This paper discusses the performance of a recently built 30 kW CPV (Concentrator PhotoVoltaic) power plant. It was constructed in a landfill area near the new Nagoya international airport (Centrair International Airport) and the terrestrial position is N34°53′4” and E136°49′34” (see Fig. 1
Fig. 1 Photograph of 30 kW CPV system consists of 6 trackers.
Usually, CPV power plants are constructed in dry and inland area to receive better direct solar resources and to avoid degradation of components from moisture and salt. The challenge of this power plant was to show feasibility in introducing a CPV power plant in a suboptimal area; namely close to the airport, where it is exposed to dust from airplane exhaust and in a coastal area where it is subject to receiving chloride damage as well as strong wind.
It has been widely believed that CPV is not suitable for cloudy areas like Japan. This is partly correct in view of the relative advantage of dry areas. However, recent advancement of CPV technology and high-efficiency III-V multi-junction solar cells has raised the market competitiveness of CPV compared to other flat-plate PV technologies [1
1. A. W. Bett, F. Dimroth, W. Guter, R. Hoheisel, E. Oliva, S. P. Philipps, J. Schöne, G. Siefer, M. Steiner, A. Wekkeli, E. Welser, M. Meusel, W. Köstler, and G. Strobl, “Highest Efficiency Multi-Junction Solar Cell for Terrestrial and Space Applications”, Proc. 24th EUPVSEC, (2009), 1.
2. A.W. Bett, F. Dimroth, W. Guter, J. Jaus, P. Nitz, E. Oliva, G. Peharz, S. Phillips, J. Schöne, T. Schult, G. Siefer, M. Steiner and E. Welser, “Raising the Efficiency of FLATCON®Modules”, 5th ICSC (2008).
]. One of the most important purposes of this demonstration project is to see how CPV performs in such cloudy areas.
The electricity generated by this power plant is delivered to a local sewage center that services the entire sewerage treatment of the Nagoya International airport and the urban area of Tokoname city (population: 55,000). By design, the energy generated by this CPV system supplies 10% of the electricity demand on a clear-sky day.
2. Power plant ground design
Land space in Japan is limited. It was important to propose an optimum construction plan for the best utilization of the available land.
First, the multi-tracking system was optimized taking self-shadowing among trackers in consideration. Most studies consider only the irradiance losses among trackers [3
3. J. Monedero, and F. Dobon, “Two-Axis Tracking System TETRA-TRACK: System Description and Study for Optimized Space Factor”, CDROM of International Solar Concentrator Conference for the Generation of Electricity or Hydrogen, (2003)
4. P.J.Pérez, G.Almonacid, J.Aguilera, P.G.Vidal, J. de la Casa, I.Luque-Heredia, M. Doménech. “Multi-trackers Systems Calculation of Losses due to Self-shadowing”, 4th ICSC, (2008), 201.
]. This method often leads to relatively large clearance between trackers along the East–west direction. However, this conventional approach is not correct for a spectrum-sensitive III-V concentrator PV. The new calculation method anticipates effects on energy production due to nonlinear and seasonal influence on photovoltaic output, while also considering mismatch losses from shading effects, tracking durations, inherent seasonal fluctuation of efficiency, seasonal fluctuation of sunshine duration, and DNI fluctuation. After applying this new method, the tracker allocation became substantially more compact leading to a high performance ratio from a limited space (see Fig. 2
Fig. 2 Left: Ratio of irradiation utilization after influenced by the tracker in the center; Right: Correction by a new method.
Weed suppression is also important to installations in Japan. Even though trackers are arranged optimally, there is ample open space. Unfortunately, due to the relatively large fraction of diffused sunlight in Japan, it was found that stray sunshine is sufficient to grow grass. We decided to cover the ground by turf rather than gravel. In other words, CPV has the least optical impact to the ground while retaining its positive utilization for gardens and agriculture.
In contrast to modules in dry areas, CPV modules in Japan have to be prepared for a wet environment. First, all the materials in the module enclosure including receiver components and solar cells were tested in a high moisture and water environment, with risk of water condensation. Some materials commonly used in CPV receivers did not pass our tests and were replaced by alternative robust ones. Second, breathing holes were optimized to avoid water dew on the backside of the Fresnel lens. The main idea is to enhance air ventilation while prohibiting water intrusion, similar to traditional houses in East Asia.
Recently, the module performance test has become a big issue in a volume production environment. A typical approach is to use a solar simulator. Another approach is an outdoor test. However, the solar spectrum and irradiation are not stable (especially in Japan). It is also time-consuming to load modules to a tracker just for testing. The approach we took was to complete the performance test at the receiver level and to assume that the module performance is predictable by the testing of some components. This systematic approach is based on the design requirement that alignment and component dimension error always stay in the design window, and the performance and quality variation of lenses are sufficiently small. However, it is still necessary that the performance of some modules be verified by outdoor measurement.An I-V curve of a typical production module is shown in Fig. 3
Fig. 3 I-V curve of one of the production modules.
. After accounting for mismatch losses and variation of cell performance, the efficiency of the entire CPV module array was predicted as 24.1%. Later, this was confirmed by continuous operation.
It is acknowledged that soiling or dust accumulation of optics is one of the main reasons for lower performance of CPV systems in real installations. Many systems rely on periodic cleaning, or are even equipped with water showers. Since traditional roofs in East Asia use curved roof tiles to guide water flow, a dome-shaped lens surface likewise helps wash away dust, even with light rain (see Fig. 4
Fig. 4 How dust on the lens are washed by rainfall.
). This also saves water for cleaning.Among the distinct advantages of dome-shaped Fresnel lenses are:
• Reduced coma. Since the principal plane of a lens satisfying the Abbe sine condition is spherical, a spherical shape for the Fresnel lens is preferred;
• Minimum deviation makes it possible to minimize chromatic aberration;
• Minimum reflectance loss especially that of the back surface of the lens;
• Shorter focal length;
• Robustness against static load to the lens such by snowfall.
In most cases, the structure and the size of actuators in trackers are governed by wind load. This is one of the reasons why most trackers are equipped with a stow system to avoid an excessively rigid structure and heavy actuators. In this regards, an area like Japan which often encounters strong storms, requires different design rules. At the same time, due to the limited land space and for meteorological reasons, many candidates of CPV power plant in Japan lie in coastal area and have even greater chance to receive strong winds.
The first thing we had to do to design a 30 kW CPV system was to investigate wind statistics and soil condition for foundation. Almost half the days in a year recorded more than 10 m/s of maximum wind (averaged over 10 minutes) velocity (by the observation of Airport Aviation Weather Service). Considering that the instantaneous wind velocity often reaches a value that is twice as high than that of a 10 minutes average, most of the current conventional trackers, which set the maximum operation wind velocity as 45 km/h (12.5 m/s), did not seem to be applicable to such a windy area. Even worse, most of the strong wind in this area comes from the west, which means that the tracker will receive the highest wind load during the whole night, unless it assumes a night stow position, which brings additional tracking power consumption.
One compromise is to set a new and appropriate value for the stow limit which does not require excessively high structural specification but allows reasonable duration of stow. Based on the statistics of the wind velocity distribution, the appearance probability of gusts are anticipated. For example, if the wind stow limit is set to 20 m/s (72 km/h), the probability of triggering a wind stow is a little less than 0.01 and acceptable to the plant operation.
Once, the tracker detects strong wind, it moves to stow position and waits until it is safe to move back to the normal operation position. Considering the fact that it usually takes 10 minutes to reach to the stow position, it is sometimes wise to wait for a while. The longer the stow duration, the safer, but this causes a loss of operation time. For application in windy areas, it is important to optimize the wind stow strategy, referring to meteorological statistics. The necessary information is the correlation between strong wind and sunshine, as well as the autocorrelation of the wind velocity. Unfortunately, most of the strong wind on this site was due to the monsoon burst in winter and early spring, and was often experienced on clear sky days. After some probability calculations based on autocorrelation data, the total stow in a year was calculated varying both the wind velocity threshold and the stow duration (namely time duration until the tracker starts go back to the operation position). It was shown that the considering the threshold value is essential in the wind stow strategy. It was also necessary to refer to the wind velocity record of storm days. On this site, 2 hours is the safe value to avoid the onset of the revival of gusts.
On Oct. 08, 2009, a strong Typhoon descended on our site. Wind speeds as strong as 42.4 m/s (153 km/h) were recorded at the neighboring Nagoya International Airport. All the trackers experienced that wind in the stow position. As a matter of fact, there was no damage.
The robustness against strong wind was demonstrated by another 14 kW tracker of bigger size, which was installed in Miyazaki, Japan. On December 31, 2009, a win gust a strong as 47.1 m/s (170 km/h) was observed at the center height of this tracker. Again, no damage was reported.
To achieve robustness to strong wind, various improvements were made. One of these is an offset elevation structure. By shifting the rotation center of the panel lower, the panel height in stow position is reduced. It was also effective in reducing vibration due to wind load. The drawback of this design is a prolonged linear actuation and an increase of tracking power consumption. To decrease the actuation duration and to reduce power consumption, a new tracking control was developed which succeeded in tracking the sun only once a minute while retaining precision. The drivers and motors were unpowered except during actuation. The measured averaged power consumption by the motors of the 30 kW system was 10-30 W, depending on the ambient temperature and the number of revolutions (see Fig. 5
Fig. 5 Total power consumption of motors of 30 kW CPV system as a function of temperature and culmination altitude of the sun
). The motor power tends to increase with a decrease of ambient temperature and the increase of daytime. Applying linear regression, the annual averaged motor power consumption was 19.6 W, which, for a 30 kW system, corresponds to 0.07% of the rated power and 0.6% of the generated power.
Thanks to the structural design to withstandstrong winds and the advanced parameter-based error correction control, the tracking error is sufficientlysmall to accommodate the measured tracking error tolerance of the array of moduels which is plus or minus 1 degree, .
Robustness is as important as high performance. For example, it is well-known among CPV system engineers that the use of good solar cells with 5% higher efficiency is easily offset by the loss of 50% from the combination of bad trackers or installation. Daido Steel recorded a 28% module efficiency in 2003, and a 31.5 ± 1.7% corrected efficiency (29% uncorrected) in 2004, but we have not pursued higher number since then. Rather, interests have moved to robustness to maintain high performance in the field.
As shown in the I-V curve in Fig. 6
Fig. 6 I-V curve of the 5 kW array
, the fill-factor is maintained at high numerical values due to the small current mismatch losses. We attribute this positive outcome to ouradvanced module design, which tolerates larger alignment and tracking errors. Since our receiver solution provides the necessary tolerance, both module and tracker assembly processes do not require any special tools or any special alignment, to stay within the required error budget. Actually, all the modules in the 30 kW system were assembled in the warehouse, and the entire module were simply fastened to the tracker frame without the need for alignment adjustment, as is common in CPV installation. This suggests that local assembly is possible and practical with the help of well-controlled receivers made by Daido Steel.
The acceptance angle of the tracking errors, depending on the tracker, lies between ± 1 to ± 0.7 degrees. This number is one of the largest among leading CPV manufactures.
6. Energy generation
Since March 3, 2009, the 30 kW CPV system has had continuous service (see Fig. 7
Fig. 7 Efficiency and energy generation of 30 kW CPV system since grid connection. No cleaning we performed since installation.
). There have been no significant problems during this period.
Power generation varied according to climate and was high on clear-sky days but lower on cloudy days. The variation of the efficiency appears larger than for crystalline silicon systems as expected from the spectral sensitivity of 3J cells. However, a peak uncorrected efficiency as high as 24% has been observed for the 30 kW array, not counting the seasonal reduction in performance attriburable to the inherent spectral sensitivity of 3J celle. . Because of this spectral sensitivity, the higher the solar elevation, the higher the observed efficiency
The performance ratio up to now is 0.87. The performance ratio is the ratio between power generation per rated power and the irradiance per nominal irradiance rating. The performance ratio is almost unity when the direct normal irradiance is more than 700 W/m2
(see Fig. 8
Fig. 8 Histogram of performance ratio by various levels of irradiance
) This means that better performance can be expected when the installation is in dry areas rather than in cloudy Japan.
Aside from spectrum and irradiance, the influence of the temperature and the soiling of optics were also investigated.
Thanks to the lens design which was copied from traditional roofs, every rainfall effectively washed out the dust on the lenses. Thus the surface of the lens is kept clean. It is said that one of the secrets to recording high performance in CPV is frequent cleaning, but this is not true in our system. Our high performance does not rely on periodic cleaning work but relies on self-cleaning by nature (see Fig. 9
Fig. 9 Efficiency degradation as a function of the number of days after rainfall
Some systems rely on active cooling, or passive but large heat sink structures. Although we do not use heat sinks or fins, our temperature characteristics are similar to those of 3J solar cells. Like other CPV systems using active cooling and heat sinks, high performance is maintained even on hot summer days (see Fig. 10
Fig. 10 Performance ratio at the noon vs. ambient temperature
One of the well-known characteristics of 3J cells is their spectral sensitivity. Due to the spectral dependence on air mass, the output current of the top junction is limited and thus the output power drops. This influence is sometimes larger than the voltage drop by temperature. This is the reason why the energy output in winter drops (see Fig. 11
Fig. 11 Efficiency vs. air mass
A 30 kW concentrator photovoltaic power plant was constructed and has started operation. The power consumption of the tracking motors was only 19.6 W, namely 0.07% of the rated power. Improved optics reduces the mismatch losses associated with optical aberrations. An efficiency of 25.8% (STC) was achieved even in an array size as large as 23.8 m2. The system performance ratio was 0.87, and the capacity factor was 11.7%. The energy generation per rated power was 1,020 kWh/kWp. While it is true that CPV system perform better in dry and high irradiance areas, our 30 kW system installed in a cloudy area like Japan also showed satisfactory performance.