- Email: [email protected]
A multipurpose dual-axis solar tracker with two tracking strategies
Renewable Energy 72 (2014) 88e98
Contents lists available at ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
A multipurpose dual-axis solar tracker with two tracking strategies Yingxue Yao a, Yeguang Hu b, *, Shengdong Gao b, Gang Yang b, Jinguang Du b a
Shenzhen Graduate School, Harbin Institute of Technology, HIT Campus, University Town of Shenzhen, Xili, Shenzhen 518055, Guangdong Province, PR China b School of Mechanical and Electrical Engineering, Harbin Institute of Technology, 92 West Dazhi Street, Harbin 150001, Heilongjiang Province, PR China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 January 2014 Accepted 1 July 2014 Available online
This paper deals with a multipurpose dual-axis solar tracker that can be applied to solar power systems. This tracker employs a declination-clock mounting system that locates the primary axis in east-west direction. Based on this mounting system, normal tracking strategy and daily adjustment strategy are developed for flat Photovoltaic (PV) systems and Concentrating Solar Power (CSP) systems respectively. While the former strategy keeps the tracking errors smaller than the pre-specified values, the latter one simplifies the tracking process by adjusting the primary axis once a day and driving the secondary axis to rotate at a constant speed of 15 /h. Results of the accuracy test indicate that the tracking error of the normal tracking strategy is within 0.15 . The other strategy may have greater tracking errors, but its annual average cosine loss for flat PV systems is estimated to be below 1.3%. Furthermore, in the test on the output of the PV modules, it is found that the average energy efficiency of the normal tracking PV, compared with the fixed PV, is more than 23.6%. And the average energy efficiency of the daily adjusted PV is more than 31.8%. Results of the experiment show that the two tracking strategies are both feasible for the developed tracker. © 2014 Elsevier Ltd. All rights reserved.
Keywords: Solar energy Solar power system Solar tracker Automatic tracking strategy Cosine loss
1. Introduction As a kind of clean and renewable energy source, solar energy has been drawing more and more attention, especially in the field of electricity generation, due to the shortage and pollution of fossil fuels. The process of converting solar energy to electric energy is realized mainly through flat PV systems or CSP systems. The power output that these systems could produce depends on various factors, including the amount of the energy they receive from the solar radiation. Some researchers have studied the optimal angle of solar collector to increase the power output [1,2]. As the sun's position changes throughout the day, the solar tracker is a more efficient method of increasing the energy production. So the solar tracker is being studied by more and more researchers. Currently, there are mainly two types of solar trackers based on movement capability: single-axis tracker [3e5] and dual-axis tracker [6e9]. Different single-axis and dual-axis trackers have been presented by the previous studies. Clifford and Eastwood [10] presented a passive solar tracker activated by aluminium/steel bimetallic strips and controlled by a viscous damper. Poulek and
* Corresponding author. Tel.: þ86 451 86413557. E-mail address: [email protected] (Y. Hu). http://dx.doi.org/10.1016/j.renene.2014.07.002 0960-1481/© 2014 Elsevier Ltd. All rights reserved.
Libra [11] designed a simple single-axis solar tracker based on a new arrangement of auxiliary bifacial solar cell connected directly to DC motor. Kim et al. [12] proposed a single polar axis tracker with the solar collector rotating at a speed of 15 /h round the polar axis. Roth et al. [13] proposed an altitude-azimuth dual-axis tracker which, guided by a closed loop serve system, could operate automatically. Batayneh et al. [14] proposed a dual-axis sun tracking system with altitude-azimuth mounting controlled by the designed fuzzy controller. Mavromatakis and Franghiadakis [15] presented a novel single-axis azimuthal tracker with the ability to move the collector's plane in two directions through a special support structure. According to the previous studies, solar trackers have been used in different solar collector systems. C.S. Chin et al. [16] presented an active single-axis solar tracker used in flat PV systems. The experimental test showed that the efficiency over the fixed solar panel was around 20%. Chang [17] tested the flat PV system mounted on a single-axis tracker and found that the gain of the single-axis tracking panel installed with the yearly optimal angle was 17.5%, compared to a traditional fixed panel. Kim et al. [12] applied the single polar axis tracker to CPC solar collector. The study of Kacira et al. [18] showed a daily average of 34.6% gain in generated power with two-axis solar tracking compared to a fixed PV panel on a particular day in July in Sanliurfa, Turkey. Abdallah and Nijmeh [19] designed a two axes sun tracking system for PV
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
system. The system surface showed a better performance with an increase in the collected energy of up to 41.34% compared with the fixed surface. When analyzing the current literature, the authors found no available trackers that are developed properly for both flat PV systems and CSP systems. Flat PV systems and CSP systems have different demands for tracking accuracy. On the one hand, current low accuracy trackers, such as the single-axis tracker designed for flat PV systems, cannot be used in CSP systems because their low accuracy will lead to a great loss of the solar energy intercepted by the receiver. On the other hand, if a tracker, designed for highaccuracy tracking purposes, is used in a flat PV system, there will be extra operational costs imposed by control process. This paper takes into account the two types of demands for accuracy and develops a dual-axis tracker employing declination-clock mounting system (This mounting system has been used by Bakos [20] for the efficiency improvement of parabolic trough collector (PTC) in practice. However no available analysis is made for the principle of this mounting system in the literature. The main new contributions of this paper compared to [20] are summarized as follows: The tracking formulas for the declination-clock mounting system have been derived and also a much simpler tracking strategy has been proposed for the flat PV system; A linkage mechanism, which provides with a simple and light-weight structure, has been designed for the motion of the primary axis; The belt transmission is employed for the motion of the secondary axis, which can reduce the dynamic load occurring in the mechanism; Mechanical groupcontrol, which allow the solar tracker to accommodate more collectors to track the sun, is achieved in the proposed structure; A sun position sensor with high resolution is designed and employed in the presented solar tracker.). This mounting system would allow the presented tracker in this paper to move under two tracking strategies, namely normal tracking strategy and daily adjustment strategy. Normal tracking strategy shows a good tracking performance with the error remaining below 0.15 . Daily adjustment strategy is used for simple tracking purpose instead of highaccuracy tracking. Its annual average cosine loss for flat PV systems is calculated to be below 1.3%. 2. Description of the solar tracker 2.1. Analysis of declination-clock mounting system The developed tracker employs a mounting system named as declination-clock mounting which has fine motion ability. The
89
declination-clock mounting system has two rotation axes (Fig. 1(b)): primary axis, located in east-west direction and secondary axis, perpendicular to the primary axis and able to rotate around it. Rotation angles for the primary and secondary axis are defined as declination angle and clock angle respectively. The declination-clock system is obtained from the pseudo-azimuthal system [21] by inverting the order of the rotation axes e regarding the latter system, the primary axis is for the azimuthal motion (which, in terms of the former system, generates a corresponding clock angle), while the secondary axis is for elevation (which, in terms of the former system, generates a corresponding declination angle). The tracking schematics for the developed tracker and the pseudo-azimuthal tracker are shown in Fig. 1(a) and (b) for comparison. The rotation angles of the tracker are determined by the sun's position. So the unit normal vector of the solar collector is given by
n ¼ vs
(1)
where n is the unit normal vector of the solar collector; vs is the unit vector incident to the sun. In general, the vector vs can be determined by the solar azimuth angle gs and altitude angle as [22]. Two coordinate systems, namely XYZ-O and IJK-O, are established for the analysis of declination-clock mounting system, (Fig. 2). XYZ-O is located on the ground and IJK-O is on the declination-clock tracker. The vector vs can be expressed in the coordinate system of XYZ-O as
vs ¼ ðsin gs cos as ; cos gs cos as ; sin as Þ
(2)
According to the relations in the figure, vector n can be expressed in the coordinate system of IJK-O as
nIJKO ¼ ðsin qCL ; cos qCL ; 0Þ
(3)
Then through the rotation transformation from the system IJK-O to the system XYZ-O, the vector n can be expressed in the coordinate system of XYZ-O as
nXYZO ¼ ðsin qCL ; sin qDE cos qCL ; cos qDE cos qCL Þ
(4)
So the formulas of the angles qDE and qCL can be expressed as
tan qDE ¼ cos gs cot as sin qCL ¼ sin gs cos as
(5)
Similarly, through the rotation transformation, the angles qAZ and qEL for pseudo-azimuthal system can be derived from
Fig. 1. Comparison of tracking schematics of: (a) pseudo-azimuthal mounting; (b) declination-clock mounting.
90
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
2.2. Design of the solar tracker
Fig. 2. Schematics of vector n in the declination-clock mounting system.
tan qAZ ¼ sin gs cot as sin qEL ¼ cos gs cos as
(6)
The location is set in Harbin city, where the latitude is 45.75 north. Based on Eqs. (5) and (6), the theoretical curves showing the two rotation angles for both the pseudo-azimuthal mounting and declination-clock mounting are plotted in Fig. 3(a) and (b). It should be noted that the black curves in Fig. 3 indicate the rotation angles of the primary axes, and the red (in web version) curves indicate the rotation angles of the secondary axes. So here the declination angle and the azimuth angle are brought into comparison. The rotation speed of the primary axis for declination-clock mounting is very small in most times of the day (only a rise or decline in the morning or evening), especially when it comes close to the spring equinox or autumn equinox. For the pseudo-azimuthal mounting, by contrast, the primary axis rotates at a considerable speed during the tracking period. Besides, the rotation speed of the secondary axis for the declination-clock mounting system is almost constant, whereas for the pseudo-azimuthal system, the rotation speed of the secondary axis undergoes a considerable change. Apparently, the declinationclock mounting system shows a steadier performance in mechanical motion. Furthermore, based on the features of this mounting system, this paper proposes a very simple tracking strategy, which is detailed in Section 3.
For the designed solar tracker (Fig. 4(a)), the solar collectors are installed on the support frames (13). The linkages, consisting of a linear actuator (3, 4) and four connecting rods (1, 2, 5, 6) connecting with the two ends of the actuator symmetrically, are used to drive the solar collector array to rotate around the primary axis. When DC motor drives the linear actuator to move, the rotation support (7), which holds the solar collector array, will rotate around the primary axis. The rotation of the secondary axis is implemented by the linear actuator (9, 12) installed on the rotation support. The linear actuator connects with the belt (10). When the linear actuator pulls the belt to reciprocate, the two related pulleys (8, 11) will drive support frames (13) to rotate around the secondary axis. Every frame is installed with a pulley and all of the pulleys are connected to the belt in sequence. The distance between the axes of two adjacent pulleys is 605 mm. Thus through efficient belt transmission, the solar collectors will rotate around the secondary axis simultaneously, and thus mechanical group-control for the solar collector array is realized. A corresponding physical prototype has been produced for tracking experiment, which will be detailed in Section 4. If, however, economy issues are taken into consideration, the above-mentioned structure can be further simplified: the four connecting rods can be omitted and the linear actuator can be directly connected to the base and to the rotation support (Fig. 4(b)). Although this simplified structure is not as robust as the original one which provides more connecting joints for the base and the rotation support, it can be applied under conditions of low load. In this simplified structure, the connections between the rotation support and the base are moved to the short lateral frames of the rotation support in order to eliminate the imbalance of this mechanical structure for the declination motion. In the same idea, the driving mechanism for the motion of the secondary axis is mounted on the other side of the rotation support. The physical prototype for this model is not put into production due to the same tracking function with the original one. The mechanisms, which drive the solar collectors to rotate around the primary axis, constitute a spatial kinematic chain (see Fig. 5). The number of degrees of freedom (DOF) for this mechanism can be calculated by
Fig. 3. Comparison of rotation angles for: (a) pseudo-azimuthal mounting; (b) declination-clock mounting.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
91
Fig. 4. Structure of the solar tracker: (a) the model with linkage mechanism; (b) the model without the linkage mechanism. (1, 2, 5, 6) connecting rods, (3) the linear actuator (piston head) for declination motion, (4) the linear actuator (piston cylinder) for declination motion, (7) rotation support holding solar collector array, (8, 11) pulleys, (9) linear actuator (piston head) for secondary motion, (10) belt, (12) linear actuator (piston cylinder) for secondary motion, (13) support frame for solar collectors, (14) solar panel, (15) secondary axis, (16) primary axis.
DOF ¼ 6n
X
r ¼ 6n ð5P1 þ 4P2 þ 3P3 þ 2P4 þ P5 Þ
(7)
P where n is the number of moving links in the mechanism; r is the number of restrictions; P1 ~ P5 are respectively the numbers of pairs of class I ~ V. Apparently, the redundant constraint exists in the presented mechanism since the axis of revolute joint K coincides with that of joint J. Thus the revolute joint K should be taken away for the calculation of DOF. Due to the symmetry of the linkages, the restrictions, imposed by the universal joints H and I on linkages' translation along the axis z (perpendicular to the plane of the linkages), are redundant since there have been the same restrictions imposed by the joints B and A. Besides, the universal joint A has restricted the rotation of the plane of linkages around the axis x (parallel to the plane of linkages and meanwhile perpendicular to the primary axis), so that the same restrictions imposed by the joints B, H and I are redundant. Thus the universal joint B is equivalent to the pair of class III. The joints H and I are both equivalent to the pairs of class IV. Therefore for the presented mechanism, n ¼ 7, P1 ¼ 6, P2 ¼ 1, P3 ¼ 1, P4 ¼ 2, P5 ¼ 0. The degree of freedom of the mechanism can be obtained, DOF ¼ 6 7 5 6 4 1 3 1 2 2 ¼ 1. The main advantages of the proposed solar tracker are as follows:
The linkages have lower moment of inertia compared to gear transmission systems. Additionally, linkages function as both a support and a drive mechanism for solar collectors. By means of mechanical group-control, the solar tracker can accommodate as many collectors as possible to track the sun. So it provides a low-cost installation for solar collector array. Similar to the parallel mechanism, the two rotation axes of the developed tracker are independent of each other on the mechanism and coupled together in motion. This is different from the traditional trackers which usually adopt the serial mechanism characterized by the coupled mechanism and independent motion. Due to this characteristic, the structure of the developed tracker is a simplified one and meanwhile the stiffness is strengthened. The belt transmission for the rotation of the secondary axis is characterized by the simple structure and smooth motion. Furthermore, it is able to reduce the dynamic load occurring in the mechanism. A higher layout density is achieved in the filed for the tracking PV system owing to the declination-clock mounting and thus the field efficiency will somewhat increase [23]. The disadvantages involve: Somewhat kinematical imprecision due to the machining error (but no great impact on the tracking result, see Section 4). The dynamic loads occurring in the mechanism are transmitted in the primary axis actuating system, which causes certain damage to the linear actuator. The permanent deformation of the belts may arise due to the belts' long-time exposure to the weather (the synchronous steel belts will be used instead in the future researches).
3. Automatic tracking strategies
Fig. 5. Kinematic diagram of the primary axis driving mechanism. (A, B, H, I) universal joints (pairs of class II), (C, D, F, G, J, K) revolute joints (pairs of class I), (E) prismatic joint (pairs of class I).
Two automatic tracking strategies, which are suitable for different solar power generation systems, are developed in this study: one is normal tracking strategy, i.e. rotating both the primary and secondary axes to keep the tracking error smaller than a prespecified value; the other one is daily adjustment on the primary axis, i.e. keeping the primary axis fixed during the working period and making an adjustment on it at the end of the tracking. The two tracking strategies are detailed below.
92
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
3.1. Normal tracking strategy
3.2. Daily adjustment on primary axis
3.1.1. Basic idea of normal tracking If the tracking error is larger than a certain value, the performance of Concentrating Photovoltaic (CPV) systems or other CSP systems (such as parabolic trough and dish collector systems) will be significantly affected. This paper proposes a strategy of normal tracking with high tracking accuracy. This strategy keeps the tracking errors smaller than the pre-specified values. It is a hybrid strategy that combines time-based control and sensor-based control. Time-based control employs the mathematical formulas to determine the sun's position [24]. Due to the installation error and the time error, this method alone cannot guarantee high enough accuracy. The single method of sensor-based control has a good tracking performance in clear weather conditions, but it has serious problems in cloudy weather conditions [24]. The proposed hybrid strategy avoids the above problems. In this strategy, the tracker is controlled to reach the calculated position based on the local time and then the tracking error is corrected based on the feedback signal from sun position sensor.
3.2.1. Basic idea of daily adjustment strategy When it comes to flat PV systems, a little decline of accuracy does not have a great impact on power output. Thus there is no great need for high-precision tracking, since the high-precision tracking will lead to the extra operation costs. Therefore a simple and low-cost tracking strategy is developed. As was mentioned in Section 2.2, for most of the tracking duration, the declination angle changes very slowly and the clock angle varies at an almost constant speed. Based on this phenomenon a simple tracking strategy, similar to the tracking law of the polar dual-axis tracker, is proposed. The declination angle is adjusted once a day so that the solar beam is perpendicular to the solar panels at noon every day. The secondary axis rotates at a constant speed of 15 /h every day. This speed is the theoretical rotation speed derived on the spring equinox or autumn equinox. So the equations concerning the two rotation angles are as follows
3.1.2. Control method of normal tracking As is shown in Fig. 6, the microprocessor can calculate the sun's position based on the solar time. Besides, the microprocessor can also receive signals from the sun position sensor. The motors for the primary axis and the secondary axis are controlled by the microprocessor, which realizes the motion of the mechanical system. The personal computer is connected to the microprocessor to monitor the tracking accuracy in real-time. The flow chart of this tracking strategy is shown in Fig. 7. At the beginning of tracking, the program waits for the starting time of the operation. Here the starting time is programmed to be 30 min later than the sunrise time, since it is difficult for the sensor to capture the sun's position just at sunrise. The tracker aims at the sun based on the sun position sensor. Then the system calculates constantly the errors of both the declination angle and the clock angle, namely DqDE and DqCL. If either of the two errors, i.e. DqDE and DqCL, is beyond the pre-specified value of DDE or DCL, then the system will adjust the axis to correct the error. After these adjustments, the system will detect further errors of declination angle and clock angle, i.e. dqDE and dqCL, with the help of the sensor. Again, if either of the two errors is beyond the pre-specified value, then it will be corrected. The above tracking process continues until sunset. After that, the microprocessor calculates the time of sunrise and sunset for the next day. Finally the tracker returns to the initial position of the next day. In this duration, the microprocessor sends the acquired data from the sensor constantly to the computer to monitor the pointing error. It should be noted that, in cloudy weather conditions, the sensor's output will decrease below a certain level so that only the time-based control is available.
q0DE ¼ 90 anoon s q0CL ¼ u
(8)
where anoon is solar altitude angle at noon; u is the hour angle s [22]. Apparently, due to the inaccuracy of the tracking formulas shown in Eq. (8), there exists a tracking error Dq in daily adjustment strategy. We call this type of error tracking principle error. Considering that high operating time of the motor and large transmission ratios [21] in the case of continuous tracking, a stepwise rotation method is applied to the secondary axis. The secondary axis is rotated through 0.5 every step (the motion step duration Dt is less than 5 s) and the time interval between the two adjacent steps is 2 min. Then the angular displacement per hour for the secondary axis equals to 15 . Thus the secondary axis rotates at an average angular speed of 15 /h which is consistent the constant speed shown in Eq. (8). It should be noted that the error caused by the stepwise rotation method is below 0.5 , and in the case of daily adjustment strategy its effect on the tracking accuracy is much smaller than the tracking principle error. When the tracker moves under the daily adjustment strategy, the unit normal vector n0 of the solar collector can be expressed as
n0XYZ-O ¼ sin q0CL ; sin q0DE cos q0CL ; cos q0DE cos q0CL
(9)
where n0XYZ-O refers to n0 in the coordinate system of XYZ-O. The tracking principle error Dq is derived from
Fig. 6. Block diagram of the control system.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
93
cosDq ¼ nXYZO $n0XYZ-O ¼sin qCL sinu þ sin qDE cos qCL cos anoon cosu s cosu þ cos qDE cos qCL sin anoon s (10) In flat PV systems, tracking error will result in the loss of the energy received by the surface. This loss can be denoted by hcos, namely, the cosine loss. The cosine loss of the inclined surface can be calculated by
hcos ¼ 1 cosDq
(11)
Based on Eqs. (10) and (11), curves showing the tracking principle error Dq as well as the corresponding cosine loss on typical days (equinox and solstice) are drawn in Fig. 8. The maximum error occurs on the summer solstice or winter solstice (i.e. d ¼ ±23.45 ). The further the time deviates from the noon, the bigger the error becomes. The error reaches about 23 at 6:00 or 18:00 and the corresponding cosine loss is 8.2%. There is no tracking principle error on the equinox day. The annual average cosine loss hcos could be used to evaluate the tracking performance of a solar tracker. The value is calculated by
Z Ein hcos dt Z Ein dt
hcos ¼
(12)
where Ein is the direct incident energy. The scattered energy and the ground-reflected energy are not considered because they do not have a great effect on the cosine loss. In general, the solar radiation reaches the maximum at midday. So Eq. (12) can be changed into
Z hcos ¼
Z Ein hcos dt Z Ein dt
hcos dt T
(13)
where T is the total sunshine time in a year. Through calculation, the annual average cosine loss of daily adjustment strategy is only 1.3%. The result is applicable not only to the city of Harbin but also to other installation sites. 3.2.2. Control method of daily adjustment strategy The mechanical error of the tracker is generally much smaller than tracking principle error. Thus a high-cost solar position sensor [25] is not necessary. The detailed tracking process is shown in Fig. 9. When the tracking starts, the secondary axis is
Fig. 7. Flow chart of normal tracking strategy.
Fig. 8. Tracking principle error and corresponding cosine loss for daily adjustment.
94
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
3.3. Operation of the control system The control system for the solar tracking contains four main modules, as is shown in Fig. 10. The “central processing module” undertakes information processing. It is based on the ARM S3C2440X microprocessor. This high-performance microprocessor is not only energy-saving but also suitable for embedded control. In this module, the analog to digital converter (ADC) is used to collect the analog signal from the sensor. The real-time clock (RTC) is used to provide the precise time for the microprocessor. The serial port and the I/O port are used to connect this module to the modules of “personal computer” and “drive circuit” respectively. The “sensor module” is used to detect the sun's position (only used in the normal tracking strategy). This module contains a fourquadrant silicon photocell and a related processing circuit. It is installed in an opaque tube with a small circular aperture on its top. Fig. 11(a) shows the sensing device. When the solar beam is incident onto the device, a circular spot, with its size equal to the tube's aperture, will appear on the surface of the photocell (Fig. 11(b)). The current of the photocell is linear with the spot size. The spot size on the four quadrants will be varies as the sun moves, which consequently produces different currents for each quadrant. Then the signal processing circuit will amplify the weak current to signals collectable by “central processing module”. The resolution for the designed sensor is 105 and the measurement range is 1.7. Besides detecting the sun's position, the sensor is also able to judge the solar intensity by adding together the four quadrants' currents. So in this study, this sensor is also used to detect weather conditions. 4. Experiment results and discussion Both the tracking accuracy and the performance of PV systems under the two automatic tracking strategies were tested. The experimental device is installed on the roof of Zhizao Building in Harbin Institute of Technology. 4.1. Accuracy tests on two automatic tracking strategies
Fig. 9. Flow chart of daily adjustment strategy.
adjusted every 2 min. After the end of the tracking, the system calculates the time of sunrise and sunset for the next day. And then the system adjusts the primary axis and secondary axis for next day's tracking.
The tracking error of the normal tracking strategy stems mainly from the sensor detecting error and the tracking step angles prespecified in the software. The step angles, namely DDE and DCL, should not be set too small. Otherwise the tracking process will become unstable. In the test, DDE and DCL are both set as 0.1. The
Fig. 10. Construction of the control system.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
95
Fig. 11. Device of sun position sensor: (a) picture of the sun position sensor; (b) schematic of light spot on four-quadrant photocell.
Fig. 12. Tracking error of normal tracking on September 15th.
tracking error monitored on September 15th (a clear day) is shown in Fig. 12. The curve fluctuates slightly around 0.1 and the value stays below 0.15 . Besides, the general trend of the curve keeps stable during the tracking period. As the tracking error of the daily adjustment strategy may exceed the measurement range of the sun position sensor, a simple measurement device is used to record the tracking error, as is shown in Fig. 13. The minimum error that can be detected by this device is 0.2 . The tracking error can be derived from the projection length of the slender rod on the paper. The test was carried out on July 17th and September 12th (Fig. 14). For both days, the experimental error at noon is almost zero. While the maximum error of July 17th reaches 24.8 at half past eighteen, the maximum error of September 12th is only 4 . The experimental result is identical with the theoretical calculation.
maximized. PV modules used in the experiment are identical and the model of the PV modules is DT010P-12. They reach the peak power of 10 W when the solar radiation is 1000 W/m2 and the temperature is 25 . The output of PV systems is controlled by using
4.2. Performance tests on PV systems The performance of PV systems under the two tracking strategies is tested in comparison with that of a fixed solar panel (Fig. 15). The fixed solar panel faces the south. Its tilt angle is set as such a value that it is perpendicular to the solar beam at noon on testing days. In this way, the level of solar radiation it receives at noon is
Fig. 13. Error measurement device for daily adjustment tracking strategy.
96
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
Fig. 14. Tracking errors of daily adjustment strategy on July 17th and September 12th.
September 12th. The tilt angles of the fixed solar panel were 24.6 , 25.4 , 42.4 , 43.6 respectively on the four days. The output was recorded every 10 min from 8:00 to 19:00. Fig. 16(a) and (b) show the result of tests on the normal tracking strategy. The maximum power output of both the tracking and fixed PV is 12 W on July 21st and 11.4 W on September 15th. Tracking PV system has significantly higher power output than the fixed one for most times of the tracking duration. Fig. 17(a) and (b) show the result of tests on the strategy of daily adjustment. The maximum power output of both the tracking and fixed PV is about 11.6 W on July 17th and 9.7 W on September 12th. The output of the tracking PV system is also much higher than the fixed one. The effect of tracking on the PV performance can be measured by the energy efficiency h [21], which is expressed as
h¼
Fig. 15. Comparative experiment on the tracking PV module versus the fixed one.
the maximum power point tracking (MPPT) method. The experiments on the normal tracking strategy were carried out on July 21st as well as on September 15th. The experiments on the other strategy were carried out on the two days of July 17th and
½ET ðEF þ EC Þ 100% EF
(14)
where ET is the energy produced by the tracking PV system; EF is the energy produced by the fixed PV system; EC is the energy consumption by the tracking. We can get the energy production of a single PV module by multiplying the sum of the recorded data by the time interval of 10 min. Note that the solar tracker could accommodate as many as six identical PV modules. Therefore ET and EF are both 6 times of the
Fig. 16. Output of the normal tracking and fixed PV module on the days of (a) July 21st; (b) September 15th.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
97
Fig. 17. Output of the daily adjusted and fixed PV module on the days of (a) July 17th; (b) September 12th.
corresponding energy production of a single PV module (see Tables 1 and 2). The exact energy consumption EC is difficult to obtain but its upper limit can be obtained by
EC Pr DT
(15)
where Pr is the rated power of the motor and its value is 2.4 W; DT is the operating time of the motor. For normal tracking strategy, the operating time should be less than 11 h (from 8:00 to 19:00). In this way, the upper limit of the energy consumption by the normal tracking strategy is obtained, EC 2:4 11 2 ¼ 52:8 Wh/day (for the rotation of both the primary axis and secondary axis). For daily than 0.5 h which is adjustment strategy, the operating time is less obtained by multiplying the number N N ¼ 1160 ¼ 330 of 2 tracking steps by the motion step duration Dt ðDt 5sÞ. Thus the upper limit of the energy consumption by the daily adjustment strategy is also obtained, EC 2:4 0:5 ¼ 1:2 Wh/day (only for the rotation of the secondary axis). In this way, the lower limits of energy efficiencies for both tracking strategies are obtained according to Eq. (14) (Tables 1 and 2). Results show that the average energy efficiency of the normal tracking PV, on the two days, is more than 23.6% and the average energy efficiency of the daily adjusted PV is more than 31.8%. Both tracking strategies show an obvious increase in PV output compared to the fixed one. Note that the energy efficiency of the daily adjusted PV module may have the possibility to be higher than that of the normal tracking PV module. This is because the energy consumption of the daily adjusted tracking is much less than that of the normal tracking.
Table 1 Energy efficiency of the normal tracking PV module. Date
ET (Wh/day)
EF (Wh/day)
EC (Wh/day)
h (%)
July 21st September 15th Average
113.4 6 99.2 6 106.3 6
85.4 6 72.5 6 79.0 6
57.6 57.6 57.6
22.5 24.7 23.6
Table 2 Energy efficiency of the daily adjusted PV module. Date
ET (Wh/day)
EF (Wh/day)
EC (Wh/day)
h (%)
July 17th September 12th Average
108.1 6 78.1 6 93.1 6
80.0 6 60.6 6 70.3 6
1.2 1.2 1.2
34.9 28.6 31.8
5. Conclusion A dual-axis solar tracker has been presented in this paper. It employs a declination-clock mounting system. On the basis of the above work, this study has reached the following conclusions: Normal tracking strategy combines together time-based control and sensor-based control so that the tracker could work in different weather conditions. Experiment results show that the tracking error of the normal tracking strategy is below 0.15 . So it could provide high tracking accuracy for the solar power systems. Daily adjustment strategy has simplified the complicated tracking process. Additionally, the primary axis doesn't move in the tracking period so that the energy consumption by the tracking is reduced. Thus this strategy provides a low-cost operation for flat PV systems, since it avoids the complicated control systems as well as the energy consumption by the rotation of the primary axis. Furthermore, the annual average cosine loss of the daily adjustment strategy is calculated as below 1.3%, a little loss on energy output but a sharp reduction on the cost. Therefore daily adjustment is a tracking strategy suitable for flat PV systems. The performance results of PV systems show that both tracking strategies based on the developed tracker are feasible and effective for solar power systems. Under different strategies, the tracker can be applied appropriately to flat PV systems and CSP systems. The presented tracker will be further researched for its future development. With its group-control capability as well as its good mechanical performance, the tracker, after being improved in the future, can be considered to be applied to the heliostat field for Solar Power Tower (SPT) systems. References [1] Tang RS, Wu T. Optimal tilt-angles for solar collectors used in China. Appl Energy 2004;79(3):239e48. [2] Chang TP. The sun's apparent position and the optimal tilt angle of a solar collector in the northern hemisphere. Sol Energy 2009;83(8):1274e84. [3] Kalogirou SA. Design and construction of a one-axis sun-tracking system. Sol Energy 1996;57(6):465e9. [4] Abouzeid M. Use of a reluctance stepper motor for solar tracking based on a programmable logic array (PLA) controller. Renew Energy 2001;23(3e4): 551e60. [5] Abu-Khader MM, Badran OO, Abdallah S. Evaluating multi-axes sun-tracking system at different modes of operation in Jordan. Renew Sustain Energy Rev 2008;12(3):864e73. [6] Neville RC. Solar energy collectors orientation and sun tracking mode. Sol Energy 1978;20(1):7e11.
98
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
[7] Sungur C. Multi-axes sun-tracking system with PLC control for photovoltaic panels in Turkey. Renew Energy 2009;34(4):1119e25. [8] Arbab H, Jazi B, Rezagholizadeh M. A computer tracking system of solar dish with two-axis degree freedoms based on picture processing of bar shadow. Renew Energy 2009;34(4):1114e8. [9] Song J, Yang Y, Zhu Y, Zhou J. A high precision tracking system based on a hybrid strategy designed for concentrated sunlight transmission via fibers. Renew Energy 2013;57:12e9. [10] Clifford MJ, Eastwood D. Design of a novel passive solar tracker. Sol Energy 2004;77(3):269e80. [11] Poulek V, Libra M. A very simple solar tracker for space and terrestrial applications. Sol Energy Mater Sol Cells 2000;60(2):99e103. [12] Kim Y, Han GY, Seo T. An evaluation on thermal performance of CPC solar collector. Int Commun Heat Mass Transf 2008;35(4):446e57. [13] Roth P, Georgiev A, Boudinov H. Design and construction of a system for suntracking. Renew Energy 2004;29(3):393e402. [14] Batayneh W, Owais A, Nairoukh M. An intelligent fuzzy based tracking controller for a dual-axis solar PV system. Automation Constr 2013;29:100e6. [15] Mavromatakis F, Franghiadakis Y. A highly efficient novel azimuthal heliotrope. Sol Energy 2008;82(4):336e42. [16] Chin CS, Babu A, McBride W. Design, modeling and testing of a standalone single axis active solar tracker using MATLAB/Simulink. Renew Energy 2011;36(11):3075e90.
[17] Chang TP. Output energy of a photovoltaic module mounted on a single-axis tracking system. Appl Energy 2009;86(10):2071e8. [18] Kacira M, Simsek M, Babur Y, Demirkol S. Determining optimum tilt angles and orientations of photovoltaic panels in Sanliurfa, Turkey. Renew Energy 2004;29(8):1265e75. [19] Abdallah K, Nijmeh S. Two axes sun tracking system with PLC control. Energy Convers Manag 2004;45(11e12):1931e9. [20] Bakos GC. Design and construction of a two-axis sun tracking system for parabolic trough collector (PTC) efficiency improvement. Renew Energy 2006;31(15):2411e21. [21] Alexandru CA. Novel open-loop tracking strategy for photovoltaic systems. Sci World J 2013;2013:1e12. [22] Alata M, Al-Nimr MA, Qaroush Y. Developing a multipurpose sun tracking system using fuzzy control. Energy Convers Manag 2005;46(7e8): 1229e45. [23] Schramek P, Mills DR. Heliostats for maximum ground coverage. Energy 2004;29(5e6):701e13. [24] Mousazadeh H, Keyhani A, Javadi A, Mobli H, Abrinia K, Sharifi A. A review of principle and sun-tracking methods for maximizing solar systems output. Renew Sustain Energy Rev 2009;13(8):1800e18. [25] Agee JT, Jimoh AA. Flat controller design for hardware-cost reduction in polaraxis photovoltaic systems. Sol Energy 2012;86(1):452e62.
Contents lists available at ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
A multipurpose dual-axis solar tracker with two tracking strategies Yingxue Yao a, Yeguang Hu b, *, Shengdong Gao b, Gang Yang b, Jinguang Du b a
Shenzhen Graduate School, Harbin Institute of Technology, HIT Campus, University Town of Shenzhen, Xili, Shenzhen 518055, Guangdong Province, PR China b School of Mechanical and Electrical Engineering, Harbin Institute of Technology, 92 West Dazhi Street, Harbin 150001, Heilongjiang Province, PR China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 January 2014 Accepted 1 July 2014 Available online
This paper deals with a multipurpose dual-axis solar tracker that can be applied to solar power systems. This tracker employs a declination-clock mounting system that locates the primary axis in east-west direction. Based on this mounting system, normal tracking strategy and daily adjustment strategy are developed for flat Photovoltaic (PV) systems and Concentrating Solar Power (CSP) systems respectively. While the former strategy keeps the tracking errors smaller than the pre-specified values, the latter one simplifies the tracking process by adjusting the primary axis once a day and driving the secondary axis to rotate at a constant speed of 15 /h. Results of the accuracy test indicate that the tracking error of the normal tracking strategy is within 0.15 . The other strategy may have greater tracking errors, but its annual average cosine loss for flat PV systems is estimated to be below 1.3%. Furthermore, in the test on the output of the PV modules, it is found that the average energy efficiency of the normal tracking PV, compared with the fixed PV, is more than 23.6%. And the average energy efficiency of the daily adjusted PV is more than 31.8%. Results of the experiment show that the two tracking strategies are both feasible for the developed tracker. © 2014 Elsevier Ltd. All rights reserved.
Keywords: Solar energy Solar power system Solar tracker Automatic tracking strategy Cosine loss
1. Introduction As a kind of clean and renewable energy source, solar energy has been drawing more and more attention, especially in the field of electricity generation, due to the shortage and pollution of fossil fuels. The process of converting solar energy to electric energy is realized mainly through flat PV systems or CSP systems. The power output that these systems could produce depends on various factors, including the amount of the energy they receive from the solar radiation. Some researchers have studied the optimal angle of solar collector to increase the power output [1,2]. As the sun's position changes throughout the day, the solar tracker is a more efficient method of increasing the energy production. So the solar tracker is being studied by more and more researchers. Currently, there are mainly two types of solar trackers based on movement capability: single-axis tracker [3e5] and dual-axis tracker [6e9]. Different single-axis and dual-axis trackers have been presented by the previous studies. Clifford and Eastwood [10] presented a passive solar tracker activated by aluminium/steel bimetallic strips and controlled by a viscous damper. Poulek and
* Corresponding author. Tel.: þ86 451 86413557. E-mail address: [email protected] (Y. Hu). http://dx.doi.org/10.1016/j.renene.2014.07.002 0960-1481/© 2014 Elsevier Ltd. All rights reserved.
Libra [11] designed a simple single-axis solar tracker based on a new arrangement of auxiliary bifacial solar cell connected directly to DC motor. Kim et al. [12] proposed a single polar axis tracker with the solar collector rotating at a speed of 15 /h round the polar axis. Roth et al. [13] proposed an altitude-azimuth dual-axis tracker which, guided by a closed loop serve system, could operate automatically. Batayneh et al. [14] proposed a dual-axis sun tracking system with altitude-azimuth mounting controlled by the designed fuzzy controller. Mavromatakis and Franghiadakis [15] presented a novel single-axis azimuthal tracker with the ability to move the collector's plane in two directions through a special support structure. According to the previous studies, solar trackers have been used in different solar collector systems. C.S. Chin et al. [16] presented an active single-axis solar tracker used in flat PV systems. The experimental test showed that the efficiency over the fixed solar panel was around 20%. Chang [17] tested the flat PV system mounted on a single-axis tracker and found that the gain of the single-axis tracking panel installed with the yearly optimal angle was 17.5%, compared to a traditional fixed panel. Kim et al. [12] applied the single polar axis tracker to CPC solar collector. The study of Kacira et al. [18] showed a daily average of 34.6% gain in generated power with two-axis solar tracking compared to a fixed PV panel on a particular day in July in Sanliurfa, Turkey. Abdallah and Nijmeh [19] designed a two axes sun tracking system for PV
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
system. The system surface showed a better performance with an increase in the collected energy of up to 41.34% compared with the fixed surface. When analyzing the current literature, the authors found no available trackers that are developed properly for both flat PV systems and CSP systems. Flat PV systems and CSP systems have different demands for tracking accuracy. On the one hand, current low accuracy trackers, such as the single-axis tracker designed for flat PV systems, cannot be used in CSP systems because their low accuracy will lead to a great loss of the solar energy intercepted by the receiver. On the other hand, if a tracker, designed for highaccuracy tracking purposes, is used in a flat PV system, there will be extra operational costs imposed by control process. This paper takes into account the two types of demands for accuracy and develops a dual-axis tracker employing declination-clock mounting system (This mounting system has been used by Bakos [20] for the efficiency improvement of parabolic trough collector (PTC) in practice. However no available analysis is made for the principle of this mounting system in the literature. The main new contributions of this paper compared to [20] are summarized as follows: The tracking formulas for the declination-clock mounting system have been derived and also a much simpler tracking strategy has been proposed for the flat PV system; A linkage mechanism, which provides with a simple and light-weight structure, has been designed for the motion of the primary axis; The belt transmission is employed for the motion of the secondary axis, which can reduce the dynamic load occurring in the mechanism; Mechanical groupcontrol, which allow the solar tracker to accommodate more collectors to track the sun, is achieved in the proposed structure; A sun position sensor with high resolution is designed and employed in the presented solar tracker.). This mounting system would allow the presented tracker in this paper to move under two tracking strategies, namely normal tracking strategy and daily adjustment strategy. Normal tracking strategy shows a good tracking performance with the error remaining below 0.15 . Daily adjustment strategy is used for simple tracking purpose instead of highaccuracy tracking. Its annual average cosine loss for flat PV systems is calculated to be below 1.3%. 2. Description of the solar tracker 2.1. Analysis of declination-clock mounting system The developed tracker employs a mounting system named as declination-clock mounting which has fine motion ability. The
89
declination-clock mounting system has two rotation axes (Fig. 1(b)): primary axis, located in east-west direction and secondary axis, perpendicular to the primary axis and able to rotate around it. Rotation angles for the primary and secondary axis are defined as declination angle and clock angle respectively. The declination-clock system is obtained from the pseudo-azimuthal system [21] by inverting the order of the rotation axes e regarding the latter system, the primary axis is for the azimuthal motion (which, in terms of the former system, generates a corresponding clock angle), while the secondary axis is for elevation (which, in terms of the former system, generates a corresponding declination angle). The tracking schematics for the developed tracker and the pseudo-azimuthal tracker are shown in Fig. 1(a) and (b) for comparison. The rotation angles of the tracker are determined by the sun's position. So the unit normal vector of the solar collector is given by
n ¼ vs
(1)
where n is the unit normal vector of the solar collector; vs is the unit vector incident to the sun. In general, the vector vs can be determined by the solar azimuth angle gs and altitude angle as [22]. Two coordinate systems, namely XYZ-O and IJK-O, are established for the analysis of declination-clock mounting system, (Fig. 2). XYZ-O is located on the ground and IJK-O is on the declination-clock tracker. The vector vs can be expressed in the coordinate system of XYZ-O as
vs ¼ ðsin gs cos as ; cos gs cos as ; sin as Þ
(2)
According to the relations in the figure, vector n can be expressed in the coordinate system of IJK-O as
nIJKO ¼ ðsin qCL ; cos qCL ; 0Þ
(3)
Then through the rotation transformation from the system IJK-O to the system XYZ-O, the vector n can be expressed in the coordinate system of XYZ-O as
nXYZO ¼ ðsin qCL ; sin qDE cos qCL ; cos qDE cos qCL Þ
(4)
So the formulas of the angles qDE and qCL can be expressed as
tan qDE ¼ cos gs cot as sin qCL ¼ sin gs cos as
(5)
Similarly, through the rotation transformation, the angles qAZ and qEL for pseudo-azimuthal system can be derived from
Fig. 1. Comparison of tracking schematics of: (a) pseudo-azimuthal mounting; (b) declination-clock mounting.
90
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
2.2. Design of the solar tracker
Fig. 2. Schematics of vector n in the declination-clock mounting system.
tan qAZ ¼ sin gs cot as sin qEL ¼ cos gs cos as
(6)
The location is set in Harbin city, where the latitude is 45.75 north. Based on Eqs. (5) and (6), the theoretical curves showing the two rotation angles for both the pseudo-azimuthal mounting and declination-clock mounting are plotted in Fig. 3(a) and (b). It should be noted that the black curves in Fig. 3 indicate the rotation angles of the primary axes, and the red (in web version) curves indicate the rotation angles of the secondary axes. So here the declination angle and the azimuth angle are brought into comparison. The rotation speed of the primary axis for declination-clock mounting is very small in most times of the day (only a rise or decline in the morning or evening), especially when it comes close to the spring equinox or autumn equinox. For the pseudo-azimuthal mounting, by contrast, the primary axis rotates at a considerable speed during the tracking period. Besides, the rotation speed of the secondary axis for the declination-clock mounting system is almost constant, whereas for the pseudo-azimuthal system, the rotation speed of the secondary axis undergoes a considerable change. Apparently, the declinationclock mounting system shows a steadier performance in mechanical motion. Furthermore, based on the features of this mounting system, this paper proposes a very simple tracking strategy, which is detailed in Section 3.
For the designed solar tracker (Fig. 4(a)), the solar collectors are installed on the support frames (13). The linkages, consisting of a linear actuator (3, 4) and four connecting rods (1, 2, 5, 6) connecting with the two ends of the actuator symmetrically, are used to drive the solar collector array to rotate around the primary axis. When DC motor drives the linear actuator to move, the rotation support (7), which holds the solar collector array, will rotate around the primary axis. The rotation of the secondary axis is implemented by the linear actuator (9, 12) installed on the rotation support. The linear actuator connects with the belt (10). When the linear actuator pulls the belt to reciprocate, the two related pulleys (8, 11) will drive support frames (13) to rotate around the secondary axis. Every frame is installed with a pulley and all of the pulleys are connected to the belt in sequence. The distance between the axes of two adjacent pulleys is 605 mm. Thus through efficient belt transmission, the solar collectors will rotate around the secondary axis simultaneously, and thus mechanical group-control for the solar collector array is realized. A corresponding physical prototype has been produced for tracking experiment, which will be detailed in Section 4. If, however, economy issues are taken into consideration, the above-mentioned structure can be further simplified: the four connecting rods can be omitted and the linear actuator can be directly connected to the base and to the rotation support (Fig. 4(b)). Although this simplified structure is not as robust as the original one which provides more connecting joints for the base and the rotation support, it can be applied under conditions of low load. In this simplified structure, the connections between the rotation support and the base are moved to the short lateral frames of the rotation support in order to eliminate the imbalance of this mechanical structure for the declination motion. In the same idea, the driving mechanism for the motion of the secondary axis is mounted on the other side of the rotation support. The physical prototype for this model is not put into production due to the same tracking function with the original one. The mechanisms, which drive the solar collectors to rotate around the primary axis, constitute a spatial kinematic chain (see Fig. 5). The number of degrees of freedom (DOF) for this mechanism can be calculated by
Fig. 3. Comparison of rotation angles for: (a) pseudo-azimuthal mounting; (b) declination-clock mounting.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
91
Fig. 4. Structure of the solar tracker: (a) the model with linkage mechanism; (b) the model without the linkage mechanism. (1, 2, 5, 6) connecting rods, (3) the linear actuator (piston head) for declination motion, (4) the linear actuator (piston cylinder) for declination motion, (7) rotation support holding solar collector array, (8, 11) pulleys, (9) linear actuator (piston head) for secondary motion, (10) belt, (12) linear actuator (piston cylinder) for secondary motion, (13) support frame for solar collectors, (14) solar panel, (15) secondary axis, (16) primary axis.
DOF ¼ 6n
X
r ¼ 6n ð5P1 þ 4P2 þ 3P3 þ 2P4 þ P5 Þ
(7)
P where n is the number of moving links in the mechanism; r is the number of restrictions; P1 ~ P5 are respectively the numbers of pairs of class I ~ V. Apparently, the redundant constraint exists in the presented mechanism since the axis of revolute joint K coincides with that of joint J. Thus the revolute joint K should be taken away for the calculation of DOF. Due to the symmetry of the linkages, the restrictions, imposed by the universal joints H and I on linkages' translation along the axis z (perpendicular to the plane of the linkages), are redundant since there have been the same restrictions imposed by the joints B and A. Besides, the universal joint A has restricted the rotation of the plane of linkages around the axis x (parallel to the plane of linkages and meanwhile perpendicular to the primary axis), so that the same restrictions imposed by the joints B, H and I are redundant. Thus the universal joint B is equivalent to the pair of class III. The joints H and I are both equivalent to the pairs of class IV. Therefore for the presented mechanism, n ¼ 7, P1 ¼ 6, P2 ¼ 1, P3 ¼ 1, P4 ¼ 2, P5 ¼ 0. The degree of freedom of the mechanism can be obtained, DOF ¼ 6 7 5 6 4 1 3 1 2 2 ¼ 1. The main advantages of the proposed solar tracker are as follows:
The linkages have lower moment of inertia compared to gear transmission systems. Additionally, linkages function as both a support and a drive mechanism for solar collectors. By means of mechanical group-control, the solar tracker can accommodate as many collectors as possible to track the sun. So it provides a low-cost installation for solar collector array. Similar to the parallel mechanism, the two rotation axes of the developed tracker are independent of each other on the mechanism and coupled together in motion. This is different from the traditional trackers which usually adopt the serial mechanism characterized by the coupled mechanism and independent motion. Due to this characteristic, the structure of the developed tracker is a simplified one and meanwhile the stiffness is strengthened. The belt transmission for the rotation of the secondary axis is characterized by the simple structure and smooth motion. Furthermore, it is able to reduce the dynamic load occurring in the mechanism. A higher layout density is achieved in the filed for the tracking PV system owing to the declination-clock mounting and thus the field efficiency will somewhat increase [23]. The disadvantages involve: Somewhat kinematical imprecision due to the machining error (but no great impact on the tracking result, see Section 4). The dynamic loads occurring in the mechanism are transmitted in the primary axis actuating system, which causes certain damage to the linear actuator. The permanent deformation of the belts may arise due to the belts' long-time exposure to the weather (the synchronous steel belts will be used instead in the future researches).
3. Automatic tracking strategies
Fig. 5. Kinematic diagram of the primary axis driving mechanism. (A, B, H, I) universal joints (pairs of class II), (C, D, F, G, J, K) revolute joints (pairs of class I), (E) prismatic joint (pairs of class I).
Two automatic tracking strategies, which are suitable for different solar power generation systems, are developed in this study: one is normal tracking strategy, i.e. rotating both the primary and secondary axes to keep the tracking error smaller than a prespecified value; the other one is daily adjustment on the primary axis, i.e. keeping the primary axis fixed during the working period and making an adjustment on it at the end of the tracking. The two tracking strategies are detailed below.
92
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
3.1. Normal tracking strategy
3.2. Daily adjustment on primary axis
3.1.1. Basic idea of normal tracking If the tracking error is larger than a certain value, the performance of Concentrating Photovoltaic (CPV) systems or other CSP systems (such as parabolic trough and dish collector systems) will be significantly affected. This paper proposes a strategy of normal tracking with high tracking accuracy. This strategy keeps the tracking errors smaller than the pre-specified values. It is a hybrid strategy that combines time-based control and sensor-based control. Time-based control employs the mathematical formulas to determine the sun's position [24]. Due to the installation error and the time error, this method alone cannot guarantee high enough accuracy. The single method of sensor-based control has a good tracking performance in clear weather conditions, but it has serious problems in cloudy weather conditions [24]. The proposed hybrid strategy avoids the above problems. In this strategy, the tracker is controlled to reach the calculated position based on the local time and then the tracking error is corrected based on the feedback signal from sun position sensor.
3.2.1. Basic idea of daily adjustment strategy When it comes to flat PV systems, a little decline of accuracy does not have a great impact on power output. Thus there is no great need for high-precision tracking, since the high-precision tracking will lead to the extra operation costs. Therefore a simple and low-cost tracking strategy is developed. As was mentioned in Section 2.2, for most of the tracking duration, the declination angle changes very slowly and the clock angle varies at an almost constant speed. Based on this phenomenon a simple tracking strategy, similar to the tracking law of the polar dual-axis tracker, is proposed. The declination angle is adjusted once a day so that the solar beam is perpendicular to the solar panels at noon every day. The secondary axis rotates at a constant speed of 15 /h every day. This speed is the theoretical rotation speed derived on the spring equinox or autumn equinox. So the equations concerning the two rotation angles are as follows
3.1.2. Control method of normal tracking As is shown in Fig. 6, the microprocessor can calculate the sun's position based on the solar time. Besides, the microprocessor can also receive signals from the sun position sensor. The motors for the primary axis and the secondary axis are controlled by the microprocessor, which realizes the motion of the mechanical system. The personal computer is connected to the microprocessor to monitor the tracking accuracy in real-time. The flow chart of this tracking strategy is shown in Fig. 7. At the beginning of tracking, the program waits for the starting time of the operation. Here the starting time is programmed to be 30 min later than the sunrise time, since it is difficult for the sensor to capture the sun's position just at sunrise. The tracker aims at the sun based on the sun position sensor. Then the system calculates constantly the errors of both the declination angle and the clock angle, namely DqDE and DqCL. If either of the two errors, i.e. DqDE and DqCL, is beyond the pre-specified value of DDE or DCL, then the system will adjust the axis to correct the error. After these adjustments, the system will detect further errors of declination angle and clock angle, i.e. dqDE and dqCL, with the help of the sensor. Again, if either of the two errors is beyond the pre-specified value, then it will be corrected. The above tracking process continues until sunset. After that, the microprocessor calculates the time of sunrise and sunset for the next day. Finally the tracker returns to the initial position of the next day. In this duration, the microprocessor sends the acquired data from the sensor constantly to the computer to monitor the pointing error. It should be noted that, in cloudy weather conditions, the sensor's output will decrease below a certain level so that only the time-based control is available.
q0DE ¼ 90 anoon s q0CL ¼ u
(8)
where anoon is solar altitude angle at noon; u is the hour angle s [22]. Apparently, due to the inaccuracy of the tracking formulas shown in Eq. (8), there exists a tracking error Dq in daily adjustment strategy. We call this type of error tracking principle error. Considering that high operating time of the motor and large transmission ratios [21] in the case of continuous tracking, a stepwise rotation method is applied to the secondary axis. The secondary axis is rotated through 0.5 every step (the motion step duration Dt is less than 5 s) and the time interval between the two adjacent steps is 2 min. Then the angular displacement per hour for the secondary axis equals to 15 . Thus the secondary axis rotates at an average angular speed of 15 /h which is consistent the constant speed shown in Eq. (8). It should be noted that the error caused by the stepwise rotation method is below 0.5 , and in the case of daily adjustment strategy its effect on the tracking accuracy is much smaller than the tracking principle error. When the tracker moves under the daily adjustment strategy, the unit normal vector n0 of the solar collector can be expressed as
n0XYZ-O ¼ sin q0CL ; sin q0DE cos q0CL ; cos q0DE cos q0CL
(9)
where n0XYZ-O refers to n0 in the coordinate system of XYZ-O. The tracking principle error Dq is derived from
Fig. 6. Block diagram of the control system.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
93
cosDq ¼ nXYZO $n0XYZ-O ¼sin qCL sinu þ sin qDE cos qCL cos anoon cosu s cosu þ cos qDE cos qCL sin anoon s (10) In flat PV systems, tracking error will result in the loss of the energy received by the surface. This loss can be denoted by hcos, namely, the cosine loss. The cosine loss of the inclined surface can be calculated by
hcos ¼ 1 cosDq
(11)
Based on Eqs. (10) and (11), curves showing the tracking principle error Dq as well as the corresponding cosine loss on typical days (equinox and solstice) are drawn in Fig. 8. The maximum error occurs on the summer solstice or winter solstice (i.e. d ¼ ±23.45 ). The further the time deviates from the noon, the bigger the error becomes. The error reaches about 23 at 6:00 or 18:00 and the corresponding cosine loss is 8.2%. There is no tracking principle error on the equinox day. The annual average cosine loss hcos could be used to evaluate the tracking performance of a solar tracker. The value is calculated by
Z Ein hcos dt Z Ein dt
hcos ¼
(12)
where Ein is the direct incident energy. The scattered energy and the ground-reflected energy are not considered because they do not have a great effect on the cosine loss. In general, the solar radiation reaches the maximum at midday. So Eq. (12) can be changed into
Z hcos ¼
Z Ein hcos dt Z Ein dt
hcos dt T
(13)
where T is the total sunshine time in a year. Through calculation, the annual average cosine loss of daily adjustment strategy is only 1.3%. The result is applicable not only to the city of Harbin but also to other installation sites. 3.2.2. Control method of daily adjustment strategy The mechanical error of the tracker is generally much smaller than tracking principle error. Thus a high-cost solar position sensor [25] is not necessary. The detailed tracking process is shown in Fig. 9. When the tracking starts, the secondary axis is
Fig. 7. Flow chart of normal tracking strategy.
Fig. 8. Tracking principle error and corresponding cosine loss for daily adjustment.
94
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
3.3. Operation of the control system The control system for the solar tracking contains four main modules, as is shown in Fig. 10. The “central processing module” undertakes information processing. It is based on the ARM S3C2440X microprocessor. This high-performance microprocessor is not only energy-saving but also suitable for embedded control. In this module, the analog to digital converter (ADC) is used to collect the analog signal from the sensor. The real-time clock (RTC) is used to provide the precise time for the microprocessor. The serial port and the I/O port are used to connect this module to the modules of “personal computer” and “drive circuit” respectively. The “sensor module” is used to detect the sun's position (only used in the normal tracking strategy). This module contains a fourquadrant silicon photocell and a related processing circuit. It is installed in an opaque tube with a small circular aperture on its top. Fig. 11(a) shows the sensing device. When the solar beam is incident onto the device, a circular spot, with its size equal to the tube's aperture, will appear on the surface of the photocell (Fig. 11(b)). The current of the photocell is linear with the spot size. The spot size on the four quadrants will be varies as the sun moves, which consequently produces different currents for each quadrant. Then the signal processing circuit will amplify the weak current to signals collectable by “central processing module”. The resolution for the designed sensor is 105 and the measurement range is 1.7. Besides detecting the sun's position, the sensor is also able to judge the solar intensity by adding together the four quadrants' currents. So in this study, this sensor is also used to detect weather conditions. 4. Experiment results and discussion Both the tracking accuracy and the performance of PV systems under the two automatic tracking strategies were tested. The experimental device is installed on the roof of Zhizao Building in Harbin Institute of Technology. 4.1. Accuracy tests on two automatic tracking strategies
Fig. 9. Flow chart of daily adjustment strategy.
adjusted every 2 min. After the end of the tracking, the system calculates the time of sunrise and sunset for the next day. And then the system adjusts the primary axis and secondary axis for next day's tracking.
The tracking error of the normal tracking strategy stems mainly from the sensor detecting error and the tracking step angles prespecified in the software. The step angles, namely DDE and DCL, should not be set too small. Otherwise the tracking process will become unstable. In the test, DDE and DCL are both set as 0.1. The
Fig. 10. Construction of the control system.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
95
Fig. 11. Device of sun position sensor: (a) picture of the sun position sensor; (b) schematic of light spot on four-quadrant photocell.
Fig. 12. Tracking error of normal tracking on September 15th.
tracking error monitored on September 15th (a clear day) is shown in Fig. 12. The curve fluctuates slightly around 0.1 and the value stays below 0.15 . Besides, the general trend of the curve keeps stable during the tracking period. As the tracking error of the daily adjustment strategy may exceed the measurement range of the sun position sensor, a simple measurement device is used to record the tracking error, as is shown in Fig. 13. The minimum error that can be detected by this device is 0.2 . The tracking error can be derived from the projection length of the slender rod on the paper. The test was carried out on July 17th and September 12th (Fig. 14). For both days, the experimental error at noon is almost zero. While the maximum error of July 17th reaches 24.8 at half past eighteen, the maximum error of September 12th is only 4 . The experimental result is identical with the theoretical calculation.
maximized. PV modules used in the experiment are identical and the model of the PV modules is DT010P-12. They reach the peak power of 10 W when the solar radiation is 1000 W/m2 and the temperature is 25 . The output of PV systems is controlled by using
4.2. Performance tests on PV systems The performance of PV systems under the two tracking strategies is tested in comparison with that of a fixed solar panel (Fig. 15). The fixed solar panel faces the south. Its tilt angle is set as such a value that it is perpendicular to the solar beam at noon on testing days. In this way, the level of solar radiation it receives at noon is
Fig. 13. Error measurement device for daily adjustment tracking strategy.
96
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
Fig. 14. Tracking errors of daily adjustment strategy on July 17th and September 12th.
September 12th. The tilt angles of the fixed solar panel were 24.6 , 25.4 , 42.4 , 43.6 respectively on the four days. The output was recorded every 10 min from 8:00 to 19:00. Fig. 16(a) and (b) show the result of tests on the normal tracking strategy. The maximum power output of both the tracking and fixed PV is 12 W on July 21st and 11.4 W on September 15th. Tracking PV system has significantly higher power output than the fixed one for most times of the tracking duration. Fig. 17(a) and (b) show the result of tests on the strategy of daily adjustment. The maximum power output of both the tracking and fixed PV is about 11.6 W on July 17th and 9.7 W on September 12th. The output of the tracking PV system is also much higher than the fixed one. The effect of tracking on the PV performance can be measured by the energy efficiency h [21], which is expressed as
h¼
Fig. 15. Comparative experiment on the tracking PV module versus the fixed one.
the maximum power point tracking (MPPT) method. The experiments on the normal tracking strategy were carried out on July 21st as well as on September 15th. The experiments on the other strategy were carried out on the two days of July 17th and
½ET ðEF þ EC Þ 100% EF
(14)
where ET is the energy produced by the tracking PV system; EF is the energy produced by the fixed PV system; EC is the energy consumption by the tracking. We can get the energy production of a single PV module by multiplying the sum of the recorded data by the time interval of 10 min. Note that the solar tracker could accommodate as many as six identical PV modules. Therefore ET and EF are both 6 times of the
Fig. 16. Output of the normal tracking and fixed PV module on the days of (a) July 21st; (b) September 15th.
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
97
Fig. 17. Output of the daily adjusted and fixed PV module on the days of (a) July 17th; (b) September 12th.
corresponding energy production of a single PV module (see Tables 1 and 2). The exact energy consumption EC is difficult to obtain but its upper limit can be obtained by
EC Pr DT
(15)
where Pr is the rated power of the motor and its value is 2.4 W; DT is the operating time of the motor. For normal tracking strategy, the operating time should be less than 11 h (from 8:00 to 19:00). In this way, the upper limit of the energy consumption by the normal tracking strategy is obtained, EC 2:4 11 2 ¼ 52:8 Wh/day (for the rotation of both the primary axis and secondary axis). For daily than 0.5 h which is adjustment strategy, the operating time is less obtained by multiplying the number N N ¼ 1160 ¼ 330 of 2 tracking steps by the motion step duration Dt ðDt 5sÞ. Thus the upper limit of the energy consumption by the daily adjustment strategy is also obtained, EC 2:4 0:5 ¼ 1:2 Wh/day (only for the rotation of the secondary axis). In this way, the lower limits of energy efficiencies for both tracking strategies are obtained according to Eq. (14) (Tables 1 and 2). Results show that the average energy efficiency of the normal tracking PV, on the two days, is more than 23.6% and the average energy efficiency of the daily adjusted PV is more than 31.8%. Both tracking strategies show an obvious increase in PV output compared to the fixed one. Note that the energy efficiency of the daily adjusted PV module may have the possibility to be higher than that of the normal tracking PV module. This is because the energy consumption of the daily adjusted tracking is much less than that of the normal tracking.
Table 1 Energy efficiency of the normal tracking PV module. Date
ET (Wh/day)
EF (Wh/day)
EC (Wh/day)
h (%)
July 21st September 15th Average
113.4 6 99.2 6 106.3 6
85.4 6 72.5 6 79.0 6
57.6 57.6 57.6
22.5 24.7 23.6
Table 2 Energy efficiency of the daily adjusted PV module. Date
ET (Wh/day)
EF (Wh/day)
EC (Wh/day)
h (%)
July 17th September 12th Average
108.1 6 78.1 6 93.1 6
80.0 6 60.6 6 70.3 6
1.2 1.2 1.2
34.9 28.6 31.8
5. Conclusion A dual-axis solar tracker has been presented in this paper. It employs a declination-clock mounting system. On the basis of the above work, this study has reached the following conclusions: Normal tracking strategy combines together time-based control and sensor-based control so that the tracker could work in different weather conditions. Experiment results show that the tracking error of the normal tracking strategy is below 0.15 . So it could provide high tracking accuracy for the solar power systems. Daily adjustment strategy has simplified the complicated tracking process. Additionally, the primary axis doesn't move in the tracking period so that the energy consumption by the tracking is reduced. Thus this strategy provides a low-cost operation for flat PV systems, since it avoids the complicated control systems as well as the energy consumption by the rotation of the primary axis. Furthermore, the annual average cosine loss of the daily adjustment strategy is calculated as below 1.3%, a little loss on energy output but a sharp reduction on the cost. Therefore daily adjustment is a tracking strategy suitable for flat PV systems. The performance results of PV systems show that both tracking strategies based on the developed tracker are feasible and effective for solar power systems. Under different strategies, the tracker can be applied appropriately to flat PV systems and CSP systems. The presented tracker will be further researched for its future development. With its group-control capability as well as its good mechanical performance, the tracker, after being improved in the future, can be considered to be applied to the heliostat field for Solar Power Tower (SPT) systems. References [1] Tang RS, Wu T. Optimal tilt-angles for solar collectors used in China. Appl Energy 2004;79(3):239e48. [2] Chang TP. The sun's apparent position and the optimal tilt angle of a solar collector in the northern hemisphere. Sol Energy 2009;83(8):1274e84. [3] Kalogirou SA. Design and construction of a one-axis sun-tracking system. Sol Energy 1996;57(6):465e9. [4] Abouzeid M. Use of a reluctance stepper motor for solar tracking based on a programmable logic array (PLA) controller. Renew Energy 2001;23(3e4): 551e60. [5] Abu-Khader MM, Badran OO, Abdallah S. Evaluating multi-axes sun-tracking system at different modes of operation in Jordan. Renew Sustain Energy Rev 2008;12(3):864e73. [6] Neville RC. Solar energy collectors orientation and sun tracking mode. Sol Energy 1978;20(1):7e11.
98
Y. Yao et al. / Renewable Energy 72 (2014) 88e98
[7] Sungur C. Multi-axes sun-tracking system with PLC control for photovoltaic panels in Turkey. Renew Energy 2009;34(4):1119e25. [8] Arbab H, Jazi B, Rezagholizadeh M. A computer tracking system of solar dish with two-axis degree freedoms based on picture processing of bar shadow. Renew Energy 2009;34(4):1114e8. [9] Song J, Yang Y, Zhu Y, Zhou J. A high precision tracking system based on a hybrid strategy designed for concentrated sunlight transmission via fibers. Renew Energy 2013;57:12e9. [10] Clifford MJ, Eastwood D. Design of a novel passive solar tracker. Sol Energy 2004;77(3):269e80. [11] Poulek V, Libra M. A very simple solar tracker for space and terrestrial applications. Sol Energy Mater Sol Cells 2000;60(2):99e103. [12] Kim Y, Han GY, Seo T. An evaluation on thermal performance of CPC solar collector. Int Commun Heat Mass Transf 2008;35(4):446e57. [13] Roth P, Georgiev A, Boudinov H. Design and construction of a system for suntracking. Renew Energy 2004;29(3):393e402. [14] Batayneh W, Owais A, Nairoukh M. An intelligent fuzzy based tracking controller for a dual-axis solar PV system. Automation Constr 2013;29:100e6. [15] Mavromatakis F, Franghiadakis Y. A highly efficient novel azimuthal heliotrope. Sol Energy 2008;82(4):336e42. [16] Chin CS, Babu A, McBride W. Design, modeling and testing of a standalone single axis active solar tracker using MATLAB/Simulink. Renew Energy 2011;36(11):3075e90.
[17] Chang TP. Output energy of a photovoltaic module mounted on a single-axis tracking system. Appl Energy 2009;86(10):2071e8. [18] Kacira M, Simsek M, Babur Y, Demirkol S. Determining optimum tilt angles and orientations of photovoltaic panels in Sanliurfa, Turkey. Renew Energy 2004;29(8):1265e75. [19] Abdallah K, Nijmeh S. Two axes sun tracking system with PLC control. Energy Convers Manag 2004;45(11e12):1931e9. [20] Bakos GC. Design and construction of a two-axis sun tracking system for parabolic trough collector (PTC) efficiency improvement. Renew Energy 2006;31(15):2411e21. [21] Alexandru CA. Novel open-loop tracking strategy for photovoltaic systems. Sci World J 2013;2013:1e12. [22] Alata M, Al-Nimr MA, Qaroush Y. Developing a multipurpose sun tracking system using fuzzy control. Energy Convers Manag 2005;46(7e8): 1229e45. [23] Schramek P, Mills DR. Heliostats for maximum ground coverage. Energy 2004;29(5e6):701e13. [24] Mousazadeh H, Keyhani A, Javadi A, Mobli H, Abrinia K, Sharifi A. A review of principle and sun-tracking methods for maximizing solar systems output. Renew Sustain Energy Rev 2009;13(8):1800e18. [25] Agee JT, Jimoh AA. Flat controller design for hardware-cost reduction in polaraxis photovoltaic systems. Sol Energy 2012;86(1):452e62.