Solar Energy Advances 1 (2021) 100004

Contents lists available at ScienceDirect

Solar Energy Advances

journal homepage: www.elsevier.com/locate/seja

Accurate shading factor and mismatch loss analysis of bifacial HSAT

systems through ray-tracing modeling
Changrui Zhao∗, Jianwei Xiao, Yuanjie Yu, Jean-Nicolas Jaubert
CSI Solar Co. Ltd. 199 Lushan Road, SND, Suzhou, Jiangsu, China, 215129

a r t i c l e i n f o a b s t r a c t

Keywords: We have developed models using Rhino (3D modeler) and DIVA (Ray-tracing modeler), allowing system-level
Bifacial PV systems raytracing simulations which include nearby shade obstructions from the photovoltaic racking structures. The
Fixed racking ray-tracing simulated irradiance profiles are coupled with cell-level IV curve modeling to provide insight into
Horizontal single axis tracker HSAT
the shading factor and mismatch loss for different racking configurations. The paper focuses on reporting results
Mismatch loss
for horizontal single axis tracker (HSAT) and fixed racking bifacial system, included common module portrait
Ray-tracing modeling
Shading factor and landscape configurations, as well as demonstrating the distance effect of the module above the torque tube,
providing important guidance for further optimizing such mounting systems and accurate parameters required
in PVsyst modeling.

1. Introduction individual rays entering the module [10]. But the view factor model is
not yet practical if structure shading is considered, e.g., torque tube,
With recent accelerated developments in bifacial photovoltaic (PV) post, mounting rail, when shading factor and mismatch loss calculated
technology, the proportion of bifacial PV systems deployed annually is with backside irradiance data modeled through view factor method are
expected to grow significantly in the next few years. Global installed bi- used as inputs in PVsyst, the predicted annual rear-side irradiance gain
facial solar accumulative capacity has reached 8.8 GWdc by first half of will be overestimated [11]. The accurate result can be achieved with
2019 [1], bifacial modules and systems are expected to expand to about ray tracing method.
30% global market share in 2021 and conquering 60% market share The shading factor and mismatch loss of bifacial systems has been
within 10 years [2]. Especially the combinations of bifacial and track- evaluated with ray tracing model in some previous studies [8,12-16],
ing systems significantly enhance the energy yield of a PV system and in which a variety of structure effects have been incorporated. These
reduce the levelized cost of electricity (LCOE) [3,4]. As the industry be- modeling studies were generally based on traditional full cells module
comes familiar with bifacial technology, accurate energy yield modeling (e.g.,60 cells or 72 cells in series), and mismatch loss is only calculated
of bifacial PV systems naturally becomes a main concern. Strong com- within a single operated module. Furthermore, model validation against
mitment from research institutes and independent engineers have en- the field measured data and comparing one-in-portrait (1P) to two-in-
abled the development and validation of several accurate energy yield portrait (2P) tracker architectures were also reported. The study showed
prediction software programs, e.g., PVsyst, SAM and PVLib [5–7]. Al- that the shading and mismatch losses are impacted by the time of the
though these programs can simulate the energy yield of bifacial system, day and the weather condition (amount of direct versus diffuse light),
a few key factors affecting bifacial PV energy generation have not yet so the annual weather data is a necessary modelling input, worth to be
to be fully quantified, among these are the shading factor and mismatch further evaluated.
loss values for different system mounting configurations, which presents In this article, we evaluated the shading factor and mismatch loss
uncertainties that prevent further investment [8]. of four representative PV systems and scenarios including HSAT sys-
The front and rear side irradiance of each cell and module are re- tem one-in-portrait (1P) and two-in-portrait (2P), fixed racking system
quired when shading factor and mismatch loss are calculated. There are two-in-portrait (2P) and three-in-landscape (3L), based on commercially
two methods for modeling irradiance, view factors model, also termed representative twin half-cut cells PV modules (144 cells, 72 cells in se-
shape and configuration factors, quantify the fraction of irradiance re- ries and 2 strings in parallel). The mismatch loss is calculated within a
flected from one surface that reaches a receiving surface [9] and ray row or system, and hence, account for not only cell-to-cell mismatch,
tracing model, which simulates multipath reflection and absorption of but also module-to-module mismatch.

∗
Corresponding author.
E-mail address: [email protected] (C. Zhao).

https://doi.org/10.1016/j.seja.2021.100004
Received 13 July 2021; Received in revised form 30 September 2021; Accepted 30 September 2021
Available online 12 October 2021
2667-1131/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
C. Zhao, J. Xiao, Y. Yu et al. Solar Energy Advances 1 (2021) 100004

Fig. 1. The conventional HSAT system one-in-portrait (1P) of 3D model, in-

cluded 3D shading effects of all the relevant physical structures such as posts,
torque tube, and clamps.
Fig. 2. The electrical layout of twin half-cut cells PV module is series-parallel-
series (S-P-S) design (a), compared to series design of traditional full cells mod-
2. Model details ule (b).

2.1. System configuration 3D model

Various horizontal single axis tracker (HSAT) and fixed racking sys-
tem configuration are built in the 3-dimensional (3D) modeling software
Rhinoceros® [17]. Especially including the racking system detailed fea-
tures, for example the torque tube, clamp, rail, bearing and post shading
effect are taken into consideration (Fig. 1). The PV module and system
structure can also be directly imported as a .step file. In this article, all
modeling and analysis are based on frameless glass–glass bifacial mod-
ule with 144 solar cells. All galvanized steel structures are treated as a
partial scatterer (0.5 specularity, 0.5 roughness), the ground is flat and
treated as an ideal scatterer (0.0 specularity, 0.0 roughness). For the
HSAT system one-in-portrait (1P) mounted bifacial twin half-cut cells PV
modules, the decentralized mid-module junction boxes will be directly
located over the torque tube, this is an additional benefit compared to
traditional full cell modules.

2.2. Optical model

The bifacial optical model uses the Backward Ray-tracing Method,

in which the EnergyPlus weather (EPW) data is an input in DIVA-for-
Rhino. Using a daylighting and energy modeling plug-in, DIVA® [18],
the program computes a complete sky radiance distribution based on Fig. 3. The modeling and calculation process. (a) annual shading factor. (b)
Perez model and a cumulative sky approach [19], based on any pair annual mismatch loss.
of global horizontal irradiance (GHI) and diffuse horizontal irradiance
(DHI). For tracker systems, the tracking angle data is needed which is
the output from CASSYS developed by Canadian solar. CASSYS open- 3. Modeling and calculation methodology
source codes is available on Github [20].
The methodology for modeling and calculation of shading factor and
mismatch loss is described in Fig. 3. Three input files were needed, dif-
ferent system 3D models, location EnergyPlus weather (EPW) data file,
2.3. Module electrical model and the tracking angle data for horizontal single axis tracker modeled
system.
The module electrical model was based on a free open-source soft- For the shading factor calculation, backwards ray tracing tool de-
ware PVMismatch tool developed by Sunpower [21], this two-diode veloped based on Rhino and DIVA software were used to obtain hourly
electrical model allow different irradiance values to be assigned to each front and rear irradiance profile for each cell and module with or with-
cell, so can be used to calculate mismatch caused by back-side nonuni- out structure shading, for example torque tube, post, rail, clamp shad-
formity. We utilize PVMismatch tool to extract 2-diode parameters re- ing. Then we calculated the annual shading factor according to Eq. (1).
quired for modeling from batch cell data in mass production. The mod- Where Grear is modeled rear irradiance. N is the number of hourly data
ule electrical layout is shown in Fig. 2, made of 2 sets of 3 strings of points throughout the whole year. The modeled result can be input as
half-cells, mounted in parallel. Each pair of string shares the same by- the shading factor in PVsyst.
pass diode. Based on the physical characteristics of semiconductor, the
synthetic irradiance for bifacial model is represented by front irradi- Yearly shading factor =
(𝑁 ) (𝑁 )
ance and back irradiance multiplied by the bifaciality factor. The sec- ∑ ∑
ondary development was made for calculating different bifacial module 1− 𝐺rear with shading ∕ 𝐺rear without shading (1)
1 1
type and system mismatch loss. The system with nineteen modules, each
module has 144 half-cut cells, can be simulated to calculate module-to- For mismatch loss calculation, we random sampled some cells from a
module mismatch within a system. The cell temperature variation with lot of cell production data to compose module and system, the hourly ar-
irradiance is not accounted for in current modeling, is set to fixed value ray IV and maximum power are modeled combining with modeled front
of 25°C. and rear irradiance profile, then we can calculate hourly and annual

2
C. Zhao, J. Xiao, Y. Yu et al. Solar Energy Advances 1 (2021) 100004

Fig. 4. The modeled result of method B 36-day is representative samples in the

method A 365-day.

mismatch loss according to Eq. (2). But this mismatch loss is electrical
mismatch loss and not directly equivalent to PVSyst mismatch loss fac-
tor. When constructing the energy yield model in PVSyst, the mismatch
loss factor was set to 0%, and the modeled mismatch loss was input into
the module mismatch power loss.

Yearly mismatch loss =

(𝑁 ) (𝑁 )
∑ ∑
Array power ∕ Cells Power − 1 (2)
1 1 Fig. 5. The comparison between full cells model (a), 2736 cells are modeled
and calculated and representative cells model (b), 216 cells in central modules
Instead of basing the model on 365 days, we selected a sample of 36 are modeled and calculated.
days by picking three days from each month as representative to reduce
computation time. The computation time of full hourly simulations for
effect has only a marginal effect on the shading factor and mismatch
HSAT one-in-portrait (1P) configuration over a year of climate condi-
loss of long row and large system.
tions will be approximately 80 hours on a regular desktop computer, is
able to be reduced to 8 hours using representative 36 days. The method
B 36-day modeled result is compared with method A 365-day, modeled 3. Model validation
shading factor and the mismatch loss was respectively 7.1% and 0.61%
for method B, 7.0% and 0.60% for method A (Fig. 4). The comparison The main factor is rear irradiance model for calculating shading fac-
between the two methods shows very small deviation and demonstrates tor and mismatch loss. Although this model and validation has been
method B 36-day is a time efficient method that can provide accurate performed before in [22], which was conducted with a few modules in
results, so all annual modeling results in this paper are based on method a specific setup. To establish modeling accuracy, further validation of
B. a real fixed racking bifacial system is investigated and presented with
The comparison between full cells model and representative cells data from a test site located at Suzhou, China. The test site consists
model was also studied on HSAT one-in-portrait (1P) configuration to of eighteen test arrays in three rows, the target array with 12 bifacial
evaluate the effect of row edge (Fig. 5). Full cells model means each modules located in the middle array of the middle row in 2 portrait
cell in nineteen modules of one row is simulated and assigned modeled configuration. Eight SR05 pyranometer sensors (second class with un-
irradiance to calculate shading factor and mismatch loss. Representa- certainty 6.2% (k=2)) are deployed on backside of array as marked in
tive cells model is only one column cells of central nine modules are Fig. 6. The white gravel as enhanced Albedo material was laid under test
selected for modeling. Meteonorm 7.2 weather data was used to run array, which Albedo was measured and analyzed on site with two SR05
simulations for a sunny day on April 28 (DHI/GHI ratio 0.12) and a pyranometers. Time-series data at the original temporal resolution 60
cloudy day on July 20 (DHI/GHI ratio 0.80), the modeled shading fac- seconds is reformatted to hourly data, measured average daily Albedo
tor and mismatch loss was respectively 6.5% and 0.17% for full cells 60.5% over a month period (Aug 8th - Aug 30th , 2020), the diffuse ratio
model, 6.0% and 0.21% for representative cells model on the sunny (DHI/GHI) was 0.717.
day, the modeled shading factor and mismatch loss was respectively The measured and modeled values for the eight irradiance sensors in
6.6% and 0.18% for full cells model, 6.0% and 0.24% for representa- horizontal and vertical direction are plotted in Fig. 7. The model tended
tive cells model on the cloudy day. The mismatch loss is approximately to underestimate measured values by 3.2-13.4 Wm-2. The mean bias er-
consistent with [16]. Within the distance of 4m at the south of row, 2m ror (MBE), and root mean square error (RMSE) can be seen in Table 1.
at the north of row, system experience edge brightening, but the edge The result shows a bit high deviation. This is firstly because albedo devi-

3
C. Zhao, J. Xiao, Y. Yu et al. Solar Energy Advances 1 (2021) 100004

Table 1
MBE And RMSE for the eight sensors modeled.

Sensor location MBE_abs [w/m2] MBE_rel % RMSE [w/m2] RMSE_rel %

S1(East most) 4.5 8.1 10.3 18.3

S2 7.9 12.7 10.3 16.4
S8 7.3 11.8 9.6 15.4
S7(West most) 5.7 9.9 8.2 14.2
S3(North most) 3.2 3.1 10.0 12.0
S4 6.3 7.6 10.5 10.0
S5 10.2 17.4 12.5 21.3
S6(South most) 13.4 15.2 18.0 20.3

Fig. 6. A photo and 3D image showing the fixed system. Eight SR05 pyranome-
ter sensors are mounted on the back of test array.

ations from possible spectral mismatch, the value tested by pyranometer

is lower 6.6% mentioned in [23], secondly the complex racking struc-
tures in the site lead to the real area of white gravel laid under test Fig. 7. Measured vs. modeled values for eight sensors in horizontal (a) and
array is affected and thirdly the uncertainty of pyranometers used in vertical (b) direction from Aug 8th to Aug 30th , 2020.
test is higher than secondary standard sensor. Hence, the compassion
can be improved by well consider above these factors.
To evaluate uncertainties in the model result, one day weather data
was applied, 500 times modeling and calculation of shading factor was
implemented based on a HSAT system one-in-portrait (1P). The result
distribution is plotted in Fig. 8. The mean value was 6.81%, standard
deviation was 0.53%, while the upper and lower 95% mean are re-
spectively 6.86% and 6.77%. It is a reasonable and acceptable accuracy
range compared to the value of 7.8% in [12] based on similar system
configuration and structure.

4. Modeling results

Four various system racking type were analyzed and modeled, HSAT
system 1 portrait and 2 portrait, fixed racking 2 portrait and 3 landscape. Fig. 8. The distribution of this deviation for 500 modeling results.

4.1. HSAT system

Two factors were taken into consideration for the HSAT 1 portrait
4.1.1. HSAT 1 portrait model system configuration, the distance 50 mm between the torque
The HSAT system one-in-portrait (1P) simulation considers a PV ar- tube and module back side, array height 2m and 1.5m. Shading factor
ray of five rows, each row has one string of nineteen modules. The con- varied from 5.6% to 8.1%, mismatch loss from 0.23% to 0.94% during
figuration has a row pitch of 5m, equivalent to a ground-coverage ratio the albedo range 0.2∼0.6 (Fig. 9). The height 2m has similar trend and
of 40.2%. Its torque tube has a diameter of 12.1 cm. result with 1.5m height, also can increase module backside irradiation,

4
C. Zhao, J. Xiao, Y. Yu et al. Solar Energy Advances 1 (2021) 100004

Table 2
Shading & Mismatch Loss Factor Comparison- HSAT 1P.

Parameter Nextracker ATI CSI

Shading factor 12.3% 7.8% 5.6%

Mismatch loss factor 3.5% 2.4% 3.2%
Angle -60° +60° -52° +52° -60° +60°
Height(m) 1.5 1.6 1.5
Albedo 0.28 0.3 0.2
GCR 0.33 0.35 0.4
Weather data Annual Annual Annual (36 days)
Location Fremont, USA Sunny climate Golmud, CN
Structure model accounting for rear-shading Torque tube Mounting rails Bearings Piers Torque tube Clamps Piers Torque tube Mounting rails Bearings Piers

Fig. 10. The U shape connection (left) and I shape connection (right).

Fig. 9. HSAT 1 portrait shading factor and mismatch loss under different
Albedo.

but has no obvious contribution to improving module backside irradi-

ance uniformity.
The modeling input parameters and results presented on white pa-
pers [12,13] of tracker suppliers are summarized and listed in Table 2.
The mismatch loss result in this paper was converted to mismatch loss
factor in PVsyst so that the comparison can be made with results in white
paper. The structure shading and mismatch loss factors from Nextracker,
were determined by iterative process through constantly changing the
model inputs and comparing the modeled to measured results. The pos-
sible reason of large shading factor is the module adopted in modeling
is whole cell instead of half-cut cell module. The best-fit values were Fig. 11. HSAT 2 portrait shading factor and mismatch loss under different
determined to represent annual weather data which were used in mod- Albedo and tube-module distance.
eling by ATI. Considering the similar system configuration, the modeled
result is comparable to ATI value.
The result shows shading factor is not sensitive to tube-module dis-
tance when the Albedo range is greater than 0.4. Mismatch loss is almost
4.1.2. HSAT 2 portrait same under different tube-module distance and Albedo (Fig. 11).
The HSAT system two-in-portrait (2P) simulation considers a PV ar- The rail length 3.5m and 2.5m are also modeled and compared. The
ray of five rows, each row has two strings, nineteen modules per string. shading factor increased about 0.8% when rail length changed from
The configuration has a row pitch of 12.6m, equivalent to a ground- 2.5m to 3.5m under different Albedo. Mismatch Loss is not sensitive
coverage ratio of 33.1%. Its modules are separated by 15 cm in the EW to rail length under different albedo condition.
direction and its torque tube is 2.35 m above the ground, and has a According to tube-module distance and rail length effect modeling
square cross section of dimensions 15 × 15 cm. result, HSAT 2P backside irradiance is still affected by torque tube, rail,
The distance 50 and 100 mm between torque tube and module back post, so shading factor cannot be zero percent.
side are modeled and compared. For 2P system, when calculating the The modeling input parameters and results presented on white pa-
mismatch loss, the connection topology should be considered. I shape pers [12,14] of tracker suppliers are summarized and listed in Table 3.
connection is a reasonable connection topology because of less mis- The study shows that the shading and mismatch losses tend to vary with
match loss than U shape, so all 2P mismatch cases in this paper are the time of the day and the weather condition (amount of direct versus
calculated based on the I shape connection (Fig. 10). diffuse light) [16], only one sunny day is obviously insufficient used for

5
C. Zhao, J. Xiao, Y. Yu et al. Solar Energy Advances 1 (2021) 100004

Table 3
Shading & Mismatch Loss Factor Comparison- HSAT 2P.

Parameter Soltec ATI CSI

Shading factor 0.7% 6.5% 5.7%

Mismatch loss factor 1.1% 2.5% 2.8%
Angle -60° +60° 52° +52° -60° +60°
Height(m) 2.35 2.4 2.35
Albedo 0.28 0.3 0.2
GCR 0.33 0.35 0.33
Weather data One sunny day Annual Annual (36 days)
Location Livermore, USA Sunny climate Golmud, CN
Structure model accounting for rear-shading Torque tube Torque tube Clamps Piers Torque tube Mounting rails Bearings Piers

Fig. 12. Typical system configuration of 1P (left) and 2P (right).

Fig. 14. Fixed racking 2 portrait shading factor and mismatch loss under dif-
ferent Albedo and height.

Fig. 13. Shading factor and mismatch loss comparison under different Albedo.

simulation. Considering the similar system configuration, the modeled

result is comparable to ATI value.

4.13. Comparing HSAT 1 portrait to HSAT 2 portrait

A comparison between HSAT 1 portrait and HSAT 2 portrait was
made by choosing individual typical system configuration (Fig. 12). For
shading factor, HSAT 2 portrait had smaller values than HSAT 1 portrait
under Albedo range 0.2∼0.6. For mismatch loss, HSAT 2 portrait had
more advantages than HSAT 1 portrait with increased Albedo (Fig. 13).

4.2. Fixed racking system

The fixed racking system two-in-portrait (2P) simulation considers

a PV array of five rows, each row has two strings, nineteen modules
per string. The configuration has a row pitch of 9.0m, equivalent to a
ground-coverage ratio of 44.9%. The system is at a fixed tilt angle of
32° and an azimuth orientation of 180° (facing south). For fixed racking
2 portrait, height 0.5m showed similar trend and result with 1.0m un-
der different Albedo range for both shading factor and mismatch loss. Fig. 15. Fixed racking 3 landscape shading factor and mismatch loss under
According to the rear side irradiance distribution, increased height can different Albedo and height.

6
C. Zhao, J. Xiao, Y. Yu et al. Solar Energy Advances 1 (2021) 100004

Acknowledgment

The authors gratefully acknowledge Dr. Joshua S. Stein of the Sandia

National Laboratories, Dr. Silvana Ayala Pelaez of the National Renew-
able Energy Laboratory and Frederic Rivollier of the Canadian Solar,
Inc. for technical suggestions and discussions.

References

[1] X. Sun, Global Bifacial Module Market Report 2019, 2019 [Online]. Available:
https://www.woodmac.com/reports/power-markets-global-bifacial-module-market
-report-2019-348173 .
[2] S. Chunduri, TaiyangNews Report on Advanced Module Technologies 2019,
2019 [Online]. Available:http://taiyangnews.info/reports/advanced-module-
technologies-2019/.
[3] Rodrı´guez-Gallegos, et al., Global techno-economic performance of bifa-
cial and tracking photovoltaic systems, Joule (2020) [Online] Available:,
doi:10.1016/j.joule.2020.05.005.
[4] M Bazilian, I Onyeji, M Liebreich, I MacGill, J Chase, J Shah, D Gielen,
D Arent, D Landfear, S Zhengrong, Re-considering the economics of photo-
voltaic power, Renew. Energy 1 (53) (2013) 329–338 [Online] Available:,
doi:10.1016/j.renene.2012.11.029.
[5] Bruno Wittmer, André Mermoud, Yield simulations for horizontal axis trackers with
bifacial PV modules in PVsyst, 35th European Photovoltaic Solar Energy Conference
and Exhibition, 2018 doi: 10.4229/35thEUPVSEC20182018-6CV.2.19.
[6] Freeman, J.M., DiOrio, N.A., Blair, N.J., Neises, T.W., Wagner, .J., Gilman, P., and
Janzou, S.T., "System Advisor Model (SAM) General Description (Version 2017.9.5),"
Fig. 16. The shading factor and mismatch loss range for various system. doi: 10.2172/1440404.
[7] T. Gurupira, A.J. Rix, PV Simulation Software Comparisons: Pvsyst, NREL SAM
and PVLib, 2017 [Online] Available: https://www.researchgate.net/publication/
313249367 .
obviously boost backside irradiance (Fig. 14). The shading factor of fixed
[8] S.A. Pelaez, C. Deline, P. Greenberg, J.S. Stein, R.K. Kostuk, Model and validation of
racking was much higher than HSAT system. single-axis tracking with bifacial PV, IEEE J. Photovolt. 9 (3) (2019) 715–721 May,
The fixed racking system three-in-landscape (3L) simulation consid- doi:10.1109/JPHOTOV.2019.2892872.
ers a PV array of five rows, each row has three strings, nineteen modules [9] C.W. Hansen, J.S. Stein, C. Deline, S. Macalpine, B. Marion, A. Asgharzadeh,
F. Toor, Analysis of irradiance models for bifacial PV module, in: 43rd
per string. The configuration has a row pitch of 6.8m, equivalent to a IEEE Photovoltaic Specialist Conference. Portland, Oregon, 2016, p. 138 143,
ground-coverage ratio of 44.4%. The system is at a fixed tilt angle of 32° doi:10.1109/PVSC.2016.7749564.
and an azimuth orientation of 180° (facing south). For the fixed racking [10] P.G. Loutzenhiser, H. Manz, C. Felsmann, P.A. Strachan, T. Frank, G.M. Maxwell,
Empirical validation of models to compute solar irradiance on inclined sur-
3 landscape, height 0.5m had higher value than 1.0m under different faces for building energy simulation, Solar Energy 81 (2) (2006) 254–267,
Albedo ranges for both shading factor and mismatch loss (Fig. 15). The doi:10.1016/j.solener.2006.03.009.
rear side irradiance distribution demonstrates increased height can ob- [11] S. Ayala Pelaez, C. Deline, S. MacAlpine, B. Marion, J. Stein, R. Kostuk, Comparison
of bifacial solar irradiance model predictions with field validation, IEEE J. Photovolt.
viously increase backside irradiance. 9 (1) (2019) 82–88, doi:10.1109/JPHOTOV.2018.2877000.
[12] Field Testing Meets Modeling: Validated Data on Bifacial Solar Performance, 2020
5. Conclusion [Online] Available: https://arraytechinc.com/field-testing-meets-modeling-white-
paper/.
[13] White Paper: Quantifying Your Bifacial Gains–Using Calibrated PVsyst Model In-
The two plots in Fig. 16 show the shading factor and mismatch loss put Parameters to Accurately Predict In-field Performance, 2020 [Online] Available:
range for various PV system during Albedo range 0.2∼0.6. https://info.nextracker.com/en/quantifying-your-bifacial-gains .
[14] White Paper: Bifacial Gain and Production Analysis at BiTEC, 2018 [Online] Avail-
For HSAT 1 portrait, shading factor and mismatch loss is insensitive
able: https://lab.soltec.com/bifacial-trackers/ .
to array height, while mismatch loss is sensitive to Albedo. For HSAT 2 [15] White Paper: Determining PVSyst Bifacial Inputs, 2021 [Online] Available:
portrait, backside irradiance is still affected by torque tube, rail, post, so https://www.pvlighthouse.com.au/cms/lectures/white-papers .
the shading factor cannot be 0%, and has more advantage than 1 por- [16] K.R. McIntosh, M.D. Abbott, B.A. Sudbury, J. Meydbray, Mismatch loss in bifacial
modules due to nonuniform illumination in 1-D tracking systems, IEEE J. Photovolt.
trait under Albedo≥0.4. For fixed racking, shading factor is much higher 9 (6) (2019) 1504–1512, doi:10.1109/JPHOTOV.2019.2937217.
than HSAT system. Above conclusions are based on shading factor and [17] Rhinoceros 3D, 2021 [Online] Available: http://www.rhino3d.com.
mismatch loss, the optimal system configuration needs to be determined [18] Diva For Rhino, 2021 [Online] Available: http://diva4rhino.com.
[19] D. Robinson, A. Stone, Irradiation modelling made simple: the cumulative sky ap-
by energy yield and LCOE. proach and its applications, The 21st Conference on Passive and Low Energy Archi-
Further validation is required to establish the modeled rear irradi- tecture, 2004.
ance accuracy by using secondary standard pyranometer instead of sec- [20] CASSYS, 2021 [Online] Available: https://github.com/CanadianSolar/CASSYS.
[21] PVMismatch, tool (2018), [Online] Available: https://github.com/SunPower/
ond class SR05 which has higher uncertainty measurements. The rep- PVMismatch.
resentative modules and cell rows are selected and modeled to reduce [22] A. Pelaez Silvana, C. Deline, J.S. Stein, B. Marion, K. Anderson, M. Muller, Ef-
computation time, the method of running the simulation on high perfor- fect of Torque-Tube Parameters on Rear-Irradiance and Rear-Shading Loss for Bi-
facial PV Performance on Single-Axis Tracking Systems, 2019 [Online] Available:
mance computing (HPC) needs to be further explored to further improve https://www.nrel.gov/docs/fy20osti/73203.pdf .
modeling accuracy in future. [23] M. Gostein, B. Marion, Bill Stueve, Spectral Effects in Albedo and Rearside Irra-
diance Measurement for Bifacial Performance Estimation:Preprint, 2020 [Online]
Available: https://www.nrel.gov/docs/fy20osti/75927.pdf .
Declaration of Competing interests

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence
the work reported in this paper.