Adjustment of Anemometer Readings for Energy
Transcription
Adjustment of Anemometer Readings for Energy
Adjustment of Anemometer Readings for Energy Production Estimates WINDPOWER – June 2008 – Houston, Texas Matthew Filippelli, Julien Bouget, Michael Brower, and Dan Bernadett AWS Truewind, LLC | 463 New Karner Road | Albany, NY 12205 www.awstruewind.com Kathleen Moore Integrated Environmental Data, LLC | 255 Fuller Road, Suite 289 | Albany, NY 12203 www.iedat.com Headquarters: Albany, NY • Mapping • • • • Energy Assessment Project Engineering Performance Evaluation Forecasting Industry Leader & Consultant for ~20,000 MW in 60 countries Full spectrum of wind plant design, development & evaluation services Consultant to several NY and US agencies & utilities for wind & PV systems Established in 1983; 65 employees Introduction • There is growing evidence that the energy production of many wind projects has been overestimated. No single cause is likely to blame; rather a number of factors probably contributed. • One plausible source of trouble is dissimilarity between wind resource assessment methodology and wind turbine power performance verification. Objective & Scope • The following parameters were studied to characterize potential sources of measurement error and/or energy production overestimation. 1. Effect of Turbulence Intensity 2. Effect of Off-horizontal flow 3. Effect of unusual shear and other weather anomalies • The usefulness of new and existing models in improving wind resource assessment and energy production was also explored. • These results and related literature were examined to assess the feasibility of wind speed measurement correction and to identify potential improvements in wind resource assessment methodology. Power Curve Measurement • Procedure for power curve measurement and certification has been standardized – IEC 61400-12-1 • • • No standard for wind resource assessment procedure Best practices are generally followed, but discrepancies still exist in execution These differences can potentially lead to measurement errors and/or energy over-estimation Wind Measurement Equipment • Resource assessment practice generally attempts to follow IEC recommendations on several factors to lower uncertainty in measured data – Separation from tower – Boom orientation • • Risoe P2546A • NRG #40 One of the biggest differences is choice in anemometer IEC standards call for a Class 1 Anemometer (or currently: 1.7A or 2.5B) NRG #40 Anemometer has been North American standard for several years Anemometer Choice • • • Both instruments were chosen as they represent the majority of devices deployed for their intended use As the Max 40 is the current standard for wind resource assessment in North America, its sensitivity to turbulence and offhorizontal flow will be examined The Risoe will be used as a reference for this work. While it is not believed be free from all error, its performance properties are well documented and it meets the IEC criteria for power performance verification Turbulence Assessment Database • • Archived and current data from 5 resource assessment towers across eastern North America were analyzed for this component. Each station equipped a Risoe P2546A / NRG Max 40 pair at one monitoring height Tower A B C D E Table 1: Turbulence Analysis Case Database Monitoring Tower / Mount Risoe NRG Height (AGL) type Orientation Orientation 40 m Lattice/ Side 180° T 300° T 40 m Lattice/ Side 285° T 165° T 40 m Lattice/ Side 32° T 212° T 40 m Lattice/ Side 22° T 202° T 140 m Lattice/ Top 180° T 0° T POR (months) 5 30 12 3 9 Data Filtering • The following filtering criteria were used for each station: – Only valid data: remove icing, obvious malfunction, etc. – For side mount booms, use 20° direction sector bisecting sensors – For top mount booms, use 40° direction sector bisecting sensors from prevailing orientation – Only speeds above 4 m/s Table 2: Turbulence Analysis Filter Criteria and Summary Statistics Tower Direction Wind NRG Risoe Mean of Ratio Data Sector Speeds TI TI NRG/Risoe Points 230° – 250° T ≥4 m/s 6.6% 6.9% 0.980 4200 A 225° – 245° T ≥4 m/s 9.8% 10.0% 1.003 8500 B 300° – 320° T ≥4 m/s 19.6% 21.4% 1.020 3100 C 280° – 300° T ≥4 m/s 23.5% 25.3% 1.003 600 D 250° – 290° T ≥4 m/s 11.7% 12.1% 0.982 7000 E Methodology • The following parameters were calculated or used in the analysis . – – – • P R TI 40 m wind speed from NRG anemometer. Designated “Primary” (m/s) 40 m wind speed from Risoe Anemometer. Designated “Redundant” (m/s) Turbulence Intensity as defined 40 m NRG anemometer – σP/P (%) The following procedure was used to assess the impact of horizontal turbulence intensity on wind speed measurements. – – – – – – The Risoe sensor was used as the reference anemometer, as it has been shown to have little sensitivity to turbulence intensity P/R and TI were calculated for each valid 10-minute data record The P/R values were binned and averaged in 1% TI segments for the total data set (P≥4.0 m/s), and for three separate speeds (P= 5, 8, and 12 m/s) Frequency distributions of TI were calculated for each data set identified above Linear regressions were fit for each set of data at each tower Following are the graphical representations of the results for each station Tower A Results Tower B Results Tower C Results Tower D Results Tower E Results Database Comparison with Reference Reference curve is NRG data from [1] Turbulence Intensity Results • • • • • All towers, with the exception of Tower D, showed a strong correlation between horizontal turbulence intensity and the NRG/Risoe (P/R) speed ratio. The positive slopes of the regression lines show that the NRG anemometers indicate progressively higher relative speeds under more turbulent conditions. This is evidence of turbulent over-speeding. Based on these field results and the previous work, the NRG varies from reference speed at an average rate of 0.0975% per 1% TI. There appear to be a site-specific variations in this relationship with wind speed. The field and reference data indicate that this response by the NRG anemometer to turbulent flow could induce over-estimates of wind speed on the order of 1% to 2%+, which can translate into overestimations of 1.8% to 3.6%+ in energy estimates. Turbulence Intensity Discussion • • • There appears good agreement between the field-measured slopes and previously published research results; the differences in offsets may be partially attributable to individual sensor pair characteristics and/or other environmental factors. Tower C’s slope is steeper than the other three towers and the research data. This higher sensitivity to turbulence may be attributable to the terrain at this site. The area is rugged, heavily forested and has with a high incidence of off-horizontal flow, the implications of which are discussed in the next section. Given the strong correlation between increasing relative speed output and TI measured by the NRG anemometer, it is envisioned that a linear correction could be implemented as a partial treatment for these conditions. Off-Horizontal Flow • • • In sloped terrain and complex flow regimes, a turbine’s output will be driven by the horizontal component of the wind As such, the potential for errors in energy production estimates grows when the resource assessment anemometer does not respond to off-horizontal flows as a turbine does. Numerous field and wind tunnel studies have produced a body of data on t he inclined flow response of both the NRG and Risoe anemometers Risoe P2456A manufacturer specifications Off-Horizontal Flow Database • • Archived and current data from 3 resource assessment stations across eastern North America were analyzed for this component. Each station is equipped with an NRG Max 40 anemometer pair at one monitoring height supplemented with an RM Young model 27107T vertical propeller anemometer. One of the stations, Tower E from the previous section, also has a Risoe P2546A anemometer installed. Tower E F G Table 3: Off-Horizontal Flow Data Base Tower RM Young Risoe NRG Type Height/Bearing Height/Bearing Height/Bearing Lat. 132 m / 0° T 140 m /180° T 140 m / 0° T Lat. 74 m / 208° T N/A 76 m / 88° T 76 m / 208° T Lat. 47 m / 240° T N/A 49 m / 173° T 49 m / 241°T POR (months) 9 3 3 Tower F Overview Tower F – 80m Tower G Overview Tower G – 50m Approximation Off-Horizontal Flow Angle Vout 1 0.02 5,5 radians 3.6 Voutput Vtrue 1 0.02 Vtrue 1 5 / 180 5 V Vvert V 3.6 sin( ) vert vert 1 Vtrue 5 Vtrue V output 3 . 6 1 5 1 sin Angular response characteristics of NRG anemometer with vector scalar wind speed definition [2] Vvert 3.6 1 Vtrue 5 A Vvert Vtrue A 3.6 A 1 5 Calculation of Over-Speeding 3.6 Voutput Vtrue 1 5 cos( ) Vhorizontal Vtrue cos( ) Voutput 3 . 6 1 5 Over / underspeeding Voutput Vhorizontal Vhorizontal Procedure: • The formulae presented were applied to each 10-minute valid data record from Towers F and G • Data were averaged in 10° bins and plotted Tower F Results Tower G Results Off-Horizontal Flow Discussion • • • This process illustrates that with onsite vertical speed measurement and wind tunnel results, off-horizontal flows can be assessed with standard wind resource assessment sensors. The resulting upslope angle calculations are consistent with the expected terrain-induced flow distortion, and will be useful for other aspects of project development (e.g. turbine suitability). The calculated speed deviations exhibit sizeable range for both towers – Tower F deviations range from -1.19% to 2.71%. Several direction sectors have calculated over-speeding of greater than 2.5%, which could yield energy over-estimation on the order of 4.5% – Tower G deviations range from -1.51% to 3.7%. Several direction sectors have calculated over-speeding of greater than 3%, which could yield energy over-estimation on the order of 5.4% • These effects can be further compounded by turbulent over speeding of the anemometry, particularly in rugged terrain. Vertical Turbulence Intensity • • • Previous field tests evaluating the performance of cup anemometry indicated that the vertical turbulence intensity had a large influence in the differences between cup anemometer measurements. It was also shown that corrections to anemometer measurements using this relationship could reduce variation of the wind speed ratios [3]. Tower E’s equipment configuration facilitated investigation into the significance of this parameter on the ratio between the NRG and Risoe anemometers. The following procedure was employed to assess the influence of vertical turbulence intensity on wind speed measurements. – – – – The vertical turbulence intensity is defined as such: σW/P (also σW/U) The Risoe sensor was again used as the reference anemometer P/R and σW/P were calculated for each valid 10-minute data record The P/R values were binned and averaged in 1% σW/P segments for the total data set (P≥4.0 m/s) – Frequency distributions of σW/P were also calculated – A linear regression was fit to the resulting data set – Following is a graphical representation of the results Tower E Results Vertical Turbulence Intensity Results • • • As shown in [2], a distinct relationship is apparent between the vertical turbulence intensity and the wind speed ratio between the NRG and Risoe Anemometer. Based on these results, the NRG varies from reference speed at an average rate of 0.218% per 1% σW/P. The slope of this regression agrees with the earlier horizontal turbulence studies, suggesting that the NRG instrument indicates relatively higher wind speed values as the vertical turbulence intensity increases. Unusual Shear and Other Events • • The effects of unusual shear and other transient weather events can be difficult to capture with single point measurements from anemometry. Appropriate equipment is atmospheric profiler – Sodar – Lidar • Several types of events can have detrimental effects on turbine and park performance. Directional Shear • • Can manifest in several different ways with different time scales. Potential turbine- and park-scale performance implications – – – Strong, transient directional shear event in the mountains of western US [4] • Directional shear across the rotor plane causes losses from aerodynamic inefficiency, 2% to 3% losses estimated for shears of 10° to 18° In extreme cases, can cause excessive loading and down time On a park scale, performance can degrade if optimized to the wrong wind rose (typically measured below hub height and not adjusted for increased elevation). Commonly difficult to detect on a resource assessment tower with only two vanes. Other Meteorologically Driven Shear Conditions • • • Examples include time-varying horizontal shear, low level jets, frontal passages, stability, and local circulations. Can have a potentially significant negative impact on energy production to a slight positive effect depending upon phenomena Certain conditions may be difficult to identify with standard meteorological towers, but may be more visible with high-sensitivity sensors (e.g. 3D sonics, ΔT, etc.) [4] [5] Terrain and Monitoring Condition Driven Shear • • • • Conditions driven by slope steepness, complexity, land cover, exposure, etc. Here, anemometer scrutiny and potential correction may be useful Over-speeding on mast anemometers may lead to significant errors in absolute speed measurements and shear calculations Aside from complementary remotesensing campaigns, resource assessment in complex terrain would also benefit from vertical wind speed measurements on the masts Usefulness of Models • • • A High-resolution, CFD model was sought to help describe the wind regime through the complex terrain of Towers F and G’s project area. Meteodyn WT was run to assess its usefulness in providing 3D flow field information, including offhorizontal wind data and turbulence. The results were checked against the measurements taken at Towers F and G. Tower F Overview Tower F Tower G Overview Tower G Tower F Results • Inclined flow output from model compared to measurement at 80 m Tower G Results • Inclined flow output from model compared at 80 m compared to measured data at 50 m. Conclusions • • • • There are clear correlations between turbulence intensity and the wind speed ratios of equally exposed NRG and Risoe anemometers. The evidence resembles turbulent overspeeding, and appears to be treatable through linear corrections. Off-horizontal flow angles and the resulting speed deviations can be calculated and treated through the analysis of standard resource assessment anemometry and vertical speed measurements. Anemometer adjustments and careful resource assessment practices may help mitigate errors in shear estimates due to speed measurement inaccuracy. The use of high-resolution, CFD-based wind models show promise for assisting with the characterization of flow through project sites, particularly in complex terrain. References [1] Papadopoulos, K. H. ‘Effects of Turbulence and Flow inclination on the Performance of Cup anemometers in the Field,’ Boundary Layer Meteorology, 101: 77-107, 2001. [2] Pedersen, T. F., ‘Power Curve Measurements under Influence of Skew Airflow and Turbulence,’ 2002. [3] Albers, A., ‘Open Field Cup Anemometry,’DEWI Magazine, No. 19, August 2001. [4] Moore, K. and B. Bailey, ‘Classifying Rotor Span Shear Profile Variability and Improving Wind Turbine Production Prediction,’ [5] Banta, R. M., et al., “Nocturnal Low-level Jet Characteristics observed during CASES-99,’