NREL - Comparison of Projections to Acutal Performance in th

Comparison of Projections to Actual Performance in the DOE-EPRI Wind Turbine Verification Program August 2000 • NREL/CP-500-28608 H. Rhoads, J. VandenBosche, T. McCoy, and A. Compton Global Energy Concepts, LLC Brian Smith National Renewable Energy Laboratory Presented at the American Wind Energy Association’s WindPower 2000 Palm Springs, California April 30–May 5, 2000 National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute •••• Battelle •••• Bechtel Contract No. DE-AC36-99-GO10337 NOTICE The submitted manuscript has been offered by an employee of the Midwest Research Institute (MRI), a contractor of the US Government under Contract No. DE-AC36-99GO10337. Accordingly, the US Government and MRI retain a nonexclusive royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for US Government purposes. This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available electronically at http://www.doe.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 phone: 865.576.8401 fax: 865.576.5728 email: reports@adonis.osti.gov Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 phone: 800.553.6847 fax: 703.605.6900 email: orders@ntis.fedworld.gov online ordering: http://www.ntis.gov/ordering.htm Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste 1 COMPARISON OF PROJECTIONS TO ACTUAL PERFORMANCE IN THE DOE-EPRI WIND TURBINE VERIFICATION PROGRAM Heather Rhoads, John VandenBosche, Tim McCoy and Alex Compton Global Energy Concepts, LLC 5729 Lakeview Dr. NE, Suite 100 Kirkland, WA 98033 USA 425-822-9008 gec@globalenergyconcepts.com Brian Smith National Renewable Energy Laboratory 1617 Cole Boulevard Golden, CO 80401-3393 USA 303-384-6911 Brian_Smith@nrel.gov Abstract As part of the U.S. Department of Energy/Electric Power Research Institute (DOE-EPRI) Wind Turbine Verification Program (TVP), Global Energy Concepts (GEC) worked with participating utilities to develop a set of performance projections for their projects based on historical site atmospheric conditions, turbine performance data, operation and maintenance (O&M) strategies, and assumptions about various energy losses. After a preliminary operation period at each project, GEC compared the actual performance to projections and evaluated the accuracy of the data and assumptions that formed the performance projections. This paper presents a comparison of 1999 power output, turbine availability, and other performance characteristics to the projections for TVP projects in Texas, Vermont, Iowa, Nebraska, Wisconsin, and Alaska. Factors that were overestimated or underestimated are quantified. Actual wind speeds are compared to projections based on long-term historical measurements. Turbine power curve measurements are compared with data provided by the manufacturers, and loss assumptions are evaluated for accuracy. Overall, the projects performed well, particularly new commercial turbines in the first few years of operation. However, some sites experienced below average wind resources and greater than expected losses. The TVP project owners successfully developed and constructed wind power plants that are now in full commercial operation, serving a total of approximately 12,000 households. Introduction and Background The U.S. DOE and EPRI began the TVP in 1992 to evaluate prototype advanced wind turbines and to provide a bridge from development programs to commercial purchases. The TVP is intended to help utilities learn about wind power through first-hand experience, and to build, test, and operate enough new wind turbines to gain statistically significant performance data. Other TVP objectives include verifying the performance, reliability, maintainability, and cost of new wind turbine designs and system components in commercial utility environments; and providing other utilities and stakeholders with information about wind technology, the project development process, and the operation of wind power plants from the perspective of utility owners and operators. 2 EPRI and DOE selected TVP projects based on site and wind resource documentation, geographic and climatic diversity among selected hosts, evidence of intent to include wind power as a generation resource, the relevance of the project to the future use of wind power, and prospects for sufficient funding to achieve project implementation. Figure 1 shows the turbine configurations and approximate locations of all seven TVP projects, ranging from 0.66 megawatts (MW) at Kotzebue to 34.32 MW at Big Spring. Figure 2 shows a timeline of the projects’ operation and TVP reporting periods. With a total installed capacity of 51.98 MW and expected energy of nearly 168.1 million kilowatt hours (kWh) per year, the seven TVP projects together serve approximately 12,000 households. Figure 1. Locations and descriptions of DOE-EPRI wind turbine verification projects This paper presents turbine production at the seven TVP projects during 1999 compared to historical performance, when possible, and the long-term projected annual output. It also describes the methodology GEC used to develop the projections and loss assumptions. Key factors affecting performance, including the sites’ wind resources, the turbines’ power curves, and the projects’ availability and other energy losses, are examined, and lessons learned are summarized. Results Between January and December 1999, the seven TVP projects together produced a total of more than 120.3 million kWh of electricity, which represents a 28.6% capacity factor based on a combined annual average 48.03 MW of installed capacity.1 The overall calculated TVP system availability, which takes into account all downtime, averaged 91.3% across all projects on a per-turbine basis during 1999.2 Annual project capacity factors ranged from 10.6% at Kotzebue to 33.9% at Algona; 1999 TVP project availability ranged from 82.8% at Fort Davis to 96.5% at Big Spring. TVP Map 4 - 00 Kotzebue, AK Kotzebue Electric Assn. 0.66 MW 10 AOC 15/50 66 kW Springview, NE NPPD/KBR RPPD * 1.5 MW 2 Zond Z - 50 750 kW Algona, IA CFU / AMU * 2.25 MW 3 Zond Z - 50 750 kW Glenmore, WI Wisconsin PS * 1.2 MW 2 Tacke 6 00e 600 kW Searsburg, VT Green Mtn. Power 6.05 MW 11 Zond Z - 40 FS 550 kW Initial TVP Projects TVP III Projects “Associate” TVP Projects * Consortium of Utility Owners Big Spring, TX York / TXU Electric & Gas 34.32 MW 42 Vestas V47 660 kW 4 Vestas V66 1.65 MW Fort Davis, TX Central & South West 6.0 MW 12 Zond Z - 40A 500 kW 3 Figure 2. Timeline of TVP projects’ installation, operation, and reporting periods Figures 3 through 5 compare 1999 production to the long-term projection for each TVP project using three different performance measures. Although the Searsburg Z-40FS turbines produced the greatest surplus over the prediction, the Z-50s at Algona produced the highest output per turbine among all of the TVP projects during the 12-month period. A primary reason for lower than predicted production at Kotzebue, Glenmore, and Big Spring during 1999 was the substantially lower wind energy available compared to expectations based on long-term wind resource measurements. Lower than expected turbine availability also decreased production at Fort Davis, Glenmore, and Springview. Figure 3. Long-term projected turbine output achieved in 1999 Sp rin gview , Nebraska B ig S pring , T exas Ft. D av is , T exas Searsburg , Verm ont K otzeb ue, Alaska 2000 2001 G lenm ore , Wiscons in Alg ona , Iow a 19991995 1996 1997 1998 Phase 1 Phases 2 & 3 V47s V66s T urbines O nline Commercial O pe ration T VP R eporting Pe riod 0 500 1,000 1,500 2,000 2,500 Ft. Davis Z-40A Searsburg Z-40FS Kotzebue Phase 1 AOC 15/50 Glenm ore Tacke 600e (46 m ) A lgona Z- 50 Springview* Z-50 B ig Spring V47 & V66 * Springview was released for full operation in late Jan-99 M W h / T ur bi ne 1999 Actual Turbine Output Projected Output 95% 116% 68% 94% 88% 92%107% 4 Figure 4. Long-term projected vs. 1999 specific yield Figure 5. Long-term projected and 1999 actual capacity factors The wide range of turbine and project sizes included in the TVP presents a challenge for meaningful comparisons on the same scale. Because the turbines’ rotor diameters affect the total collection area and therefore the amount of wind energy available for capture, swept-area yield is considered a good way to compare performance between projects of different configurations. The diagonal line in Figure 4 shows where the actual 1999 swept yield is equal to the projection. Most projects were very close, and Vermont and Iowa surpassed the expectation. The variation between projects in specific yield is because some sites have higher average wind speeds than others. The wind resources at Kotzebue and Fort Davis are relatively low, whereas Springview and Big Spring have the most energetic wind resources in the TVP. 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% Ft. Davis (500 kW) Searsburg (550 kW) Kotzebue Phase 1 (66 kW) Glenmore* (600 kW) Algona* (750 kW) Springview (750 kW) Big Spring Phase 1 (660 kW) * Glenmore and Algona capacity scaled for months turbines online 1998 1999 Projection 0 250 500 750 1000 1250 1500 0 250 500 750 1000 1250 1500 Projected Swept Area Yie ld (kW h/m 2) A ct ua l S w ep t A re a Y ie ld ( kW h/ m 2 ) Springview Algona Searsburg Glenmore Kotzebue Phase 1 Ft. Davis Big Spring Phase 1 5 The V47 Phase 1 turbines at Big Spring achieved the highest swept-area yield during 1999, and the V66s are expected to achieve the highest swept-area yield over the long-term, based on their specific power rating of 0.48 kW/m2. Methods As TVP’s technical support contractor, GEC downloads, processes, and analyzes 10-minute turbine production and meteorological (met) data from Supervisory Control and Data Acquisition (SCADA) systems and other data loggers at each project. Wind direction and wind speeds are typically measured at two and three levels, respectively, with redundant anemometers at hub height. Because of the research nature of the program, the TVP projects are heavily instrumented relative to typical wind power projects of their size. In cooperation with participating utilities, various energy projections were prepared and published for each TVP project during initial site evaluations and since installation. Significant uncertainty was associated with many of the original projections as they were based on limited wind resource data and early, theoretical power curves for the turbines. Now that substantial operational experience has been gained at most of the TVP projects, GEC has developed a set of new performance projections based on historical site atmospheric conditions, turbine performance data, O&M strategies, and more informed assumptions about energy losses. GEC utilized standard industry procedures to calculate long-term annual expected energy output in a consistent manner. Figure 6 illustrates the major steps involved in determining net turbine and project energy. For this analysis, GEC used 1999 validated met data scaled to the sites’ historical mean wind speed, when possible, compared to local long- term airport wind data. For all projects except Big Spring, GEC developed complete annual met data sets for 1999 following the guidelines developed by the National Renewable Energy Laboratory (NREL) for the Utility Wind Resource Assessment Program (UWRAP)3 Met data processing methods included validation for sensor accuracy and icing and replacement of missing periods with redundant sensors, adjacent records, or the monthly diurnal average to develop complete annual data sets. For Kotzebue and Springview, data were not available at hub height for much of the year, so concurrent monthly shear factors were applied to data collected at lower heights. Because the overall data recovery was high and because good relationships were found between the various sensors used at each project, the reconstructed wind speed data sets provide reasonably accurate representations of Figure 6. TVP Method for Calculating the wind characteristics at the sites during 1999. Performance Projections Establish Site Pow er Curve Estim ate Losses Process Wind Data Adjust for shear, Determ ine frequency distribution Gross Turbine Energy Apply Losses Net Turbine & Project Energy 6 Frequency distributions (the hours of occurrence at each wind speed) were calculated in 0.5 meter/second (m/s) bins for the 1999 and long-term scaled data sets. When possible, we used independently measured power curves were used and adjusted to the annual site density, which was determined by long-term annual site temperature and elevation. The frequency distribution (hours in each bin) was multiplied by the site power curve (kW in each bin) to calculate gross energy (kWh). Availability and other loss assumptions were multiplied to determine the cumulative estimated losses and then applied to the gross energy to calculate net turbine energy. Because TVP production is reported from measurements taken at the turbines, projections reported here do not include line losses within the array or to the substation. The predicted project net energy is simply the predicted net turbine energy multiplied by the number of turbines in each project. GEC developed new loss assumptions for this analysis, shown in Figure 7, based on performance to date and each project’s operational strategy. As the only commercially owned facility in the program, Big Spring is expected to have the highest turbine availability over the life of the project. Algona and Springview are also expected to have reliable turbine operation due to conscientious maintenance by the host utilities. Although though the Ft. Davis site operations have been diligent at repairing their turbines, greater availability losses are expected to continue there because of aileron design problems. The Glenmore turbines are a lower priority for the host utility and turbine vendor, which is substantial availability losses. Figure 7. Current TVP loss assumptions Array, weather-related, control and turbulence, and blade soiling losses were estimated based on turbine layout, design, and site considerations. Fort Davis experiences frequent lightning storms, but an effective mitigation approach has been developed so no additional significant lightning-related downtime is expected in the future. Although Kotzebue has the coldest climate, weather-related losses are not expected to be significant; as very little lightning or ice accumulation on the blades has been experienced. However, substantial continuing control-related losses are expected with the AOC 15/50 turbines at Kotzebue related to “slow start” difficulties coming online during winds just above the rated cut-in speed. 0% 5% 10% 15% 20% Ft. D avis Searsburg K otz ebue Glenmore Algona Springview B ig Spring P er ce nt L os s TVP Availability Array W eather-Related Control & Turbulence Blade Soiling Line Loss 7 Line loss estimates were based on utility meter measurements for Fort Davis, Searsburg, Kotzebue, and Big Spring, and estimated based on interconnection configurations for Glenmore, Algona, and Springview. The seven TVP projects came online over a four-year period, so the calendar years used for comparison reflect varied periods of operating experience. Kotzebue’s Phases 2 and 3, Springview, and Big Spring turbineswere not fully commercial during all 12 months of 1999, so additional operational data will allow for a more complete performance analysis. Partial data for 1998 for Glenmore and Algona is included for comparison. Sensitivity Analysis Wind Resource TVP evaluations are being conducted in a variety of terrain types including mountains, plains, desert, and coastal tundra; in atmospheric conditions ranging from arid to arctic; and in fairly low to relatively high wind resources. Figure 8 shows the 1998, 1999, and historic hub-height mean wind speed for the TVP sites, as well as historical mean wind speeds adjusted to 40 m (131 ft) for comparison of the sites’ wind resources. Note that the average wind speed at the turbines may be different than at the projects’ met towers, particularly at sites with complex terrain, such as Fort Davis, Searsburg, and Big Spring,and large numbers of turbines. Figure 8. 1999 and Historical Wind Resources at TVP Sites The 1999 average annual hub height wind speeds ranged from 5.4 m/s (12.1 mph) at Kotzebue to 8.2 m/s (18.3 mph) at Springview. All of the projects except Fort Davis experienced below-average mean wind speeds during 1998, and Kotzebue was the only site with lower winds in 1999 than in 1998. The annual average wind speed at Glenmore was also slightly lower than its long-term estimate, but the 1999 averages at Fort Davis, Searsburg, Algona, Springview, and Big Spring were higher than the expected long-term averages. Fort Davis has the lowest historical average wind speed at 40 m, but Kotzebue has 0 1 2 3 4 5 6 7 8 9 10 Ft. Davis (40 m) Searsburg (40 m) Kotzebue (26 m) Glenmore* (60 m) Algona** (50 m) Springview (65 m) Big Spring (65 m) * G lenm ore 1998 va lue inc ludes on ly May-Dec da ta ; ** A lgona 1998 va lue includes only Sept-Dec data A ve ra ge W in d S pe ed (m /s ) 1998 Hub Height 1999 Hub Height Historical Hub Height Historical 40 m 8 the lowest historical wind speed at hub height (26.5 m). Big Spring has the highest long-term average wind speed, both at 40 m and hub height. With the 1999 average wind speed 10% below the historical value, Kotzebue experienced the greatest deviation from its expected long-term average. As shown in Figure 9, significantly fewer hours in the high-wind-speed bins resulted in 27% less energy available during 1999 than the historical average. Figure 9. Kotzebue 1999 and long-term wind resource Power Curve Turbine power performance has a considerable impact on production. The warranted and measured power curves for Glenmore are shown in Figure 10. Typically, the warranted curve is conservative, resulting in low estimates, but the Tacke 600e measured curve was significantly lower than warranted as the result of blade modifications to reduce vibrations. Unfortunately, the part of the power curve most affected is in the highest frequency wi