Performance Results of a Low - Concentration Photovoltaic System Based on High Efficiency Back Contact Cells by Zachary S. Judkins, Keith W. Johnston, Charles Almy, Ryan J. Linderman, Brian Wares, Nicholas A. Barton, Matt Dawson, and Jack Peurach 2010 EXECUTIVE SUMMARY As the solar industry scales towards installation capacities in the range of 30 to 50 gigawatts per year, SunPower has begun to evaluate concentration in order to leverage cell production capacity. SunPower is currently focusing on research and development investments in low concentration photovoltaic (LCPV) systems, which can reduce capital expense per watt by up to 90% while minimizing design changes in SunPower’s high efficiency back contact solar cells. Herein, we demonstrate a linear 7X geometric, parabolic-section, single-axis concentration photovoltaic system. We have installed an LCPV array at the National Solar Thermal Test Facility at Sandia National Labs in Albuquerque, New Mexico. Preliminary performance data indicates a direct resource efficiency of 18.4% and a misalignment tolerance of ±1.25˚. EXECUTIVE SUMMARY 2 INTRODUCTION SECTION 1 In response to dramatic increases in demand for photovoltaic systems, cell manufacturers have to increase fabrication capacity. However, this approach involves high capital expenditures and long lead times. In order to meet short term demand at minimal cost, concentration photovoltaics (CPV) shows tremendous promise. In CPV systems, optical components such as mirrors and lenses are used to focus sunlight onto photovoltaic cells. CPV is not a new concept, but up to this point it has been unsuccessful in achieving significant market penetration [1]. The most often cited advantage of CPV is the replacement of expensive photovoltaic materials with large areas of glass, plastic, or metal [2]. By scaling down the contribution of the semiconductor to the total system cost (by a factor equal to the concentration ratio) and instead relying on commoditized materials, it is argued that a lower and more stable cost can be achieved. While this is true to some degree, rapidly falling silicon prices and module costs weaken this argument [3]. A second advantage of CPV is the ability to scale capacity. This is particularly true for low concentration photovoltaics (LCPV), wherein crystalline silicon cells are typically illuminated with intensities less than 20 suns [2]. If the cell and module manufacturing for an LCPV solar cell were not vastly different from the existing processes, a fraction of production capacity could be converted to a large scale LCPV production facility (e.g., SunPower’s 600MW of cell production capacity could conceivably be converted into 4200MW at a concentration of 7X). This ability to leverage production could dramatically alter the capital and time required to ramp production. LCPV systems offer many advantages compared to other CPV technologies. They typically only require single-axis tracking, show less sensitivity to tracking errors, are insensitive to changes in the incident spectrum, and can capture a larger fraction of the diffuse and circumsolar content. These factors, coupled with a lower perceived risk, highlight the feasibility of this technology [3]. The back contact SunPower cell technology is particularly well-suited to LCPV by virtue of several factors. These include the absence of front surface metallization shading, low series resistance due to thick metallization on the backside, superior temperature coefficients, and production efficiencies approaching 23% at one-sun [4]. Under concentration, cell efficiency can be driven even higher with an appropriate design [5]. These superior efficiencies provide a larger budget for optical components, thermal management, and acceptance angle to further reduce system costs. Herein, we briefly review the SunPower cell technology and its suitability for LCPV. Following this, an overview of the LCPV system architecture is provided; this includes a detailed description of our Alpha-2 LCPV array at Sandia National Labs (SNL). Finally, performance results are presented along with an explanation of the test methodology. 1. R. M. Swanson, “The promise of concentrators,” Progress in Photovoltaics: Research and Applications, 2000, pp. 93-111.; 3. R. M. Swanson, “The promise of concentrators – update 2010,” presented at CPV-6, Freiburg, Germany, 2010.; 2. S. Kurtz, “Opportunities and challenges for development of a mature concentrating photovoltaic power industry,” Technical Report, NREL/TP-520- 43208, 2009.; 3. R. M. Swanson, “The promise of concentrators – update 2010,” presented at CPV-6, Freiburg, Germany, 2010.; 4. W. P. Mulligan et al., “Manufacture of solar cells with 21% efficiency”, Proceedings of the 19th European Photovoltaic Solar Energy Conference, 2004, pp. 387-390.; 5. M. M. Bunea et al., “Simulation and characterization of high efficiency back contact cells for low-concentration photovoltaics,” presented at the 35th IEEE Photovoltaic Specialists Conference, Honolulu, Hawaii, USA, 2010. INTRODUCTION SECTION 1 3 SECTION 2 Cell Technology In a previous study, we outlined how the standard SunPower back contact solar cell technology performs as a function of cell size, illumination band width, illumination intensity, and position of the illumination on the cell [5]. This paper highlighted a quarter-wafer cell design, which is minimally-modified standard SunPower cell cut into four quadrants. This concentrator cell differs from our standard design in a few subtle ways: 1) the bus bars run the full length of the cell at constant width to handle increased current; 2) the metal finger pitch is denser, and 3) the contact spacing is slightly tighter. Despite these changes, these cells can be produced on the standard SunPower cell production lines. The quarter-wafer cell was chosen for our LCPV system because it maintains a manageable current (and therefore I2R losses) under concentration, provides a target area large enough for passive thermal management, and enables an ample acceptance angle for our tracker and mechanical tolerance stack-up. Figure 1 below shows the quarter-wafer cell efficiency as a function of intensity from 0 – 20 suns; for comparison, designs based on a third-wafer and half-wafer are also included [5]. 24.0 between the aperture area of the module and the cell area, was chosen to be seven in an effort to maximize cell efficiency. However, other factors such as optical efficiency, focal band size, direct normal irradiance (DNI), and soiling rates also impact the operating point on the efficiency-intensity curve. All of these factors were considered in optimizing the geometric concentration ratio for annual system yield. 23.5 23.0 22.5 22.0 21.5 21.0 20.5 Third-wafer 20.0 Quarter -wafer 19.5 Half-wafer Receiver And Optical Designs 19.0 0 2 4 6 8 10 12 14 16 18 20 Intensity [Suns] Figure 1: Measured efficiency-intensity curve for the quarter-wafer cell. Also shown are the corresponding curves for the third-wafer and half wafer designs. One-sun is defined as 1000W/m2. The quarter-wafer cell exhibits a relatively flat efficiency peak from 4 – 7 suns, which was an important design consideration for our system architecture. The geometric concentration ratio, defined as the ratio The receiver is comprised of 24 quarter-wafer cells connected in series. There are three bypass diodes in the circuit, and each is connected across a grouping of eight cells. The receiver package is similar to that used in a standard one-sun module with a few minor changes to mitigate the concentrated ultraviolet content and to enhance the reliability of the laminate. A folded fin heat-sink is mechanically bonded and thermally coupled to the laminate back sheet. A monopole junction box is attached to the back side at either end of the receiver. Isometric views of the front and back of the receiver are shown below in Figure 2 and Figure 3, respectively. 5. M. M. Bunea et al., “Simulation and characterization of high efficiency back contact cells for low-concentration photovoltaics,” presented at the 35th IEEE Photovoltaic Specialists Conference, Honolulu, Hawaii, USA, 2010. CELL TECHNOLOGY SECTION 2 4 SECTION 2 System Architecture Figure 2: Front-side view of the LCPV receiver. Having deployed hundreds of megawatts of singleaxis tracking systems around the world, SunPower has extensive experience in building low-cost and highly reliable trackers. In an effort to leverage this experience, our LCPV system is based on our horizontal single-axis T-0 tracker technology. Like T-0, the LCPV tracker is supported by steel piers, a central torque tube, and cross-struts to secure the modular LCPV mirror-receiver assemblies, which we call blades. The tracker design is depicted below in Figure 3: Back-side view of the LCPV receiver. Figure 4: View of tracker after addition of the cross-struts. The concentrating optic used in this system is a parabolic section linear glass mirror. A silver metallization has been used in order to maximize optical efficiency. The mirror panels each have an aperture area of 0.44m x 1.55m, and they have a focal length of 0.21m. In the system, each receiver is positioned closer to the mirror than the focal line which widens the flux band and maximizes cell efficiency. The cross-strut is designed such that the blades are held in opposing directions on the east and west halves of the north-south oriented torque tube. This design, which is symmetric about the rotational axis and gradually elevates at a constant slope outwards from the center to the edges of the tracker, has several key advantages: 6. Y. Tsuno, et al., as presented at the 25th European Photovoltaic Solar Energy Conference, Valencia, Spain, 6-9 Sept. 2010.; 7. D. Cunningham, et al., Proceedings of the 35th Photovoltaic Specialists Conference, 22 June 2010, Honolulu, Hawaii. SYSTEM ARCHITECTURE SECTION 2 5 SECTION 2 1. It allows the center of rotation to be placed at the center of gravity of the combined torque tube, cross-strut, and blade assembly. This eliminates twist in the torque tube due to the center of gravity being offset from the rotational axis and therefore enables more accurate tracking. 2. It creates space for a larger heat sink with correspondingly larger volumetric air flow rate and a lower receiver operating temperature. 3. It raises the receiver position relative to the mirror, increasing optical efficiency and misalignment tolerance. 4. It enables better utilization of the tracker and land. Almost all of the tracker area is filled with mirror aperture, lowering system and BOS costs by minimizing light loss through the aperture of the tracker. 5. It reduces the wind profile when mirrors are rotated above the rotational axis, which significantly decreases wind torque loads and structure requirements. 6. It eliminates the need for long jumper cables for electrical connections between rows in the string. The mirror and receiver are pre-assembled into a blade and shipped to the field as a single unit. It is quickly snapped into the cross-strut and can be easily handled by one person. Figure 5 shows a cross-sectional view and Figure 6 shows an elevated view of six blades attached to two cross-struts. Figure 5: Cross sectional view of the cross-strut, blades, and receivers. Figure 6: Overhead view of six mirror-receiver assemblies attached to cross-struts on torque tube. 4. M. Garcia, et al., ―Estimation of photovoltaic module yearly temperature and performance based on Nominal Operation Cell Temperature calculations‖, Renewable Energy 29 (2004) 1997-2010. SYSTEM ARCHTECTURE SECTION 1 6 SECTION 2 As shown in Figure 7, a single tracker consists of two torque tubes driven by a center-mounted actuator. There are 10 cross-struts supporting 54 blades mounted on each torque tube; an additional crossstrut is mounted on the southern end to support one 0.5m mirror for each row. This southern overbuild is necessary for horizontal single-axis trackers due to non-zero solar zenith angles. The power rating of the tracker is approximately 12kW. The Alpha-2 system consists of the northern halves of two full trackers as shown in Figure 8. Construction of the array was completed in August of this year. The maximum rotation angle of the trackers is 75˚ and the ground coverage ratio of the site is 35%. In an effort to keep the system under 600V (which allows for the use of standard production inverters) each half-tracker consists of two electrical strings (i.e., four strings total). A larger-scale version of this system would operate at a higher voltage and would consist of one electrical string per half-tracker. The tracker controller for this system is identical to that of our standard T-0 product. We have avoided the need to use an active tracking algorithm; instead, this system uses standard solar position models to determine rotational position. Figure 7: The full LCPV tracker. For viewing of the piers, the blades are not shown. ALPHA-2 SYSTEM The Alpha-2 System was installed at Sandia National Laboratory (SNL) in Albuquerque, NM. In addition to being located in a geographical location with a strong solar resource, the Photovoltaic Systems Evaluation Laboratory (PSEL) at SNL has an impeccable reputation as a first class PV characterization and evaluation facility. PSEL has been involved in this project for over a year and has provided invaluable support by participating in design reviews, advising on system characterization, helping with the design of sensors, and contributing to the failure mode and effects analysis and reliability program. Figure 8: Alpha-2 LCPV system ALPHA-2 SYSTEM SECTION 2 7 SECTION 2 Preliminary Performance Results The Alpha-2 System’s instrumentation provides a full suite of metrology tools to evaluate performance, ambient conditions, and system characteristics. We are currently monitoring the following: 1. Direct normal irradiance, global horizontal irradiance, diffuse horizontal irradiance, and plane of array irradiance. 2. Ambient temperature, wind speed, wind direction, and relative humidity. 3. In-situ reference cells that have been laminated into receivers. These allow us to monitor optical efficiency and soiling rates. 4. Cell, heat sink, and backsheet temperatures in 48 locations throughout the array. 5. DC current and voltage for each of the 4 strings. To date, we have quantified array level efficiency and acceptance angle. We are working with PSEL to further characterize the system. In particular, we will extract the coefficients for the Sandia Photovoltaic Array Performance Model, which will allow us to better understand the annual energy yield of the system. Efficiency data was extracted from current-voltage (I-V) curves obtained using a Daystar DS-100C. Before efficiency values were extracted, the I-V curves were translated to 850W/m2 direct plane-of-array irradiance and 25˚C cell temperature. Direct plane-ofarray irradiance was determined by measuring DNI with an Eppley normal incidence pyroheliometer and multiplying it by the cosine of the angle between the plane-of-array normal and the solar zenith angle. The temperature of the array was determined by a direct measurement with thermocouples and corroborated with calculations based on the open circuit voltage of the array. The best-performing strings showed a direct resource efficiency of 18.4%, which translates to a global efficiency of 15.6%. Table 1 below summarizes the efficiency results. Global Efficiency Direct Resource Efficiency Instantaneous Direct/Global Fraction String 1 15.6% 18.4% 90.2% String 2 15.6% 18.3% 91.5% String 3 15.6% 18.4% 89.1% String 4 15.4% 18.2% 90.2% Table 1: Measured string level efficiency data for the Alpha-2 System. Another important factor in the performance of an LCPV system is the acceptance angle. We measured acceptance angle by rotating the tracker to the West of the on-sun position and allowing the sun to pass through while acquiring rapid I-V measurements. We then analyzed power vs. misalignment, rather than Isc vs. misalignment which is only a measure of the optical acceptance angle. We achieved an angular range of ±1.25˚ in which the power loss was less than 5%. These data are depicted below in Figure 9. Preliminary thermal data has been promising. Depending on wind conditions, the cells have been observed to operate on average between 25°C and 40°C above ambient temperatures. Several thermal design optimizations are in progress with the objective of further enhancing the thermal performance of the receiver package. These will be discussed in a later study. 5. Sample NPV calculation is for example purposes only and is based on the following assumptions: $0.12 current rate of electricity, 3 % annual rate escalation, $0.01 REC value, $0.05 performance based solar incentive over 5 years, 8% discount rate, 25 project life cycle. P R E L I M I N A R Y P E R F O R M A N C E R E S U LT SECTION 2 8 SECTION 2 0% -5% -10% -15% -20% -25% -30% -35% -40% -4 -3 -2 -1 0 1 2 3 4 Rotational M isalignment [Degrees] Figure 9: Acceptance angle of a single string. There is a 1˚ angular range (0˚ to -1˚) in which the power loss is less than the accuracy of the measurement. In order to target specific areas for further development and optimization, we have created a comprehensive power loss model for this LCPV system. The modeling is based on spectrometric measurements, flash test data, and simulations. The main loss mechanisms are displayed in Figure 10 below. In the figure above, the receiver optical losses include reflection from the front surface of the glass and absorption in the various layers of the laminate. The mirror losses are associated with the imperfect reflection and absorption in the glass superstrate. The receiver area fill factor refers to the non-active area of the receiver (i.e., the space between the cells) that is illuminated by the flux band. The cell efficiency losses due to localized regions of high intensity and nonnormal illumination are grouped into a single category. Electrical losses include any I2R power dissipation in the cell busbar, interconnects, and junction boxes. The opto-electric mismatch includes any optical imperfections that can cause an inconsistency in illumination intensity between cells. Cell mismatch is the inherent performance loss due to binning. The model predicted a total direct resource efficiency which was within 2% of the measured data. The total optical efficiency of the system can be calculated using the receiver optical, mirror, and flux profile and angle losses. This was determined to be 88%, which will be validated using the Alpha-2 System’s metrology tools. 5% 4% 3% 2% 1% 0% Figure 10: Major loss mechanisms for the LCPV system. This breakdown does not include thermal effects, tracking losses, or solar resource capture. 5. D. DeGraaff, et al., ―Qualification, Manufacturing, and Reliability Testing Methodologies for Depolying High-Reliability Solar Modules,‖ these proceedings. P R E L I M I N A R Y P E R F O R M A N C E R E S U LT SECTION 2 9 CONCLUSIONS SECTION 3 We have demonstrated a linear 7X geometric LCPV system that utilizes the superior performance of SunPower cells under concentration. The design leverages the standard SunPower module and single-axis tracker technologies to achieve a low-cost and deployable product. Slight modifications to the module design were made to withstand higher ultraviolet content and thermal loads. This system was shown to achieve a direct resource efficiency of 18.4%, which is in agreement with modeled values. CONCLUSIONS SECTION 3 10 SUNPOWER CORPORATION 3939 North 1st Street San Jose, CA 95134 1.800.SUNPOWER (1.800.786.7693) sunpowercorp.com SUNPOWER and the SUNPOWER logo are trademarks or registered trademarks of SunPower Corporation. © February 2011 SunPower Corporation. All rights reserved. 11