Wind Patterns over a Heliostat Field - Science Direct

Available online at www.sciencedirect.com

ScienceDirect Energy Procedia 49 (2014) 229 – 238

SolarPACES 2013

Wind patterns over a heliostat field J. Smenta, C. K. Hoa a

Sandia National Laboratories,Concentrating Solar Technologies Department, P.O. Box 5800-0350, Albuquerque, NM 87185-, USA

Abstract Heliostats constitute a major portion of the direct costs of a concentrating solar power tower plant. As a result, a significant amount of effort is being focused on designing and developing cheaper heliostats. The optical and structural performance of these new and existing heliostats under dynamic wind loads must be characterized and understood in order to meet both cost and performance objectives. This paper presents the second phase of a U.S. DOE-sponsored program at Sandia National Laboratories’ National Solar Thermal Test Facility (NSTTF) that includes dynamic testing and analysis of multiple full-scale heliostats. The objectives of these tests and analyses are to characterize and understand some differences in the impacts of dynamic wind loads on heliostat strain and cyclic fatigue between perimeter and inner-field heliostats. A weather tower with three tri-axial ultrasonic anemometers has been erected just outside the field to measure the approaching boundary winds, while a portable tower was set up to characterize wind velocities and turbulence between subsequent rows within the field. Anemometers have also been mounted to some heliostats to gather close range measurement of turbulence and wind frequencies and to provide a point of comparison for computational fluid dynamics models of wind flow over the field. This paper presents mean wind speeds and wind loads on heliostats as a function of field position. The calculated mean wind loads were used to assess the mean wind-load reduction correlation of Peterka [1]. © 2013 J. Sment. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

© 2013 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection andpeer peer-review scientific conference committee of SolarPACES 2013 under responsibility Selection and review by by thethe scientific conference committee of SolarPACES 2013 under responsibility of PSE AG.of PSE AG. Final manuscript published as received without editorial corrections. Keywords: heliostat field; wind loads; turbulence

1. Introduction A 2007 cost reduction report by Sandia National Laboratories estimates that heliostats constitute more than 50% of the direct costs of a concentrating solar power tower plant [2]. Furthermore, the same report estimates that the drive and pedestal mechanisms account for 30% of the total stretched membrane heliostat cost. Wind forces in many types of heliostats are transferred directly into the azimuth drive. As a result, an effort is being made to reduce

1876-6102 © 2013 J. Sment. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer review by the scientific conference committee of SolarPACES 2013 under responsibility of PSE AG. Final manuscript published as received without editorial corrections. doi:10.1016/j.egypro.2014.03.025

230

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

material costs associated with azimuth drives which according to Winsmith, a large manufacturer of azimuth drives, may be over-built particularly in the inner field [2]. Dynamic wind loads on heliostats must be characterized and understood in order to meet both cost and performance objectives. Previous studies have focused on static wind loads using scaled models in wind-tunnel tests [2-4]. A few studies have evaluated dynamic effects (e.g., vortex shedding, vibrations) on heliostats or inclined flat plates, but nearly all of the published results have been on smallscale models [5-7]. This paper presents wind velocity data from a U.S. DOE-sponsored program at Sandia National Laboratories’ National Solar Thermal Test Facility (NSTTF) that has included dynamic testing and analysis of fullscale heliostats [8]. The objectives of these tests and analyses are to characterize and understand the trends of mean and peak wind loads as a function of heliostat row position. Nomenclature A Azimuth Cd CFX Column Elevation F F-Group h href H-Group P-Group Row STOW U Uref ρ

surface area of facets on a heliostat angle between front of heliostat and 0° North increasing Eastward up to 359° wind drag coefficient (non-specific component) wind drag Coefficient of Force in the x (longitudinal) direction heliostats aligned West to East the angle between a line normal to the plane of the facets and the horizontal land surface total wind force on heliostat the group of three anemometers mounted on a tower adjacent to the open west field height of anemometer above ground surface height of reference anemometer above ground surface, 10 meters the group of four anemometers mounted directly above adjacent heliostats the group of two anemometers mounted on a portable tower located between heliostats heliostats aligned North to South position of heliostat when not in use, -84° elevation, and 270° azimuth speed component of wind velocity speed as measured by reference anemometer at href, adjacent to the open field air density at elevation

2. Approach A five-row section of the NSTTF heliostat field has been instrumented with anemometers to measure wind speeds at various locations in the field. The anemometry strategy employs 10 three dimensional ultrasonic anemometers capable of measuring wind speed, temperature, and direction in both azimuth and elevation. The sampling rate is set at 32 Hz. The anemometers are grouped into a field group, a portable wind tower group, and a heliostat mounted group (Figure 1).

231

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

Figure 1: Anemometer Groups: F-Group in green (left) are used to measure the boundary wind conditions and are at 10 m, 7 m, and 4 m above ground. The H-Group in blue are mounted 7 m above ground or 2.2 m above the center torque tube. The P-Group anemometers (hidden in this photo) are mounted to the portable wind tower at 7 m and 4 m above ground.

2.1. Field Group Three anemometers adjacent to the flat open area west of the heliostat field constitute the “field group” which is primarily used to characterize meteorological boundary wind conditions approaching the heliostat field from the Southwest to West to Northwest directions. The anemometer mounted 10 meters above the heliostat field tarmac defines the reference meteorological wind speed (Uref), azimuthal direction, and elevation. The second anemometer is mounted at 7 meters which is the same height as the upper portable tower anemometer and the heliostat-mounted anemometers for quick comparison of wind velocities across the heliostat field. The third field anemometer at 4 meters gives velocity information at the height of the center of the heliostats’ facet geometry. In addition to the boundary wind velocity parameters, the data from the field group is used to normalize wind speeds measured throughout the rows. If a reference velocity (Uref) of an atmospheric boundary layer wind is known at a reference height (href), a power law is used to estimate the approaching wind velocity (Uz) at any height (hz) where Uz

§ h U ref ¨ z ¨h © ref

· ¸ ¸ ¹

D

(1)

and α is defined as the slope of a curve fit through the data points of the log of anemometer heights vs. the log of the ratio of velocity at desired height to the reference velocity. The boundary turbulence intensity of each velocity component (speed, azimuth, elevation) is defined by the standard deviation of the respective component divided by the average of that same component. 2.2. Portable Tower Group Two anemometers constitute the “Portable Tower Group”. The first anemometer is at the same height as the field anemometer F2 and by extension the heliostat mounted anemometers. The primary intention for this tower is to allow inter-row measurements of wind velocity used to calculate wind forces on the subsequent row of heliostats. The portable tower is positioned upwind of the heliostat being measured and is quickly moved to subsequent rows (Figure 2).

232

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

Figure 2: Portable anemometer tower will be relocated throughout the field in order to measure variations in wind velocity, and turbulence intensity as a function of row position.

2.3. Heliostat-Mounted Group The “Heliostat Group” is comprised of 5 anemometers mounted on pipe affixed to the vertical I-beam. One anemometer is mounted on each adjacent heliostat to the fifth row from the west perimeter. The primary function of this group is to provide velocity measurements in the close vicinity of the heliostats. This group has the advantage of providing data at all instrumented rows simultaneously while the portable group must be moved row to row, a process that can take 30 minutes or more. The disadvantage is that the approaching wind velocities cannot be measured so while measurements in close proximity to the heliostat are insightful, wind-loading formulas that are functions of upwind velocities are not applicable.

Figure 3: Heliostat-mounted anemometers used to measure wind in close proximity to heliostat.

3. Peterka Models A tenet of the wind-load reduction correlation of Peterka et al states “The presence of heliostats causes a decrease within the field of mean wind speed over the height of the heliostats as a consequence of wind impingement on upwind heliostats. The reduction in mean wind is accompanied by an increase in turbulent kinetic energy (gustiness) of the wind.” [9]. This study attempts to validate this assertion by measuring the mean wind speed and turbulence over five subsequent rows of heliostats. Jon Peterka’s wind load reduction theory suggests that wind load coefficients are a function of “generalized blockage area” (the projected solid area blocked by upstream heliostats divided by the ground area occupied by the

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

233

same heliostats) up to the fourth row whereby the coefficients then level off such that the fifth row and beyond have relatively constant values [1]. A significant and consistent decrease in wind velocities as measured on the portable anemometer tower correlated with in-field row position could be associated with the existence of a wind load reduction phenomenon. Results from a selected test by Peterka are shown in Figure 4. The scale-model heliostat was placed in a boundary wind tunnel and was pointed at 270° azimuth. The elevation was 0° (where the plane of the facets were vertical). The wind direction was 265° resulting in a 5° angle of attack. The coordinate system used in Peterka’s model is also shown in the same figure. Only the lateral head-on forces Fx associated with the coefficient CFX are being compared to measurements at the NSTTF.

Figure 4: Wind Load Curves from Peterka report [9].

4. Test Procedure 4.1. Meteorological Wind Profile The meteorological boundary conditions measured by the three field group anemometers mounted at 10 m, 7 m, and 4 m are summarized in Figure 5. The mean wind speed was 11 m/s (26 mph) and was approaching from the northwest at a mean heading of 310° and a relatively flat (1°) elevation. There was a high degree of turbulence in the approaching wind, particularly in the lateral or azimuthal component. The power law coefficient is very steep and the value may be affected by the relatively low heights of the anemometers. The slope would likely decrease at higher elevations to reflect a value near 1.5 that is consistent with open field boundary winds. 100 mean Speed = 11 m/s

Elevation (m)

mean Direction = 310o mean Elevation= 1o long. Turbulence = 0.165 lat. Turbulene = 0.261 vert. Turbulence = 0.105 power law coef. = 5.60 10

y=5.60x+0.98

1 0.5

0.8

1.3

2.0

U/U

ref

Figure 5: Summary of meteorological boundary wind conditions as measured by on Western edge of NSTTF heliostat field March 21, 2013.

234

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

4.2. Heliostat Positions The scope of the testing was to observe wind velocity trends over 5 rows of heliostats in a single column. The heliostats were positioned in 270° azimuth (facing west) and set to 0° (vertical facet plane), 45°, and STOW which is face-down and nearly horizontal at -84°. Additionally, the adjacent columns were placed into the same position to ensure inclusion of lateral wind velocity effects. Figure 6 shows an overview of the test block heliostat positions. Data was logged for approximately five minutes before breaking to move the portable weather station to each subsequent row. The process was then repeated with the heliostats in 45° elevation. See Figure 7.

Figure 6: Heliostat test block on northwest corner of NSTTF field. Heliostats are vertical and facing west. While only the test column of heliostats is being studied, adjacent columns are in position to simulate a “full-field” effect.

Figure 7: Heliostats are in 45° elevation in test column of heliostats. Anemometers are seen mounted to heliostats. Fork lift is moving portable weather station to subsequent row.

5. Results and Discussion The data was cropped so that only wind with an approaching velocity between 8 and 14 m/s and between 272° and 277° as measured by the 10 m field reference anemometer was included. This range was chosen to resemble the 5° angle of attack used in the Peterka study above. The wind load is calculated using a standard drag force formula, F

1 UU 2 C d A 2

(2)

where ߩ was the air density at elevation ؆ 1, ‫ ܣ‬was the 37m area of the NSTTF heliostats adjusted for angle of attack, the wind speed ܷ was measured immediately upstream (West) of each heliostat. For example ROW1 was based on wind measured by the field tower west of the heliostat field, ROW2 was measured by the 7 m anemometer

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

235

on the portable tower located West of row 2 and thus between rows 1 and 2 etc. The value of Cd was adopted from a 2006 study by Wu et al which presents drag coefficients for heliostats as a function of relative position to approaching wind [4]. For vertical, 45°, and horizontal heliostats ‫ܥ‬ௗ is 1.2, 1, and 0.25 respectively. 5.1. Mean Wind Loads as a Function of Row The coefficients for the x component (CFX), or mean head-on force in the 0 degree elevation position from the Peterka report were normalized to the value associated with “0 rows upstream”. (See Figure 4) The mean calculated NSTTF wind forces were normalized to the ROW1 wind force values which also correspond to 0 rows upstream. The normalized mean values of these wind speeds are plotted with Peterka’s normalized CFX in Figure 8. The values of the Peterka model are on average 13% higher with a standard deviation of 0.33. The inter-row stratification of mean velocities was shown in this test to be 30% higher at the 7 m level (center heliostat) than the 4 m with a standard deviation of 0.19. ,

1

7 m P-Group 4 m P-Group Peterka CFX

Mean Normalized Wind Load

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

ROW1

ROW2

ROW3 ROW4 Field Position

ROW5

ROW6

Figure 8: Normalized mean wind loads on March 21, 2013 shown with normalized coefficients of force from Peterka's results. Approaching wind heading between 272° and 277° azimuth at 8 – 14 m/s. Test heliostats’ headings at 270° azimuth and 0° elevation. Wind elevation is assumed to be 0°.

The plots in Figure 9 show the wind load reduction curves of vertical and 45° heliostat elevations as measured by both the P-Group and H-Group. The curve associated with 45° elevation is generally 40-60% lower than the vertical—a trend consistent in both the vertical and 45° configurations. The 45° curve follows a different trajectory. While the vertical curve has an a quasi-exponential decay, the 45° curve exhibits a slight bump in velocity at the second row after which it tapers off with a quasi-exponential decay. CFD models were produced with SolidWorks Flow Simulation and also exhibited this second row bump. Figure 10 is a wind velocity cut-plot with velocity probes at the location of the heliostat mounted anemometers. These probes represent an instantaneous calculation measured at a single point and as such, comparison to the measured mean velocities which have normalized standard deviations on the order of 50% (see turbulence study in section 5.2) may be coincidental.

236

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238 ,

1

Mean Normalized Wind Load

0.9 0.8 0.7

7mP-Group Vert. H-Group Vert.

0.6

7mP-Group 45o H-Group 45o

0.5

CFD 45o

0.4 0.3 0.2 0.1 0

ROW1

ROW2

ROW3 ROW4 Field Position

ROW5

ROW6

Figure 9: Normalized wind loads at different heliostat elevation angles as a function of row position on March 21, 2013. Approaching wind heading was between 272° and 277° at 8 – 14 m/s. Test heliostats’ headings are 270° azimuth and elevation as shown. P-Group measurements are taken upwind of heliostat. H-Group measurements are taken on the heliostat. Row 6 did not have a heliostat mounted anemometer and was in STOW position during the test. The CFD points are created from the probe values in Figure 10 normalized to 40 mph.

Figure 10: SolidWorks Flow Simulation with velocity probes at relative locations of heliostat mounted anemometers used in Figure 9 are shown just above the heliostats. Peak velocities which occurred just below the lower edge of the heliostats were probed for informational purposes only and are shown just below the figure.

5.2. Mean Longitudinal Turbulence Intensity as a Function of Row The trend in mean turbulence shows a large leap in intensity followed by a flattening-off period. In the vertical orientation the leap in intensity was delayed by one row. Figure 11 shows the turbulence intensity at two different heliostat orientations (vertical and 45°) as measured by the P-Group anemometers. The boundary longitudinal turbulence is below 10%. The 45° turbulence jumps immediately after the first row and has reached 47% between the first and second rows whereupon it remains relatively steady near 50%. The vertical group also shows a jump from 10% to 50% but the leap did not happen until after the second row of heliostats.

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

237

,

0.7

Mean Turbulence Intensity

0.6 0.5 0.4 0.3 7 m vertical 4m vertical

0.2

7 m 45o 0.1

4 m 45o

0

Boundary ROW1-2 ROW2-3 ROW3-4 ROW4-5 ROW5-6 Field Position

Figure 11: Longitudinal turbulence intensity measured between rows as a function of heliostat elevation, March 21, 2013. Approaching wind heading was between 272° and 277° at 8 – 14 m/s. Test heliostats were facing 270° in azimuth and elevation is as shown.

Figure 12 overlays the turbulence intensity measured between the rows by the P-Group anemometers and measured at the rows in close proximity to the heliostats by the H-Group anemometers. The jump in turbulence from ~10% to 50% is observed in both the vertical and 45° orientations and is shown to happen between rows since it is measured by the P-Group before the H-Group. The H-Group measurements were recorded at the same time as the corresponding P-Group measurements. For example the measurement for the H-Group at row 3 was recorded at the same time as the H-Group measurement for Row 3.5 (between rows 3 and 4). The STOW configuration as measured by the H-Group shows relatively flat turbulence intensity throughout.

Turbulence Intensity

Longitudinal turbulence by row and heliostat elevation angle P-Vert. (7m)

0,6

H-Vert (7m)

0,4

P-45° (7m)

0,2

H-45° (7m)

0

0

2

4 Row

6

H-STOW (7m) Boundary

Figure 12: Longitudinal turbulence intensity at the row is measured by the heliostat mounted anemometers (H). Turbulence measured between rows is measured by portable tower (P).

6. Summary Wind loads on heliostats were calculated from velocity measured at a point between subsequent rows. A standard ଵ drag force equation ߩܷ ଶ ‫ܥܣ‬ௗ was used by adopting the drag coefficients from Wu [4]. The decrease in wind ଶ loading on a heliostat as a function of row bears resemblance to the CFX vs. row curve in the Peterka report [9]. The mean loads measured at the 7 m level are higher than those measured at the 4 m level indicating that wind load values are dependent on location of measurement. Wind loads were measured at the 7 m and 4 m heights and in close proximity to the heliostats at the 7 m height. Measurements in close proximity to the heliostats showed relatively higher values but followed a similar curve attenuating in a quasi-exponential fashion as a function of row. The 45° sequences showed an idiosyncratic “bump” where wind loads increased from the first to second row. A CFD model showed a similar increase although a direct comparison between mean measured wind speeds and CFD point probes is questionable.

238

J. Sment and C.K. Ho / Energy Procedia 49 (2014) 229 – 238

Turbulence intensities had a standard pattern of jumping from boundary values to a level near 50% upon entering the field. A delayed jump effect was observed where turbulence jumped immediately after the first row when the heliostats were in the 45° orientation while in the vertical orientation the jump in turbulence intensity was delayed by one row jumping to 50% after passing the second row of heliostats. Future work is needed to determine whether these patterns occur regularly. Limits on project duration and the natural availability of wind velocities within the useable window restricted the scope of this study. The results were based on measurements taken over a consecutive 4 hour period of time. Only data between 8 and 14 m/s and 272277° azimuth was included in the results. Acknowledgments The authors would like to thank Adam Moya and Zachary Payne for their significant contributions to the data acquisition system. Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.

References [1] Peterka, J.A. and R.G. Derickson, 1992, Wind Load Design Methods for Ground-Based Heliostats and Parabolic Dish Collectors, SAND927009, Sandia National Laboratories, Albuquerque, NM. [2] Kolb, G.J., C.K. Ho, T.R. Mancini, and J.A. Gary, 2011, Power Tower Technology Roadmap and Cost Reduction Plan, SAND2011-2419, Sandia National Laboratories, Albuquerque, NM. [3] Pfahl, A., M. Buselmeier, and M. Zaschke, 2011, Wind loads on heliostats and photovoltaic trackers of various aspect ratios, Solar Energy, (9), p. 2185-2201. [4] Wu, Z.Y., B. Gong, Z.F. Wang, Z.N. Li, and C.C. Zang, 2010, An experimental and numerical study of the gap effect on wind load on heliostat, Renewable Energy, 35(4), p. 797-806. [5] Chen, J.M. and Y.C. Fang, 1996, Strouhal numbers of inclined flat plates, Journal of Wind Engineering and Industrial Aerodynamics, 61(23), p. 99-112. [6] Gong, B., Z.N. Li, Z.F. Wang, and Y.G. Wang, 2012, Wind-induced dynamic response of Heliostat, Renewable Energy, 38(1), p. 206-213. [7] Wang, Y.G., Z.N. Li, B. Gong, and Q.S. Li, 2009, Wind Pressure and Wind-induced Vibration of Heliostat, Advances in Concrete and Structures, 400-402, p. 935-940. [8] Griffith, D.T., A.C. Moya, C.K. Ho, and P.S. Hunter, 2011, Structural Dynamics Testing and Analysis for Design Evaluation and Monitoring of Heliostats, in Proceedings of ASME 2011 5th International Conference on Energy Sustainability & 9th Fuel Cell Science, Engineering and Technology Conference, ESFuelCell2011-54222, August 7-10, 2011, Washington, DC. [9] Peterka, J. A., Hosoya, N., Bienkiewicz, B., & Cermak, J. (1986). Wind Load Reduction for Heliostats. Golden, CO: Solar Energy Research Institute.