the point of interconnection, it is assumed that losses on the grid side up to the point of interconnection will be calculated and attributed to the project. Thus, if the revenue meter is located on the high voltage side of the project sub-station, it is assumed that the losses in the high voltage line between the project sub-station and the point of interconnection will be attributed to the project. • Power Consumption of Extreme Weather Package: This loss is intended to account for the energy consumed by the equipment included in an extreme weather package, if the turbines are so equipped. Power consumption for site lighting, O&M facilities, and other site facilities not Shovel Creek Page 44 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 associated with the turbines are not included as loss items and should be considered in the project’s financial modeling. Turbine Performance • Sub-Optimal Operation: This factor accounts for shortfalls from ideal performance due to suboptimal turbine settings. Typical examples include yaw misalignments, control anemometer calibration, blade pitch inaccuracies or misalignments, and other control setting issues. A value of 1.5% for this loss is typical in the absence of details regarding the operational plans for the project. • Power Curve Adjustment: This loss accounts for expected turbine performance relative to the modeled performance using the advertised power curve.24 It is comprised of two components: the default or base loss and a variable component. The variable component is calculated in Openwind for each project based on site wind shear and turbulence conditions and more than 90 power curve test time series’ results and then added to the base component. For the base component, UL Solutions had at least six 3rd party IEC tests from three separate sites in its database for V150-4.5 MW and V136-4.3 MW within these performance family. The loss was based on the difference of these measured results and the advertised power curve using the wind speed frequency distribution for the project in this report. The loss calculated for this category is within UL’s expectations for the region, site conditions and turbine model. • High Wind Control Hysteresis: For most turbines, once the wind speed exceeds the turbine’s design cut-out speed and the machine shuts down, the control software waits until the speed drops below a lower speed threshold (the reset-from-cut-out speed) before allowing the turbine to restart. This loss accounts for the energy lost in this hysteresis loop. It is calculated from wind data collected at the site and the manufacturer’s specified cut-out and reset-from-cut-out speeds. • Inclined Flow: This loss has been included to account for the estimated impact of inclined (non-horizontal) flow on power production. Environmental • Icing: This loss reflects decreased rotor aerodynamic efficiency caused by the accumulation of ice on the turbines during plant operation, as well as turbine shutdowns caused by excessive ice accumulation. The icing losses are estimated from site weather data, including the expected frequency and duration of freezing precipitation and rime ice formation. • Blade Degradation: This loss reflects changes to the aerodynamic efficiency of the turbine blades over time and consists of long- and short-term components. Long-term impacts result from normal wear and are caused by factors such as the permanent effects of sun exposure, wind-blown sand, and the freeze/thaw cycle of moisture within micro-cracks on the blades. These factors typically affect the leading edge of the blade and result in performance degradation over time. The long-term component values are presented in Table 13. Short-term effects generally result from the accretion of insects and dirt. This factor is estimated from the expected dust and insect accumulation in the area and the frequency of precipitation, which cleans the blades. • Low/High Temperature Shutdown: This loss value is calculated based on the energy that will be lost when the turbine shuts down due to temperatures outside the operating design envelope or de-rates as part of an operational strategy. • Lightning: Lightning can damage turbine components and cause electrical faults resulting in shutdowns. This loss is estimated from meteorological data indicating the likely frequency of lightning strikes at the site. Curtailments/Operational Strategies • Directional Curtailment: If turbines are spaced closer than three rotor diameters from each other, a directional curtailment strategy may be imposed by the manufacturer to limit the fatigue losses on the affected turbines caused by wake-induced turbulence. For such layouts, UL Solutions estimates a representative loss until a detailed curtailment strategy is specified by the manufacturer. At that time, a more detailed calculation of this loss can be performed. Shovel Creek Page 45 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 • Production Limit Curtailment: If the wind farm is forced to curtail production, loss of revenue could result from the sale of energy and or loss of production incentives. Typically, UL Solutions does not have sufficient information to assign a value to this loss. Consequently, it is typically set to zero unless loss data is supplied by the client. • Environmental Curtailment: If the wind farm is required to comply with certain operational standards due to environmental constraints, an environmental curtailment loss may be estimated. Production may be curtailed due to habitat concerns, noise restraints, shadow flicker, and other such environmental issues. Typically, UL Solutions does not have sufficient information to assign a value to this loss. Consequently, it is normally set to zero unless specific restrictions are supplied by the client. • Operational Strategies: If the wind farm is subject to any periodic up-rating, down-rating, optimization or shut-down not captured in the power curve or availability assumptions. Typically, UL Solutions does not have sufficient information to assign a value to this loss. Consequently, it is normally set to zero unless specific restrictions are supplied by the client. Shovel Creek Page 46 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 APPENDIX B – MONTHLY DIURNAL MATRIX Shovel Creek Page 47 of 47 Ref. No.: PR-177068 Issue: A Status: Final 1 Notice to third parties Golden Valley Electric Association 16 April 2025 Table B.1: Shovel Creek P50 Monthly Diurnal Energy Matrix (MWh) 10 Year Evaluation Period Hour Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average Total 0 60.62 42.24 46.84 42.51 39.75 36.64 37.34 40.89 44.20 47.24 46.59 55.52 45.08 16,466 1 61.26 45.69 48.07 42.84 39.06 34.27 38.88 40.47 42.63 50.50 47.52 58.05 45.81 16,734 2 59.00 48.67 50.17 44.00 36.56 39.97 39.43 35.28 43.39 54.31 43.66 59.90 46.21 16,880 3 58.78 52.14 45.05 42.30 39.34 34.35 37.37 35.10 45.45 48.81 44.52 58.56 45.13 16,485 4 57.63 50.70 45.84 41.70 38.42 32.50 38.05 32.94 45.81 46.26 48.46 58.46 44.71 16,332 5 58.65 48.99 46.66 43.05 36.34 35.11 36.53 33.94 45.98 44.16 47.80 59.60 44.72 16,335 6 55.32 49.71 45.57 39.78 31.34 28.79 33.22 32.16 42.25 40.52 51.97 61.90 42.68 15,588 7 52.90 50.56 49.85 36.51 24.42 25.14 27.46 30.96 37.57 46.64 50.91 60.53 41.09 15,007 8 53.46 49.18 52.01 31.28 22.73 23.98 24.37 32.54 39.65 45.75 52.63 57.89 40.43 14,767 9 56.25 53.98 57.01 31.96 23.08 27.84 25.81 33.95 41.64 45.33 53.36 59.12 42.40 15,486 10 54.23 55.60 53.08 28.26 25.03 26.91 25.27 33.06 42.37 42.30 54.04 65.10 42.05 15,358 11 54.73 57.91 49.12 26.62 23.77 26.44 24.98 35.78 40.58 34.83 53.68 66.29 41.15 15,030 12 55.36 56.29 51.20 26.86 26.07 25.69 24.95 40.61 39.10 32.19 50.33 60.39 40.69 14,864 13 47.37 44.65 46.65 27.01 29.06 26.14 25.71 40.62 38.79 34.97 45.87 61.51 39.04 14,258 14 53.39 42.07 44.79 29.41 27.44 23.44 26.58 40.38 35.02 33.15 43.76 62.49 38.53 14,072 15 50.81 41.44 43.08 30.69 25.81 24.27 28.73 35.02 36.25 32.90 48.55 56.91 37.88 13,834 16 55.01 40.85 45.37 30.58 26.73 19.55 27.87 32.72 38.84 35.38 57.71 54.62 38.78 14,164 17 54.40 39.41 44.17 31.85 28.61 22.77 31.68 34.44 38.53 38.05 56.91 56.86 39.83 14,549 18 52.12 42.18 44.13 33.60 32.12 23.45 24.86 36.54 42.88 41.67 56.23 58.44 40.69 14,863 19 54.53 41.42 44.95 37.94 33.59 22.17 19.94 37.51 46.86 46.30 54.22 55.86 41.28 15,079 20 55.68 44.04 41.87 40.84 32.93 26.79 24.53 39.12 46.89 48.11 53.26 55.69 42.47 15,513 21 53.39 51.18 41.46 42.07 34.37 27.24 24.28 38.31 48.61 50.66 53.81 53.47 43.18 15,772 22 57.01 49.43 37.90 42.87 40.01 27.56 29.91 39.08 48.89 51.66 53.83 57.75 44.64 16,304 23 55.48 47.06 46.68 41.48 39.77 33.02 37.59 39.97 44.90 50.12 47.42 56.49 45.02 16,443 Average 55.31 47.73 46.73 36.08 31.51 28.08 29.81 36.31 42.38 43.41 50.71 58.81 42.23 Total 41,149 32,358 34,767 25,981 23,446 20,220 22,176 27,013 30,512 32,296 36,511 43,753 370,183