Content uploaded by Michael Gostein
Author content
All content in this area was uploaded by Michael Gostein on Jun 09, 2023
Content may be subject to copyright.
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/354922307
Measuring Irradiance for Bifacial PV Systems
Conference Paper · June 2021
DOI: 10.1109/PVSC43889.2021.9518601
CITATIONS
11
READS
704
10 authors, including:
Some of the authors of this publication are also working on these related projects:
A Special Issue of Energies (ISSN 1996-1073, IF 2.707) on "Increasing the Lifetime of Photovoltaics Systems: Advanced Materials, Monitoring, O&M and Energy Modeling" View
project
IEA PVPS Task 16: Solar resource for high penetration and large scale applications View project
Michael Gostein
Atonometrics
72 PUBLICATIONS1,156 CITATIONS
SEE PROFILE
Silvana Ovaitt
National Renewable Energy Laboratory
36 PUBLICATIONS367 CITATIONS
SEE PROFILE
Chris Deline
National Renewable Energy Laboratory
119 PUBLICATIONS2,942 CITATIONS
SEE PROFILE
Aron Habte
National Renewable Energy Laboratory
117 PUBLICATIONS1,792 CITATIONS
SEE PROFILE
All content following this page was uploaded by Michael Gostein on 29 September 2021.
The user has requested enhancement of the downloaded file.
Measuring Irradiance for Bifacial PV Systems
Michael Gostein1, Silvana Ayala Pelaez2, Chris Deline2, Aron Habte2, Clifford W. Hansen3, Bill Marion2,
Jeff Newmiller4, Manajit Sengupta2, Joshua S. Stein3, Itai Suez5
1Atonometrics, Austin, TX, 78757, USA; 2National Renewable Energy Laboratory, Golden, CO, 80401,
USA; 3Sandia National Laboratories, Albuquerque, NM, 87123, USA; 4DNV, Oakland, CA, 94612, USA;
5Silfab Solar, Mississauga, ON, L5T2Y3, Canada
Abstract — The advent of bifacial PV systems drives new
requirements for irradiance measurement at PV projects for
monitoring and assessment purposes. While there are several
approaches, there is still no uniform guidance for what irradiance
parameters to measure and for the optimal selection and
placement of irradiance sensors at bifacial arrays. Standards are
emerging to address these topics but are not yet available. In this
paper we review approaches to bifacial irradiance monitoring
which are being discussed in the research literature and pursued
in early systems, to provide a preliminary guide and framework
for developers planning bifacial projects.
Index Terms — photovoltaic systems, bifacial, irradiance
measurement, performance analysis, solar energy
I. INTRODUCTION
With the adoption of bifacial modules in PV systems,
developers of PV power plants face new challenges in deciding
how best to measure irradiance for system monitoring and
assessment against performance guarantees. There is a growing
body of work examining methods for standardizing power
output measurements of bifacial PV modules and predicting
bifacial PV array performance under given conditions of front
and rear irradiance, primarily as an aid to system design
[1][2][3][4][5][6]. However, for the construction of new PV
power plants employing bifacial modules, there is still no
uniform guidance on the type, quantity, and placement of
irradiance sensors in a monitoring system or for the use of
irradiance data for monitoring and assessment in performance
contracts.
Standards are emerging to address these topics. IEC
61724-1, which covers PV system performance monitoring, is
currently undergoing revision to Edition 2. A major new
component of this revision is the addition of requirements and
recommendations for irradiance measurement in bifacial
systems. The draft document also includes definitions of
metrics for bifacial systems. In other work, proposals for
adapting PV system capacity testing to bifacial systems have
been explored [7]. A new effort is currently underway to update
IEC 61724-2 and IEC 61724-3, which cover capacity and
energy testing, for the bifacial era.
In this paper we will provide a review of various options and
recommendations for bifacial system irradiance monitoring
which are being discussed in the research literature as well as
the international standards development community, while
being pursued in practice on early adopters’ and prototype
research and development systems. The intent of this work is
to provide a preliminary guide and framework for discussion to
assist researchers, owner’s engineers, and developers that are
actively planning projects.
II. CHALLENGES
A. Overview
Ultimately the goal of measuring or modeling both the front
and rear-side irradiance of a bifacial PV system is to arrive at a
time-dependent value of the solar resource which can be used
to predict PV system power output and/or ensure that the
system is functioning as intended.
Irradiance reaching the front-side and rear-side plane-of-
array (POA) of bifacial PV modules includes many
components, as illustrated in Fig. 1. These include direct
radiation from the sun, diffuse radiation from the sky, and
ground-reflected radiation. These components have different
relative contributions on the front and rear sides of bifacial
modules.
Fig. 1. Contributions to illumination of front and rear of bifacial PV modules,
including direct, diffuse, and ground-reflected radiation.
Sandia National Laboratories is a multimission laboratory managed and
operated by National Technology and Engineering Solutions of Sandia, LLC,
a wholly owned subsidiary of Honeywell International Inc., for the U.S.
Department of Energy’s Na
tional Nuclear Security Administration under
contract DE
-NA0003525.
This work was supported by the U.S. Department of Energy under
Contract No. DE
-AC36-08-GO28308 with the National Renewable Energy
Laboratory (NREL). Funding provided by the U.S. Department
of Energy’s
Office of Energy Efficiency and Renewable Energy (EERE) under Solar
Energy Technologies Office (SETO) Agreement Number 34910.
Preprint: 2021 IEEE Photovoltaic Specialists Conference (PVSC)
While methods of front-side irradiance measurements for
monitoring the performance of monofacial solar PV systems
are already well established, for bifacial systems industry
consensus on measurement methods has not yet been reached.
There are multiple challenges to consider. These include spatial
and time-varying non-uniformity of rear-side irradiance, the
relative impact of diffuse versus direct radiation, uncertainty in
ground reflection, and spectral effects. System design
parameters affecting these uncertainties include ground
clearance height, view factor, ground coverage ratio, tracking,
rear-side shading caused by racking components such as torque
tubes and rails, and variability in the reflective properties of the
ground surface due to angular and spectral dependencies.
Choices of the sensor types, locations, and quantities in
combination with the unique parameters of the design itself
will define how much uncertainty will exist in the assessment
of a bifacial system’s performance.
B. Non-uniformity
The non-uniformity of rear-side irradiance and its impact on
power generation are especially significant factors affecting
both measurement and modeling, increasing uncertainty in
performance prediction.
Fig. 2 illustrates some key aspects of rear-side irradiance
non-uniformity, using a simulation of rear-side irradiance for
each cell in 20-module row at the center of a 7-row single-axis
tracking array. Simulations are performed for sunny conditions
in Golden, Colorado, for two different days and times. Both
simulations show irradiance at the row-ends of only ~50-75%
that at the row-center as well as significant transverse non-
uniformity and torque-tube shading along the center line. The
irradiance profile will change throughout the day: note in the
9 am simulation the enhanced irradiance on the eastern side
towards the morning sun. The changing non-uniformity
patterns displayed in Fig. 2 complicate the potential placement
of rear-side irradiance sensors, since no single sensor location
represents conditions along the entire row.
Some modeling software packages simplify the rear-side
non-uniformity by predicting only an average rear-side
irradiance value. However, Fig. 3 shows an example of how
Fig.
3. Rear-side irradiance versus east-west position across north-
south
oriented module
rows, showing irradiance non-uniformity. Adapted from [4]
.
Fig.
4: Measured non-uniformity of rear irradiance, showing spatial, temporal, and seasonal changes. Four rear-facing reference cells along collector width of
row
(
West to East) indicate lower rear irradiance values (y-axis, W/m2) in the center (Center-West, Center-East) than the edges (West, East), with hour-of-day (x-
axis)
variation
and seasonal variation (dashes = July 1, solid = September 1, dotted = December 25, faint lines = selected clear sky days in 2020).
Fig.
2: Modeled rear irradiance for each cell in a single-
axis tracker row at
the center of a
7-row x 20-module array. Modeled for
January 1st at noon and
June 1st at 9 am
in sunny conditions at Golden, CO. Non-
uniformities are
seen both near the row edges and across the collection width perpendicular
to th
e row. Irradiance scale is in W/m2. Rows are north-south.
much the actual irradiance across the rear side of bifacial
modules can deviate from the average. In this example, for a
single axis tracking system with modules on either side of a
torque tube, the irradiance is approximately 20% lower near the
center rotation axis than at the module edges.
Fig. 4 illustrates another example of rear-side irradiance
spatial and temporal variability. In the example shown by the
figure, rear-side irradiance measurements on a single axis
tracking system show variability from one transverse edge of
the modules to the other based on both time of day as well as
day of year. The figure shows up to ~30% greater irradiance on
the module edge towards the sun (east or west) during morning
or afternoon, with this non-uniformity varying throughout the
year.
Spatial non-uniformity is also a function of system design.
Fig. 5 illustrates the impact of ground clearance height on the
rear-side irradiance transverse to a module row, with higher
clearance height leading to greater irradiance and significantly
reduced non-uniformity – but also greater costs for racking.
C. Diffuse fraction
Diffuse light fraction adds another complication. In a
simplistic assumption, average rear-side irradiance could be
considered proportional to front-side irradiance; however,
Fig. 6 shows that this assumption is incorrect. Each point in the
plot is an hourly modeled spatially averaged irradiance
combination comparing rear and front POA irradiance from a
typical year of data for a horizontal single axis tracking array
located in Wisconsin. The green and red circles represent data
from days with high and low diffuse irradiance, respectively.
Depending on whether the performance of this system is
evaluated on days with high or low diffuse fraction, the
proportionality of rear to front POA irradiance can vary on the
order of a factor of up to 4, with higher diffuse fraction leading
to relatively greater rear-side irradiance. Since this can
correspond to deviations of 20% relative to a model that
ignores rear POA, short-term performance evaluation methods
should not neglect rear POA or diffuse fraction.
D. Albedo
Albedo, the reflectivity of the ground surface, is of course
critical to bifacial PV system performance. The detailed
angular and spectral distribution of ground-reflected radiation
is complex [8][9]. However, as a simplification, most PV
performance modeling software considers the albedo of a
ground surface as a single number representing the ratio of total
reflected light to total incident light.
For modeling bifacial systems, albedo values for ground
surfaces similar to what is expected to be on site are often used.
These values are often obtained from nearby weather stations
with albedometers or from satellite sources [10][11][12]. In
cases where no data are available, standard average values of
albedo can be used [13]. Some weather files such as the
National Solar Radiation Database [11] include albedo values
to be used for generalized modeling. Data sets of ground albedo
and associated meteorological data developed by using existing
Fig.
6. Impact of different diffuse irradiance fraction conditions, showing non
-
proportionality of rear
-side and front-
side POA irradiance. Green: sample high
diffuse fraction days. Red: sample low diffuse fraction days.
Fig.
5: Rear-side irradiance non-
uniformity is reduced as ground clearance
height is increased. The
figure compares
measurements, view factor model,
and Radiance model for irradiance transverse to a module row.
0
50
100
150
200
250
300
350
0 100 200 300 400 500 600
Rear-side Irradiance (W/m
2
)
Rear-side Sensor Position (mm)
Measured
View factor model
Radiance model
Clearance 0.15 meters
Fig. 7. Relationships between albedometer installation height, viewed ground
radius, field
of view, and the fraction of the ground-reflected irradiance (
GRI)
received by the albedometer.
measurement network data and data contributed by the PV
industry are available for download from NREL’s DuraMAT
website at https://datahub.duramat.org/project/about/albedo-
study [14]. The user’s guide includes plots of monthly and
hourly albedos to illustrate seasonal and diurnal variations in
albedo for a location.
PV modeling software typically supports monthly values for
albedo, which may change seasonally due to ground moisture
content, snow, or changes in vegetation [8][9]. More detailed
projections include hourly values and are often suggested to
capture performance variations between low sun elevation
angles and diffuse light conditions. Alternatively, long-term
energy predictions may average over seasonal variations
[15][16].
Albedo values used in modeling should be for identically
prepared ground surfaces, including any dust mitigation
schemes.
As a complication, during initial powering of a system the
terrain will usually not reflect expected topography, for
example due to trampled grass or moved flora or ground. Care
should be taken to correct for the projections with on-site
measurements at deployment. Afterwards, keeping
albedometers on site is recommended for continued
performance evaluation of the bifacial system.
Measuring on-site albedo is beneficial for bifacial systems
performance evaluation. Installation of an albedometer
requires a ground surface representative of the site and the
appropriate size for the field of view of the albedometer. Fig. 7
illustrates the relationships between albedometer installation
height, viewed ground radius, field of view, and the fraction of
the ground-reflected irradiance (GRI) received by the
albedometer for uniform isotropic reflection. For an
albedometer installation height of 1.5 m and a manufacturer’s
recommendation that the viewed ground radius be 10 times the
mounting height, the field of view is 170° and 99% of the GRI
is received from within this field of view. IEC 61724-1 is less
stringent with a ±80° or 160° field-of-view recommendation.
Within this field-of-view, the albedometer would receive about
93% of the GRI and the viewed ground radius is reduced from
15 m to 8.3 m. A representative ground surface directly under
the albedometer is also important because 50% of the GRI is
received from within a ground radius equal to the mounting
height.
In some cases, ground surface may not be uniform within a
region conveniently measured by an albedometer. This can
occur for example in systems with albedo-boosting reflector
materials, pavers, or fabrics; in built environments with
complex structures, such as parking shade structures; or in
emerging projects combining agricultural and PV land use. In
these cases, one approach is to use an average value of albedo,
perhaps derived from measurements on a grid.
E. Spectral response
The choice of irradiance sensor may have a significant effect
on the determination of effective irradiance and the accuracy
of performance predictions because of spectrally responsive
albedo. Spectral reflection of ground surfaces can vary
significantly [17], both from one surface to another and also
seasonally due to variations in soil moisture content,
vegetation, snow, and other factors. Therefore, measured rear
irradiance with sensors with different spectral responsivities
than PV devices can provide a range of results. At one site,
measurements with broadband pyranometers showed a 20%
higher response relative to reference cells for eight months of
rear-irradiance measurements [18]. Simulations for a set of
nine representative ground surfaces found that spectral
mismatch relative to a bifacial module was distributed over a
range of ±9.2% for thermopile pyranometers versus only
±3.7% for a typical PV reference cell [19], as shown in Fig. 8.
However, selection of broadband pyranometers versus
reference cells may also be impacted by intended purpose. For
validating rear-irradiance optical models which are spectrally
agnostic, broadband sensor measurements show better model
agreement [20]. However, for assessing rear-side contribution
to module power production, reference cells provide a closer
match to module performance because of their better spectral
match [19].
In modeling, the effect of spectral mismatch effects on rear
irradiance is still being evaluated. Furthermore, for bifacial
tandem cells and modules, a better understanding of the front
and rear spectral response is required [21].
III. PERFORMANCE EQUATIONS
Existing methods and standards for plant performance
assessment, such as capacity tests and performance ratios, need
to be adapted to bifacial systems.
One approach to this adaptation, proposed for capacity tests
in [7], is to modify traditional performance metrics for
monofacial systems, replacing the in-plane front-side
Fig.
8. Influence of spectral albedo.
Variation in albedo spectral mismatch
versus representative bifacial PV module, for PV reference cell and broadband
pyrano
meter, over a range of ground surfaces. [19]
irradiance with an effective irradiance which
includes the contributions of both front- and rear-side POA
irradiance to system power. This effective irradiance can be
written
= + (1)
where is the rear-side plane-of-array irradiance and
is the bifaciality factor of maximum power defined in
IEC 60904-1-2, the ratio between power produced by rear-side
versus front-side irradiance contributions. The term is
obtained from PV module datasheets and should be measured
according to IEC 60904-1-2 or similar methods.
The above definition is chosen to enable module power
output to be written in terms of as
= 1 + ( ) (2)
where is the module’s front-side power rating at the
reference condition (e.g., STC), is the temperature
coefficient of power, is temperature, and is the reference
temperature.
We use the term “effective” irradiance for because it is
quantified in terms of its effect on total module power output,
via eq. (2), and depends on module characteristics, via ,
rather than on irradiance alone. Note that, more generally, eq.
(1) could be elaborated to consider other characteristics that
affect module power output, such as spectral and angular
response, non-uniformity, structural shading, or electrical
mismatch, by incorporating additional correction or mismatch
factors in determination of effective irradiance, according to its
impact in eq. (2).
The draft of IEC 61724-1 Edition 2 reorganizes eq. (1) in the
algebraically equivalent form
=
1 +
(3)
=
(
4)
to define the bifacial irradiance factor BIF representing the
terms in parenthesis in eq. (3). Eq. (4) is equivalent to eq. (1),
with additional nomenclature. Note that is not a constant
but varies according to the ratio
.
Substituting for in performance metrics, such as
capacity tests and performance ratios, allows familiar
monofacial equations for these metrics to be adapted for
bifacial systems. For example, the draft of IEC 61724-1 Ed. 2
substitutes the product for in equations for
performance ratio to obtain bifacial adaptations.
More detailed methods are being developed for more
advanced assessment beyond the performance ratio metric,
such as proposed in [8]. These issues are under a very early
stage of review for bifacial updates to IEC 61724-2 and
61724-3.
Although Eqs. (1)-(4) provide a way to express the combined
effect of front and rear irradiance on system power output, to
use these it is still necessary to determine values for .
Various methods to obtain these data are discussed in the next
section.
IV. APPROACHES TO IRRADIANCE MONITORING
Several different approaches to determining rear-side
irradiance are being discussed in the research literature and
international standards community while being pursued in
practice on early systems. These different approaches address
the challenges summarized in section II while making different
tradeoffs. Here we review four approaches. Irradiance sensor
placements for these approaches are schematically indicated in
Fig. 9, while instrument options for them are listed in Table I
together with potential advantages of different choices within
each instrument category.
Approach 1: Modeling Rear-Side Irradiance
One approach, denoted “Option 1” in the draft of IEC
61724-1 Ed. 2, prescribes that the monitoring system should
measure global horizontal irradiance (GHI), front-side POA,
horizontal albedo and optionally also diffuse irradiance, and
that these data are to be used with an optical model, such as a
view-factor or ray-tracing model, to estimate the rear-side
irradiance. The rear-side irradiance is then used in a
performance model or in performance metrics via Eqs. (1)-(4).
One advantage of this approach is that it uses the optical
model to deal with the complexities of rear-side irradiance non-
uniformities – including within-module, within-row, and row-
to-row uniformities – while keeping the measurement system
relatively simple. The complexity of rear-side irradiance
Fig. 9.
Schematic options for irradiance sensor layout for approaches 1, 2,
3
and 4 discussed in section IV.
Reference module
Diffusometer
(optional)
Away from row edge, avoid posts
Open area, on
representative
ground surface
1
3
4
3
Rear-facing
POAs
2
Front-facing
POAs All
Unobstructed
Front and Rear-Facing
Albedometer
Away from
module
edge
uniformity can be handled at varying levels of detail depending
on the model selection.
Furthermore, this approach allows measured irradiance and
albedo data to be directly compared with pre-construction
resource assessment data when comparing actual performance
to design expectations.
However, since the approach relies on modeling to estimate
rear-side irradiance, it does not provide any detailed
information on the nature of possible deviations of the system
from the model assumptions, which could be necessary for
problem resolution as well as design validation and
improvement. Furthermore, actual albedo may vary throughout
the site due to differences in ground surface; therefore,
albedometers should be placed above representative ground
surfaces maintained identically to the ground surfaces
underneath the modules.
Approach 2: Measuring Rear-Side Irradiance
“Option 2” in the draft of IEC 61724-1 Ed. 2 provides for
directly measuring in-plane rear-side irradiance at multiple
points distributed throughout the array, as shown in Fig. 9.
The relative advantages and disadvantages in comparison to
“Option 1” discussed above are reversed. The rear-side
irradiance (at the measurement points) is directly determined,
allowing a precise comparison with design models and
potentially greater confidence in assessing rear-side power
contribution. In addition, measurement of rear irradiance at
multiple points allows averaging over variations in ground
surface throughout the site. However, rear-side irradiance
cannot be directly compared to any pre-construction resource
data. Also, non-uniformities in rear-side irradiance and
variations in such non-uniformities make sensor placement
critical.
Optimal positioning of rear-side irradiance sensors is not yet
known. The draft of IEC 61724-1 Ed. 2 requires at least three
rear-side irradiance measurement points per met station at the
plant and instructs that these should be placed in representative
locations throughout the plant and arranged to capture the
effects of rear-side non-uniformity, while avoiding row-ends
and unusual shading or reflections. However, the document
does not prescribe specific locations of the sensors along the
rear side. Guidance on this is evolving. Also, three rear-side
measurement points are a minimum requirement in
IEC 61724-1, with a greater number of measurement points
frequently recommended.
For addressing non-uniformity along the module’s width,
[22] suggests locating the rear irradiance sensors along the
module’s collector width at 20% from the edges. This
T
ABLE
I:
M
EASURED
P
ARAMETERS AND
I
NSTRUMENTS
F
OR
E
ACH
A
PPROACH IN
S
ECTION
IV
Approach
Measured
Parameter
Instruments Advantages Notes
All
Global Horizontal
Irradiance
Front-Facing
Plane of Array
Irradiance
• Broadband pyranometer • Connection to satellite &
historical resource data
• Reference cells
• Cost
• Spectral, angular, time
response matching PV
modules
1, 3
Albedo
• Broadband albedometers
• Connection to broadband
tabulated albedo data,
models
• Locate albedometers near the array with unobstructed
view over similar ground type to the array
• Reference cell albedometers
• Cost
• Spectral match
Diffuse Horizontal
Irradiance
(optional)
• Tracked pyrheliometer or
shaded pyranometer
• Rotating shadow-band
• Accuracy
• Directly measure DHI, or measure DNI and GHI,
from which DHI is calculated
• Photodiode array with
patterned shade
• Reference cell array with
varying orientations
• Cost
2, 3, 4 Rear
Irradiance
• Broadband pyranometers
• Many models predict
broadband irradiance
•
Avoid edge effects: 5m from row edges; 20% distance
module edges
• Avoid module shading by sensors
• Keep sensors near module plane; avoid obstructions
• Reference cells
• Cost
• Spectral, angular, time
response matching PV
modules
4 Effective
Irradiance
• Bifacial reference module,
measured at Pmax (preferred)
or Isc
• Measure impact of non-
uniform shading
• Directly measures effective irradiance, eq. (1)-(4)
All
Module
Temperature
• Back-of-module temperature
sensors
• Take care not to shade bifacial rear side; sensor and
tapes must cover <10% of any cell’s area
placement provides the most similar measurement to the
single-value average rear irradiance models. To avoid edge
effects, modeling has shown that a placement on center rows at
least 5 meters from edges is ideal [23].
The large number of rear-side irradiance sensors required for
this approach may favor the use of lower-cost irradiance
sensors.
The ratio of co-located front and rear irradiance sensors has
also been suggested for evaluating with more detail the real-
time optical bifacial gains of a system [24].
Approach 3: Measuring Unobstructed Rear Irradiance
Since the non-uniformity of irradiance on the rear of the
array with both time and position complicates the mapping of
point measurements to the average rear-side irradiance, this
approach intentionally moves the measurement location to the
end of the tracker/row where measurements have fewer near-
field obstructions. This option means the raw measurements
are more stable but are biased relative to the spatially averaged
rear irradiance available to the array, so the bias must be
corrected for using a model. In practice, the model used for
performance verification is adapted to produce estimates of the
irradiance in this exposed location rather than averaged across
the rear of the array, and the performance target is derived
relative to this proxy measurement.
As depicted in Fig. 9, a rear-facing irradiance sensor located
outboard from the rear of the array is exposed to a relatively
unobstructed rear irradiance measurement. This is not
physically representative of the light reaching the rear POA,
but it is a stable location for measurement comparable with
typical front-side POA measurements. It avoids impacts from
diffuse shading on the array as well as the speckled, time-
varying shade patterns that occur across the rear of the array
even before structural shading is considered.
Since the irradiance components necessary to model average
rear irradiance based on an unobstructed sensor reading are not
easily available, use of this approach requires that the
performance expectation be formulated relative to the
unobstructed sensor rather than the modeled average input
irradiance. On a fixed-tilt array modeled in PVsyst, the
expected sensor input may be extracted by adding the structural
obstruction term BackShd to the computed average
irradiance reaching the surface GlobBak. For a horizontal
north-south-axis single-axis tracker, a location that does not
change height would be along the torque tube axis, and
mounting at the equatorial end will minimize the amount of
shade on the ground for the most stable measurements. The
fractions of the field of view occupied by shaded and unshaded
ground can be calculated and used to adapt the PVsyst
intermediate rear irradiance components obtained using
relevant (measured) ground albedo to predict the irradiance at
the sensor location.
This approach combines the benefits of the two previously
discussed approaches – while also requiring more irradiance
sensors since both sets of measurements are performed.
However, it minimizes the impact of measurement errors
arising from rear-side time-varying shade patterns.
Approach 4: Bifacial Reference Devices
An alternative to separately measuring both front-side and
rear-side in-plane irradiance is to use a bifacial reference
device that simultaneously measures both quantities, as
described in the draft of IEC 61724-1 Ed. 2. Such a bifacial
reference device can be either a bifacial cell or a full-size
bifacial module connected to an I-V measurement system. It
should have identical response as the modules that are to be
monitored, especially including identical bifaciality.
Bifacial reference cells, especially cells matched to a
particular bifacial module, are not generally commercially
available and must be supplied by the module manufacture.
Full-size bifacial modules identical to those in the array can
be used, advantageously exactly matching module properties.
A significant benefit of using a bifacial reference module is
that it intrinsically captures within-module rear-side irradiance
non-uniformity, including shading effects that arise from the
racking structure. If the module’s power is measured, not just
its short-circuit current, the result also provides the impact of
irradiance non-uniformity on power production due to
electrical mismatch.
With currently available in-field PV module I-V tracing
units, use of a reference module takes up an otherwise
productive slot in the array. However, automated I-V tracers
that can isolate a single module from a string to measure an I-V
curve and then switch back to regular operation are being
developed and tested.
Due to the current lack of field calibration methods, bifacial
reference modules may require calibration in a laboratory
according to IEC 60904-1-2, and this step adds cost and
logistical difficulties during construction, unless such modules
are available as part of the due-diligence activities prior to
construction.
A potential method for field calibration of a bifacial
reference module comprises calibrating the module against a
front-side POA reference cell while the module rear side is
covered and using eq. (2) to determine a value for .
Subsequently, after uncovering the module rear side, module
power is used to measure effective irradiance . The rear-
side contribution is estimated using the module bifaciality,
which can be taken from the module datasheet or possibly
determined by repeating the field calibration with the module
mounted with its rear side facing front. Field calibration
methods still require more development and validation.
V. SELECTING AN APPROACH
Selection of one of the measurement approaches described
in section IV will depend on goals for uncertainty and cost as
well as the user’s application. Additionally, the user should
consider whether their objective is to measure only inputs
(solar resource) and outputs (power), or also intermediate
variables (rear-side irradiance).
The uncertainty of the measurement approach depends on
both the instrumental uncertainty of the selected sensors and
systematic uncertainties introduced by the many challenges
discussed in section II. It is not yet possible to provide a general
uncertainty estimate, but some examples are instructive. In one
case analyzed by one of the authors, instrumental uncertainties
involved in measuring rear-side irradiance on a single-axis
tracking system were found to contribute 11.5% uncertainty in
the predicted rear-side contribution, corresponding to ~2.3%
uncertainty in total output; however, systematic uncertainties,
for example arising from irradiance non-uniformity, were not
estimated. Systematic uncertainties can be significant. In
another case analyzed by one of the authors, a single-axis
tracker system was found to produce almost 15% more energy
than expected when the performance was predicted using only
the overall incident solar resource and albedo data without
measuring any rear irradiance quantities.
From these examples, we propose that the uncertainty of
approach 1 in section IV, in which rear irradiance is not
measured, should be considered greater than the uncertainty of
approaches 2, 3, or 4.
The different approaches have different requirements for
modeling. For system assessment relative to contractual
performance guarantees, it is common to compare measured
performance with expected performance for the measured
conditions. For this purpose, it should be noted that approaches
1 and 3 require access to an irradiance model to estimate rear
irradiance from other measured quantities, while approaches 2
and 4 can be used without an irradiance model since they
directly measure the rear-side contributions. These factors may
influence the selection of a measurement approach.
REFERENCES
[1] T. S. Liang et al., “A review of crystalline silicon bifacial
photovoltaic performance characterisation and simulation,” Energy
and Environmental Science, vol. 12, no. 1. Royal Society of
Chemistry, pp. 116–148, Jan. 01, 2019, doi: 10.1039/c8ee02184h.
[2] C. W. Hansen et al., “Analysis of irradiance models for bifacial PV
modules,” in 2017 IEEE 44th Photovoltaic Specialist Conference,
PVSC 2017, 2017, pp. 1–6, doi: 10.1109/PVSC.2017.8366322.
[3] C. Deline, S. Macalpine, B. Marion, F. Toor, A. Asgharzadeh, and J.
S. Stein, “Assessment of Bifacial Photovoltaic Module Power Rating
Methodologies-Inside and Out,” IEEE J. Photovoltaics, vol. 7, no. 2,
pp. 575–580, 2017, doi: 10.1109/JPHOTOV.2017.2650565.
[4] S. A. Pelaez, C. Deline, S. M. Macalpine, B. Marion, J. S. Stein, and
R. K. Kostuk, “Comparison of Bifacial Solar Irradiance Model
Predictions with Field Validation,” IEEE J. Photovoltaics, vol. 9, no.
1, pp. 82–88, 2019, doi: 10.1109/JPHOTOV.2018.2877000.
[5] E. Ozkalay, J. Lopez-Garcia, L. Pinero-Prieto, A. Gracia-Amillo,
and R. P. Kenny, “Evaluation of the non-uniformity of rear-side
irradiance in outdoor mounted bifacial silicon PV modules,” in AIP
Conference Proceedings, Aug. 2019, vol. 2147, no. 1, p. 020011,
doi: 10.1063/1.5123816.
[6] C. Monokroussos et al., “Rear-side spectral irradiance at 1 sun and
application to bifacial module power rating,” Prog. Photovoltaics
Res. Appl., vol. 28, no. 8, pp. 755–766, Aug. 2020, doi:
10.1002/pip.3268.
[7] M. Waters, C. A. Deline, J. Kemnitz, and J. Webber, “Suggested
Modifications for Bifacial Capacity Testing,” in 2019 IEEE 46th
Photovoltaic Specialists Conference (PVSC), 2019, pp. 3519–3524,
doi: 10.1109/PVSC40753.2019.9198974.
[8] B. Marion, “Ground Albedo Measurements and Modeling Bill
Marion 2018 Bifacial PV Workshop • Publication number or
conference,” 2018, Accessed: Mar. 23, 2021. [Online]. Available:
https://pubs.er.usgs.gov/publication/ds1035.
[9] B. Marion, “Albedo from satellite and ground-based observations,”
2019.
[10] G. Maclaurin, M. Sengupta, Y. Xie, and N. Gilroy, “Development of
a MODIS-Derived Surface Albedo Data Set: An Improved Model
Input for Processing the NSRDB,” 2016, Accessed: May 10, 2021.
[Online]. Available: https://www.osti.gov/biblio/1335471.
[11] M. Sengupta, Y. Xie, A. Lopez, A. Habte, G. Maclaurin, and J.
Shelby, “The National Solar Radiation Data Base (NSRDB),”
Renewable and Sustainable Energy Reviews, vol. 89. Elsevier Ltd,
pp. 51–60, Jun. 01, 2018, doi: 10.1016/j.rser.2018.03.003.
[12] M. Sengupta, A. Habte, S. Wilbert, C. Gueymard, and J. Remund,
“Best Practices Handbook for the Collection and Use of Solar
Resource Data for Solar Energy Applications: Third Edition,”
Golden, CO (United States), Apr. 2021. doi: 10.2172/1778700.
[13] “Albedo (1 month) | NASA,” May 2021.
[14] B. Marion, “Measured and satellite-derived albedo data for
estimating bifacial photovoltaic system performance,” Sol. Energy,
vol. 215, pp. 321–327, Feb. 2021, doi:
10.1016/j.solener.2020.12.050.
[15] M. T. Patel, M. Ryyan Khan, A. Alnuaimi, O. Albadwawwi, J. J.
John, and M. A. Alam, “Implications of Seasonal and Spatial Albedo
Variation on the Energy Output of Bifacial Solar Farms: A Global
Perspective,” in Conference Record of the IEEE Photovoltaic
Specialists Conference, Jun. 2019, pp. 2264–2267, doi:
10.1109/PVSC40753.2019.8981163.
[16] R. Bailey, P. Keelin, R. Perez, J. Robinson, G. Bender, and J. Chard,
“Investigations of Site-Specific, Long Term Average Albedo
Determination for Accurate Bifacial System Energy Modeling,” in
Conference Record of the IEEE Photovoltaic Specialists Conference,
Jun. 2019, pp. 2268–2274, doi: 10.1109/PVSC40753.2019.8980744.
[17] T. C. R. Russell, R. Saive, A. Augusto, S. G. Bowden, and H. A.
Atwater, “The Influence of Spectral Albedo on Bifacial Solar Cells:
A Theoretical and Experimental Study,” IEEE J. Photovoltaics, vol.
7, no. 6, pp. 1611–1618, 2017, doi:
10.1109/JPHOTOV.2017.2756068.
[18] S. Ayala Pelaez et al., “Ultimate bifacial showdown: 75kW field
results,” in 7th bifiPV (virtual), 2020, pp. 1–35.
[19] M. Gostein, B. Marion, and B. Stueve, “Spectral Effects in Albedo
and Rearside Irradiance Measurement for Bifacial Performance
Estimation,” in 2020 47th IEEE Photovoltaic Specialists Conference
(PVSC), Jan. 2020, pp. 0515–0519, doi:
10.1109/pvsc45281.2020.9300518.
[20] M. Monarch and S. Ayala Pelaez, “Analysis and Validation of
Spectral Irradiance Simulations for Methodology Results
Conclusion and Future Work,” (in progress).
[21] A. Onno et al., “Predicted Power Output of Silicon-Based Bifacial
Tandem Photovoltaic Systems Predicted Power Output of Silicon-
Based Bifacial Tandem Photovoltaic Systems,” Joule, pp. 1–17, doi:
10.1016/j.joule.2019.12.017.
[22] S. Ayala Pelaez, C. Deline, J. S. Stein, B. Marion, K. Anderson, and
M. Muller, “Effect of torque-tube parameters on rear-irradiance and
rear-shading loss for bifacial PV performance on single-axis tracking
systems,” 2019.
[23] S. Ayala Pelaez, C. Deline, P. Greenberg, J. S. Stein, and R. K.
Kostuk, “Model and validation of single-axis tracking with bifacial
PV,” IEEE J. Photovoltaics, vol. 9, no. 3, pp. 715–721, 2019, doi:
10.1109/JPHOTOV.2019.2892872.
[24] S. Ayala Pelaez et al., “Field-Array Benchmark of Commercial
Bifacial PV Technologies with Publicly Available Data,” in 47th
IEEE PVSC Proceedings, 2020, pp. 1757–1759, doi:
10.1109/PVSC45281.2020.9300379.
View publication stats
