Increased Energy Production of First Solar Horizontal Single-Axis Tracking PV Systems Without Backtracking
Increased Energy Production of First Solar Horizontal Single-Axis Tracking PV Systems Without Backtracking
Increased Energy Production of First Solar Horizontal Single-Axis Tracking PV Systems Without Backtracking
ABSTRACT — Backtracking (“BT”) is a shade avoidance producing energy. The PV cells in First Solar modules are
algorithm for single-axis horizontal tracker PV systems which oriented in a way such that non-uniform shading of the cells is
has historically been regarded as superior to all other tracking very unlikely to occur. When the module is oriented with the
algorithms. Because their internal electrical structure does not shadow perpendicular to the cells, the power loss is roughly
create a large shading penalty, First Solar modules have the
option of not backtracking (also termed “Truetracking” or “TT”)
linear with shading, which allows for the possibility to TT.
while most c-Si modules do not. Experiments on multiple PV This paper presents information regarding energy
systems presented in this paper show that TT yields 1.7% - 2.5% advantages of TT with First Solar modules, including
higher annual energy than BT with First Solar modules. First empirical results from small and commercial-scale systems.
Solar and independent third parties have shown that PVSyst, the Additionally, specifics of predicting energy output of BT and
industry standard PV prediction tool, over-predicts energy for TT First Solar systems with commercially available
systems in BT mode due to an incorrect assumption about diffuse simulation tools such as PVSyst, PV*SOL, and PVWatts is
shading. Additionally, other publically available PV prediction discussed. Finally, ISIS – First Solar’s internal prediction tool,
tools have limitations that prevent them from accurately is used to predict energy gain of TT over BT.
predicting tracking PV system performance. First Solar has
developed its own PV system energy prediction tool, ISIS, which
appropriately accounts for diffuse shading and other differences II. EXPERIMENT ON A 100-KW ARRAY
between BT and TT. ISIS predicts more energy for First Solar
systems in TT mode. Consequently, the default tracking Two test systems are installed in North America. These
configuration for First Solar utility scale systems is TT. systems are identical by all measures (DC/AC, inverter type,
Index Terms — First Solar, Backtracking, shading, PV system row spacing, etc.) except for the tracking algorithm that they
modeling employ. Both systems began their life in BT mode in 2011.
Since that time, they have been periodically switched between
I. INTRODUCTION BT and TT to test the hypothesis that TT produces more
energy with First Solar PV modules.
Backtracking is an algorithm for horizontal single-axis Both systems were backtracking until System A was
trackers that has been historically regarded by crystalline changed to TT on June 16, 2011. Clear days before and after
silicon (“c-Si”) PV system developers as superior to all other this tracking algorithm change are presented in Fig 1. Notice
tracking algorithms because it optimizes energy yield and the relative change in position of the system performance in
reduces the use of diodes which reduces diode failure [1]. In the morning of these two days. When both systems are BT,
backtracking (“BT”) mode, trackers follow the path of the sun their performance is nearly identical. However, after System A
in the middle of the day, aiming to minimize the angle-of- is put into TT mode, its performance exceeds that of the BT
incidence between the sun and the PV modules. However, in System B in the morning. The daily energy difference between
the morning and the evening, as the sun is rising and setting, these systems was 3.1% on this day.
trackers in BT mode no longer directly follow the sun, but Multiple tracking algorithm reversals were conducted
rather move into a shade avoidance tracking algorithm. This throughout the subsequent year, with Systems A and B
technique requires the modules to track “backwards” toward a switching between TT and BT. Each time the tracking
horizontal orientation just enough that no direct beam shading algorithms changed, a reversal in the relative power in the
occurs. In a Truetracking (TT) mode, trackers follow the path morning and afternoon was observed. The ratio of the daily
of the sun as it traverses the sky, minimizing the angle-of- energy produced by the TT system to that of the BT system
incidence between the sun and the panels as closely as the was calculated for one consecutive year (Fig. 2). Values above
single axis of rotation will allow during all hours of the day, 100% represent days where the TT system produced more
regardless of shade patterns. Trackers behave identically in the energy. This calculation is system-agnostic in that the BT
middle of the day for both BT and TT systems. system was switched between System A and System B
BT is a necessity for most c-Si PV modules because they multiple times during this test. The measured annual energy
use bypass diodes to protect the cell strings in the module. gain of TT for this experiment was 1.7%.
Depending on mounting orientation, panels with bypass
diodes have significantly reduced energy output if an entire
row of cells becomes shaded because the bypass diodes will
force entire strings of cells within the module to cease
106%
Not Backtracking/
104% a 2.0% loss when compared to the median system on site. All
102%
analyses of this system suggest that TT provides an energy
gain of 2.0% to 2.5%.
100%
98%
IV. PREDICTED ENERGY DIFFERENCES BETWEEN
96%
BACKTRACKING AND TRUETRACKING
94%
Predicting these small energy differences requires
detailed calculations of direct, diffuse and albedo shading for
each timestamp, and the impact of these shading factors on the
Fig 2. Ratio of TT to BT energy production measured on 100- energy production. No commercially available PV system
kW test systems. Annual gain of TT was 1.7%. simulation tool appropriately models these factors.
PVSyst (v5.56 and earlier) [3], the industry standard
III. MULTI-MW SYSTEM EXPERIMENT software for predicting PV system energy predicts that a First
Solar BT system produces more energy than a TT system. The
A similar experiment was conducted on a commercial- root cause of the predicted energy being inconsistent with the
scale system in North America with an AC nameplate greater measured results presented in this paper is, simply put, that
than 10 MW. All arrays in this system began their life in BT PVSyst incorrectly sets diffuse and albedo shading losses to
mode. In the autumn of 2012, two of the arrays were changed 0% (only beam shading should be set to 0%). When
to TT mode (TT arrays subsequently referred to as “System
simulating a tracker in TT mode, PVSyst does model a diffuse
C” and “System D”). System C was changed to TT mode on
shading loss. The combination of these factors means that
August 30, 2012; System D was changed on September 18,
predictions run in BT mode will have artificially low shading
2012. Daily power profiles of the two TT arrays were then
compared to the median of the BT arrays on this site. Before losses, thus over-predicting energy. A similar result was
the change, these two arrays performed very similarly to the observed by Vanicek and Stein [2]. This error is true
median array in the morning and afternoon. Improvements in regardless of the module used in the prediction, and thus all
the morning power profiles compared to the median BT BT systems modeled in PVSyst will result in an overestimate
inverter are immediately noticeable after the change to TT of energy. Fig 3 depicts a comparison of the same system
mode, as is shown in Table I. The ratio of daily AC energy predicted in PVSyst, once in TT mode and once in BT mode.
production of the two TT arrays to that of the median BT array In the middle of the day, despite the two systems operating
on site was calculated before the change (Fig. 6). The median identically (as is evidenced by the lack of difference in angle
ratios before the algorithm change were compared to those of incidence), PVSyst predicts 1% - 2% more energy for the
after the algorithm change to calculate the energy gain due to BT system.
TT. On average, the two systems showed a TT energy gain of
2.1% over the median BT array.
On December 3, 2012 all arrays on site except for two
(“System E” and “System F”) were set to TT mode. Systems E
Predicted Daily Energy Gain (TT/BT)
30 1.05
1.04
25 1.03
1.02
(deg)
20
1.01
15 1
0.99
10
0.98
5 0.97
0.96
0
6 7 8 9 10 11 12 13 14 15 16 17
Hour of Day
Fig 7. Daily energy gain of two arrays that remained in BT mode after the rest of the arrays on site were transitioned to TT on
December 3, 2012. Horizontal black lines show median daily gain values before and after the change. The difference between
these two lines represents the energy advantage of TT over BT. In this case, the measured energy advantage of TT is 2.0%.
REFERENCES
V. CONCLUSIONS
[1] Pancino, David, et al. “BT: A Novel Strategy for Tracking PV
Experiments run on both kW- and MW-scale PV systems Systems”. IEEE. 1991.
show that TT with First Solar modules yields 1.7% - 2.5% [2] Vanicek, Philipp and Stein, Sebastian. “Simulation of the Impact
higher annual energy than identical BT systems. TT is a viable of Diffuse Shading on the Yields of Large Scale Single Axis
option for First Solar systems but not for most c-Si systems Tracked PV-Plants”, 24th European Photovoltaic Solar
Energy Conference, 21-25 September 2009, DOI
due to intrinsically different electrical structure in First Solar 10.4229/24thEUPVSEC2009-4BV.1.39
modules (i.e., PV cell design). No commercially available PV [3] PVSyst 5.56, http://www.pvsyst.com/en/, Feb 4, 2013
prediction tool appropriately models the complex phenomena [4] PV*SOL Pro 4.0 R2, http://www.solardesign.co.uk/pv.php, Feb 4,
involved in BT. First Solar has developed its own PV system 2013
energy prediction tool which appropriately accounts for [5] PVWatts, http://www.nrel.gov/rredc/pvwatts.html, Feb 4,
2013.
diffuse shading and predicts 1% - 2% more energy for systems
in TT mode with First Solar modules.
AC Power (kW)
AC Power (kW)
400
August 28, 2012
300
300
All systems BT
200 200
Median System
100 System C 100
System D
0 0
600
500
September 9, 2012 500
400
AC Power (kW)
AC Power (kW)
Median System and 400
System C BT 300
300
System D
200 200
TT Median System Median System
100 System C 100 System C
System D System D
500 500
400 400
September 20, 2012
AC Power (kW)
AC Power (kW)