NREL/SR-520-23942 • UC Category: 1270

Photovoltaic Mg.dule Energy

Rating Procedure
Final Subcontract Report

C.M. Whitaker and J.D. Newmiller

Endecon Engineering
San Ramon, California

NREL technical monitor: B. Kroposki

National Renewable Energy Laboratory

1617 Cole Boulevard
Golden, Colorado 80401-3393
A national laboratory of
the U.S. Department ofEnergy
Managed by Midwest Research Institute
for the U.S. Department ofEnergy
under Contract No. DE-AC36-83CH10093

Prepared under Subcontract No. AAl-4-14192-01

January 1998
 This publication was reproduced from the best available camera-ready copy
submitted by the subcontractor and received no editorial review at NREL.

NOTICE

This report was prepared as an account of work sponsored by an agency of the United States government.
Neither the United States government nor any agency thereof, nor any of their employees, makes any
warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness,
or usefulness of any infonnation, apparatus, product, or process disclosed, or represents that its use would
not infringe privately owned rights. Reference herein to any specific commercial product, process, or service
by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United States government or any agency thereof. The views
and opinions of authors expressed herein do not necessarily state or reflect those of the United States
government or any agency thereof.

Available to DOE and DOE contractors from:

Office of Scientific and Technical Information ( OSTI)
P.O. Box62
Oak Ridge, TN 37831
Prices available by calling (423) 576-8401

Available to the public from:

National Technical information Service (NTIS)
U.S. Department of Commerce
5285 Port Royal Road
Springfield, VA 22161
(703) 487-4650

,�
t• .: Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste
A cknowledgments

The authors wish to express their gratitude to all of the individuals who provided ideas,
support, and guidance to this project. The following people deserve special recognition.
Benjamin Kroposki of NREL performed most of the outdoor measurements, compiled
data from a number of sources, made sure the data got to us, and took the torch at the
end of the project. Daryl Myers of NREL provided irradiance model code, evaluated
the NSRDB data and selected the reference days. Keith Emery of NREL oversaw most
of the indoor measurements and provided code and guidance for the spectral and
incidence angle models. Laxrni Mrig of NREL was the project manager and TRC
Chairman. Jerry Anderson of Sunset Technologies provided one of the performance
models as well as a number of great ideas that have been incorporated in this
procedure. Tim Townsend of Endecon Engineering and Howard Wenger of Pacific
Energy Group also provided a good deal of useful input and feedback.

We would also like to thank the members of the Technical Review Committee for their
interest, knowledge, guidance, and patience. The members of the TRC were:

Scott Albright Brian Farmer Laxmi Mrig

Jerry Anderson Vahan Garboushian Daryl Myers
Gobind Atrnaram Subhendu Guha Mark O'Neill
Moneer Azzam Robert Hammond Don Osborn
Joe Burdick Mandy Herner Carl Osterwald
Eric Daniels Steve Hogan Miles Russell
Alan Delahoy Neil Kaminar Mike Stern
Steve Durand David King Chuck Whitaker
Pete Eckert Ben Kroposki John Wohlgemuth
Sam Edwards Leslie Libby Teddy Zhou
Keith Emery Alex Maish
 Table of Contents
Section Page

1. Introduction .... ... . . .. . ... . ... . . ...

. . . . . .. . . . . . . ... .... ... . . . ... .. . .
.... ... . . .. . . . .. . . ....... . . .. . ...... ...
. . .. . . .... .. 1..

1.1 Background ........................................................................................................... 1

1 .2 Approach ............................................................................................................... 3

2. Module Energy Rating Computation .... ... .

.. .......... . . ..
... . ......... ... . .. . .. . . . . ..... ..
. .... ..
. . 4
2.1· Algorithms ............................................................................................................. 5
2.1.1 Energy Computation ................................................................................... 5
2.1.2 DC Power Model .......................................................................................... 6
2.1.2.1 Linear Irradiance-Only (Myers) Model ................................................. 6
2.1.2.2Interpolation Model .................................................................................. 7
2.1.2.3 Anderson Model ....................................................................................... 8
2.1.2.4Blaesser Model. .......... ...
. . ........ ..
. ................ . .......................... . .................... 9
2.1.2.5Lumped Four Parameter (L4P) Model.. ............................................... 11
2.1.3 Spectral Model ............................................................................................ 14
2.1.3.1 Effective Irradiance ................................................................................. 15
2.1.3.2Normalization and Incident Surface Translation ............................... 16
2.1.3.3Estimated Spectrum with Cloud-Cover .............................................. 17
2.1.3.4Estimated Clear Sky Spectrum .............................................................. 20
2.1.4 Module Thermal Model ............................................................................ 26
2.1.4.1 General Approach................................................................................... 27
2.1.4.2Module INOCT (°C) ............................................................................... 31
2.1.4.3Module and Anemometer Heights (m) ............................................... 32
2.1.4.4Irradiance (W /m2) 32
•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••

2.1 .4.5Ambient Temperature (°C) .................................................................... 32

2.1.4.6Wind Speed (m/s) ................................................................................... 32
2.1.4.7Date and Time ......................................................................................... 32
2.1.4.8Limitations ............................................................................................... 32
2.1.5 Optical Model ............................................................................................. 33
2.1.6 Irradiance Model ........................................................................................ 34
2.1.6.1 True Local Solar Time ............................................................................ 35
2.1.6.2Local Solar Position ................................................................................ 35

October 1997 Module Energy Rating Methodology

 Table of Contents (cont.)
Section Page
2.1.6.3Plane-of-Array Incidence Angle ........................................................... 36
2.1.6.4Plane-of-Array Incident Beam and Diffuse Irradiance ...................... 36
2.2 Required Input Data .......................................................................................... 38
2.2.1 Module Characteristics .............................................................................. 38
2.2.2 Evaluation Characteristics ........................................................................ 40
2.2.2.1 Module Load Type .................................................................................. 40
2.2.2.2Location .................................................................................................... 40
2.2.2.3 Environmental Data ................................................................................ 40

3. Module Characterization Procedures ................................................................ 42

3.1 Module Testing ................................................................................................... 42

3.1.1 Power Model ............................................................................................... 42
3.1.1.1 Laboratory Testing (Anderson, Blaesser, Interpolation) ................... 42
3.1.1.20utdoor Testing (Blaesser, Myers) ....................................................... 44
3.1.2 Spectral Model ............................................................................................ 45
3.1.3 Thermal Model ........................................................................................... 45
3.1 .4 Optical Model ............................................................................................. 45
3.2 Data Reduction ................................................................................................... 46
3.2.1 Power Model ............................................................................................... 46
3.2.1.1Anderson Model ..................................................................................... 46
3.2.1.2Myers Model ............................................................................................ 47
3.2.1.3Interpolation Model ................................................................................ 48
3.2.1.4Blaesser Model. ........................................................................................ 48
3.2.2 Spectral Model ............................................................................................ 48
3.2.3 Thermal Model ........................................................................................... 48
3.2.4 Optical Model ............................................................................................. 49

4. Weather Data .......................................................................................................... 50

4.1 Hot Sunny: Phoenix, AZ, June 15, 1976 .......................................................... 53

4.2 Cold Sunny: Alamosa, CO, February 11, 1961 ............................................... 54
4.3 Hot Cloudy: Brownsville, TX, July 4, 1983 ..................................................... 55
4.4 Cold Cloudy: Buffalo, NY, December 6, 1985 ................................................ 56

Module Energy Rating Methodology ii

 Table of Contents (cont.)
Section Page
4.5 NICE: Sacramento, CA, May 4, 1967 ............................................................... 57

5. Results ..................................................................................................................... 58

5.1 Testing .................................................................................................................. 58

5.1.1 Laboratory Testing ..................................................................................... 59
5.1.2 Outdoor Testing ......................................................................................... 62
5.2 Validation Analysis ............................................................................................ 65
5.2.1 Power Models ............................................................................................. 65
5.2.1.1Anderson Model ..................................................................................... 74
5.2.1.2Blaesser Model. ........................................................................................ 75
5.2.1 .3 Interpolation Model ................................................................................ 75
5.2.1.4Myers Model ............................................................................................ 75
5.2.2 Spectral Model ............................................................................................ 76
5.2.3 Thermal Model ........................................................................................... 81
5.2 .4 Reflection Model.
........................................................................................ 86
5.2.5 Irradiance Model ........................................................................................ 86
5.3 Module Energy Rating Results ......................................................................... 86

6. Conclusions and Recommendations . .. ........................................................ . ..... 91

6.1 Testing .................................................................................................................. 92

6.1.1 Indoor Tests ................................................................................................. 92
6.1.2 Outdoor Tests .............................................................................................. 93
6.2 Validation Analysis ............................................................................................ 93
6.2.1 Power Models ............................................................................................. 93
6.2.1.1 Anderson Model ..................................................................................... 94
6.2.1 .2Blaesser Model. ........................................................................................ 94
6.2.1.3Interpolation Model ................................................................................ 95
6.2.1 .4Myers Model ............................................................................................ 95
6.2.2 Thermal Model ........................................................................................... 95
6.2.3 Spectral Model ............................................................................................ 95
6.2.4 Optical Model ............................................................................................. 96
6.2.5 Irradiance Model ........................................................................................ 97

Module Energy Rating Methodology 111

 Table of Contents (cont.)
Section Page
6.3 MER Calculations .............. . . . . .. . . . . . .
. . . .. ... .. . . . ... .................. . ... .. ..
.. . . .. . ... . ................ 97

7. References ............................................................................................................... 98

Appendix A Indoor Module Testing Data ............................................................ 100

Appendix B Additional MER Power Model Validation and

Comparison Plots ........................................................................................................ 102

Module Energy Rating Methodology iv

1. INTRODUCTION

In May 1994, the National Renewable Energy Laboratory (NREL) initiated an effort to
develop a consensus-based approach to rating photovoltaic modules. This new
approach was intended to address the limitations of the de-facto standard module
power rating at Standard Test Conditions (STC). Using technical input from a number
of sources and under the guidance of an industry-based Technical Review Committee,
the approach described in this document was developed.

This document describes testing and computation procedures used to generate a

photovoltaic Module Energy Rating (MER). The MER consists of 10 estimates the
amount of energy a single module of a particular type (make and model) will produce
in one day. Module energy values are calculated for each of 5 different sets of weather
conditions (defined by location and date), and 2 load-types. Since reproduction of these
exact testing conditions in the field or laboratory is not feasible, limited testing and
modeling procedures and assumptions are specified.

1.1 Backg round

Presently, manufacturers supply a module's rating (power, open circuit voltage, short
circuit current, peak power voltage and current) at STC, the module's Nominal
Operating Cell Temperature (NOCT2), and possibly voltage and current temperature
coefficients. With these parameters, a user can translate the module rating to another
set of conditions. Translation accuracy is strongly dependent on the translation range
and on equation complexity.

Pacific Gas and Electric Company and others found that most PV modules mounted
outdoors rarely, if ever, produce their rated power. This discrepancy is due, in large
part, to the fact that under high irradiance conditions (>500 W /m2), PV modules
typically operate much hotter than the 25°C cell temperature specified by STC. Based
on this experience, systems procured by the Photovoltaics for Utility Scale Applications
(PVUSA) project are rated at PVUSA Test Conditions (PTC3), which are more indicative

1 STC: 1000 W /m2 irradiance with an Air Mass 1.5 spectrum and 25°C cell temperature.

2 NOCT: 800 W /m2 irradiance, 1 m/s wind speed, 20°C ambient temperature, single module, open
circuited.
3 PTC: 1000 W /m2irradiance, 1 m/s wind speed, and 20°C ambient temperature.

Module Energy Rating Methodology 1

than STC of peak performance conditions in many other locations. By specifying
ambient conditions rather than module conditions, PTC also allows for a more realistic
comparison of modules and array designs with different thermal characteristics.

While PTC is a step in the right direction, a single point rating does not account for
variations in performance with changing conditions. Specifically, different technologies
have different temperature coefficients (change of electrical output due to changes in
device temperature), thermal characteristics (change in device temperature due to
changes in ambient conditions), spectral response characteristics, and solar angle of
incidence response characteristics.

The need for something beyond an STC power rating goes back at least 15 years to
Gay's [1,2] AM/PM approach. Reference [2 ] accurately describes the issues that need to
be addressed. AM/PM is simply the module energy produced for a standard day as
defined by profiles of irradiance, ambient temperature, and air mass. Through
AM/PM, Gay was trying to characterize the module's thermal response to ambient
conditions and its power production as a function of light intensity, spectral content and
module temperature.

Ideally, a performance rating should allow the user to compare not only similar
products from different manufacturers, but different technologies as well, and should
provide a realistic performance measure for the installation region. The ideal rating
would provide values that could be used by designers to quickly generate first-cut
system designs. The information necessary to convey the module rating would be
simple enough to be included on the module's label. The procedure for generating the
rating would be well defined and repeatable. Because the rating predicts performance
for real operating conditions, the industry would be compelled to optimize their cell
and module designs for real conditions rather than for STC. However, because the
conditions really are representative, the rating would not unduly tax the industry.
Rather, it might show that one technology works best in one climate region, but not as
well in another region. Finally, because the rating accurately describes a module's
performance, the method would be unanimously accepted by the general PV
community including manufacturers, installers, users, and researchers. In summary, the
rating should:
• provide relative comparison of different manufacturers and technologies
• provide realistic performance for the contemplated site

Module Energy Rating Methodology 2

 • be easy to use, simple to convey, and accurate
• have the blessing_ of the PV community

1.2 A pproach

As stated above, the module energy rating is based on modeled performance under give
sets of weather conditions and two load types.

The five sets of weather conditions provide the basic range of environmental conditions
anticipated for typical uses of PV modules in the US, and allow for comparison of
module performance under a wide variety of conditions. The National Solar Resource
Database (NSRDB) was selected as the source of the weather data. By doing so, entire
daily weather profiles can be specified simply by indicating a city name and a location.

Two load types are assumed, corresponding to the two most common loads connected
to PV modules: maximum power tracking for grid-tied applications and fixed voltage
for battery charging.

For purposes of rating comparison, a 14.4-Volt (V) operating voltage per battery (2.4
V /battery-cell) is assumed. The 14.4 V value may be divided by the recommended
number of modules and multiplied by the recommended number of batteries to obtain
a fixed voltage for purposes of rating. If the manufacturer does not recommend this
module for battery application, then the modules need not be rated for fixed voltage. It
is assumed that no charge regulation occurs; that is, the module operates at 14.4 V
whenever there is sufficient sun.

The information presented here is the results of the combined efforts of Endecon
Engineering, the National Renewable Energy Laboratory, and Sunset Technology, as
well as the members of the Technical Review Committee (TRC). The 35 TRC members
represent PV manufacturers, system integrators, government and academic researchers
and industry consultants. This group guided the technical and philosophical approach
and reviewed results.

The following sections describe the MER computations and the corresponding module
testing procedures to obtain the module-specific input data.

Module Energy Rating Methodology 3

2. MODULE ENERGY RATING COMPUTATION

The Module Energy Rating consists of measuring module characteristics, defining sets
of weather and load conditions, and estimating module performance under those
conditions. This section describes the last of these three processes. The selection of
performance estimating (modeling) tools must be completed first as they define which
weather parameters and which module characteristics are necessary.

Based on the experience of the project team and the guidance of the TRC, five modeling
areas were considered essential to adequately estimate module performance:

1) Irradiance: estimating the magnitudes of the various components of sunlight

striking the module

2) Thermal: estimating the module temperature based on module characteristics

and weather conditions

3) Spectral: estimating the response of the module to changes in the solar

spectrum

4) Optical: accounting for optical effects, such as reflection loss at high angles of
solar incidence

5) Power: estimating module output based on a characterization of the module,

and the defined weather data and the results of the other models

Figure 2-1 shows conceptually the flow of data and the relationships between the
various models. We will start at the right side of the diagram and work our way to the
left. First, the algorithms used in each of the computational process (the circles in
Figure 2-1) are described. Next, the input variables (the rectangles in the figure)
required for each process are presented.

Module Energy Rating Methodology 4

 Module Module Module
Orientation Optical Spectral Power Model
Response Response Coefficients

Weather
Data

Module
Thermal Load
Characteristics

Figure 2-1 Model Flow Diagram

2.1 A lgorithms

This section discusses the computation processes shown in Figure 2-1. The model
inputs are described in the next section.

2.1.1 E nergy Computation

Energy is computed from the daily power production curves by numerical integration
according to the following formula (Euler integration):
n

E =llt· L� (2-1)
i=l

where
E Module output energy (Wh)
Llt = Data sampling interval (hours)
Pi Power at the ith sample time (W)

Module Energy Rating Methodology 5

This method is as accurate for daily power integration as trapezoidal integration would
be, because the power at the beginning and end of the interval (midnight) is zero.

2.1.2 DC Power Model

The p ower output of the module is computed from the plane-of-array irradiance
(adjusted for spectral and incidence-angle effects), the module temperature, and
appropriate coefficients. Ultimately, this methodology will specify only one power
model. Each model is required to estimate power output for maximum power
operation and ampere-hours for fixed voltage operating conditions. Five models have
been investigated: linear (Myers), interpolation, Anderson, Blaesser, and the lumped
four-parameter model.

The first model is a simple linear-fit model that only depends on irradiance. This model
is included to help estimate the value of the refinements used in the other power and
minor effects models.

The second model uses the raw data supplied by the module characterization process (a
matrix of IV curves dependent on irradiance and temperature) and extracts appropriate
maximum power and fixed voltage points for each irradiance/temperature
combination. The resulting two tables are used to linearly interpolate the power
behavior as desired.

The third and fourth models use translation models (developed by J. Anderson and G.
Blaesser, respectively) to translate an IV curve taken at reference irradiance and
temperature conditions to the actual conditions.

The fifth model is the only analytical model that was investigated. Suggested by T.
Townsend, this model described the module as a single, lumped diode and a series
resistance.

The following sections describe these models in more detail.

2.1.2.1 Linear lrradiance-Only (Myers) Model

This model was proposed by D. Myers of NREL. Myers noted that the errors inherent
"in the measurement of the input parameters to various power models might be larger
than the magnitude of the contributions of any parameters other than irradiance. His
proposal is to use the following equation:

Module Energy Rating Methodology 6

 P=a·H+b (2-2)
where
P = power out of module (W)
H = Plane-of-array irradiance (WI m2)
a, b = Least-squares fit coefficients (m2, W)

using field measured data as the basis for determining the a and b coefficients. Field
measured data must be used because the effect of irradiance on module temperature in
field conditions must be included in the coefficients for this equation to work best.

Note that this model does not predict the shape of the IV curve at all. Therefore, this
model will only be compared to the other models for the maximum power case.

2.1.2.2 Interpolation Model

Determining model coefficients for a module often requires data from an

irradiance/ temperature cross-sensitivity test where module IV curves are measured
under various controlled irradiance and temperature conditions. Rather than
calculating performance coefficients, the interpolation model uses the sensitivity data
directly. It does not attempt to simplify or fit any particular equation to the data; rather,
it uses the actual maximum power and fixed voltage points from the test as a table to
interpolate into. The interpolation scheme used is bilinear interpolation, as described in
[Press, et. al., 1986]. Assuming the tabular data for both the maximum power and the
fixed voltage cases are obtained at the same combinations of irradiance and
temperature, the equations are:

pmax (H,Tm)=(1- t) (1-u)· pmax(j,k)

·

+ t. (1-u). pfv(J+I,k)
(2-3)
+ t · pmax(J+l,k+l)
U •

+ (1- t) · U •
pmax(J,k+l)

I fv (H, Tm) = (1- t)· (1-u) · 1 fvCJ,k)

+ t. (1-u) . J fvCJ+l,k)
+ . f fv(j+I,k+I)
t U
(2-4)
·

+ (1- t) f fv(j,k+I)
· U ·

Module Energy Rating Methodology 7

 (2-5)

(2-6)

where
H = the irradiance at which the output power is desired
Tm = the module temperature at which the output power is desired
pmax(n,m) = the maximum power output corresponding to the nth value of
irradiance and the mth value of temperature
[fo(n,m) = the fixed voltage current output corresponding to the nth value of
irradiance and the mth value of temperature
J = the index of the largest tested irradiance value less than H
k = the index of the largest tested temperature value less than Tm
t,u = H and Tm interpolation grid fraction, respectively
2.1.2.3 Anderson Model

Jerry Anderson of Sunset Technology has offered the following IV curve translation
equations as modifications to those under development in IEEE SCC21 [Anderson,
1994].

(2-7)

(2-8)
Isc2 = Hz/H l
Iscl 1+a(T1 -T2) (2-9)
\1oc2 - 1
l!ocl [ 1+ fJ( Tl -rz)] [1+ 5ln(H l /Hz )]
---

X
(2-10)
where
I = current (A)
H = irradiance (W /m2)

Module Energy Rating Methodology 8

 v = voltage (V)
T = module temperature, (C)
a = temperature coefficient for current (C1)
J3 = temperature coefficient for voltage (C1)
0 = irradiance coefficient for voltage (unitless)

and where subscripts

sc = short circuit
oc = open circuit
1,2 = at conditions 1 or 2
1 = ith point on the IV curve

Coefficients a, J3, and o are determined experimentally, and are normalized at

condition 2. Cell or module temperature is needed, as no conversion from ambient
conditions is provided. Each point on an IV curve measured at conditions 1 are
translated to conditions 2 using equations (2-7) and (2-8). Short circuit current and open
circuit voltage are converted directly with equations (2-9) and (2-10); peak power is
found by applying the translations to the maximum power point of the reference IV
curve.

Note that the equations are not symmetric; that is, if the reference and target conditions
are exchanged in the equations, the appropriate values for the coefficients will be
different. Anderson argues that the difference between these two versions of the
coefficients is small ( <1%) for reasonable translations.

2.1.2.4 Blaesser Model

Gerd Blaesser of the European Solar Test Institute (ESTI) in Ispra Italy, introduced a
slightly different approach [Blaesser, 1995] than Anderson. The basic equations are:

Ir = I(HI,r J HI ) (2-11)
Vr=V +DV (2-12)
ir=Jr/Isc,r' i=Iji5,1
v r=vrfVoc,r I
v=vfVoc Dv = DVfVoc,r
I
(2-13)
(2-14)

Module Energy Rating Methodology 9

 i= i r
(2-15)
v ( v,-Dv) /( 1-Dv)
= (2-16)
Dv = a·ln(HI IHI ) + b( Tamb - TI ) +c. HI
,r
(2-17)

(2-18)

(2-19)

where
I = current, A
= normalized current
H = irradiance, WI m2
v = voltage, V
v = normalized voltage
a, b, c = coefficients
T = temperature (ambient or cell), C
FF = fill factor
p = power, W

and where subscripts

I = in-plane (for irradiance)
r = at reference conditions
amb = ambient (temperature)
m = at maximum power point

and no subscript implies measured values.

Like a, jJ, and Jin the Anderson equations, coefficients a, b, and care determined
experimentally (default values: a = 0.06, b = 0.004°C1 , and c=0.12 m2 /kW) and equations
(2-17), (2-16), (2-18), (2-11), and (2-14) are used to calculate Dv, vmt FF, Isc, and Vocat the
new conditions given values of V0c, Vm' ISCI and Im at some initial conditions. A new peak
power may be computed using (2-19).

Module Energy Rating Methodology 10

Specifically, the maximum power may be computed using:

Pm = Im,r · :I,rI (vm,r

·
- Dv · Voc,r ) (2-20)

The fixed voltage current may be computed by translating the entire reference IV curve
to the specified conditions and interpolating, but this process is computation intensive.
A more computationally efficient alternative is to translate the fixed voltage from the
specified conditions to the reference conditions, interpolate to find the corresponding
reference current, and translating that value back to the specified conditions. That is,

HI
I
fv -Ifv,r
-
·­
(2-21)
HI,r

I
_
I (
= j+l,r - j,r Vtv,r - VJ,r + IJ,r (2-22)
IJv,r
V ;+l,r V ,r J
·
)
(2-23)

and where subscripts

1 = index of IV pair whose voltage is the maximum value less than Vfv,r
j+ 1 = index of IV pair whose voltage is the minimum value greater than
vfv,r
The Blaesser approach assumes a current multiplier equal to the ratio of measured and
reference irradiance (ignoring temperature effects) and a voltage offset proportional to
the natural log of the irradiance ratio (direct effect of irradiance on voltage), a delta T
and a constant irradiance term (relating module voltage to irradiance). The model is
conveniently based on ambient temperature but converts to (or from) a cell
temperature. Wind speed could be added as a term to equation (2-17) though Blaesser
says it doesn't improve the model. This model was derived based on the need to
translate field data (where ambient temperature is much easier to measure than module
temperature) to STC.

2.1.2.5 Lumped Four Parameter (L4P) M odel

The L4P model is derived from an electrical model of a single-bandgap solar cell;
specifically, a light-induced current source in parallel with a single diode and series
resistance. This model has been discussed by Townsend (1989), Kreith (1978) and

Module Energy Rating Methodology 11

briefly by Rauschenbach (1980). This model is sometimes modeled with an additional
shunt resistance; this form of the model is considered more appropriate for a-Si devices.

The defining equations for this model are:

V =
r k
.

q
.
Tc .
In( IL I-

I0 ·NCP
+ lJ -
I .
R". (2-24)

(2-25)

r = A·NCS (2-26)
where
v = Voltage on translated IV curve, as a function of current I (A)
I = Current on translated IV curve (A)
IL = Light-induced current (A)
IL,rcf = Light-induced current at reference conditions (A)
Ia
= Reverse saturation current (A)
Tc = Cell temperature (C)
Tc,rcf = Cell temperature at reference conditions (C)
H = Irradiance (W /m2)
Hrcf = Irradiance at reference conditions (W /m2)
f-Lisc = Short circuit current temperature sensitivity (A/ C)
Rs = Series resistance (Q)
y = Multi-cell diode ideality factor (unitless)
A = Single-cell diode ideality factor (unitless)
NCS = number of cells in series
NCP = number of cells in parallel
k = Boltzmann constant (1.380622*10-23)
q = Electron charge constant (1.6021917*10-19)

The four parameters for this model are Iu I0, y, and Rs. Of these four, one is trivial to
obtain, while the remaining three are not. IL,ref may be simply approxima�ed by Isc,ref
Townsend offers several techniques for obtaining the remaining parameters using
conventional module characterization parameters and reference curve characteristics.

Module Energy Rating Methodology 12

One "simplified" method he offered involves solving for an intermediate parameter A
using peak power current and voltage and short circuit current and open circuit voltage,
using this to obtain y, Rs and I0 as follows:

-----'--( ---'--)
Imp
+ln l- f,..c
A= (2-27)
2. vmp- voc

q
r=A·k·T;, (2-28)

]
A: · 1n(1- -) + Voc - Vmn
Imp
I·'� r
Rs = (2-29)
Imp

I �·exp(-A· Voc )
o = NCP
(2-30)

where
I sc = Short circuit current (A)
V oc = Open circuit voltage (V)
Imp = Current at maximum power point of curve (A)
Vmp = Voltage at maximum power point of curve (V)

Townsend indicated that this technique sometimes yields "unrealistic" results, and
offered an alternative solution technique that iteratively computes the
voltage/temperature sensitivity flvoc,A from Rs until the "known" value of flvoc is
obtained, thus, indirectly obtaining the correct value for Rs. The equations for this
approach are:

(
In l-
Imp
1,\"C
) (2-31)
A

f.lvoc,A
r ·k l 1
=- q
-.
�J Isc
-I-0
+
1;, f-lisc
·
Isc
-
q · &c
3- �r_:· """"k·- NCs
·

_Tc_j
l
(2-32)

where
&c = Bandgap energy (1.12 for silicon)

Module Energy Rating Methodology 13

and I0 and rare computed as in the "simplified" method. Unfortunately, for our
exploratory tests this technique yielded "unrealistic" negative resistances. These results
appear to reflect the low sensitivity of the I-V curve measurements to the parameters of
interest; and conversely, the high sensitivity of the parameters on the I-V curve
measurements. For example, the uncertainties in l0 and Rs due to uncertainties in the I-V
curve measurements are greater than the values of l0 and Rs.

Two drawbacks to this approach have been suggested: namely, the restriction to
application to single bandgap devices; and the difficulty of estimating the parameters
accurately. The difficulty of estimating the parameters accurately is of concern because
the model behavior is fairly strongly dependent on these values. Both the fact that the
model behavior is so sensitive and that the resistance estimates are sensitive to
measurement error in the testing procedure suggest that this model may not produce
repeatable results.

Because of the difficulty encountered in obtaining reasonable estimates of the model

parameters, this model was not compared to the other models. Further investigation
into the reasons why the parameters were so difficult to obtain could either allow this
model to be used, or perhaps provide insight into difficulties observed with other
models. For example, it is possible that a non-linear curve fitting algorithm could be
applied that would tend to be less sensitive to random experimental error by more fully
utilizing the test data. In addition, this model may be a good basis for extension to
multi-bandgap devices, should detailed models be required for them.

2.1.3 Spectral Model

The spectral model has two major components. First, the spectral model simulates
irradiance spectra appropriate to the specified weather conditions for each time
interval. These spectra are then combined with a measured module spectral response
function to obtain an equivalent AM1.5 broadband irradiance value for input to the
power model.

This modeling approach ignores the dependence of the IV curve shape on the incident
spectral distribution. While this dependence may be an issue for multi-bandgap
devices, it is not an issue for most PV technology currently in production. None of the
empirical power models investigated here or suggested by the Technical Review
Committee (TRC) handle the spectral IV-curve-shape effect, so our approach represents

Module Energy Rating Methodology 14

a practical compromise. Some analytical models have been proposed [Emery and
Osterwald, 1988] that could handle these shape effects, but no practical module testing
procedures have been developed yet to obtain the appropriate model parameters.

The general approach used is to compute the ratio of two weighted-average spectral
responses (SR) of the module: the numerator SR weighted with the actual (simulated)
spectrum, and the denominator SR weighted with the ASTM E892 (AM1.5) spectrum to
which the laboratory tests are referenced. This approach has been discussed by Emery
and Osterwald (1988), King and Hansen (1991), and Seaman (1981).

In order to make adjustments based on field and test spectra, these spectra need to be
available. Since these data are very rare, a model is used to estimate the spectrum
corresponding to the weather data set. To accomplish this, the model uses the station
pressure, dewpoint temperature, solar geometry, diffuse horizontal and plane-of-array
(broadband) irradiances, and the beam (broadband) irradiance. In addition, a fixed
dataset is required that contains the extraterrestrial spectrum, typical clear sky spectral
absorption characteristics, and a set of empirically derived spectral cloud-cover model
coefficients. The standard (AM1.5) testing spectrum is required for the effective
irradiance calculation.

The spectral model chosen is that implemented by the SEDES2 code by Nann and
Bakenfelder (described by Nann and Riordan (1991)), which is in turn derived from the
SPCTRL2 model by Bird and Riordan (1986) with an empirically derived modifier
added to account for cloud cover. The cloud cover modifier is a function of wavelength,
air mass, and the ratio of the actual broadband global horizontal irradiance to the
predicted clear-sky wavelength.

The major steps that are used in the SEDES2 model are estimation of clear-sky spectra
(direct and diffuse horizontal), estimation of the (potentially) cloudy-sky spectra,
magnitude normalization and estimation of plane-of-array diffuse spectra, and
estimation of the effective irradiance. These topics are covered in the following sections
in reverse order, to emphasize results and required inputs in a "top-down" manner. For
discussion of the origins of, assumptions in, and derivations of these equations, refer to
the papers described above.

2.1.3.1 Effective lrradiance

The effective (usable) irradiance for a specific module is given by

Module Energy Rating Methodology 15

 HPOA,Ejf = SCF . JAM . HPOA (2-33)

where
lAM = incidence angle modifier (see page 34)
HPaA = plane-of-array irradiance (see page 37)

and the spectral correction factor, SCF, is given by

( JsR(.<}Hro(A.)dA (A.)dA.)
]HPOA A,
(2-34)
SCF -
( JsR(A.)·dH(A.)dA.
- std(A.)dA.)
--r-;.----�

]Hst

where
A- = irradiance wavelength (J...lm)
SR(A.) = spectral response of the module (AIW)
HPOA(A-) = plane-of-array irradiance spectrum (WIm2I J...lm)
H (A.)
std
= standard (ASTM E892 AM1.5 irradiance spectrum) (WIm2 I J...lm).

The integrals may be evaluated using Euler numerical integration. This computation is
simplified if all of these functions of wavelength (represented as explicit arrays of data
rather than as closed-form formulas) are evaluated at the same wavelengths.

2.1.3.2 Nor malization and Incident Surface Translation

The spectrum generated by the SEDES2 model does not contain exactly the same
broadband power as the measured (input) irradiances do. This is because cloud cover
modifiers change the shape of the clear-sky spectrum based solely on wavelength and
the broadband measured-to-predicted irradiance ratio. That is, for each wavelength the
magnitude is multiplied by a coefficient that does not depend on the rest of the actual
(or estimated) spectrum. Since the coefficients are derived from actual data, they may
be expected to work reasonably well as long as the colors in view of the site being
simulated are similar to those where the data for the coefficients were obtained. This is
not an entirely unreasonable assumption, but we have observed a 3-7 percent difference
between the integrated SEDES2 spectrum and the input broadband irradiance.

Module Energy Rating Methodology 16

Assuming the cloud cover modifiers do in fact improve the shape of the estimated
actual spectrum, the magnitude of the modified spectrum should be corrected to agree
with the broadband irradiance before it is used. However, for the purposes of the
Module Energy Rating (MER), the shape of the spectrum is all that is needed. As long
as the broadband irradiance is not substituted for the integrated spectrum in the SCF
calculation, the scaling makes no difference.

To obtain global plane-of-array spectral irradiance from direct normal and diffuse
horizontal spectral irradiances, the diffuse spectral distributions are assumed to be
isotropic. That is, the color is isotropic while intensity may not be. This reduces the
problem to a simple scaling computation, as follows:

HPOA (A.)= Hd (A.) · cos(B) + Hs,POA (A.) (2-35)

Hs,POA (A)= Hs,H (A) · /{

H POA (2-36)
s,H
where
HiA.) = direct (beam) spectral irradiance (W lm2fJ.tm)
Hs,POA (A,) = scattered (diffuse) plane-of-array spectral irradiance (W 1m2 I J..Lm)
Hs,H(A,) = scattered (diffuse) horizontal spectral irradiance (W 1m2 I �-tm)
Hs,POA,meas = scattered (diffuse) broadband plane-of-array irradiance (W 1m2),
measured or estimated using Perez model
Hs,H,meas = scattered (diffuse) broadband horizontal irradiance (W 1m2),
measured
e = incidence angle (page 36)

2.1.3.3 Estimated Spectrum with Cloud-Cover

To account for the most general weather conditions, the SEDES2 model estimates the
spectral irradiance using:

Hd (;t)= Hd,clear( A.) · CCM(A.,NGH,z) (2-37)

Hs,H( A.)= Hs,H,clear ( A.) · CCM(A.,NGH, z) (2-38)

Module Energy Rating Methodology 17

 CCM(...i, NGH, z) = (Al(...i) + ���;?)
+(Bl(...i) + :;s�;�) NGH · (2-39)
+(Cl( ) + C2(A.) 1
cos(z)) NGH
./!,
2
·

where
Hd,clear( A) = direct clear sky spectrum (page 20)
Hs,H,clear( A) = scattered horizontal clear sky spectrum (page 20)
NGH = normalized global horizontal insolation (page 21)
z = sun zenith angle (see page 35)
CCM(...i,NGH,z) = cloud cover modifier function
Al(...i), A2(...i), Bl(...i),
B2(...i) , Cl(...i), C2(A,) = cloud cover coefficients (Tables 2-1, 2-2)

Module Energy Rating Methodology 18

 Table 2-1. SEDES2 Cloud Cover Modifier Coefficients
A. Al(A.) A2(..i) B1(..i) B2(..i) C1(..i), C2(..i)
320 1.285724 0.306791 -0.29613 -0.58516 0.020632 0.20915
330 1.235103 0.262007 -0.28377 -0.53864 0.010728 0.206493
340 1.206166 0.250204 -0.25258 -0.51989 0.004315 0.204614
350 1.139737 0.242676 -0.19222 -0.49821 -0.01184 0.201329
360 1.091643 0.244214 -0.13386 -0.48722 -0.0272 0.200767
370 1.033731 0.251496 -0.07915 -0.48133 -0.04285 0.202966
380 0.997179 0.243862 -0.0655 -0.45039 -0.03607 0.191915
390 0.997948 0.227502 -0.08976 -0.40715 -0.01039 0.17371
400 0.990572 0.205403 -0.12091 -0.35735 0.018082 0.152083
410 0.984024 0.193105 -0.13671 -0.32748 0.034395 0.140702
420 0.971385 0.177868 -0.15584 -0.29288 0.051752 0.127548
430 0.97645 0.159398 -0.18434 -0.25421 0.072127 0.112706
440 0.973204 0.142079 -0.20773 -0.21836 0.088689 0.098569
450 0.979785 0.129315 -0.22806 -0.19197 0.103365 0.08717
460 0.98578 0.119208 -0.24438 -0.1714 0.117445 0.076708
470 0.99861 0.109176 -0.26163 -0.15113 0.132599 0.066066
480 1.005317 0.099677 -0.27866 -0.13004 0.147219 0.055762
490 1.019677 0.089575 -0.30482 -0.10709 0.16626 0.045127
500 1.024404 0.080517 -0.32229 -0.0875 0.179513 0.036474
510 1.031585 0.069067 -0.34795 -0.06441 0.196865 0.025472
520 1.049367 0.056443 -0.38233 -0.04055 0.218811 0.013729
530 1.063939 0.046316 -0.40907 -0.02121 0.236117 0.004201
540 1.071553 0.0383 -0.42769 -0.00587 0.248413 -0.00299
550 1.070387 0.031852 -0.43045 0.004491 0.251829 -0.00768
560 1.062834 0.026342 -0.41879 0.011999 0.246651 -0.01046
570 1.045843 0.024689 -0.37226 0.009432 0.223078 -0.00801
580 1.037469 0.023472 -0.33927 0.008971 0.207507 -0.0069
590 1.026083 0.023298 -0.3141 0.008151 0.195734 -0.00518
600 1.040383 0.015681 -0.34917 0.024337 0.218887 -0.01426
610 1.05082 0.006659 -0.38518 0.041762 0.241564 -0.02411
620 1.051636 0.000294 -0.39171 0.051032 0.246386 -0.02902
630 1.040294 -0.00264 -0.36449 0.050866 0.230633 -0.02769
640 1.04091- -0.00243 -0.35577 0.051709 0.225536 -0.02653
650 1.040678 -0.00316 -0.34746 0.053757 0.221069 -0.02611
660 1.065054 -0.00775 -0.38644 0.068596 0.246247 -0.0347
670 1.081709 -0.0102 -0.40061 0.077291 0.257483 -0.04034
680 1.077241 -0.00697 -0.36968 0.071587 0.240555 -0.03716
690 1.040409 -0.00413 -0.28523 0.052308 0.187536 -0.02455
700 1.016405 -0.00067 -0.23359 0.036039 0.150176 -0.01227
710 1.006519 -0.00416 -0.21335 0.030742 0.130581 -0.00725
720 1.015005 -0.00986 -0.20643 0.033451 0.120013 -0.00709
730 1.11212 -0.03985 -0.3703 0.086799 0.198931 -0.03506
740 1.259638 -0.07938 -0.63633 0.167885 0.336038 -0.08023
750 1.359701 -0.10681 -0.82757 0.227298 0.435026 -0.11411
760 I 1.364134 -0.10886 -0.84101 0.233641 0.440063 -0.11907
770 1.413504 -0.12491 -0.91952 0.262684 0.480399 -0.13497
780 1.472111 -0.14378 -1.00406 0.291321 0.524579 -0.14918
790 1.460142 -0.14248 -0.96339 0.280995 0.499939 -0.14149
800 1.397082 -0.12613 -0.83251 0.242547 0.428312 -0.11892
810 1.303223 -0.09812 -0.64065 0.184689 0.325405 -0.08646
820 1.231193 -0.08347 -0.50422 0.149742 0.253542 -0.06661
830 1.278968 -0.09801 -0.59564 0.179143 0.30194 -0.08288
840 1.394604 -0.12999 -0.82226 0.248595 0.42466 -0.12262
850 1.48684 -0.15767 -1.02211 0.309727 0.533828 -0.15811
860 1.533058 -0.17332 -1.12535 0.343352 0.589583 -0.17738
870 1.54842 -0.17691 -1.14042 0.350555 0.597075 -0.18138

Module Energy Rating Methodology 19

 Table 2-1. SEDES2 Cloud Cover Modifier Coefficients (cont.)
A. Al(A.) A2(2) B1(2) B2(2) C1(2), C2(2)
880 1.509161 -0.16271 -1.02979 0.319605 0.536668 -0.1636
890 1.398192 -0.1247 -0.77108 0.242984 0.400871 -0.1215
900 1.176119 -0.06824 -0.34215 0.120864 0.176265 -0.05349
910 0.986845 -0.01315 0.01589 0.015608 -0.00784 0.003255
920 0.830406 0.03159 0.284686 -0.06127 -0.14019 0.042905
930 0.611229 0.097008 0.607703 -0.15086 -0.28158 0.08258
940 0.369127 0.137444 0.920399 -0.22796 -0.42836 0.122108
950 0.306382 0.132255 1.017929 -0.25108 -0.50619 0.144856
960 0.427639 0.084802 0.85788 -0.20327 -0.46987 0.132757
970 0.650115 0.034501 0.600522 -0.12507 -0.37126 0.097659
980 0.843689 -0.01411 0.352464 -0.04375 -0.26576 0.058199
990 1.018712 -0.05584 0.115209 0.032978 -0.16069 0.01951
1000 1.110714 -0.08242 -0.02662 0.08182 -0.09732 -0.00507
1010 1.158305 -0.09845 -0.10842 0.111704 -0.0598 -0.02013
1020 1.187785 -0.10971 -0.17215 0.13436 -0.02617 -0.03236
1030 1 .216623 -0.12039 -0.24681 0.157773 0.018206 -0.04635
1040 1 .242954 -0.13007 -0.3248 0.179514 0.068458 -0.06071
1050 1 .242954 -0.13007 -0.3248 0�179514 0.068458 -0.06071

The cloud cover coefficients describe how the actual spectrum tends to deviate from the
clear sky spectrum for each wavelength as the air mass (approximated by 1 / cos(z)) and
normalized global horizontal irradiance vary. These coefficients were derived from
measured data using statistical analyses as described by Nann and Riordan, 1991.

2.1.3.4 Estimated Clear Sky Spectrum

The clear sky spectrum may be estimated by computing the absorption and scattering
effects of the atmosphere on the (relatively constant) extraterrestrial spectrum using
local weather conditions as input. However, the non-isotropic field of view and non­
uniform atmospheric conditions (that are not reflected in the local weather conditions,
but may affect the light as it passes through the atmosphere) significantly reduce the
accuracy of this approach. If the atmosphere is assumed to behave as a neutral density
filter (spectrally independent scaling), then the first cut estimate of the spectral
irradiance may be "corrected" by scaling it back to the measured broadband irradiance.
Any errors in spectral shape introduced by this assumption may be assumed to be
corrected by the empirically derived cloud cover modifier. Thus, the clear sky
horizontal scattered (diffuse) spectra Hs,H, cicar(1) and the direct (beam) spectra Hd, ciear(1)
may be computed using the equations:

Hs,H,ciear (1) = (Hd ,est (1) cos(z) + Hs,H ,est (1)) · NGH - Hd,ciear (1)
· (2-40)

(2-41)

Module Energy Rating Methodology 20

where the est subscript indicates first cut spectral estimates from the extraterrestrial
spectrum (assumed constant) affected by atmospheric absorptance, scattering and
reflection as described below. The normalized direct and global horizontal irradiances
NDIR and NGH represent scaling factors to convert the magnitudes of the first cut
estimates to match the broadband measurements:

HH,meas - Hs,H,meas
NDJR = (2-42)
Hd,est . coS: z)
HH,meas
NGH = (2-43)
H.<,H,est + Hd,est . cos(z)
where
= global horizontal broadband measured irradiance (W /m2)
H = scattered (diffuse) horizontal broadband measured irradiance
(W /m2)
s,H,meas

and the estimated broadband irradiances are given by:

(2-44)

(2-45)

The first-cut direct irradiance spectra are obtained by multiplying the estimated
extraterrestrial spectrum by the transmittances of the optically significant components
of the atmosphere:

(2-46)

where the extraterrestrial spectrum is a standard spectrum modified in intensity as a

function of the time of year:

1.0001 1· . .
( i
Hexo ( A.) = H0 ( /L) · l +0.034221 · cos (a D ) + 0.00128 · sin(aD } · · j (2-47)
+0.000719 · cos (2a D ) + 0.000077 · s in(2a D )
where H0(/L) is given in Table 2-2, and a0 is the day angle given on page 35.

Module Energy Rating Methodology 21

The transmittances due to Rayleigh scattering (TJI!.)), aerosols (Ta(A.)), water vapor
(Tw(A.)), ozone (To(A.)), and uniform gases (Tu (A.)) are given by:
( -m l
(A.) = exp 4
T, l
A- (1 15.64�6 - y
• )j (2-48)

� ( A.) = exp(-ra (A.) · m) (2-49)

=
l( -0.2385 · A) A.) · W · m l
Tw ( A.) exp
(1 + 20.07 Aw (A.) · W · m)0.45
• j (2-50)

� (A.) = exp(-A a (A.) · 03 mo )

· (2-51)

T, ( A.) = exp
l( -l.41 · Au(A.) · m lj ( 2-52)
(1 + 1 18.3 · Au (A.) · m)o.45
where
m = air mass (page 37)
mP = pressure-corrected air mass ( = m · 1 �3 )
AjA.) = absorption spectrum of water (given in Table 2-2)
Ao(A.) = absorption spectrum of ozone (given in Table 2-2)
Au (A.) = absorption spectrum of uniform gases (given in Table 2-2)

and the aerosol spectral turbidity ( r.(A-)), water amount (W), ozone amount (03) and
ozone mass (m0) are estimated using:

(2-53)

W= exp (0.06930 · :Z:.ewpc - 0.0756) (2-54)

r r 150· · · l1
I I II
03 = 0.001 ·1 235 + sin(1.28 1�0 ¢ ) 2 · I +40 · sin(0.9865 · 1�0 (Nday - 30)) . . I I
· • • (2-55)

l l +20 · sin(3 · 1�o · B03 ) Jj

1 + RR
= -;======
m (2-56)
o �cos(z) 2 + 2 RR ·

Module Energy Rating Methodology 22

where
a = Exponent for rural aerosol turbidity model ( = 1.14)
J3 = scaling factor for rural aerosol turbidity model (= 0�5�: )
Tdewpt = dewpoint temperature (Celsius)
Nthy = day of year (1 = January 1)
¢ = latitude of location (degrees in northern hemisphere)
Ba3 = adjusted longitude in degrees east of Greenwich. If negative (west
of meridian), then no adjustment. If positive, then add 20 degrees.
RR = height of ozone layer in earth radii ( =22/ 6370)

When the dewpoint temperature must be estimated from ambient temperature and
relative humidity, the following formulas were used in SEDES2:

_J
f · (39.9 8 + 24.83 · ln(Pw ) + 0.8927 · ln( Pw Y) ; Tamb < 0
(2-57)
p lf · (47.047 + 30.5 79 · ln(Pw ) + 1.8893 · ln( Pw Y) ; �mb ;?: 0
Tdew t
-

(2-58)
1 827.17843 - 71208.271 2

pws = 0.02953 · 1 0
8.42926609-
(2-59)
T.mb+273.1 5 ( Tamb +273 .15)

where
Tamb = ambient temperature (Celsius)
Pws = water vapor saturation pressure (in-Hg)
RH = relative humidity (percent)

The diffuse horizontal estimated spectrum (Hs,H,,l?t,) ) may be computed with:

Hs,H,est (/L) = Cs (/L) · (Hr(/L) + Ha(/L) + Hg (/L) ) (2-60)

s
C ( /L) =
{1
( /L + 0.5 5)
1 .8
; /L < 0.45
(2-61)
; /L :?:: 0.45
where
C5( A) = adjustment factor for short wavelength light

and

Module Energy Rating Methodology 23

 Hr ( /L) = t · Hexo ( /L) · cos(z) · I: (/L) · 1', (/L) · Tw ( /L) · �a ( /L) · (1 - T,. ( /L) o.95 ) (2-62)

Ha ( /L) = Hexo ( /L) · cos(z) · I: ( /L) · 1', ( /L) · Tw (/L) · �a ( /L) · 1',. ( /L) !.5 (1 - �s ( /L) ) · Fs ( z)
• (2-63)

�(� · �(�
Hg (/L) = 1 - r ( /L) r ( /L) · (Hd,est(/L) · cos(z) + Hr(/L) + H)/L) )
- (2-64)
s g
where
H,(/L) = Rayleigh scattering diffuse component (W 1m2 I f.liTl)
Ha(/L) = aerosol scattering diffuse component (WI m2 I f.liTl)
H//L) = ground-reflected scattering diffuse component (W 1m2 I f.liTl)
rs(/L) = sky reflectivity
rg(/L) = ground albedo (assumed constant = 0 .2)
To( /L) = ozone transmittance
T" ( /L) = uniform gasses transmittance
Tw(/L) = water vapor transmittance
Taa (/L) = transmittance due to aerosol absorptance
Tas( /L) = transmittance due to aerosol scattering
T,(/L) = Rayleigh transmittance
Fs(z) = ratio of forward to total scattering.

The aerosol transmittance components due to scattering ( Tas(/L)) and absorptance ( Taa (/L)),
the aerosol single-scattering albedo(lV(/L)), and the ratio of forward to total scattering are
given by:

�s ( /L) = exp(-m( /L) · Ta ( /L) · m) (2-65)

� ( /L)
�a (/L) = �)/L) (2-66)

m( /L) = m0.4 • exp(-m'· [ ln(/4)]2 ) (2-67)

F_,(z) = 1 - + · exp((AFS + BFS · cos(z) ) · cos(z)) (2-68)

where
ra(/L) = aerosol turbidity (page 22)

Module Energy Rating Methodology 24

 m = air mass
TJ:L) = total aerosol transmittance (page 22)
= aerosol single scattering albedo at 0 . 4 1-lffi wavelength (assuming
rural model, use 0.945)
= single scattering wavelength variation factor (assuming rural
model, use 0.095)

and

AFS = ALG · (1.459 + ALG · ( 0.1595 + ALG 0.4129)) · (2-69)

BFS = ALG (0.0783 + ALG (-0.3 824 - ALG 0.5 874))

· · · (2-70)

ALG = In( 1 - ASF) (2-71)

where
ASF = aerosol symmetry factor (for rural modet assume 0.65)

The sky reflectivity is computed using:

+ F'.J T'r (A,) · (1 - T'as (A,))]

�,(A,) = T' u (A,) · T' w (A,) · T'aa (A,) T-T;(-<) (1 - · (2-72)

where the primed transmittances are the regular atmospheric transmittance terms
evaluated at air mass 1 . 8, and

F = 1-
1
.,.
I
2 · exp
(AFS+1.8 BFSJ
1.8
(2-73)

Module Energy Rating Methodology 25

 Table 2-2. Standard spectrum and atmospheric optical characteristics.

'}.. HO Aw Ao Au '}.. HO Aw Ao Au '}.. HO Aw Ao Au

300 535.9 0 10 0 740 1298 0.061 0.01 0 1520 292.8 0.16 0 0
305 558.3 0 4.8 0 752.5 1269 0.0008 0.008 0 1539 275.5 0.002 0 0.005
310 622 0 2.7 0 757.5 1245 0.0001 0.007 0 1558 272.1 0.0005 0 0.13
315 692.7 0 1.35 0 762.5 1223 0.00001 0.006 4 1578 259.3 0.0001 0 0.04
320 715.1 0 0.8 0 767.5 1205 0.00001 0.005 0.35 1592 246.9 0.00001 0 0.06
325 832.9 0 0.38 0 780 1183 0.0006 0 0 1610 244 0.0001 0 0.13
330 961.9 0 0.16 0 800 1148 0.036 0 0 1630 243.5 0.001 0 0.001
335 931.9 0 0.075 0 816 1091 1.6 0 0 1646 234.8 O.Dl 0 0.0014
340 900.6 0 0.04 0 823.7 1062 2.5 0 0 1678 220.5 0.036 0 0.0001
345 911.3 0 0.019 0 831.5 1038 0.5 0 0 1740 190.8 1.1 0 0.00001
350 975.5 0 0.007 0 840 1022 0.155 0 0 1800 171.1 130 0 0.00001
360 975.9 0 0 0 860 998.7 0.00001 0 0 1860 144.5 1000 0 0.0001
370 1119.9 0 0 0 880 947.2 0.0026 0 0 1920 135.7 500 0 0.001
380 1103.8 0 0 0 905 893.2 7 0 0 1960 123 100 0 4.3
390 1033.8 0 0 0 915 868.2 5 0 0 1985 123.8 4 0 0.2
400 1479.1 0 0 0 925 829.7 5 0 0 2005 113 2.9 0 21
410 1701.3 0 0 0 930 830.3 27 0 0 2035 108.5 1 0 0.13
420 174D.4 0 0 0 937 814 55 0 0 2065 97.5 0.4 0 1
430 1587.2 0 0 0 948 786.9 45 0 0 2100 92.4 0.22 0 0.08
440 1837 0 0 0 965 768.3 4 0 0 2148 82.4 0.25 0 0.001
450 2005 0 0.003 0 980 767 1.48 0 0 2198 74.6 0.33 0 0.00038
460 2043 0 0.006 0 993.5 757.6 0.1 0 0 2270 68.3 0.5 0 0.001
470 1987 0 0.009 0 1040 688.1 0.00001 0 0 2360 63.8 4 0 0.0005
480 2027 0 0.014 0 1070 640.7 0.001 0 0 2450 49.5 80 0 0.00015
490 1896 0 0.021 0 1100 606.2 3.2 0 0 2500 48.5 310 0 0.00014
500 1909 0 0.03 0 1120 585.9 115 0 0 2600 38.6 15000 0 0.00066
510 1927 0 0.04 0 1130 570.2 70 0 0 2700 36.6 22000 0 100
520 1831 0 0.048 0 1145 564.1 75 0 0 2800 32 8000 0 150
530 1891 0 0.063 0 1161 544.2 10 0 0 2900 28.1 650 0 0.13
540 1898 0 0.075 0 1170 533.4 5 0 0 3000 24.8 240 0 0.0095
550 1892 0 0.085 0 1200 501.6 2 0 0 3100 22.1 230 0 0.001
570 1840 0 0.12 0 1240 477.5 0.002 0 0.05 3200 19.6 100 0 0.8
593 1768 0.075 0.119 0 1270 442.7 0.002 0 0.3 3300 17.5 120 0 1.9
610 1728 0 0.12 0 1290 440 0.1 0 0.02 3400 15.7 19.5 0 1.3
630 1658 0 0.09 0 1320 416.8 4 0 0.0002 3500 14.1 3.6 0 0.075
656 1524 0 0.065 0 1350 391.4 200 0 0.00011 3600 12.7 3.1 0 0.01
667.6 1531 0 0.051 0 1395 358.9 1000 0 0.00001 3700 11.5 2.5 0 0.00195
690 1420 0.016 0.028 0.15 1442.5 327.5 185 0 0.05 3800 10.4 1.4 0 0.004
710 1399 0.0125 O.D18 0 1462.5 317.5 80 0 0.011 3900 9.5 0.17 0 0.29
718 1374 1.8 O.D15 0 1477 307.3 80 0 0.005 4000 8.6 0.0045 0 0.025
724.4 1373 2.5 0.012 0 1497 300.4 12 0 0.0006

2.1.4 Module Thermal Model

Another explicit input to several of the contemplated PV performance models is module

temperature. Module output power varies by roughly 0.5%;oC change in temperature.
This module temperature coefficient is less for most thin-film devices and for higher

Module Energy Rating Methodology 26

efficiency devices1• The temperature of a PV module depends on module construction,
module mounting, ambient conditions (in order of decreasing importance: ambient
temperature, irradiance, wind speed, wind direction, humidity, barometric pressure,
elevation, etc.), and conversion efficiency (solar power converted to electricity is
unavailable for heating up the module).

The thermal model used in PVFORM [Fuentes, 1987] was developed and evaluated
using measured array data and takes into account module construction, mounting, and
operation in a term called Installed Nominal Operating Cell Temperature, INOCT.
INOCT is based on JPL's NOCT which is defined as the cell temperature of the module
under conditions of 800 WI m2 irradiance, 20 oc ambient temperature, 1 ml s wind
speed, with a single open-circuited module mounted in an open rack (or per the
manufacturer's requirements). INOCT has the same ambient conditions, but the
module is mounted and operated per the system designer's intention. As such, one
module can have a wide range of INOCTs because of the possibility of different
mounting configurations and operating points (peak power, fixed voltage, etc.)

2.1.4.1 General Approach

The Fuentes model uses a simplified heat balance approach. Convective and radiative
heat transfer equations are developed and a number of assumptions are made.
Included in these assumptions are a tilt angle of 30° (for convective heat transfer
coefficient), module efficiency of 8%, reflectance of 0.10, absorptance of 0.83, emittance
of 0.84, a thermal mass (m·c) of 11 kJ I m2 °C, and a convection distance of 0.5m (enough
to allow turbulence to occur, but a small value relative to an array). Also, Fuentes
assumes that modules with INOCT > 48°C are thermally coupled to a roof or some
other structure and increases the thermal mass proportional to INOCT.

The heat balance with simplifying assumptions yields the following expression for
module temperature:

(he · Ta + hrs · Ts + hrg · Tg + a · H0 + a · !lif I L). (1 - ) + a · !lif

e
L

Tmod = + Tmod0 · e
L
(2-74)
h c + hrs + hrg

' The absolute value in Watts;oC or Volts;oc can decrease and the divisor, reference Watts or Volts,
increases. Therefore the %JDC is reduced.

Module Energy Rating Methodology 27

where
Tm = module temperature, K ·

Ta = ambient temperature, K
Ts = sky temperature, K
Tg = ground or roof temperature, K
he = overall convective heat transfer coefficient (W /m2-K)
hrs = radiative h/t coefficient to the sky, (W /m2-K)
hrg = radiative h/t coefficient to the ground or roof, (W /m2-K)
a = module absorptivity ( <3.5 pm)
Ho = plane of array irradiance from previous time step, W /m2
&i = change in POA irradiance from previous time step, W /m2

and

L= -(!ze + h's + hrg ) ilt j(m . e) (2-75)

where
m = module mass per unit surface area (kg/m2)
e = overall module specific heat (J /kg-K)
m·e = 11,000 J/m2-K (for INOCT 48 ac), •

1.1E4·[1+(INOCT-48)/12] (for INOCT > 48°C)

Llt = time step (s)

The overall convection coefficient is calculated as the cube root of the sum of the cubes
of the forced and free convection coefficients:

h e = fzV}forced3 + hfree3 '{J / 3 (2-76)

For forced convection the following equations are used:

hforced = St . p . ep • wm (2-77)

(2-78)

Re = -­
wm·Dh
v (2-79)

Module Energy Rating Methodology 28

 - 0.0003484-101 325
p
-
Tfilm (2-80)

To.76
v = 2.4237 · 1 0-7 • 1;' (2-81)

Tm + Ta
Tfilm = ---- (2-82)
2

where
St = Stanton number
c ·f.J
Pr = Prandtl number (= -T-= 0.71 for air)
Re Reynolds number
p = density of air
cP = specific heat of air ( = 1007 J/kg-°C)
wm = wind speed at the module height (m/s)
D!t module hydraulic diameter1 (= 0.5m)
v kinematic viscosity of air
Tfilm film temperature

and
a b c
0.67 0.86 -0.5
0.4 0.028 -0.2

For free convection,

Nu · k
hfree = ---;:;­ (2-83)
h

Nu = 0.21 · (Gr · Pr
)0.32
(2-84)

1 Strictly speaking, the hydraulic diameter is intended to be used for fluid flow in a tube with a non­
circular cross section. Fuentes assumed typical module dimensions of 1.2m x 0.3m. Using the standard
Dh 4A/P 4*(1.2*.3)/(2*1.2+2*0.3) 0.48m ±0.5m gives the stated value. For convection on a flat
= = =

surface, Dh is replaced with x, the distance from the leading edge. Since it is unclear from which
direction the wind is corning (wind direction was not incorporated into this model) and the module
orientation is unspecified, it appears from the validation results that a value of 0.5 m is a good
approximation.

Module Energy Rating Methodology 29

 g (Tm - Ta) Dh3 sin(m)
Gr = '�"
(2-85)
· · •

vz Tfitm
.

•
0.84
k = 2 1 695 · 10-4 · Tfilm (2-86)

where
Nu = Nusselt number
Gr = Grashof number
g = gravitational constant (=9 .8 m/ s2)
cp = module tilt angle, assumed to be 30°
k = thermal conductivity of air

For front surface radiation, sky temperature is estimated from ambient temperature
using

Ts = 0. 6 8 · (0.0552 · Ta l .5 )+ 0.32 · Ta (2-87)

which assumes the sky is clear 68% of the time and cloudy 32% of the time.

The model assumes that the roof or ground temperature under the array is somewhere
between module and ambient temperatures. Ratios for the total to front surface
convection coefficients and, subsequently, the ratio of the roof or ground to module
temperature rise above ambient are calculated.

Wind speed is corrected from the measurement height (typically 10m) to the average
array height using

(2-88)

where
wm = wind speed at module height (m/s)
w. = wind speed at anemometer height (m/ s)
Ym = average module height (m)
Y. = anemometer height (m)

Module Energy Rating Methodology 30

The Sandia report provides source code for calculating module temperature based on
weather data and the input parameters described in the following sections.

2.1 .4.2 Module INOCT (°C)

As discussed previously, the INOCT is the NOCT in its installed configuration. The
Sandia document provides code for estimating INOCT from a set of measured data.
This routine starts with a guess for INOCT, uses the model and the measured ambient
data to estimate module temperature for each point in the data set, calculates the
residual (weighted with irradiance), adds the weighted residual to the INOCT guess
and repeats until the residual is less than 0.1 °C. This approach will probably give the
best INOCT value since it is based on measured data and optimizes for the model's
assumptions.

For the module energy rating, we have assumed the same mounting configuration as
assumed for the standard NOCT measurement-open rack, or per manufacturer's
specifications. However, NOCT is based on open circuited modules, which will tend to
run hotter than modules under operation. The temperature difference due to operating
the module is proportional to the irradiance level and the module operating efficiency.
We can estimate the temperature difference by starting out with a very simple energy
balance:

H · ra - Um · (Tm - Ta) - H · 77 = 0 (2-89)

where
H = plane of array irradiance, W /m2
ra = module effective transmittance-absorptance product
Um = overall module heat loss coefficient, W-m2 ;oc
17 = module efficiency at H and Tm

Assuming a ra of 0.9, plugging in the NOCT conditions (800 W /m2 irradiance, 20°C
ambient temperature, wind speed is ignored, 0.0% module efficiency), and solving for
Um gives

800 · 0.9
Um = (2-90)
NOCT - 20

Module Energy Rating Methodology 31

Finally, substituting Eqn. (2-89) and NOCT conditions into Eqn. (2-89) and solving for
Tm ( =INOCT) gives
0 .9 - 17
INOC T = (NOCT - 20) · + 20 (2-91)
0.9

Equation (2-91) provides the INOCT given measured values of NOCT and 17 at NOCT.
Note that these results will be somewhat conservative for fixed voltage operation.

2.1.4.3 Module and Anemometer Heights (m)

We need to make an assumption regarding module height for consistency and

comparative purposes. A module height of 2m might be an appropriate compromise
between ground mounted and rooftop arrays. The anemometer height for the National
Solar Resource Database (NSRDB) is 10 m.

2 .1.4.4 lrradiance (W/m 2)

The full spectrum plane of array irradiance (the output of the irradiance model
described in section 3.2.6) is used for the thermal model.

2.1.4.5 Ambient Temperature (°C)

One of the specified input weather parameters.

2.1.4.6 Wind Speed (m/s)

One of the specified input weather parameters.

2.1.4.7 Date and Time

The Fuentes thermal model is dynamic in the time domain. The date and time are used
to determine the time interval from the last reading.

2.1.4.8 Limitations

This model was developed using typical 1980-vintage flat plate modules. It was
verified using various residential roof-mounted arrays-from direct mount to standoff
(1-9 inches between the roof and array for airflow) to rack mount.

Module Energy Rating Methodology 32

While the basic approach wouldn't obviate its use for concentrator modules, some of
the simplifying assumptions and generalized heat transfer parameters might be
inappropriate. A concentrator module is typically designed to transfer most of its heat
through the back (i.e. a heat sink), whereas a flat plate module loses a significant
amount of it's heat from the front surface. The model ignores module conductivity and
assumes that the front and rear surfaces are at the temperature of the cell. This is, at
best, less true with a concentrator module. There can be a 5 to 30 C difference between
cell and rear surface temperature in a concentrator module.

The model starts with a "measured" module temperature under fairly typical peak
conditions and adjusts it based on actual ambient conditions. In general, a concentrator
module will respond in the same manner as a flat plate module to changes in ambient
conditions. A separate thermal model for concentrators may be required.

2.1 .5 Optical Model

The reflectivity model uses the plane-of-array beam irradiance incidence angle to obtain
an incidence angle modifier (lAM) which accounts for reflection and other effects.
Reflectivity may be affected by conditions other than the incidence angle, such as
module materials, soiling and incident spectrum. Of these additional effects, only
module materials (as represented by module test data) were considered for this
application.

For best accuracy, this effect should be applied to the beam component separately from
the diffuse component. For an isotropic diffuse component1, the effect of reflection may
be integrated over the field of view to obtain a constant net effect. Depending on how
the diffuse value is obtained, this constant may already be accounted for by instrument
calibration. Deriving coefficients to match the measured effect on total plane-of-array
irradiance is equivalent to assuming a fixed ratio of beam to diffuse irradiance. Such an
assumption cannot be supported for application to both clear and cloudy days.

The IAM may be computed using the empirical curve fit suggested by Gaul and Rabl
[Wenger, personal communication; see also Whitaker et. al. (1991)]:

1 Note that the Perez model assumes the diffuse component includes circumsolar and horizon
components. While the circumsolar component will vary along with the beam component, for fixed
orientation the horizon component will also integrate to a constant.

Module Energy Rating Methodology 33

 (2-92)

where
B = Plane-of-array incidence angle (radians)
a-e = Empirical curve-fit coefficient

Based on PVUSA data, Wenger suggests:

a = 90.8392
b = 2.2081E-2
c = -1.13092E-3
d = 2.5439E-5
e = -2.1088E-7

However, the TRC pointed out that since these coefficients were obtained using global
plane-of-array pyranometer data and maximum power point data, they suffer from
confounding of the irradiance components (as described above), incidence angle effects
on the pyranometer itself, and confounding of irradiance and power model effects
(since Isc was not used).

An alternative approach to quantifying the incidence angle effect was developed by D.

King [King, 1996]. This technique involved the use of an azimuth/altitude type two­
axis tracker whose azimuth was varied to obtain Isc data for a range of incidence angles
while keeping the diffuse component essentially constant. Attempts to reproduce this
procedure at NREL yielded unsatisfactory results, so no lAM function was used in this
document.

2.1 .6 lrradiance Model

The irradiance model transforms commonly available time-correlated irradiance and

other weather measurements and plane-of-array orientation into incidence angle and
full-spectrum incident beam and diffuse irradiance estimates using astronomical solar
position equations and the Perez diffuse radiation model. The steps in these
computations are the determination of the true local solar time, computation of the local
solar position, computation of the plane-of-array beam irradiance and incidence angle,
and the evaluation of the Perez diffuse radiation model. With the exception of azimuth

Module Energy Rating Methodology 34

angle (Equation (2-98)), these equations were extracted from source code provided by
Daryl Myers of NREL and represented the "NREL consensus" method for this type of
application. Equation (2-98) [Meeu, 1982] was chosen because it handles the full range
of azimuth angles in a more concise form.

2.1.6.1 True Local Solar Time

The true local solar time ( m; radians) is given by:

(2-93)

E = 0.000075
+0.001868 · cos( an ) - 0.032077 · sin( an ) (2-94)
-0.014615 · cos(2 · an ) - 0.040849 · sin(2 · an )

21C
an = · ( D - 1) (2-95)
365
L
t zl = (2-96)
15 - Z

where
E = Equation of time (radians)
ao = Day angle (radians); position of sun relative to stars
D = Day number (1 Jan = 1; 31 Dec = 365 or 366 in leap year)
t = Time at which OJ is to be computed (hours past midnight)
tzl = Local standard to local sidereal time correction (hours)
L = Site latitude (degrees; positive east of Greenwich)
z = Time zone (hours to be added to GMT for local standard time)

2.1.6.2 Local S olar Position

The local solar position (zenith angle z and azimuth angle A in radians) is given by:

z = cos-1 (cos(¢) · cos(5) · cos(m) + sin(¢ ) · sin(5)) (2-97)

A = ATAN2(cos(m) · sin(¢) - cos(¢) · tan(5),sin(m )) (2-98)

Module Energy Rating Methodology 35

 t5 = 0.00691 8
-0.399912 · cos( a 0 ) + 0.070257 · sin( a 0 )
(2-99)
-0.006758 · cos(2a 0 ) + 0.000907 · sin(2a 0 )
-0.002697 · cos( 3a 0 ) + 0.00148 · sin( 3a 0 )

where
¢ = Site latitude (radians)
t5 = Sun declination (radians)
{i) = True local solar time (radians; see Equation (2-93))
ATAN2(x,y) = four quadrant arctangent function
a0 = Day angle (radians); position of sun relative to stars

2.1.6.3 Plane-of-Array Incidence Angle

The plane-of-array (beam irradiance) incidence angle on a fixed tilted surface is given
by:

() = cos-1 (cos(z) · cos(rp) + sin(z) · cos( A - A91 ) sin( rp ))

• (2-100)

where
z = Sun zenith angle (radians)
A = Sun azimuth (radians, measured from north, positive toward east)
Aq:> = Azimuth toward which plane-of-array is facing (radians)
rp = Angle of tilt of plane-of-array relative to horizontal (radians;
positive in northernhemisphere, negative in southern hemisphere)

Note that for a fixed latitude-tilt surface, the MER methodology assumes Aq:> will be rr in
the northern hemisphere, and 0 in the southern hemisphere.

For two-axis tracking structures, the incidence angle is assumed to be zero. Alternate
tracking position algorithms are not considered here because the rating procedure is not
anticipated to allow for them.

2.1.6.4 Plane-of-Array Incident Beam and Diffuse lrradiance

The primary inputs to the power model are plane-of-array (POA) beam and diffuse
irradiance. Unfortunately, these values are not normally available in weather data, so an

Module Energy Rating Methodology 36

irradiance model is needed. The contribution of the beam irradiance to the POA
irradiance may be computed as:

HPoA ,B = HB · cos( B) (2-101)

where
H8 = Beam irradiance (W /m2)
B = Plane-of-array incidence angle (radians)

The contribution of the diffuse irradiance to the POA irradiance is somewhat more
complex. Including an assumed uniform ground albedo and a uniform sky brightness
excepting a circumsolar region and a horizon band, Perez et al. developed the model
given below [6] for the diffuse POA irradiance:

(2-102)

a = max(O,cos(B)) (2-103)

b = max(cos(85o ),cos(z)) (2-104)

F1 ( E ) = max(O, F11 ( E ) + F12 ( E ) · b. + F13 ( E ) · z) (2-105)

F2 ( E ) = F21 ( E ) + F22 ( E ) . /).. + F23 ( E ) . z) (2-106)

(2-107)

(2-108)

m = ------------=-=-1
3 (2-109)
cos(z) + 0.00094 . ( 1.6389 - z f1 • 25

where
= Diffuse horizontal irradiance (W /m2)
= Angle of tilt of p lane-of-array relative to horizontal (radians;
always positive)
a,b = terms describing the incidence-weighted solid angle sustained by
the circumsolar region as seen respectively by the tilted surface and
the horizontal (unitless)

Module Energy Rating Methodology 37

 F/ &) circumsolar "reduced" brightness coefficient (unitless)
Fi &) = horizon "reduced" brightness coefficient (unitless)
FwF2i Coefficients for computing brightness coefficients, see Table 2-3.
& = sky clearness (unitless)
K a constant: 1.041 (1 /radians3)
z Sun zenith angle (radians)
Ll Sky brightness (unitless)
m Air mass (unitless)
Ho = Extraterrestrial normal incident beam irradiance (W /m2)
Table 2-3. Perez Model Brightness Coefficients
F F, F F, F, F,.,
1:>e:>l.065 -0.0083117 0.5877285 -0.0620636 -0.0596012 0.0721249 -0.0220216
1.065<e<1.23 0.1299457 0.6825954 -0.1513752 -0.0189325 0.065965 -0.0288748
1.23<e:>l.5 0.3296958 0.4868735 -0.2210958 0.055414 -0.0639588 -0.0260542
1.5<E<l.95 0.5682053 0.1874525 -0.295129 0.1088631 -0.1519229 -0.0139754
1.95<e:>2.8 0.873028 -0.3920403 -0.3616149 0.2255647 -0.4620442 0.0012448
2.8<e:>4.5 1.1326077 -1.2367284 -0.4118494 0.2877813 -0.8230357 0.0558651
4.5<e:>6.2 1.0601591 -1 .5999137 -0.3589221 0.2642124 -1.127234 0.1310694
6.2<E 0.677747 -0.3272588 -0.2504286 0.1561313 -1 .3765031 0.2506212

2.2 Required Input Data

This section describes the input data required by the various modeling components
described in section 2.1. While input requirements were discussed briefly with each
algorithm, this section discusses the specific parameters that need to be available.

These parameters are subdivided here into module characteristics and evaluation
characteristics. The module characteristics must be obtained by reduction of module
test data, while the evaluation characteristics specify the common conditions under
which all modules should be compared. The latter values are fixed as detailed by this
rating procedure.

2.2.1 Module Characteristics

The module characteristic input parameters are listed in Table 2-4 below.

Module Energy Rating Methodology 38

 Table 2-4. MER Required Module Characteristics
Parameter Units Model Use Notes
Incidence Angle Modifier Coefficients Reflectivity 1
Spectral Response Function A/W Spectral 2
Nominal Operating Cell Temperature oc Thermal
Efficiency at NOCT % Thermal
lrrm:;er Model Coefficients Power 3
II ]jffective Fixed Voltage v Power 4

1 The coefficients for the IAM equation are described in Section 2.1.5.
2 The SR function represents the response of the particular cell technology to spectrum with as
many points as necessary or feasible to characterize the response.
3 The coefficients for the power model are described in Section 2.1.2.
4 Since the fixed 14.4 V value may be modified by assuming a recommended number of batteries
and modules, this value is somewhat adjustable.

The spectral response and quantum efficiency of a module both quantify how a module
responds to various wavelengths of incident light. The spectral response is typically
presented as a function of wavelength with units of A/W, and the quantum efficiency is
also presented as a function of wavelength but with units of electrons/photon.
Quantum efficiency may be converted to spectral response using the equation:

q
SR(A-) = - . A-· QE(A-) (2-110)
h·c

where
q = Electron charge constant (1.60219·10-19 A·s)
h = Planck's constant (6.6262·10-34 J·s)
c = Speed of light (2.997925·108 m/s)
The spectral model expects the module spectral response represented as a series of A--SR
data points over the range from 300 to 1100 run. For the purposes of the MER, scaling
need not be accurate, as only the shape of the response function is needed. Therefore, a
relative SR obtained by multiplying a relative QE by A. is adequate.

The specific characteristics required will depend on the power model chosen, but in all
cases the characteristics will be available from the results of the module testing.

Module Energy Rating Methodology 39

2.2.2 Evaluation Characteristics

The evaluation characteristics are module load type, location and environmental data.
These parameters are defined by the MER, and applied, as appropriate, to all modules.
The latter two characteristics are interrelated by the selection of specific actual days at
specific locations to represent the various conditions under which the module is to be
evaluated.

2.2.2.1 Module Load Type

The power output of any module depends on the voltage or current at which it is
operated. Two load types representative of typical installations are specified.

The first load type is an ideal maximum power tracker. This load varies the voltage and
current as necessary to maximize the power output of the module. This load represents
the ideal behavior of a grid-connected de to ac power inverter, and represents the best
energy production output obtainable from the module under the given environmental
conditions.

The second load type is an ideal voltage source. This load maintains fixed voltage
representing a battery charging application. The voltage specified for this load type is
14.4 V, which is a common upper voltage limit for 12 V lead-acid batteries. This voltage
was not universally accepted by the TRC. It was suggested that a time-varying voltage
profile might be more useful for comparative rating and as an estimate of module
output. This is a topic for further development.

2.2.2.2 Location

The location is specified using the latitude, longitude and time zone. The latitude is
expected in degrees north of the equator (south is negative), the longitude is expected in
degrees east of Greenwich Observatory (west is negative), and the time zone is expected
in hours earlier than GMT (time zones for the United States are negative).

2.2.2.3 Environmental Data

The environmental conditions described in Table 2-5 below are functions of time and
location. Time is expected to be local standard time as reported in the location's time
zone. Many of these parameters are available in the National Solar Resource Database
(NSRDB), which is available as the Solar and Meteorological Surface Observation

Module Energy Rating Methodology 40

Network Compact Disc, or CD-SAMSON. This database contains hourly data from 239
sites throughout the United States.
Table 2-5 MER Required Evaluation Parameters

Parameter Units Model Use Notes

Site Latitude 0
Irradiance 1
Site Longitude 0
Irradiance 1
Time Zone h Irradiance 2
Date yyddd Irradiance 1
Time hh.hh Irradiance 1
Global Horizontal W/m2 Irradiance 1
Irradiance
Beam (Direct Normal) W/m2 Irradiance 1
Irradiance
Ambient Temperature oc Thermal 1
Dew Point Temperature oc Spectral 1
Barometric Pressure mBar Spectral 1
Wind Speed m/s Thermal 1
Cloud Cover Modifier Spectral 3
Coefficients
Std Solar Spectrum W/m2 /pm Spectral 4
Atmospheric Absorption Spectral 4
Spectra

1 These parameters are available in the National Solar Resource Database.

2 Number of hours from Greenwich Mean Time. For example, PST is -8.
3 SEDES2 provides 74 sets of 6 coefficients: A1, A2, B1, B2, C1, and C2 to compute cloud cover
modifier values.
4 Provided with SEDES2.

Module Energy Rating Methodology 41

3. MODULE CHARACTERIZATION PROCEDURES

The module characterization procedures consist of testing the modules and reducing
the test data to the form required by the model used in the MER computations.

3.1 Module Testing

Module testing is needed to obtain model parameters that characterize the module type.
These tests obtain raw data to characterize the module with respect to the power,
spectrat thermal and reflection models.

3.1.1 Power Model

The testing required to characterize the module for the power model depends on the
model chosen. However, parameters for each power model must be extracted from data
obtained either from indoor (laboratory) or outdoor tests.

3.1 .1 . 1 Laboratory Testing (Anderson, Blaesser, Interpolation)

The following procedure is used to determine module characteristics for use in the
Anderson, Blaesser, and interpolation models.

3.1 .1 . 1 .1 Background

The temperature and irradiance dependence of a sample PV module is required to be

able to translate from standard reporting conditions to any other set of reporting
conditions or visa versa. There are a wide variety of translation equations that have
been developed or proposed. Some translation equations only need to know the
percent change per °C or per W /m2 • Other translation equations are in units of volts,
amps or W per °C or per W /m2• At least one translation method does not use any
equations but relies on interpolation to a given temperature and total irradiance based
upon a matrix of I-V measurements vs. irradiance and temperature. To satisfy the
required data for all of these translation methods, the I-V characteristics as a function of
the total irradiance and temperature must be known.

The irradiance vs. temperature dependence matrix is generated by flash testing a PV

module at different temperatures and irradiances. Temperature is controlled using a
heating pad. Irradiance is controlled by stacking layers of an attenuating medium (such
as velum) on the module.

Module Energy Rating Methodology 42

3.1 .1 .1 .2 Procedures

Setup: The normal procedure of calibrating the intensity monitor using a reference
device and spectral mismatch error is followed. (1000 W /m2, ASTM E892 AM1.5 global
reference.)

Mount module: Once the monitor is calibrated, attach a thermocouple to the back of the
module and place the module on a heating pad. The thermocouple should have some
thermal insulation between it and the heating pad so that the temperature above the
heating pad is not measured. Attach the positive and negative leads from the current
and voltage terminals on the flash tester to the module. The height of the monitor must
be adjusted following normal procedures to be level with the front surface of the
module.

Measurements: All measurements on a given sample should be performed without

moving the module or resetting the monitor calibration. If the monitor is reset then a
discontinuity of as much as 1% in the current readings will result. This may be a
problem if the data is analyzed as a single set or the coefficients are in units of Amps or
Watts per oc. Measure the I-V characteristics at STC (25°C, 1000 W /m2, ASTM E892
global reference). Step A: Wait until the measured temperature changes by less than 1°C
over a 5 minute period. This helps ensure that the measured temperature is close to the
space-charge region temperature and the temperature gradients across the module have
stabilized. Step B: Measure the I-V characteristics and record the temperature as
measured with the insulated thermocouple and number of velum sheets along with the
usual information. Step C: Place the plastic velum diffuser sheets over the module (but
not the monitor), and repeat Step B at increasing or decreasing intensity as the sheets are
removed or added. The efficiency reported by the simulator will be wrong for these
measurements because the simulator still thinks that the module is seeing one-sun.
Choose the number of sheets for each step such that about 6 irradiance levels are tested
in the range from 0.15-1 sun (for example, 0 sheet, 1 sheet, 3 sheets, etc., to 9 sheets).
The temperature should not be changing as Steps B and C are repeated so
measurements are taken at a fixed temperature with varying irradiance. Step D:
Increase the temperature 10°C. Repeat steps A, B, C and D until the maximum desired
temperature is reached. Step E: Decrease the temperature and measure only the 1-sun I­
V characteristics following steps A and B until room temperature is reached. Note that
only 3 points are needed for this step. The purpose of this step is to verify that the IV

Module Energy Rating Methodology 43

vs. temperature characteristics are not a function of whether the temperature is
increasing or decreasing.

Data Quality Control: Prior to release of the data, the I-V curves must be analyzed.
This constitutes performing a linear least squares fit to the Voc, Isc, Pmax and FF vs.
temperature. The data should be linear for the single-junction PV technologies. The
temperature coefficient in ppm/°C or % ;oc should be computed so that the
temperature coefficient for this sample can be directly compared with other samples of
the same and different technologies. This data should be saved with the file name
linked to a test report number and sample ID or the manufacturer, and sample ID. It is
also useful if the technology is in the sample name. The data in the file and record book
should be commented sufficiently to be able to recover the raw data (filename or
manufacturer, sample ID, and date).

3.1 . 1 . 1 .3 Assumption and Sources of Uncertainty

This method assumes that Isc is linear with total irradiance. That is, incident irradiance
is estimated based on the ratio of the Isc measured with n velum sheets to the Jsc
measured with no velum sheets times the irradiance measured with no velum sheets. If
the 1000 W /m2 point is used in translation equations involving irradiance coefficients,
then a plot of Isc vs. irradiance must be linear and the 1000 W /m2 point must be the
greatest irradiance on the graph. This is because the spectral mismatch error is changed
between no velum sheets on the module and one or more sheets, so the spectral
calibration must be performed with no velum sheets. It is believed that the spectral
mismatch error does not change between one and more than one sheet because the
velum acts as a scattering medium with almost no specular transmittance. This method
assumes that the space-charge region temperature has been measured and that the
temperature is uniform across the module surface. Note that, substantial additional
time may be required if temperature coefficients measurements below 20°C are required
because of the time required to control the room temperature. These additional points
should not be required if the coefficients are linear with temperature.

3.1 . 1 .2 Outdoor Testing (Biaesser, Myers)

The Blaesser and Myers models have built-in thermal models and, therefore, need to be
characterized under actual conditions. In general, this means outdoors. Data should be
collected over a 4 week period that is characterized by diverse weather conditions

Module Energy Rating Methodology 44

including a high percentage of clear days. Testing should be done at locations and times
of year where the noon-time air mass is less than 3.
The following items must be addressed:

• module mounted at latitude tilt, facing due south.

• monitor back of module temperature, HPOA' Tamb' WS, V and I.

• measure IV curves every 30-180 seconds, average values of Isc' V IPP' V P Ifv along
oc' PP' PP

with ambient conditions every 10-15 minutes, and record a snapshot IV curve along
with the averaged data. Maintain module at peak power conditions between IV
curves.

3 . 1 .2 Spectral Model

Typically, a sample cell from the module's technology will be tested by exposing it to a
series of monochromatic light sources and measuring its short circuit current output.
This will yield a Relative (external) Spectral Response curve.

3.1.3 Thermal Model

The Fuentes model requires the module's Nominal Operating Cell Temperature. A
procedure for measuring and calculating NOCT is provided by ASTM
E1036/ Annex A1, "Test Method for Electrical Performance of Non-Concentrator
Terrestrial PV Modules and Arrays Using Reference Cells." If the test is performed per
the specification (i.e. with the module open circuited), the resulting module temperature
will have to be corrected to account for operation. Alternatively,. the module can be
operated at its peak power point during the test so that the correction does not have to
be applied.

3.1.4 Optical Model

Several approaches have been suggested for testing the influence of solar angle-of­
incidence on a module. The approach favored by the Technical Review Committee was
developed by David King of Sandia [King, 1996]. It consists of mounting the module on
an azimuth-elevation type controllable 2-axis solar tracker, tracking the sun in elevation.
The tracker orientation is varied in azimuth only to maintain (approximately) constant
diffuse irradiance and albedo while varying the incidence angle of the beam irradiance.

Module Energy Rating Methodology 45

The test is performed under a very clear sky with high direct-to-global irradiance ratio.
This the AOI influence is at its peak. The other extreme would be isotropically diffuse
irradiance where the module would exhibit no AOI sensitivity.

This test does not account for the effects of soiling on the reflectivity of the modules,
since it is performed quickly with clean modules. Other approaches that used longer­
term data were considered to be subject to an uncontrollable variation in soiling, which
would complicate reproducibility of the results. Since the MER is specified for clean
modules, this should not be an issue.

3.2 Data Reduction

Data acquired by testing the modules must, in most cases, be used to compute estimates
of model parameters. This section describes the key calculations to be performed on
test data for the models described in this report.

3.2.1 Power Model

The choice of power model used in the MER computation determines the parameters
that must be computed. Since errors introduced in this model by inaccurate parameters
or inappropriate models tend to dominate the error in the final result, the determination
of the appropriate parameters should be undertaken very carefully.

3.2. 1 . 1 Anderson Model

Solving the Anderson Isc and V translation equations for the temperature and
oc

irradiance coefficients a , fJ , and 5 :

( H)sc1/H1Isc2) - 1 (3-1)
a=

v[ oc1/Voc2( 1 + 5 ln( H1 /H2) )] - 1

fi = ��--�--------�-- (3-2)
Tl -T2
[voc1/Voc2(1 + fJ(T1 -T2))] - 1
-=--

5 - -"-------' (3-3)
---:---- -
-:---

ln(H1 /H2)

Module Energy Rating Methodology 46

Equations (3-2) and (3-3) are not independent, but if data are selected where H1 =H ,
2
then the 5 term in equation (3-2) drops out leaving

(3-4)

Similarly, if the data are such that T1 = T then the f3 term drops out of equation (3-3)
2
resulting in

(3-5)

These equations should be applied to the results of an IV vs. irradiance and temperature
indoor test. While an average of all combinations of pairs of different test points may be
chosen to compute a (as well as f3 and 8 subject to the stated constraints), averaging
only a selection of test points with widely differing inputs is likely to be most accurate.
Computation of the variance and examination of a residual plot should be used to avoid
errors due to outliers or systematic errors.

3.2. 1 .2 Myers Model

This model does not use the laboratory test procedures described in Section 2.1 to obtain
its coefficients. Instead, a set of measured data points must be collected from a field
test, including plane-of-array irradiance, maximum power output, and power output at
the appropriate fixed voltage conditions. This information can be extracted from full IV
curve data, though averaging may be advisable to reduce the data storage requirements
without undersampling. For the fixed voltage condition, the module may be assumed
to never draw current if the open circuit voltage is below the specified fixed voltage.
This assumption is justified if a charge controller disconnects the battery under low
irradiance conditions.

Two linear least-squares regressions of a quality-checked subset of the field data (P vs.
m

H; P vs. H) can be performed to obtain the appropriate equation parameters. To check

tv

the quality of the data, the irradiance must be greater than 0 (value subject to change?)
and less than 1425 W /m2, and the power values should be greater or equal to 0. No
fewer than (200?) values should be used for the regression.

Module Energy Rating Methodology 47

3.2.1 .3 Interpolation M odel

The only data reduction required for this model is the extraction of the maximum
power and fixed voltage power for each of the IV curves obtained from testing, with the
corresponding irradiances and temperatures.

3.2. 1 .4 Blaesser Model

Equation (2-17) of the Blaesser model contains the regression coefficients a, b, and c.

D v = a · ln(H r I H) + b · (Tamb - Tr ) -tc · H (3-6)

Dv = D V/ Vac ,r ; D V = Vr - V = Vac ,r - Voc (3-7)

Combining these three equations results in

1 - Vocl �c,r = a · ln(Hr f H) + b · (Tamb - Tr ) +c · H (3-8)

A linear regression can be performed using this equation and measured values of Voc' H,
and Tamb' along with the reference values Voc,r' Tr, and Hr. Since the Blaesser model is
used to adjust a reference curve to actual conditions, the above reference values should
correspond to the reference IV curve to be used. Best results will most likely be
obtained using one of the IV curves obtained during the outdoor testing representing
the average conditions. It is also possible to use an IV curve measured indoors for the
reference.

3.2.2 Spectral Model

If relative quantum efficiency data must be used, multiply each RQE value by its
corresponding wavelength to obtain relative spectral response. Also, to simplify
computations, the SR should be interpolated to obtain data at the wavelengths at which
the extraterrestrial (AMO) spectrum, H01 used by the SEDES2 spectral model is specified.

3.2.3 Thermal Model

Once the NOCT is obtained using ASTM E1036, equation 2-37 is used along with the
reference module efficiency to estimate INOCT.

Module Energy Rating Methodology 48

3.2.4 Optical Model

No data reduction procedures have been defined for this model, because no model has
been selected.

Module Energy Rating Methodology 49

 4. WEATHER DATA

One of the key ingredients in the Module Energy Rating is the selection of appropriate
weather data. The weather data should define extreme conditions that will allow
differences in module design and performance to be discernible. It would be fairly easy
to define the reference days by arbitrarily assigning hourly values for each parameter.
However, it was felt that users might find real data from specific dates and locations a
bit more descriptive, i.e. a "summer day in Phoenix". To emphasize the standardization
of the weather, the National Solar Radiation Database (NSRDB) was selected as the
source of the data. These data comply with the parameter requirements stipulated in
section 2.2, and are available on CD-ROM from NREL. Thus, a complete set of hourly
weather data is specified by a site (city name) and date.

Subjective criteria for the reference days were developed as shown in Table 4-1. Data
from the NSRDB were summarized and reviewed by Daryl Myers of NREL. He then
developed a list of days generally meeting each set of criteria, though primarily based
on the time of year, irradiance and temperature criteria. From this list, Myers selected
sites providing some geographic diversity. A summary of the selected days are
provided in Table 4-2 and Figure 4-1. According to Myers, these days are representative
of extreme conditions that can actually be achieved without being artificially extreme or
unique in themselves.

Table 4-1 Criteria for Reference Days

GHI DNI Tamb Wind Speed Rel

Hum
!Month Day Peak+ Total Peak Total Peak Wtd Peak Wtd Wtd
Hrs (W/m2) (kWh/ (W /m2) (kWh/ (OC) ( OC) (m/s) (m/s) (%)
m2) m2)
!Hot Sunny Sum >1000 10-12 >35 Low Low
Cold Wint >900 0.0 Avg Low
Sunny
Hot Sum <400 >35 Avg High
Cloudy
Cold Wint 200 - 400 0.0 High High
Cloudy
NICE Sprg 800 6 20 Avg Avg

t - Did not consider latitude and time of year.

Module Energy Rating Methodology 50

 Table 4-2 Selected Reference days

GHI DNI Tamb Wind Speed Rel

Hum
Day Location Month Day Peak• Total Peak Total Peak Wtd+ Peak Wtd+ Wtd+
Hrs (W/m2) (kWh/m2) (W/m2) (kWh/m2) CC) (OC) (m/s) (m/s) (%)
Hot Phoenix, Jun 15 1080 9.14 1003 11.9 39.4 36.3 9.8 3.9 6.3
Sunny AZ
Cold Alamosa, Feb 11 671 4.34 1032 9.4 3.3 -0.6 4.6 2.2 50.9
Sunny co
Hot Brownsville Jul 15 480 3.47 81 0.3 33.3 32.1 9.3 7.9 59.8
Cloudy TX
Cold Buffalo, Dec 10 237 1.32 2 0.0 0.0 -0.1 4.6 3.7 93.4
Cloudy NY
NICE Sacramento, May 14 905 7.12 846 8.3 19.4 17.3 5.7 5.0 65.4
CA

+ - Values are weighted with GHI

During the evaluation of the NSRDB, it became obvious that the subjective criteria were
somewhat self-contradictory. For example, the hot sunny day had higher wind speed
than desired. Data for Phoenix, Bakersfield, Dagget, and Las Vegas all showed
relatively high wind speed trends at temperatures above 40 °C. Similarly, the humidity
on the Hot Cloudy day seemed low. However, high humidity is accompanied by either
low temperature (relative humidity goes up as temperature decreases) or by high
irradiance (probably driving evaporation).

Note that comparing MER values from different days can be a bit misleading. One of
the main reasons PV systems provide less energy in the winter than in the summer is
the shorter winter day, not necessarily worse weather. Latitude exacerbates this
situation: as latitude increases the differences between summer and winter day lengths
increase.

The following figures show the weather data graphically. Tabular data for each of the
selected reference days are presented in the following sections. Tables in each section
provide the actual data from the NSRDB. Only selected parameters are presented: hour
(time of day) global horizontal irradiance (GHI), direct normal irradiance (DNI), diffuse
horizontal irradiance (Diff), ambient temperature (Tamb ), wind speed (WS) and relative
humidity (RH).

Module Energy Rating Methodology 51

 --- H S -----CS - - -HC ......... cc --- N IC E
1 20 0
N G l o b a l H o ri z o n ta l Irra diance / ......
s
1000
......
a: 800 /
v./"' ...... _...... �
tJ
=:
...
(lJ
600 // .. ; ...- -- ..... "" "
..... , '
'

....
"0
ttl
ttl
400 v /: ,......� /. ..... �'\'

' ....'\...
-

1-4
1-4 200 / v,/,_ . . -- - � . .. w .. '\.\�
;,· . . -
-C.�-�- .1.""'�
- - ..

"- �
- - ... _
_
.
_
- _ __
...
_ _

0 I
---

1 20 0
N D irect N o rm a l Irra d i a n c e
s
1000 - -

-.. ��
a: 800 )� A -..._
......

tJ
...
(l)
600 // /
-./ v 1\ i\
v

"0
=:
ttl
400
l! y '\. \
Iv/ \\\
ttl
1-4
1-4 200
I

/ _:_ - '\ \"

-

0
- -
- .... "'
I

50
A m b i ent Tem p er a tu re -i---t--t::=
u
40 :: T---t-----j
:: F====t===f:::::=-
0
_,......-::: - --.. -
-

aJ
-
30 - - - -
1-4
- - - - -

....=. 20
ttl
1-4 ---- -
(lJ 10 ----
c..
s
(lJ
0 --
-
-
-
- -- , - - - - - - -- - - - - - - - -- -
-- - - - - -----

E-< -10 -- - .... _

--

- - - - _ _ ,_.,.
-20

12
<ll
Wind Speed
...... 10
s
"0
...
8
I' ,.
.......
...,_ - - ...... 1/\
(lJ
(lJ
c.. 6 / "' ......_ ""'
/'I \
-
r:J)
"0
=:
4 .... ....
, .,.
�
�
/
. . .. .... _
(-\-
. -
. . .'--"
"" �
·� '-�
,.....
, ·:.:..,. ,L
·""" -

Q ; L'
.. -·
- .... -

v----::.
- -

,__,; �- --- -- · ·
�"--
...-
. .,,. ..
_ .. . .. .. _

a: 2 - ..... ..; . ..
. .

·� - - ' f.-/
.
_ _ _

''
--

.
-

'
---
·

0 .
0 /. 2 4 6 8 10 12 14 16 18 20 22 24

Figure 4-1 Reference Day Weather Data Plots

Module Energy Rating Methodology 52

4.1 Hot Sunny: Phoenix, AZ, June 15, 1976

This day exemplifies the summer in the desert southwest: hot, dry, and clear. Modules
with good thermal characteristics or low temperature coefficients should do relatively
well on this day.

Hour GHI DNI Diff Tamb ws RH

(W /m2) (W /m2) (W /m2) (oC) (m/s) (%)
0.5 0 0 0 28.9 0 15
1.5 0 0 0 27.2 2.1 15
2.5 0 0 0 25.6 3.1 21
3.5 0 0 0 24.4 3.1 18
4.5 0 0 0 24.4 3.6 20
5.5 32 166 15 24.4 3.1 16
6.5 172 624 35 25.6 1.5 15
7.5 388 802 53 27.8 4.6 16
8.5 608 893 70 31.7 4.1 11
9.5 805 949 84 33.9 4.1 7
10.5 959 982 93 35 6.2 5
11.5 1049 992 98 36 3.6 6
12.5 1080 995 101 37.2 4.1 6
13.5 1061 1003 99 38.3 2.6 4
14.5 960 982 93 38.9 2.6 5
15.5 807 950 84 39.4 3.1 5
16.5 617 906 71 39.4 3.1 4
1 7.5 397 821 53 38.9 5.7 5
18.5 175 636 35 38.3 9.8 5
19.5 34 175 15 36.1 3.1 7
20.5 0 0 0 35 3.1 8
21.5 0 0 0 34.4 1.5 10
22.5 0 0 0 33.3 1.5 12
23.5 0 0 0 31.7 3.6 15

Module Energy Rating Methodology 53

 4.2 Cold Sunny: Alamosa, CO, February 1 1 , 1 961

With the extremely high DNI and low temperature, this day should produce peak
.
p ower values. However, since it is in the winter, the short day length will limit module
energy.

Hour GHI DNI Diff Tamb ws RH

(W /m2) (W/m2) (W /m2) (oC) (m/s) (%)
0.5 0 0 0
1 .5 0 0 0
2.5 0 0 0
3.5 0 0 0 -15.6 1 79
4.5 0 0 0 -15.6 1 79
5.5 0 0 0 -16.1 1.5 75
6.5 0 0 0 -16.1 1.5 75
7.5 67 548 12 -14.4 1.5 76
8.5 251 882 22 -10 1.5 73
9.5 434 979 31 -6.1 2.1 68
10.5 566 1008 38 -1.7 1.5 53
11.5 650 1032 42 -0.6 1.5 56
12.5 671 1027 49 1.7 1.5 47
13.5 624 1027 41 2.2 2.6 48
14.5 523 1012 35 3.3 2.6 32
15.5 357 925 31 1.7 4.6 42
16.5 165 774 20 1.1 3.1 38
17.5 34 226 9 -3.9 3.1 62
18.5 0 0 0 -6.1 2.1 71
19.5 0 0 0 -7.8 3.6 70
20.5 0 0 0
21.5 0 0 0
22.5 0 0 0
23.5 0 0 0

Module Energy Rating Methodology 54

'_1 .3 Hot Cloudy: Brownsville, TX, July 4, 1 983

The medium irradiance levels and high temperatures of these conditions will
emphasize low temperature sensitivity.

I Hour GHI DNI Diff Tamb ws RH

(W /m2) (W /m2) (W /m2) (oC) (m/s) (% )
0.5 0 0 0 27.8 5.2 82
1.5 0 0 0 28.3 4.1 80
2.5 0 0 0 27.8 4.1 82
3.5 0 0 0 27.8 3.6 82
4.5 0 0 0 27.8 4.1 82
5.5 5 2 5 27.8 3.6 79
6.5 71 16 68 28.9 5.7 77
7.5 172 27 162 30.0 6.7 72
8.5 340 81 294 31.7 8.2 63
9.5 381 51 343 31.7 9.3 63
10.5 480 69 419 33.3 8.2 56
11.5 405 12 393 32.8 8.8 56
12.5 367 9 358 32.2 8.8 59
8 3.5
4.5
426
329
15
5
411
325
32.8
32.2
8.2
6.7
56
54
15.5 225 2 223 32.2 7.2 5?
16.5 152 1 151 31.1 5.2 65
17.5 88 1 88 30.6 7.2 70
18.5 24 1 24 29.4 5.2 75
19.5 1 1 1 28.9 4.6 74
20.5 0 0 0 28.3 4.6 8C
21.5 0 0 0 28.3 4.1 8C
22.5 0 0 0 28.3 3.6 8C
23.5 0 0 0 27.2 3.1 85

Module Energy Rating Methodology 55

4.4 Cold C loudy: Buffalo, NY, December 6, 1 985

Cold and cloudy conditions are particularly severe for photovoltaic power generation
because of the generally lower irradiance levels. These conditions will allow
performance comparisons for wintertime carry-through capability.

Hour GHI DNI Diff Tamb ws RH

(W/m2) (W /m2) (W/m2) (oC) (m/s) (%)
0.5 0 0 0 0.0 2.1 100
1.5 0 0 0 0.0 0.0 96
2.5 0 0 0 -0.6 1.5 100
3.5 0 0 0 0.0 1.5 96
4.5 0 0 0 0.0 2.1 100
5.5 0 0 0 0.0 2.1 96
6.5 0 0 0 0.0 2.6 100
7.5 12 0 12 0.0 4.6 100
8.5 73 1 73 0.0 3.6 100
9.5 134 0 134 0.0 3.6 96
10.5 187 2 187 0.0 3.1 96
11.5 221 2 220 0.0 3.6 92
12.5 237 0 237 0.0 3.6 92
13.5 196 1 196 0.0 4.1 92
14.5 146 1 145 0.0 4.1 92
15.5 91 1 91 -0.6 3.6 92
16.5 22 0 22 -0.6 2.6 89
17.5 0 0 0 -0.6 2.1 89
18.5 0 0 0 -0.6 1 .5 89
19.5 0 0 0 -0.6 0.0 85
20.5 0 0 0 -0.6 2.1 85
21.5 0 0 0 -0.6 2.6 89
22.5 0 0 0 -0.6 3.6 89
23.5 0 0 0 -0.6 3.1 89

Module Energy Rating Methodology 56

4.5 NICE: Sacramento, CA, May 4, 1 967

NICE is an acronym for Normal Irradiance-Cool Environment, a phrase coined by Jerry

Anderson of Sunset Technology. This is intended to be the "Goldilocks" day: not too hot
and not too cold.

Hour GHI DNI Diff Tamb ws RH

(W/m2) (W/m2) (W/m2) (oC) (m/s) (% )
0.5 0 0 0 8.9 2.1 89
1.5 0 0 0 8.3 2.1 90
2.5 0 0 0 7.8 2.1 92
3.5 0 0 0 7.2 2.1 93
4.5 0 0 0 8.1 2.9 92
5.5 38 101 28 9.1 3.8 90
6.5 163 303 85 10.0 4.6 89
7.5 344 491 123 1 1.9 5.0 83
8.5 513 594 145 13.7 5.3 78
9.5 708 732 151 15.6 5.7 72
10.5 811 693 214 16.9 5.3 68
11.5 905 846 132 18.1 5.0 65
12.5 811 689 179 19.4 4.6 61
13.5 854 833 132 19.4 4.8 59
14.5 733 832 91 19.4 5.0 56
15.5 588 785 91 19.4 5.2 54
16.5 400 712 69 17.4 4.8 61
17.5 200 491 64 15.3 4.5 67
18.5 49 245 25 13.3 4.1 74
19.5 0 0 0 12.2 4.3 79
20.5 0 0 0 11.1 4.4 84
21.5 0 0 0 10.0 4.6 89
22.5 0 0 0 9.6 89
23.5 0 0 0 9.3 89

Module Energy Rating Methodology 57

5. R ESULTS

This section describes the testing of a set of five sample modules, model validation
analysis, and MER results for the sample modules. For all of these activities, the same
sample modules and calculation database are used. These common elements are
presented briefly in the following paragraphs.

The basic characteristics of the sample modules are given in Table 5-1. These modules
represent a variety of single-bandgap technologies and a range of manufacturing dates
(1988-1994). Note that in some cases 1 or more modules (Nm ) and batteries (Nba1) were
ad

combined to obtain effective per-module fixed-voltage load values that would be near
the peak power point for the module. Due to the broad range of manufacturing dates
and the fact that our choices of numbers of batteries and modules have not been
approved by the manufacturers, the data presented here are for illustrative purposes
only. So while ultimately, MER results for commercial modules are intended to be used
to compare one technology to another, these results are for model validation only.
Table 5-1. Basic Characteristics of Tested Modules

Module Technology voc,STC vmp,STC Nbat Nmod vfv Eff. psrc

(V) (V) (V) (%) (W)
1 c-Si 21.7 17.4 1 1 14.4 12.5 53.0
2 mc-Si 21.3 16.9 1 1 14.4 9.8 55.0
3 a-Si 14.9 10.3 2 3 9.6 3.6 13.7
4 CdS/CdTe 90.2 67.7 4 1 57.6 6.4 53.7
5 CdS/CuinSe 22.6 16.1 1 1 14.4 6.2 31.4
2

Most data reduction and model validation calculations were performed with Microsoft
Access. Access Basic functions were developed and used in database queries that serve
as a simple method of applying these calculations to data in interactive or automated
modes as desired. This approach proved very beneficial for evaluating individual
models, by providing the complete results of each step and allowing for the substitution
of measured data as the input to each model. Once models are finalized, this database
could provide the basis for a user-friendly MER calculation tool.

5.1 Testing

Two types of tests were performed on the five sample modules. Indoor flash tests were
performed to obtain irradiance/temperature cross sensitivity data for determining the

Module Energy Rating Methodology 58

interpolation and Anderson model parameters. Outdoor testing was performed
primarily for validating the various models, but some of this data was used as well for
generating parameters for the Myers and Blaesser models. The following sections
describe the tests performed and present the parameters obtained for the models.

5.1.1 Laboratory Testing

The laboratory testing was performed on the five sample modules at NREL on the 17th
through 20th of July 1995. A Spire flash tester, heating pad, thermocouple, and velum
were used as described in the Laboratory Testing subsection of the Module Testing
section of this document to obtain irradiance/temperature cross sensitivity data. 1 Tables
in Appendix A summarize the indoor data collected using the Spire.

Representative spectral response data from technologically similar PV cell specimens

were used because obtaining spectral response characteristics from entire modules is
technically difficult. These data are discussed in the Spectral Model validation section.

The points extracted from the IV curves for the interpolation model are presented in
Table 5-2 and 5-3.

To compute the coefficients for the Anderson model, all possible pairs of appropriate
data points were extracted from the summary IV curve data and the equations
described in the Data Reduction section were applied to those pairs. The minimum,
average, and maximum values of these computations are shown in Table 5-4. There is a
fairly wide variation in these values, though the averages for the first three modules are
similar to the sample values quoted by J. Anderson. A brief investigation into the
origins of this variation showed some patterns that could indicate problems in the data
set, although the sources of these problems were unclear.

1 It is interesting to note that for both this data and earlier test results, the open-circuit voltage extracted
from the IV curve data by the Spire data system was found to be inaccurate for several curves. It was
suggested that a blocking diode could create this problem, but the presence of such a diode was never
confirmed. The Voc data used in this analysis was estimated with the Access database using linear
least-squares fit to IV data points near Voc.

Module Energy Rating Methodology 59

 Table 5-2. Maximum Power Interpolation Table (Watts)

Temperature Irradiance
(oC) (W /m2)

Module 1 253 487 773 1000

20.00 12.80 26.03 42.20 54.13
30.29 12.42 25.10 40.36 51.78
40.45 11.70 23.87 38.20 49.78
50.18 11.53 23.23 36.97 47.95

Module 2 257 485 762 1000

19.63 13.69 26.93 43.34 56.40
30.43 13.14 25.82 41.16 53.60
40.18 12.57 24.60 39.43 51.13
50.13 12.11 23.75 37.65 48.57

Module 3 205 443 741 1000

20.10 2.35 6.04 10.20 13.69
29.28 2.54 5.99 10.25 13.64
40.08 2.52 5.98 10.15 13.61
50.13 2.45 5.90 10.07 13.46

Module 4 237 456 741 1000

19.43 12.94 25.39 41.07 54.34
30.05 12.57 24.72 40.20 53.02
40.10 12.09 23.93 38.85 51.43
50.10 11.74 23.27 37.77 50.26

Module 5 255 485 767 1000

19.90 7.10 14.61 24.25 32. 19
1 30.88 6.54 13.76 22.85 30.66
40.20 6.09 12.75 21.52 28.79
50.58 5.56 11.99 19.98 27.07

Module Energy Rating Methodology 60

 Table 5-3. Fixed Voltage Interpolation Table (Amperes)
Temperature Irradiance
(oC) (W/m2)

Module 1 253 487 773 1000

20.00 0.77 1.53 2.45 3.20
30.29 0.78 1.53 2.45 3.17
40.45 0.75 1.50 2.40 3.17
50.1 8 0.77 1 .53 2.42 3.17

Module 2 257 485 762 1000

19.63 0.84 1.60 2.57 3.40
30.43 0.84 1.60 2.56 3.39
40.18 0.83 1.60 2.56 3.39
50.13 0.82 1.59 2.56 3.34

Module 3 205 443 741 1000

20.10 0.24 0.61 1.04 1.40
29.28 0.25 0.61 1.04 1.40
40.08 0.25 0.61 1.05 1.40
50.13 0.25 0.61 1 .05 1.39

Module 4 237 456 741 1000

19.43 0.21 0.40 0.65 0.87
30.05 0.20 0.40 0.65 0.87
40.10 0.20 0.40 0.64 0.86
50.10 0.20 0.39 0.63 0.85

Module S 255 485 767 1000

19.90 0.49 0.99 1.61 2.12
30.88 0.45 0.95 1.56 2.08
40.20 0.39 0.88 1.48 1.98
50.58 0.29 0.80 1.37 1.87

Table 5-4. Ranges o f Anderson Coefficients Found

a fJ I ) 5
(oCI ) (oC ( -- )
Module Min Avg Max Min Avg Max Min Avg Max
1 -0.003718 0.000066 0.003769 -0.0038 -0.0033 -0.0029 0.0405 0.0496 0.0689
2 -0.0004 0.000333 0.000986 -0.0041 -0.0034 -0.003 0.0391 0.0536 0.0612
3 -0.000437 0.000895 0.00178 -0.0033 -0.0018 0.0045 0.0474 0.0588 0.0895
4 -0.000585 0.000167 0.000596 -0.0029 -0.0023 -0.0017 0.0349 0.0458 0.0545
5 -0.001349 -0.000021 0.001067 -0.0056 -0.0047 -0.0039 0.0888 0.1134 0.1315

Module Energy Rating Methodology 61

The average values of the computed Anderson coefficients were used for subsequent
modeling. Table 5-5 presents these values, along with the corresponding reference point
data that was used. The IV curves from which these points were derived were used as
the basis for translation to other conditions.
Table 5-5. Chosen Anderson Model Parameters

Module a fJ 6 Hl Tl vocl Jscl pmpl

CC1) (oCl) ( -- ) (W /m2) (oC) (V) (A) (W)
1 0.000066 -0.00326 0.0496 1000 20.0 22.0 3.32 54.1
2 0.000333 -0.00336 0.0536 1000 19.7 21.6 3.57 56.4
3 0.000895 -0.00176 0.0588 1000 20.1 15.1 1.76 13.7
4 0.000167 -0.00232 0.0458 1000 19.2 91.0 0.92 54.3
5 -0.000021 -0.00474 0.1134 1000 20.0 23.1 2.33 32.2

5.1 .2 Outdoor Testing

The outdoor test consisted of installing the sample modules in a latitude tilt
configuration and simultaneously recording module IV curves, weather parameters,
and global plane-of-array spectral characteristics. Data collection was performed at the
NREL Outdoor Test Facility in Golden, Colorado, from the July 28 through 31, 1995.

The weather parameters were sampled at 5-second intervals and averaged every 10
minutes. These parameters included beam irradiance (cavity radiometer), diffuse
horizontal irradiance (shaded thermopile pyranometer), plane-of-array irradiance
(thermopile pyranometer), wind speed, barometric pressure, relative humidity. Figure
5-1 shows selected parameters for the four days of testing. Note that data collection
commenced mid-morning on the 28th and only a few data points were collected in late
afternoon of the 31st.

Module Energy Rating Methodology 62

 1200 �-
---�

- DNI
' Diffuse
-
1000 - - - - - - '- - - - - - -' - - - - - - '- - - - - - -' - - - - - - '- - -

-
N. 800 -
'
- - - - - -
'
- - - - - - -
'
- - - - - -

�
-
� 600
J::
«!
. ....
"'0
«!
-t 400

200

0 +---�4-_J�-r���--�-r�L_����--��
7/28/95 7/28/95 7/29 / 95 7/29/95 7/30/95 7/30/95 7/31 / 95 7/31 /95 8 / 1 / 95
00:00 12:00 00:00 12:00 00:00 12:00 00:00 12:00 00:00

40 �------�--�====� 8
- Temperature
35 Wind 7 -

30
-
u
0
- 25
Q)
l-<
:=
....
«! 20
l-<
Q)
P..
s 15
Q)
�
10 2
' VI
5 : - - - - -:1�/(
f LV\JAV_j; -
\ 1

0 +-----+-----4---4 0
7/28/95 7/28/95 7/29 /95 7/29/95 7/30/95 7/30/95 7/31 /95 7/31 /95 8/1/95
00:00 12:00 00:00 12:00 00:00 12:00 00:00 12:00 00:00

Figure 5-1 Validation Weather Data

Module Energy Rating Methodology 63

Module measurements, for each module, included maximum power and module back
temperature sampled at 5 second intervals and averaged at 10 minute intervals, as well
as complete IV curves sampled at 10 minute intervals.

Finally, a LI-COR spectroradiometer was configured to sample in-plane global spectra

every two minutes during daylight hours in the 300 to 1100 nm. wavelength band. For
the purposes of validation computations, these spectra were subsequently averaged
point-by-point at 10 minute intervals.

This rather extensive data set was primarily collected for validation purposes, but data
from the 28th was also used to obtain model parameters for all five modules. While
some outdoor data from earlier testing was available, it only included data for the first
three modules. The parameters obtained for the Myers and Blaesser models are given
in Table 5-6 and Table 5-7.

Table 5-6. Linear Irradiance-Only (Myers) Model Parameters

Module a b
(m2) (W)
1 0.0547 0.0481
2 0.1435 0.0485
3 0.0164 0.0135
4 -0.0674 0.0429
5 -0.1111 0.0245

Table 5-7. Blaesser Model Parameters

Module a b c H/,r Tl voc,r Isc,r

(--) ( C)
o l
(m2 /W) (W /m2) ( C)
o
(V) (A)
1 0.0562 0.00250 0.0000733 1000 20.0 22.0 3.32
2 0.0486 0.00383 0.0000716 1000 19.7 21.6 3.57
3 0.0408 0.00423 0.0000368 1000 20.1 15.1 1.76
4 0.0483 0.00483 0.0000892 1000 19.2 91.0 0.92
5 0.1079 0.00578 0.0001251 1000 20.0 23.1 2.33

Module Energy Rating Methodology 64

5.2 Validation Analysis

A crucial aspect of this project was the validation of the various models used to estimate
module performance under the prescribed MER conditions. A significant effort was
expended generating the required validation data. Because of the need to evaluate each
modet extensive measurements were made.

At the outset of the project, the intent was to use existing module and array
performance data from NREL's outdoor test facility in Golden, Colorado, PG&E's
Photovoltaic Test Facility in San Ramon, CA, and the Photovoltaics for Utility Scale
Applications project in Davis, CA. Initial validation results presented to the TRC in
March 1995 suggested that the irradiance measurements from these projects, based
solely on global in-plane thermopile pyranometers, and the lack of measured spectra
would not provide the required accuracy or level of detail necessary to fully evaluate
the models. Thus, the outdoor measurements described in the Testing Section were
made.

5.2.1 Power Models

The four power models described in the DC Power Model section were evaluated using
results from the precursor models (irradiance, spectral, etc.), as well as using various
measured parameters from the validation tests.

Models were evaluated by comparing integrated modeled power over each of four
validation days to measured integrated power (daily energy), as well as comparing
individual modeled power values to measured 10-minute averages. The results of the
analysis, discussed below, are based on measured inputs: module temperature, POA
irradiance, and spectral content/ correction. A comparison of measured POA irradiance
to DNI/ diffuse derived irradiance is discussed in the Irradiance Model Section.

Peak power results for four models, five modules and four days are shown in Figure 5-
2. Similar fixed voltage results for Anderson, Blaesser, and Interpolation models are
shown in Figure 5-3. (We did not attempt to generate fixed-voltage coefficients for the
Myers model.) Though there are a few bright spots, these results are generally
unsatisfactory. Depending on the model and module, peak power model error ranged
from 2% to 48% as shown in Figure 5-4. This chart shows the aggregate error for the

Module Energy Rating Methodology 65

four days of validation data (sum of estimated energy for all four days divided by the
measured energy, quotient minus 1).

Figure 5-5 and Figure 5-6 compare modeled to measured power and model error as a
function of irradiance for each 10-minute average point for module 1 . Figure 5-7 and
Figure 5-8 show similar data for fixed voltage operation. The corresponding plots for
modules 2 through 5 are provided in Appendix B.

These results show that overall, the Anderson model performed the worst and the
Myers model performed the best for peak power prediction. Note that both the Blaesser
and Myers model coefficients were generated based on the July 28 data set, so we
would expect them to predict that day fairly well. Conversely, data for the Interpolation
model and the coefficients for the Anderson model were based on indoor
measurements.

Module Energy Rating Methodology 66

 Module Energy Rating Model Validation
Maximum Power
II Measured i� Anderson Ill Blaesser B Interpolation i� Myers
400
350 7/28/95
300
..t::
--

� 250
;;...,
bO 200
lo<
Cj)
:::: 150
rJ;,l
100
50
0
300
7/29 /95
250
:2 200
�
;;...,
bO 150
lo<
Cj)
::::
rJ;,l 100

50

0
250
7/30 /95
200
:2
� 150
;;...,
bO
lo<
Cj) 100
::::
rJ;,l
50

0
50
7/31 /95
40
:2
� 30
;;...,
bO
lo<
Cj) 20
::::
rJ;,l
10

0
1 2 3 4 5
Module
Using measured weather, spectrum, Tmod and POA; Modeling power only

Figure 5-2 MER Power Model Validation, Maximum Power

Module Energy Rating Methodology 67

 Module Energy Rating Model Validation
Fixed Voltage
II Measured E!Anderson II Blaesser Iii Interpolation
25
7/28/95
20

15

10

0
16
14
12
10
8
6
4
2
0
16
14
12
10
8
6
4
2
0
3
7/31/95
2.5

1.5

0.5

0
1 2 3 4 5
Module
Using measured weather, spectrum, Tmod and POA; Modeling power only

Figure 5-3 MER Power Model Validation, Fixed Voltage

Module Energy Rating Methodology 68

 Module Energy Rating Methodology Model Validation
Power Model Error

Module 5

Module

Figure 5-4 MER Power Model Error, Maximum Power and Fixed Voltage

Module Energy Rating Methodology 69

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 1

i
60
U I I I I
...--..

:s: 50
'--"
• Anderson I t--- I
1-< I
(1)
� 40
--�� I I
0
� I
p,.
"1j
(1)
.....
30
J�� l II '

I
<1l 20
.§ 10 •
•• r� I I I
(/)
.....

� I I I
. ft

0 �
0 10 20 30 40 50 60

...--..
60
�
1-<
so
(1)
� 40
0
p,. 30
"1j
(1)
.....
<1l 20
.§ 10
(/)
.....

�
0
0 10 20 30 40 50 60

60
•
�1-<
50 Interpolation
(1)
� 40
I
I
0
p,. 30 I
"1j
(1)
.....
<1l 20
.§ 10
(/)
.....

�
0
0 10 20 30 40 50 60

60
�1-<
50
(1)
� 40
0
p,. 30
"0
(1)
.....
<1l 20
.§ 10
(/)
.....

�
0
0 10 20 30 40 50 60
Actual Power (W)

Figure 5-5 MER Power Model Comparison, Estimated Max. Power vs Actual

Module Energy Rating Methodology 70

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 1

0 I i
. ..0 . I
I/
50%
I
30%
·. .,. �,..

•
0
o.
..
I I I
I
o Anderson 1/
;...0.
;...;..... 10% ·� 0 .o
•
" "
lo
0 •o • 0
• •
. 11
I•
I
o -.1' 0:.• o ... .lo,..
I • .,_
I
o •"�.&....,. �I'll> ,: • eo .,..,.,.,.
I
I• • •
�
0
-10% I
I
l
i
..
I :

I I
�
II I
-30% I
I
I I
I I II
-50%
0 200 400 600 800 1000 1200

•• • I
I
50%

30%
•
•
a.
·, · . ••
I I
I I I I
I o Blaesser I
;...0.
;...;..... 10%
..
'�.r
•
.
"
• . ·.
I I
I .. .. .
II
I
II I
I ..o.u.
I
I
..

·l • • 'lloo• �... ....... . . ..o..o..r

� r-•? - .. ..
0
-10% .,.. � .
. ·' � . r � '""
I
0
•

I
�

-30%
� ·.r::: �\·.
I I
I
I
I
I

I
! I I I j
-50%
0 200 400 600 800 1000 1200

I I I
I .' Interpolation I !
50%
I
I
II I
I
i II
• • ••• •• • ! i i
30%
o I i
;...0. l 0o i I I I I
;...;.....
�
10%
••:.·
!
I

JIJ. .. .
o-
•
•

• • ..! �CI. .. ..,

' .
- tHf "' •• ,...,.
J
- �-...
0 .'I.
.,
L
1-
... ... , . ........ lo•
I
I
• II
0
-10%
• I !
I
�
-30% � i
I
I
I
I
I I
I
r I
I

-50% . I I I I I

0 200 400 600 800 1000 1200

50% .. . I I
'(
yo
.

• •• II
i I
I
I
I
I
i / • MyersJ I
I
30%
\ • •• 0 ! I I
i
;...0. :."·��
'" i Oo
I
I

i I
I I
I
•
!
I l
;...;.....
�
10% ::· . '

• · "' • .
,I
•
. .
I
I • •
I
I
o 'lot- ,..... ... • .. ,.;.a j
1 ..
�
I
I
. :.·.,.
foo"1&: 0 •• 't . J .t .. . .
.
-�·
'
....
0
� -10% I 1I I
I !
I i ! I
I
-30%
I
I
-50% i i !
I
I I
0 200 400 600 800 1000 1200
Irradiance (IN/m2)

Figure 5-6 MER Power Model Comparison, Max. Power Error vs Irradiance

Module Energy Rating Methodology 71

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 1 Current at 14.4V (nominal)
4.0
J • Anderson I I II I
.......,
...c:
3.5
II · I
I

• � •
�
.

·
I

I
$ 3.0 � •
.....
�
(I)
0 • � •• 0
• • 0 .
oi 0

II � I � • •
2.5

u
.....
.....
::s 2.0
0

..... .,.
1:.
0 I
"d I
0
J ol 0 • I
(I)
.....
(1j

.§
1.5
:"1 "l�J�·· ·· .

I • • • I I
..... 1.0 0 •
00
j;.il
�� � - .� .I (o I
I
0.5
·
0.0 � ·I I
.

I
'
I I
• • I
I
•

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

4.0

3.5
J • Blaesser I v •
I
I
i
I
::2I

--< 3.0
I �I I' .
•
"-'

.....
J I · �· • ·: •
. ....
•
• I
0 I
�
(I)
.....
.....
2.5
I· i � i r. · · I I I
i • 0 0
• 0 I
I

u
::s

"d
2.0
+ � 1 ..... I I
0

• • r'
'
# 0

(I)
.....
:·1 � I !
• I' •o ,I o o o o""
0

!
1.5
(1j

.§ I I
I I I •
I •
'!
Cillo I

I
0

.....
• • I
1.0
00
j;.il
0.5 •
� 1·:� . I ..
• I
f.!.. "-�
0.0 � ·� . . I I I I
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

4.0
• Interpolation
,.
3.5
::2 ' •
I
•
$ 3.0
.....
�
(I) 2.5
.....
.....
::s
u
2.0
"d •
(I)
..... 1.5
(1j

.§
.....
1.0
•
00
j;.il l
I
0.5
II •
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Actual Current (A-h)

Figure 5-7 MER Power Model Comparison, Model Estimated A-H vs Actual

Module Energy Rating Methodology 72

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 1 Current at 14.4V (nominal)
•
.
• !I
50%
I I !
I

40%
•
.. I I
I
o Anderson Jl
•II
I
I I
I
I

I
30% • �
I
I
• •
..
I I •I
•. . .
20%
....
• • ••• II
.
•
I . I I
0
10% I
• I • • I• •
I
I• I
I
• • • ;•V• JI
....
....
w 0%
...
• : • •,:. .. � , I
I
l .o o
o • I
!
. ... . � o • •
••
A

I
0
� -10%
• . o 0 .....
...• :- ••
I

.. � •
i ' :. .. • r·
• I
• • • - . ..
• I • • I
• I ••
'
.. :i. . • •• I •
-20%
.
�
I 1 I
-30%
• • • .<t • II I • • I I
I
I
-40% • . • .. I I I

I
-50% I
..
I • I I I

0 200 400 600 800 1000 1200

50% • • I • I I • Blaesser�
• •• I I I
II I
40% 0
I I II II iI I
•. • • I

•••• I
30%
,
·I II I I I
• .
0 I I
20% I

....
0 0' I
I I I I I i
0
....
10%
....
0
.
. I
I • • I 0•
0
1 . I.
I
I
....
0% • . . r • • • •
I
0
w Oo
'• ,:..
0
I"o -...-. av-. I I I
•
o

• 0• o •• � .. o. l oo ••
0
• I 0 •
o
•
0 o
I
.,
� •

I
O o o •• o0
I � · ..
-10%
•
I
0
"' •
t
o o
. .. I
I ! I
-20%
. . . ..; . • II
.: - •

I • I
I • i •
0
I
I
I
I
. -· 0

•. .:..;. • . ..
I • •• I I
-30%
I I I I
-40%
\
•
• •• •• i I
I I I
I
I I I
I
i
-50%
0 200 400 600 800 1000 1200

• •• • I •
50% I i
40% I
I
I I
I
r 0 Interpolation
I
I
I
i ! II I I
30% • I

I I I
• I
I I
1'
20%
0 ... II
.. I I
....
0
....
10% • • • .• • I I

•
• ·I
I
I I• I
!
•I• •
....
w 0%
0 .o . .•
•
-· I I

•
. •
0
10 o o
0 0 o
o• oliro I
I
J
•••• 0 0. • i
0
• • 0 ••• • • •.• � - .,o· · : • I
1 e
�
0
• I
-10% 0
"' ' 8 . I
I 0
I !

• •• c;••• •
0
-20%
. I •
!

I � . •
• • • • :o i
. I
I

0
I
I
I
-30%
-40% •• J • I
I
! . . i !
-50%
0 200 400 600 800 1000 1200

Irradiance 0N/m2)

Figure 5-8 MER Power Model Comparison, Model A-h Error vs Irradiance

Module Energy Rating Methodology 73

The Anderson model proved to be more competitive with the other models in
predicting fixed voltage performance. Figures 5-7 and 5-8 show that the integrated
results are a bit misleading. Both of these plots suggest all of the models tend to slightly
under-predict current at moderate irradiance levels. However, at low irradiance they
tend to significantly over-predict. For example, at a measured current of 1 A the
estimated current can be as much as 3.5 A high. Over a day, that +2.5A absolute error
from one low-irradiance value offsets ten of the more typical -0.25A moderate irradiance
errors.

A final general result is that, as expected, all of these models are better at predicting
module performance under high irradiance conditions than low irradiance conditions.
They also tend to be more accurate for crystalline silicon modules than for thin film
modules.

5.2.1 .1 Anderson Model

A rather unexpected result was the relatively poor performance of the Anderson model.
Despite its use of measured module temperature and a derivation very similar to the
Blaesser model, it did not fair as well. The primary difference between these two
models is the use of indoor flash test data vs. outdoor data for coefficient derivation.
The Blaesser and Myers models also benefited from the use of nearly the same data for
both coefficient derivation and model validation

Simply because of the model's poor performance here, we suspect these results. The
fact that it tends to predict the highest energy (both peak power and fixed voltage)
suggests that this model is not capturing all of the "losses" outdoors. The one
parameter that we elected to ignore was the incidence angle modifier. The lAM reduces
the irradiance input to the power model as a function of incidence angle to account for
the modules front surface reflectivity. Since the Anderson and Interpolation models are
based on flash tests done at normal incidence, they would tend to over-predict power at
high incidence angles. The use of pyranometer measured POA irradiance rather than
DNI/Diffuse derived irradiance would reduce that error somewhat because of the
pyranometers own attenuation at high incidence angle. However, Figure 5-6 shows
consistent over prediction under high irradiance, low incidence angle conditions as well
as overcast, "no" incidence angle conditions.

Module Energy Rating Methodology 74

Another potential source of error is in the characterization data. Any spectrum or
uniformity errors associated with the sheets of velum used to attenuate the flash
intensity will be accentuated at the lowest irradiance levels where module performance
becomes non-linear. The velum used was selected because of its uniformity and some
testing was done to insure that spectral changes were minimal.

Any model can be quite sensitive to the procedures used to generate and evaluate the
data for the coefficients. This comment appears to be especially true of the Anderson
model. The procedures used here should, therefore, be further reviewed and refined.
Also, additional validation data are necessary to evaluate each of these models under
the broad range of conditions with which they are intended to be used.

5.2. 1 .2 Blaesser Model

Both the Blaesser and Myers model coefficients were generated based on the July 28
data using the measured POA irradiance without spectral correction. In spite of these
simplifications, the Blaesser model performs relatively well coming in a close second to
the Myers model in overall peak power performance. Fixed voltage estimates are lower
than Anderson estimates in all cases and are lower than the Interpolation model in most
cases.

5.2. 1 .3 Interpolation Model

The most notable behavior of the Interpolation model is its error at low irradiance
(Figure 5-6). While the error for the other modules becomes more positive at low
irradiance, the Interpolation model error becomes more negative. The weakness of the
interpolation model is that you must actually extrapolate to obtain points beyond its
measurement range. At the low irradiance end, linear extrapolation is used where the
performance is becoming non linear. Any error in the four points used for extrapolation
will be magnified for large extrapolation. Thus, a requirement for use of the
Interpolation model is the to have measurements covering the entire range of expected
weather conditions.

5.2.1 .4 Myers Model

Equally unexpected to the Anderson's poor results were the Myer's relatively good
results. Myer's contention-if the error associated with measuring and characterizing a
given parameter is greater than the contribution of that parameter, don't include it in

Module Energy Rating Methodology 75

the model-appears to ring true. Since the model is based on a regression analysis of
one of the validation days, July 28, it's not surprising that it works well for that day.
However, July 28 was the clearest day and the model works equally well for the July 31
which had roughly 1 /10 the insolation. Since the module only takes into account
irradiance, we would expect it to show some error on days with similar irradiance
levels but different ambient temperatures. For the data set utilized, lower irradiance
levels tend to occur at lower temperatures, so this effect is masked. Again, a wider
range of validation data is necessary to reach a supportable conclusion.

5.2.2 Spectral Model

A validated spectral model was not absolutely required in order to compute the MER,
because it is used to define the (otherwise unavailable) spectrum for the reference days.
The spectral model must generate reasonable spectra, but once the spectra for each
reference day are specified, the model is no longer needed1•

However, to most accurately characterize the input conditions for validating the power
models, measured spectra were used. Measured spectra and individual spectral
response characteristics were combined to obtain spectral correction factors (SCF) for
each data sample. The SCF is then combined with measured irradiance to obtain
effective irradiances for each of the modules, which were in tum used as inputs to the
power model along with measured module temperature. The spectral correction factor
computed from the measured spectral data is shown in Figure 5-9. These spectral
correction calculations were limited to the measured bandwidth, which implicitly
assumes that the actual spectra at unmeasured wavelengths were no different than the
corresponding reference spectra.

As part of the investigation of spectral effects in the modeling process, a number of

additional plots were generated. Figure 5-10 includes plots of spectral responses for the
sample modules, sample spectral shapes, and a comparison of the predicted spectrum
and measured spectrum for one sample on the cloudiest outdoor test date. For
example, comparing the spectral response shape of module 3 (a-Si) with the noon and
evening spectral shapes, one can see that the lower proportion of energy in the higher
wavelengths in the evening will make the module appear to be more efficient at

1 In reality, the model may still be necessary since the spectra vary with orientation and it would be
cumbersome to provide spectra for all conceivable orientations.

Module Energy Rating Methodology 76

converting the available irradiance. Note that the reddening effect apparent in the
evening sky is limited to direct and circumsolar diffuse irradiance. For the fixed south
facing orientation used in this analysis, the spectrum is dominated by the more blue
diffuse sky component (see above footnote). In addition, the comparison of estimated
and measured global spectra shows that the spectral model produces reasonable data.

Figure 5-11 shows the effective irradiance for the sample modules on the MER reference
days, and the net change in effective insolation (kWh-m2 I kWh-m2) that results from
application of the spectral correction factor. Note the small rise in estimated plane-of­
array irradiance near sunrise and sunset; these are artifacts of applying the Perez
translation model to averaged irradiance data. These artifacts cannot be completely
eliminated given that averaged data are used. The net change in effective insolation (or
equivalently, change in effective efficiency) for module 3 (a-Si) shows improvement for
medium and higher temperatures, with a corresponding reduction for cooler
temperatures. This effect has been observed by T. Townsend in comparisons of a-Si­
based PV systems in tropical and temperate climates [personal communication,
September 1995]. The order of magnitude of the effect is about 4% for a-Si and 1-2% for
the remaining modules. However, it is unclear from this analysis if there is a change in
spectral response due to temperature, if there is simple spectrum/ temperature
relationship, or if there is some other mechanism acting here, such as a-Si annealing.

Module Energy Rating Methodology 77

 Module Energy Rating Methodology
NREL Outdoor Test Facility

-- Module 1 - - Module 2 • • · · • ·Module 3 - - - Module 4 - - - Module 5

1.2 ,-------,--,---.

'
I
I
•
•
•
•
1.15 +------1---....--+---1
•
�
�
� I
,, (
,,,, \'
;(
•'
'
•

"
•
'
�:
.
" .
" I
'.
' •

I
. '
..
,•

'
I
. '
·' : ' '
'.
'

'. ••
;..
1.05 fll . .
'
II I
t

}'11 . I :
I

I f't .
I I . I ,.
fi t
I 4 �. I
""' I ttl. I � l�l
�w
tt�
0 :: t:
.... 1\/"'1 � ••• I �'• tI l
<;.1 \ \
•

Ill . . ..."
J;o,. I £ \'f : � : .......
=
1
0
.....
....
<;.1
(IJ
""'
��.
""'
0
u I
-
Ill
...
0.95
....
<;.1
'• ' I
(IJ
!:l.. I
r:J)

0.9
I I
I.
I I
II I
;� -----1--+------+--------+--------1
0. � +-��
I'
.I
li
0.8 +-�,------�---+-l--+---�--�

0.� +-------4---�---+---�
7/28/95 7/29/95 7/30/95 7/31/95 8/1 /95

Figure 5-9 Spectral Correction Factors Based on Measured Spectra

Module Energy Rating Methodology 78

 Module Energy Rating Validation
Spectral Model
-- 1 --- 2 _ _ _ _ _. _ 3 ---- 4 ----5
1.2
Module Spectral
1 Response -- \
- �

� .
V'.J .
"0
V"\
0.8
QJ
-�
-
I'd
. \
0.6
e
1-<
.
.
0 .
z
0.4 I \
\ \
'
. \.
0.2
' .......
-- -.....
c-- .
0
0 0.25 0.5 0.75 1 1.25 1:5
Wavelength (p.m.)

1 .2 T-;::----:--;:::---:-:�-----------�================::::;l
-- 5 I4/67 12:00:00 PM
Sample Spectral Shapes

- - - 5/4/67 6:00:00 PM
- - - - - - ASTM E892 (AM1.5)

0 0.25 0.5 0.75 1 1.25 1.5

Wavelength (p.m.)

200
180 Global Horizontal Spectra
(7/31/95 12:00)
6::t. 160
-- Estimated
� 140
e
..._ 120
...__ Measured

� 100
>..
:::: 80
rJ'J
;::::
QJ
....
60
;::::
...... 40
20
0
0 0.25 0.5 0.75 1 1.25 1.5
Wavelength (p.m.)

Figure 5-10 Spectral Response and Spectral Model Results

Module Energy Rating Methodology 79

 Module Energy Rating
Spectral Effect
-- Module 1 - - Module 2 - - - - Module 3 - - - Module 4 - - - Module 5
1200 ,-----.--.---,--�--�
Effective Irradiance

"'
800
e
�
a)
<;J
;:::
600
!U
.....
"'C
!U
;..,
;..,
-
400

0 +---���---L�+-----r---�--�--�-L--����
04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00

Effect of Spectral Response

----+--- on Effective Insolation

11 3 11 4 11 5

HC cc NICE cs HS

Figure 5-11 Effective Irradiance

Module Energy Rating Methodology 80

Figure 5-12 shows the spectral correction factor for the sample modules on the MER
reference days. The small spikes before and after sunset are associated with the
Perez/ averaged data artifacts previously discussed. Note that the a-Si module reacts
more strongly than the other modules, and sometimes increases when the other
modules are not affected or are decreasing.

5.2.3 Thermal Model

Though NOCT can determined using a standard ASTM-defined procedure, we did not
perform that procedure nor were NOCT values available for all of the modules tested.
Since the modules were relatively similar in construction a common 47 oc was used for
all modules. Figure 5-13 shows the measured module temperatures for all 5 modules.
The data-10-minute averages-shows little in the way of steady state performance, but
there are some instances where the transients are not too severe. These instances
suggest that the difference in temperature between the 5 modules is no more than about
5 oc.
Figure 5-14 and Figure 5-15 show that Fuentes model over-predicts module temperature
by 10-15 °C for all modules at higher temperatures. Both absolute and relative error is
lower at lower temperatures. This result is expected since, at least for this data set
lower temperature is associated with lower irradiance and module temperature tends
towards ambient temperature at low irradiance.

Module Energy Rating Methodology 81

 Module Energy Rating Methodology
Spectral Correction Factors
2
I I I I
· · · · · · Mod 3 - · - · Mod 4 - · · - Mod S
1 .5
I Mod 1 - - - Mod 2

�
u 1 I
I
I
........ ..
..... ..
I .... .... ......
I
rJ) I
.. ... ...

If!
I
I --�
..

0.5
I
I I I
I
( cs
I
I
•,

I I
0 I I I I I i I I
i I

2 {
;·, .
.
•:
1.5
�
u
rJ)
1
� '• -··
.

0.5 NICE
.

0
.
.
I '
lA '

2 .
.
.
. .
.
.
.
1.5 .
t '..
�
u
� .. ..
.. .. . .. ..... .. ..
•'
�
I
1
rJ)

0.5 HS

0 A I A
2 \
..
..
. .. .
.
I .
1.5
. . .
.
� . '
u
..... .... .... _,,.' _A:
1 -
rJ) ;il
.

0.5 � HC

0 '
t. I
2

1.5
� I
u 1 .... .. ..... .. ............ ..
rJ)
I .

0.5 cc

0 I � '

00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00

Figure 5-12 MER Reference Day Spectral Correction Factors

Module Energy Rating

82
 Module Energy Rating Methodology Model Validation
NREL Outdoor Test Facility
Measured Module Temperature
60 '

I I !
If --
Module 1 - - Module 2 - - - - - Module 3 - - - Module 4 - - - · Module
I
sl
I I!
:

I
50

:�
I
'
�
I
/1
I·'
I
I
40
"
I
!;
u
c

Q.l
1-<
::I
....
I

. .,.'•

I•
' •
I'

....
�
1-<
Q.l
Q.. '
30
6
Q.l ,.
�
-Q.l
::I
"tj
\ �.
0
�
20 I \
�I
\
� �
, .

�· ���
10
i
i �
�� w ��k��
\
0
'
I
I
I'
II
I
7/27/95 7/28/95 7/29/95 7/30/95 7/31 / 95 8 / 1 / 95 8 / 2 / 95

Figure 5-13 Measured Module Temperature

Module Energy Rating Methodology

83
 Module Energy Rating Methodology Model Validation
NREL Outdoor Test Facility
Estimated vs Measured Module Temperature
75
60 jModule 1 1
• ----·
45 ..fl#l.i.e: Jt�
A. ,..
• '""' �
30 • oiA��.
�

•
15
0 I
-------

75
60 jModule 2 1
--r..=---
• •
45 . �fk:rl�
�
.
u 30 ..._...,_ .... ,...3.::
,_� � :
0
...
QJ
l-<
15
� ----
.....
�
l-<
0
QJ
�
E
75
QJ
E-c 60 UModule 3 l ••
"0 45 �·
· · .._.,.�.f....� \'- �_:::.�
..
QJ
.... .
�
30 #!'>�· .;.� ·�
· :
...

.
.E
....
.....
fl)
15
� ---
0
75
60 �Module 4 l ..

z,.
�..,... ..;t:t-0·��
45 � ·:: 0 �r--
.....
•

30 0 • •o 0
o_,, • a:

15 0 ·-

0 I
------

75
60 �Module S I
� 0.. ...

45 • !b•
• .:?�..::::p--�,.--t ·--�
•\ ::--
30 •-
0 � ....�·� .

15 ..&

----
0
0 10 20 30 40 50 60
Measured Tempearture, o c
Figure 5-14 MER Thermal Model Validation, Actual Temperature vs. Estimated

Module Energy Rating Methodology

84
 Module Energy Rating Methodology Model Validation
NREL Outdoor Test Facility
Module Temperature Error
40%
30% I l �odule 1
0 •.• •
. .. .: :_� .. ..-�..
20% • • • • g• •) . . ..
.
.:..,. . ... .
,. . ..
•·"-· I> • •• •
10%
0%
d>
• • .
!
ttl· ...
.....

!40
0
..

�
..
.
.

·�::� :f\tt'!·,
0 ••
0•
•
•

0
... 'I. ..
0 ••
GilD
• 0 ••0
..
0
•
•
•
••

-10% 0 0
•
-20%

40%
...
. .. . • "- . "' . l �odule 2
30% o• 0 \0 .
.
. • 0 0 .:·.. �: }"). �,· .. •:.:·
• • •
0 0• •..t . �.._ . ... 0 ·· :.··0....• �·· ... • • 0 • ,
20% .
• •
• � lillt
- • .
10%
0%
• 0 �l • • • •
· !411.
r--.
. � to ••
�· !"f'C1>··!�·
• .. 0 • 0
;,r �" •
•• 0

•
-10%
.
0
-20%

40% �------�--1'--���-��-r-�====�
30% r-----��-r--��--�t'--�:!-���o�d�u�le�3�
20o/o +-----���--�,..�r-����4-���rt&--�
10o/o +-------���---,&
Oo/o +-------�--����-S��--�--+--4
-10o/o +-----------���------�--�------,_ ,_ �----------� __________ __________

-20o/o �-------L--�--L---J---�--�

40% 0
30% :� •J
•
l �odule 4J
••
�

• •� + 0 • . ' 0 • .... •
0, ,. 0 .,.. . . .. :� 0• 0
0

• ·_?�·
20% .. . .
�t; -�...... .. • a ·-���, • 0
00 . . .� .. .
.;:'� , _ 0.-,....• • . . ... 0 0 •
10%
0 • 0 ..
�

o6 .-.....!!"-
0 • • •
0% . • 0
· r� �
-10% ,. .
.
-20%

40%
., l �odule 5
30%
• • 0 • 8 . . . • .
.

•• 0:• 0 0 • .: . . �;;; .
20% "0 • . . d.: .. 0
•• • o, t. ""·
• I "� ""•
. �
.•
.• � . ... . • • :
0� :J:�·
. : '� --:-..,�_,.
•

10% •
• . •
:·: • ti- II' ';'·:• • .
0% • u
·� • • I "� • • •
. .

0
-10%
.
-20%
0 10 20 30 40 50 60
�odule Temperature, oc
Figure 5-15 MER Thermal Model Validation, Model Error vs. Temperature

Module Energy Rating �ethodology

85
5.2.4 Reflection Model

Based on measurements made by Keith Emery of NREL and his resulting

recommendations, we did not include an incidence angle correction in these validation
results. As mentioned in the Power Model Section, this may have had some impact on
the power model results.

5.2.5 lrradiance Model

Figure 5-16 shows a comparison between the measured POA irradiance and the value
estimated from measured DNI and diffuse using the Perez model. For the sunny day,
the calculated irradiance is consistently high, though the relative error shows a
consistent decrease until late in the afternoon. A similar trend is apparent in the second
and third days, though the random error seems to increase as the cloud cover increases.

With random errors up to 50%, the final day shows clearly that the irradiance model is
not working very well at all for cloudy conditions. This may be attributable to poorer
accuracy under cloudy conditions, or to more variable conditions, or to inappropriate
model correlation coefficients in the Perez model for overcast conditions.

5.3 Module Energy Rating Results

Using the coefficients described previously, module energy ratings were computed for
the five sample modules, and are shown in Figure 5-17 and Figure 5-18. Keeping in
mind that the MER results are intended primarily to support comparisons between
modules and that inaccuracies in the various models have been noted, the presence of
multiple model results is of secondary importance. Also, the fact that this data is only
intended for illustrative purposes and does not necessarily reflect the general trends in
PV technology bears repeating here.

Module Energy Rating Methodology 86

 Module Energy Rating Validation data
NREL Outdoor Test Facility
1200 ,
,-------r,-----r-
,
, -------,-
1 ;=========;:]
N
S 1000 -+----/(\
-!7."·\--
-- ----i,c-- J
- ---M.-��---+-1-----1�-------H
. -- Measured
- - - - - - Calculated

�
� 800 �j \\ .. I I�
+---r-�---�--��---+---��---�------� l Ii ,; �
.
600 -
]�
I
+---�gr-�1--�----J�+-----Ir --;+
. �--
l ---r-------------j
400 +---r-�--r--+-�---+--4���-4---------�
<
0
�
200 +---,_--�---+I ---r
l-��--+-'i --T--h
I � �---+----------�
0 \ I J � I J VJ. \ I �
+---�------�----;r-�-------=�-r--�------�--+, ----�----���

7/28/ 95 7/29/95 7/30/95 7/31 /95 8 / 1 /95

1200 �------�------�-------�------�-----�------�
1000 +-----�---+--+---����---j
II II I .
800 +-------�---+--�--����---j

600 +-------�---+--�����--+-�--�---j

400 +-------+--���---4--+---�

200 +-----���---+--�--+---�---j

0 ��------+---+---�

0 200 400 600 800 1000 1200

Measured Irradiance, W/m2

50% 0

Q)
v
40%
30% n
0
0
u
1 (Calc /Meas - 1) I
= 0
Q) �n
"'"
Q)
20% 0 �0 0
.�
0
0 0_
....
.
....
.
..... 10% �
� .aS
0
0
" L
� 00 o oi9-
� 0
s
0
0
Q)
0% -
- <>o
���
cfJ"'fS �0 0 0 0
n
cP o oo
CX) 0 Olb 0 0
v
=
-10% O"'
8
o "'
.L!) .OlD
O o Cf' �o 0
� cP' 8
� 0o o "' o
ra
..... -20% 0
0
1:) 0
-o o
<o
OODn o. 0
o<> o cro
"0
0 %()(!
ra
"'"
"'"
-30% 0
0 ca
0
.. Q. ' Q..O -
eo "0 8
-
-40% o"'
0
0
0
-50%
7/28/95 7/29/95 7/30/95 7/31 /95 8 / 1 /95
Figure 5-16 MER Irradiance Model Validation

Module Energy Rating Methodology 87

 Module Energy Rating Results
Maximum Power
II Anderson
I?.I Blaesser �Interpolation E3 Myers
500 .------.

:2 400 tliiib;;;;;:;;:C::::]­
� 300
>..
e.o
Q)
200
&J 100
0

500 ,-------�
,-. 400 Cold/Sunny
�- 300
>..
e.o
Q)
200 .
;:::
� 100
0

500 �------�--�==�
,-. 400
�
� 300
>..
e.o
Q)
200
&J 100
0

100
�------�====�
,-. 80 Cold/Cloudy

�- 60
>..
e.o
Q)
40
;:::
� 20
0

:2 150
�
>. 100
e.o
Q)
;:::
� 50

0
1 2 3 4 5
Module
Figure 5-17 Module Energy Rating Results, Maximum Power

Module Energy Rating Methodology

88
 Module Energy Rating Results
Fixed Voltage
• Anderson 123 Blaesser E;3 Interpolation
30 ,-------�

:2 Hot/Sunny
� 20
(!)

�
� 10
u

5 �------����--�====�
.::: 4
�
3
(!)

�
r<:l
2
.:::
u 1
0

1 2 3 4 5
Module

Figure 5-18 Module Energy Rating Results, Fixed Voltage

Module Energy Rating Methodology

89
The most obvious difference between the modules is the magnitude of their production
capacity. It is clear that for a given system load, the number of modules required to
meet an energy demand will be different for the different module types. Since other
factors (such as required Voc or economics) may also affect the system module count,
these factors should be accounted for and the product of the module count and the MER
values should be compared. If all other factors are equal, then the highest total MER
will indicate the best alternative.

Note that the MER values for maximum power loads tend to track their STC relative
power ratings, when insolation is disregarded. In theory, the advantage of the MER is
that some modules will be more strongly affected by climate than others, leading to
cases where a poorly performing module at STC may outperform its competitor(s) in a
more realistic climate. For example, modules 1 and 2 have very similar STC ratings, but
the difference between their maximum power MER ratings varies with climate. When
other factors are included, this fact may change the recommended module for a
particular application. While the differences between these two modules are small,
comparing with a more sensitive technology like module 3 in an otherwise similar
system configuration, the differences would likely be more distinct.

In contrast with the maximum power MER rating, the fixed voltage MER rating does
not track the STC power rating well, unless the module is designed for this type of
application. Comparing the fixed voltage and maximum power MER estimates for
Module 4 shows how dramatic this effect can be. Of course, this module has an open
circuit voltage of 90.2 V, so this module was probably not intended for battery charging.

Module Energy Rating Methodology 90

6. CONCLUSIONS AND RECOMMENDATIONS

This project set out to address three specific goals:

• Define a Module Energy Rating Methodology

• Define specific conditions for evaluating module performance
representative of the US climate
• Define procedures and techniques for measuring module characteristics
and estimating module performance at the representative conditions

The first of these goals has been squarely met. A general method for describing a PV
module energy rating has been developed and agreed upon by an industry­
representative Technical Review Committee.

The second goal has been met in principle with the definition of five representative
daily weather profiles. These profiles are further defined by the source of weather data
(the National Solar Radiation Database) and specific locations and dates within the
database. While these suggested weather profiles appear to adequately represent a
broad range of geographic locations and climatic conditions, they are subject to change
under further scrutiny of system owners and designers.

The third goal has not been satisfactorily met. The combined errors in characterizing
module performance, in translating indoor measurements to outdoors, in measuring
ambient conditions and module performance, and in accounting for all of the various
parameters affecting module performance exceed the accuracy required to compare one
module to another under a variety of conditions. Recommendations in the following
sections are intended to address the measurement and modeling errors encountered.

A consensus was never reached as to the level of accuracy required for the overall
performance modeling. We feel that to provide users with ratings that they can
confidently compare and make decisions on, we need to be able to model performance
to within 5 percent of actual over the full range of expected conditions. This implies
that we need to account for factors that have rather small impact on module
performance--as small as 1 percent-so that the combined error of all factors puts us
below that 5 percent level. As we all initially suspected and have shown through this
initial work, this level of accuracy will not be achieved easily.

A common complaint is that we're making this process too difficult, that it should be
easy to perform and simple to calculate. While it is true that the results should be easy

Module Energy Rating Methodology 91

to use and interpret, the steps necessary to achieve those results will likely have to be
complex. The process is not intended to be an easy to use tool, only the results are. The
process needs to be accurate and repeatable. An easy to use process with low accuracy
will provide numbers for users to compare, but the comparisons may not be valid, may
misrepresent products and will ultimately prove not useful.

While we are closer to a useful module energy rating, there is still a good deal of work
to be done. The following comments are offered to help direct further work on this
topic.

6.1 Testing

• Modules used for model validation should be representative of industry

current production. This requirement is more difficult to implement and
more critical for advanced thin film devices for which "current production"
changes rapidly.

6.1 .1 Indoor Tests

• Data used to characterize module performance must span the range of

conditions that are to be modeled, especially for the interpolation model.

•
The requirement for a broad range of measurements must be weighed against
the added error of attempting to create extreme conditions, particularly
irradiance. Most simulators are designed and calibrated for 1000 W /m2
ASTM AM 1.5 spectrum based on a specific amount of power being fed to a
specific light source placed a specific distance from the test device. Higher
irradiances can be achieved by increasing the power to the light source, which
will likely change spectral content, or by moving the light source closer to the
module, which could introduce light distribution non-uniformity.

Reducing intensity can also be done by several different methods. Reducing

the power to the light source will also change the spectral content. Moving
the light source away from the PV module could actually improve
distribution uniformity, but may not be physically possible due to test
chamber dimensions. The final option is to use neutral density filters to
attenuate the light; however, all filters are somewhat wavelength dependent
and may suffer from non-uniformities.

Module Energy Rating Methodology 92

 • Translation of indoor flash tests to outdoor measurements has been. an
ongoing battle. Confidence is fairly high if the conditions being translated
from and to are not vastly different and if crystalline silicon is used.
Unfortunately, our requirements are for the broadest of translation conditions
and for all technologies. One thought is that modules may perform
differently under direct and diffuse light conditions than they do under the
direct-only conditions in a simulator.

6.1 .2 Outdoor Tests

• Validation data must represent the range of climatic conditions that are to be
modeled.

• Sufficient data under these various conditions must be collected to reduce

statistical uncertainty.

• Highest accuracy irradiance data will likely be achieved using the cavity
radiometer and in-plane diffuse measurements. A method for accurately
measuring in-plane diffuse needs to be developed.

6.2 Validation Analysis

• All of the results presented here are based on software. There is a substantial
amount of code involved in performing these calculations and, though an
extensive effort was made to eliminate bugs, there is always the possibility
that bugs still exist.

6.2.1 Power Models

. • Model coefficients should be generated using the same types of input data as
the resulting models will use. For example, spectral corrections and incidence
angle modifiers should be applied to measured irradiance when outdoor data
are used.

• The results tend to favor models based on outdoor characterizations when

compared to outdoor validation data. This is not a wholly unexpected
conclusion, however, it was anticipated that proper accounting for the
differences between indoor and outdoor measurements could be made.

Module Energy Rating Methodology 93

 • It appears that, ultimately, different power models will be specified for
different technologies. Though we had difficulties with the diode model
analytical approach, this may prove to be the most accurate. However, these
models are very technology specific and, presently, we've only found models
for single and poly-crystalline silicon. Further research in this area is
necessary.

• None of the modeling or validation work involved concentrator modules.

There are some unique characteristics associated with concentrators that are
not dealt with in the existing models. For example, under high concentration
crystalline silicon efficiency changes much more dramatically with changes in
irradiance than under one-sun conditions. Concentrators have to be included
in this research at some point. Fortunately, the overall MER approach and the
definition of weather conditions is independent of module type; only the
modeling and characterization work needs to be expanded.

6.2. 1 . 1 Anderson Model

The work performed by Sunset Technology and documented under a separate NREL
report [Anderson, 1994] shows that this model should perform better than we have
observed. However, Anderson's validation results in that report were based on the
matrix flash simulator IV curves used to develop the coefficients, not on independent
outdoor measurements. These results may not address the applicability of the model to
outdoor data.

6.2. 1 .2 Blaesser Model

• Blaesser states that the coefficients for his model can be obtained with a few
IV curves under different conditions. His intent was to provide a simple
method to evaluate PV array performance by making some standard
measurements and translating the results to STC. His method provides a
great deal of simplicity (no module temperature measurements), and the
model has more than sufficient accuracy relative to the equipment used in
field testing. For our purposes, we would expect Anderson to work better as
it makes no assumptions about module thermal characteristics. However, the
Blaesser derivation appears to address some of those mystical outdoor
characteristics that elude indoor measurement based models.

Module Energy Rating Methodology 94

6.2.1 .3 Interpolation Model

• If there are differences between indoor and outdoor measurements that we

aren't accounting for, the interpolation model will be the most severely
affected.

• A broader range of measurement conditions is necessary to utilize this

approach.

• This model might be technology independent if the indoor vs outdoor issues

are addressed. Non linearities of multi-junction thin films do not need to be
characterized per se, the measurements will take them into account.
However, cross dependencies, such as temperature dependence of spectral
response, might require multi-dimensional test matrices.

6.2.1 .4 Myers Model

• For peak power, the simplest model was as good as any.

• It needs to be evaluated under a broader range of ambient conditions.

• Application of this model to fixed voltage estimation becomes rather complex

if a voltage profile is to be defined.

6.2.2 Thermal Model

• Potential areas of weakness within the Fuentes thermal model include the
fixed values of emittance and thermal capacitance, the determination of the
characteristic length through the use of a quasi-hydraulic diameter
calculation, the sky temperature model.

• PVUSA and Sandia are looking at alternative temperature models using

regression analysis. These should be compared in detail with the Fuentes
model.

6.2.3 Spectral Model

• Spectral effect on insolation was in the 1 to 5% range for the sample modules.
While significant, the power model accuracy will have to improve before

Module Energy Rating Methodology 95

 these effects can be distinguished. In many cases, spectral effects will not
contribute significantly to the difference between modules.

•
Multi-bandgap technologies will need significantly more work in order to
account for spectral effects. The interactions between the junctions when
spectral variation occurs can change the shape of the IV curve, necessitating
an integrated power and spectral model. No attempt to handle these
technologies has been attempted in this work.

6.2.4 Optical Model

• While some measurements were made, very little analysis was performed on
incidence angle effects.

• Differences in albedo and diffuse sky component associated with

measurements made outdoors may be of the same order of magnitude as the
incidence angle effect itself. These errors can be reduced if measurements are
made under very clear sky conditions, by varying only the module azimuth
angle, and at locations where ground reflectance and the elevation of the
horizon are constant in all directions.

• Indoor flash tests at varying incidence angles can be used to eliminate the
diffuse problem, but the relative distances between the near and far edges of
the module with respect to the light source become important.

• It has been suggested that laser ray tracing might provide more accurate
results by eliminating the diffuse problem and the problem of one edge of the
module being closer to the light source than the other edge. Care must be
taken to ensure that the beam of light strikes the same area of cell as the
incidence angle is varied (non-uniform response) and does not strike any
inactive cell area (grid lines or interconnects).

• While it is generally agreed that an lAM should be applied to the direct

component, there is less agreement as to if and how to apply an lAM to the
diffuse component. One currently used application of an lAM (King, 1996)
applies the correction to the global plane of array irradiance. This application
under high diffuse conditions needs to be evaluated.

Module Energy Rating Methodology 96

6.2.5 lrradiance Model

• The problems of early morning and late afternoon errors in the Perez model
need to be addressed. A work around was implemented for these results, but
a more thorough solution should be implemented.

6.3 MER Calculations

• For our purposes, where we are continuously changing parameters and

rerunn ing the MER calculations for multiple modules and multiple models,
the Access database is rather slow. However, it allows us to conveniently
evaluate and modify intermediate results, which would be much more
difficult using a traditional programming la nguage. There are many
optimizations that can be made to the final product to enhance its speed.
Also, a user will typically analyze only one module at a time using one power
model.

• It is clear that the fixed voltage MER will show a very different view of some
modules than the maximum power MER will. This will offer manufacturers a
significantly different benchmark to optimize their products for (should they
so choose). The results presented here show that if an appropriate number of
series modules and batteries are used, poor performance will result. The
fixed voltage performance of Module 4 is an example of this behavior.

• The selection of a fixed 14.4 battery voltage is a simplification that may not be
necessary and may be misrepresentative. A battery profile would be
preferable, but agreeing on an appropriate profile might be difficult. Also, the
difference between battery terminal voltage and the module voltage need to
be accounted for (blocking diodes, charge controller losses, wiring losses,
etc.).

• The TRC felt rather strongly that the MER should be presented in terms of
W-h or A-h, which will vary significantly with module power but which are
comprehendible by most PV buyers. Other terms such as efficiency or
normalized values like W-h/WRated and A-h/ ARated are more easily compared
but are more obscure. Ultimately the user can divide the module's cost by
appropriate MER value and compare the $/W-h or $/ A-h values directly.

Module Energy Rating Methodology 97

7. R EFERENCES

A. J. Anderson, "Photovoltaic Translation Equations: A New Approach," Final

Report for Task 1.0 of NREL Subcontract No. TAD-4-14166-01, November
1994.

R. E. Bird, Riordan, C., "Simple Solar Spectral Model for Direct and Diffuse
Irradiance on Horizontal and Tilted Planes at the Earth's Surface for
Cloudless Atmospheres," Journal of Climate and Applied Meteorology,
Volume 25, Number 1, pp. 87-97, January 1986.

G. Blaesser, personal communication, November 1995.

K. Emery, Osterwald, C. R., "Efficiency Measurements and Other Performance

Rating Methods," Current Topics in Photovoltaics, Volume 3, Academic Press
Ltd., 1988.

F. Fabero, Vela, N., Chenlo, F., "Influence of Solar Spectral Variations on the
Conversion Efficiency of a-Si and m-Si PV Devices: A Yearly and Hourly
Study," preprint from Proceedings of the 23rd Photovoltaic Specialists
European Conference, 1995.

M. K. Fuentes, "A Simplified Thermal Model for Flat-Plate Photovoltaic Arrays,"

Sandia Report SAND85-0330•UC-63, May 1987.

C. F.Gay, Rumberg, J. E., Wilson, J. H., "AM PM - All-Day Module Performance

Measurements," Proceedings of the 16th Photovoltaic Specialists Conference,
pp. 1041-1046, IEEE, 1982.

D. King, Hansen, B. R., "A Sensitivity Analysis of the Spectral Mismatch

Correction Procedure Using Wavelength Dependent Error Sources,"
Proceedings of the 22nd IEEE Photovoltaic Specialists Conference, October
1991.

D. King, "Photovoltaic Module and Array Performance Characterization

Methods for All System Operating Conditions," NREL/SNL Program
Review, AlP Press, 1996, pp.347-368.

F. Kreith, Kreider, J. F., Principles of Solar Engineering, Hemisphere Publishing

Corporation, Washington, D. C., 1978.

J. Meeus, Astronomical Formulae for Calculators, Second Edition, Willmann-Bell,

Richmond, VA, 1982.

S. Nann, Emery, K., "Spectral Effects on PV-Device Rating," Solar Energy

Materials and Solar Cells, Volume 27, pp. 189-216, 1992.

Module Energy Rating Methodology 98

S. Nann, Riordan, C., "Solar Spectral Irradiance Under Clear and Cloudy Skies:
Measurements and a Semiempirical Model," Journal of Applied Meteorology,
Volume 30, Number 4, pp. 447-462, April 1991.

R. Perez, Stewart, R., Seals, R., Guertin, T., "The Development and Verification of
the Perez Diffuse Radiation Model," Contractor Report SAND88-7030, Sandia
National Laboratories, Albuquerque, NM, 1988.

R. Perez, Seals, R., Ineichen, P., Stewart, R., Menicucci, D., "A New Simplified
Version of the Perez Diffuse Irradiance Model for Tilted Surfaces," Solar
Energy, Volume 39, Number 3, pp. 221-231, 1987.

W.H. Press, Flannery, B. P., Teukolsky, S. A., Vetterling, W. T., Numerical Recipes,
Cambridge University Press, Cambridge, 1986.

T. Townsend, "A Method for Estimating the Long-Term Performance of Direct­

Coupled Photovoltaic Systems," MS Thesis, University of Wisconsin, 1989.

H. S. Rauschenbach, Solar Cell Array Design Handbook, Van Nostrand Reinhold,

New York, 1980.

C. Seaman, "Correction for Spectral Mismatch Effects on the Calibration of a

Solar Cell When Using a Solar Simulator," DOE/JPL-1012-50, January 1981.

C. M. Whitaker, Townsend, T. U., Wenger, H.J., Iliceto, A., Chimento, G., Paletta,
F., "Effects of Irradiance and Either Factors on PV Temperature Coefficients,"
Proceedings of the 22nd IEEE Photovoltaic Specialists Conference, October
1991.

Module Energy Rating Methodology 99

APPENDIX A INDOOR MODULE TESTING DATA

Table 0-1. Summary of Indoor Testing Data (Modules 1, 2 and 3)

Module T H Sheets Voc,est Voc,Spire Isc Vmp Imp Pmp
(C) (W/m2) (V) (V) (A) (V) (A) (W)
1 20 1000 0 22 22 3.32 17.6 3.07 54.1
1 20 773 1 21.8 21.8 2.56 18 2.35 42.2
1 20 487 3 21.4 21.4 1 .62 18.1 1.44 26
1 20 253 7 20.7 20.7 0.84 18 0.71 12.8
1 30.3 1000 0 21.3 21.3 3.31 17.2 3 51.8
1 30.2 773 1 21.1 21.1 2.57 17.1 2.36 40.4
1 30.3 487 3 20.6 20.6 1 .63 17 1.48 25.1
1 30.4 253 7 19.9 19.9 0.86 17.1 0.73 12.4
1 40.2 1000 0 20.7 20.7 3.33 16.3 3.05 49.8
1 40.2 773 1 20.4 20.4 2.54 16.3 2.34 38.2
1 40.5 487 3 20 20 1 .61 16.6 1 .44 23.9
1 40.9 253 7 19.2 19.2 0.83 16.1 0.73 1 1.7
1 50 1000 0 20.1 20.1 3.35 15.6 3.08 48
1 50.2 773 1 19.8 19.8 2.58 15.9 2.32 37
1 50.2 487 3 19.3 19.3 1 .63 15.6 1.49 23.2
1 50.3 253 7 18.6 18.6 0.86 15.4 0.75 11.5
2 19.7 1000 0 21.6 21.6 3.57 17.4 3.24 56.4
2 19.6 762 1 21.4 21.4 2.72 17.7 2.45 43.3
2 19.6 485 3 20.9 20.9 1 .73 17.7 1.53 26.9
2 19.6 257 7 20.2 21.7 0.92 17.2 0.8 13.7
2 30.5 1000 0 20.9 20.9 3.58 16.5 3.25 53.6
2 30.6 762 1 20.6 20.7 2.74 16.8 2.45 41.2
2 30.3 485 3 20 20.2 1 .74 17 1.52 25.8
2 30.3 257 7 19.4 19.5 0.93 16.6 0.79 13.1
2 40.5 1000 0 20.3 20.3 3.6 15.5 3.3 51.1
2 40 762 1 19.9 20 2.75 16 2.47 39.4
2 40.1 485 3 19.4 19.5 1 .75 15.9 1.54 24.6
2 40.1 257 7 18.7 18.7 0.93 15.6 0.81 12.6
2 49.9 1000 0 19.6 19.6 3.61 15 3.24 48.6
2 49.9 762 1 19.3 19.3 2.77 14.9 2.52 37.6
2 50.2 485 3 18.8 18.8 1 .75 15.2 1 .56 23.8
2 50.5 257 7 18 18 0.94 15 0.81 12.1
3 20.1 1000 0 15.1 15.1 1 .76 10.4 1.31 13.7
3 20.1 741 1 14.9 14.9 1 .31 10.7 0.95 10.2
3 20.1 443 3 14.5 14.5 0.78 10.9 0.55 6
3 20.1 205 7 13 13 0.36 9.9 0.24 2.4
3 29.5 1000 0 14.7 14.7 1 .78 10.1 1.35 13.6
3 29.3 741 1 14.5 14.5 1 .32 10.4 0.99 10.3
3 29.3 443 3 14.1 14.1 0.79 10.3 0.58 6
3 29 205 7 13.5 13.5 0.37 10.7 0.24 2.5
3 40 1000 0 14.3 14.3 1.8 9.8 1 .39 13.6
3 40 741 1 14.1 14.1 1 .33 10.2 0.99 10.1
3 40.1 443 3 13.7 13.7 0.8 9.7 0.62 6
3 40.2 205 7 13.1 13.1 0.37 10.4 0.24 2.5
3 50.1 1000 0 14 14 1 .81 9.7 1 .38 13.5
3 50.1 741 1 13.7 13.8 1 .35 9.9 1 .02 10.1
3 50.1 443 3 13.3 13.3 0.8 9.7 0.61 5.9
3 50.2 205 7 12.6 12.6 0.37 9.7 0.25 2.4

Module Energy Rating Methodology 100

 Table 0-2. Summary of Indoor Test Data (Modules 4 and 5)
Module T H Sheets Voc,est Voc,Spire Isc Vmp Imp Pmp
2
(C) (W/m ) (V) (V) (A) (V) (A) (W)
4 19.2 1000 0 91 91 0.92 68.2 0.8 54.3
4 19.5 741 1 90.1 90.1 0.68 69.3 0.59 4 1 .1
4 19.5 456 3 88.3 88.2 0.42 68.6 0.37 25.4
4 19.5 237 7 85.7 85.7 0.22 67.3 0.19 12.9
4 30 1000 0 89.4 89.4 0.92 67.2 0.79 53
30.1 741 1 88.3 88.3 0.68 66.7 0.6 40.2
q=t
4
30.1
30
456
237
3
7
86.4
83.7
86.4
83.6
0.42
0.22
68.1
65.4
0.36
0.19
24.7
12.6
4 40 1000 0 86.9 86.9 0.92 65.1 0.79 5 1 .4
4 40.1 741 1 85.6 85.6 0.69 65.4 0.59 38.8
4 40.1 456 3 83.6 83.7 0.42 63.9 0.37 23.9
4 40.2 237 7 80.6 80.6 0.22 63.3 0.19 12.1
4 50.1 1000 0 84.7 84.8 0.92 64.2 0.78 50.3
4 50.2 741 1 83.4 83.4 0.69 63.6 0.59 37.8
4 49.9 456 3 81.2 81.2 0.42 62.1 0.37 23.3
4 50.2 237 7 78.1 78.1 0.22 60.5 0.19 1 1 .7
5 20 1000 0 23.1 23.1 2.33 16.6 1.94 32.2
5 19.9 767 1 22.6 22.6 1.79 16.4 1.48 24.3
5 19.9 485 3 21.5 21.5 1.13 15.7 0.93 14.6
5 19.8 255 7 19.9 19.9 0.6 14.3 0.5 7.1
5 30.8 1000 0 22.2 22.2 2.34 15.5 1.98 30.7
5 30.9 767 1 21.5 21.5 1.78 15.5 1.48 22.8
5 30.9 485 3 20.4 20.4 1.13 14.6 0.94 13.8
5 30.9 255 7 18.8 18.8 0.6 13.4 0.49 6.5
5 40.3 1000 0 21.1 21.1 2.34 15 1 .92 28.8
5 40.1 767 1 20.4 20.5 1.79 14.7 1 .46 21.5
5 40.2 485 3 19.3 19.3 1.13 13.8 0.92 12.7
5 40.2 255 7 17.7 1 7.7 0.6 12.5 0.49 6.1
5 50.4 1000 0 20.1 20.1 2.33 14.5 1 .86 27.1
5 50.5 767 1 19.4 19.4 1.78 13.7 1 .46 20
5 50.7 485 3 18.3 18.2 1.14 13 0.92 12
5 50.7 255 7 16.5 16.5 0.6 1 1 .5 0.48 5.6

Module Energy Rating Methodology 101

APPENDIX 8 ADDITIONAL M ER POWER MODEL VALIDATION AND COMPARISON PLOTS

In section 5.2.1, Figures 5-5 and 5-6 compared modeled power to measured power and
model error as a function of irradiance for each 10-minute average point for module 1.
Figures 5-7 and 5-8 showed similar data for fixed voltage operation. The following
figures show the corresponding plots for modules 2 through 5.

Module Energy Rating Methodology 102

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 2
..--. 60
� 50 • Anderson
;...
<J)
;s: 40
0
p.. 30
'"d
<J).
....(<:$ 20
§. 10
....
.

CTJ
�
0
0 10 20 30 40 50 60

60 II
�;.... 50 H • Blaesser I I I ----
<J)
;s: 40 I I I� I
I
I
0
p.. 30 I I I� !
'"d
<J).
....(<:$ 20 I I �·· r-
§.
....
.
10 ·-
l �i I I
II
CTJ
� .HoSWD r I
0 I I
I I I
I

0 10 20 30 40 50 60

60 I
!

�;... 50 j • Interpolation I I I i
i
� _______...
I

<J)
;s: 40 I �� I !
I
I
0
p.. 30 I �� I i
'"d
<J) .
....(<:$ 20
.�l II
I
I, I
.§
.....
CTJ
10 . �
.1 � I
�
0 -
.0
I I i
0 10 20 30 40 50 60

60
�;... 50
<J)
;s: 40
0
p.. 30
'"d
<J).
....(<:$ 20
.§
..... 10
CTJ
�
0
0 10 20 30 40 50 60

Actual Power (W)

Module Energy Rating Methodology

103
 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 2
50% • .0 1 I

I
• • I

��
•
••
..
I I I • Anderson I
� ·:�
30%
.. .I1 ·0 0 • .. •
. I • II
�
.....
0
.....
.....
10% . ..
' . C:. •
�.
••
•

•
.. .. .
.
-. v, •
• I
: '11'. . •I""' ,. ,..
• • -... .._. !" ...... .� .-JUJ olD A

I !
6'2- -10%

-30% i I
-50% I I I i
0 200 400 600 800 1000 1200

I • Blaesser I
50% I

30%
0• •
• •
•
I
i II I
.

·. 1 I
•
•• • • !
.
.....
0
..... 10% ..:.
. :""'"'
· . • 0 o •
I ..
.
I
I
.
�
.....
·. =�--.·... . i • I .. .

· ' ........
.
. . ...
.. ..
.. ,:., . • "'
I
•-:,. 0 , .
. -:.•.. .. ..... ,.
:::-'2 dl>
• 0
• .I .,. .... ..
-
0
-10% "

. .,.
.....•.�-;:"
• I
•
I I
-30% i
-50%
I
i I
I I iI
0 200 400 600 800 1000 1200

50%
i I
I
I I I
. 00 •o •0
i I
• Interpolation I
30% i
I I
.....
0
•

. ..��·J
• I
I I II
••
iI
.....
• o..•-
10%
..... i • ;. • ... to . ! , •....,.. �a. •• , •• """"""' I• "'
0

I
•
� .. .• • • • • •
I .,. • • • • ·� I I
6'2- -10%
·� I
• I II I
-30% .:,oo I

-50% ,.. I I
I
I
I
0 200 400 600 800 1000 1200

50%
... .. . !
I ! I
I
I I · Myers I
30% ·� 0. I
I I
!
:t 0 I I !
•
..... · I i
:

0
• •
0 •
..... .. .
� I· ·
.0 ..
10%
..... .�--. ... . I • ., .
�
:::-'2
•.'f.L.a&Jt.
. -,...
I
Oo�l
I
•
"' ' '
. .. 'C .a�... . . • • •
! 0

-1.-: ··f' • •iS_,..,. ��

I
-·
.• •
1
I
0
-10% I

-30% I I
i
!
I
I
I
'
I
-50% I i . I I
I

0 200 400 600 800 1000 1200

Irradiance (W /m2)

Module Energy Rating Methodology 1 04

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 2 Current at 14.4V (nominal)
4.0 I
I
II
'

J • Anderson I I
I
• III� I � I
I
0
3.5
.J:::
_.....

: !
I I I • !
I o .. ...
o l I I I t•
I o I
I
·!
I

I 00
s 3.0 0 .
0 01
I �
• i
I

'
.....
::::
Q) 2.5 I • l I I I
I fl.
.
0
I
.... i
....
;::l I • I� ��
• •
' ':. i ... • • ! i
u
"0
2.0
!•
i •
•
�·�� . . -
0
· I
I
I
I 00
0 I
I
I
I •

Q)
....
c<;j
. 1 .5 • • 'I � I ll I
.§
..... •
I
I .I . . • I 0
I I 0

.
1.0
(})
lXI : . ·
�I � �I
..... .. I I II 0
'
0.5 o II •• I" I '
i
0.0 �r I I I I
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

4.0
I I
J • Blaesser I I I I ! i I�
::2
< 3.0
I
3.5
·I • I
i
i
I
I
I
• I�
I
I
I • I
•
.L 0
I ! ..
0
..... I • 0 i
1� �
• • • •--- 0 I
"'-'

:::: I I I I !
Q)
....
....
;::l
2.5
•
II • !�
i
�·c:., v ' · I ••
I
u
"0
Q).
....
2.0

0
0
• �·�! · . . ... I
II
0

, 00
00• I I
I
• I
I
1 .5 • ••
.§
c<;j

1.0 0
!
•
�� I I 0
0
0 0 I • I I
I
I
(})
lXI
·-··
.0
. •
1�r
. .. ·
I• I
I II
.
0.5
� i"
!

i II I I I
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

4.0 I ''
J • Interpolation I I 0
I! I
' ��
3.5 I 0
::2 0
.I I �-� I
I

I 0 I
s 3.0
.....
::::
' D 0 I �·f I
'
I
I
I
..

c;
....
....
;::l
2.5
iI 0
i
I
0
�I!I I
'· I
I
! . . !
I
u
2.0
I. I !i�••I �
0 • ... 0 I
"0
c;
.... . 0
0 i • • •• I

..
I I
! ! !

c<;j
1.5
• 0 [�i '! l 0 ''
' 0 ',
E 1.0 0 . •I
(}) ,. I
lXI •.
. •
�� I
'
I I
I" I
I
0.5
I " II ...... li · · I •i I i
I
!
0.0 � � ! _I i I i

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Actual Current (A-h)

Module Energy Rating Methodology 105

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
·

Module 2 Current at 14.4V (nominal)

50% 0
40%
0 0
0
I I I o Anderson
•• 0
•
I I I
30%
20%
••
0 0
0
0
•o
t
iI
I I 0
0 0
0
l-1 10% 0 ... "'
. 0
I ! 0 I 0 0 0
I
I
00 0
• 0
l-1
l-1
• • I • I • .I • • 0
0 0
0
•: . .:.. lo •
J:Ll
::::-'2
0
0%
.
.,
... -=·
• :
. l .. ' .
. I• • I • • • •• 0 0 •
•
· I
.
-10% .. 0
•
. . ..0 ..
•
.. .. o• 0•
• •
i . .. . . . -
1
•
0•
0 I
. . .. ..0 . 0 0
-20% 0 •
• ,. .
• ...
• 0 i 0 v
I
• 00 'It
-30%
• I I I
I
I I
-40% • • • ..
• . .. 0 I II I
-50%
0 200 400 600 800 1000 1200

50% 0 • i I
40%
•
•
0 I
I
I I i I I • Blaesser �
0
.·
0 I I • I
!
30%
• 0.
.
0
l I ' 0
II
I
20%
0
l-1 • 0 I 0
i I
0
10% .. 0
l-1
l-1
0 0 o ,P
I 0 0
0 · I• 0 • 0 0 0 �.... I ·
J:Ll 0 �0 • 0 0 0 0 • "'II" I
0% I
v
o• ol
::::-'2
0 '" " . ... • • • • • I
.. . .. 1 : 0
I
I
i
-10% •

-20%
•
... .. .. I .. •
0 0
• , • •
� ' \ .. . . . o,p · · I 1
I!
•• •
8 0
I
I
' · .. 0 • • 0
_... • i 0 • i • 0 I ' i
-30% • •0
-40%
.

• .t• •
•
�·
• I 0 I I I I
-50%
•
•• • !
'
'
0 i I I
0 200 400 600 800 1000 1200

50% I
.
40%
0 0 0
.
i I
_I I
•
I' I
I .
I • Interpolation
I i
. . i ! i
30%
I i I
.0 ! 0
. ! !
i !
20% 0 i i
l-1
.
0 •
.
..0 i i .
I
0
I. 0
0 I
10%
. .
l-1
'"'
J:Ll
.0 . i I 0
•I
•
•
•
•
.
0 •" o a��·
.... .
0% ... • •
• . . I' 0
•
. ! ! • • •
::::-'2
0
. 0 • · · • I 0 0
lI• o . �
: • : ' ... l•.
-10% i
• '
•
... II
• 0

ti.f-
!•
..
•
-20% . I 0
• 0 i .
•
I !I iI
•
-30%
-40%
I ...
•· t�• ..•
.
!
i
.
I I
•0
I
- !
-50%
• . . i • :

0 200 400 600 800 1000 1200

Irradiance (W /m2)

106
Module Energy Rating Methodology
 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 3
16
_......_

H •
I: I I I --
�!-<
14 Anderson
I I I �
i
I

12 i I
I l�
(!)
� I I
0
10 I I
p... I I I I � I I
8 I
"'d I I i .� I I
....
(!) 6
(lj
I .. l� I I I I
.§
....
Ul
4
•
I
� I
...L:.: I
I I I
I
2 .......
�
0 I I I I I
0 2 4 6 8 10 12 14 16

I
I
16
: I I I 1
I

--
H I I I
_......_

• I I
� 14 Blaesser
I I I I �� �
!-< 12
(!) I
� ! I I I l �...f , ,....... !
I'
0 I
10
p...
I
I I I I �-· �� .... . .. I
• ��.o: I
8
I I� l...P :" ., I
I I
"'d I . _,. I I I
....
�· � ..,r- v I I
(!) 6
(lj
I I I I i I I
.§.... 4
��I · i I I I I
I
Ul 2 !
� i I, I I !I I I
0 I I

0 2 4 6 8 10 12 14 16

16 I I

-i • Interpolation : I I I i
II� J_ ___
�!-<
14
I I i I
I I
(!)
12
I� I
� I I I
iI j
0
10
I � -1
I I I I
p... I I ! I
I
8
"'d
....
(!) I I :� r I I I

(lj
6 I
I 1 �1 I I

.§.... 4
I �I i I !
Ul I
�...
2
�
0 I I I I I I

0 2 4 6 8 10 12 14 16

16
� -1 • Myers a1 l1 I i
I
I I j_ __ I

I I� II
14
!-<
I ! I 1
12 I I
(!)
�
I I I
I I �- I I

0
10
i i I
I I �� I
p... I
I I
I
8 :
"'d '· i �� I I I
....
(!)
iI
6 � I
(lj
I I
I �I I
s
.....
....
4 l! ••� i i i I
! !
Ul 2 I I I
I I
� ! I I ! I

0
0 2 4 6 8 10 12 14 16

Actual Power (W)

Module Energy Rating Methodology

1 07
 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 3

• •• II
•
• � •0 .
50%
• .
I
I I I I • Anderson I
I
30%
,� I I
·- ·
;..., ..� ... . I I
0 .. ..
;...,
;...,
10%
·- ��.. .: J •• • · · -t :flll,ll!t •• •• . ... .....
• I
...1 . ..
>J::l -�·
I I
__

·-· .. o..- ...

.Ao

I
-10%
0
�
-30%
I
I I I
I

-50% I I I
0 200 400 600 800 1000 1200

I
50%

30%
•
•••
i
;
'
I iI I • Blaesser I
;..., .., • I
I
.
0
;...,
;...,
10%
� "'- • 0 •I I I
•. •. . • • o • I
I
>J::l I .. ... II " L 0 • .... . ...
0
� -10% · .....
..: ...�· -
o l •
.& j o • 0
r
-· · · ....,......,... ...: ·· ·· .........o • v
..
• 0 .. . . .

� ,...0..,.,.
..
·
-30% ...... .. .
i
-50% i I I
0 200 400 600 800 1000 1200

50% I
I '

30%
I
I
I I • Interpolation I
;...,
0
;...,
;...,
10%
I
i I
I
! ••
I
.. ._ft. ...
I
I

.I • "
>J::l . .. "' l ..
v
.tl'o "T · ��. 0oo11 00 '1P 1 -oo-• I
...
I
0
-10% .. . .:. .. . I
If'
I I
�
.. I
.•:Y. I !
-30% , .,. I
I
0�- I II
-" I
-50%
0 200 400 600 800 1000 1200

50%
..:
·
·..
0
• .
..
I I !
! I · Myers I
0 � • • • •. I
30%
;...,
0 ...
0
I
. oo:i ...
-
i I
I I
;...,
;...,
10% . ..,.. . ·' .
0
o i
I
0
0 • -•
.... .
: . � • • - . .. ��
! • .,.....,_&. . ·..· ·· 'If> �--· ··· ....
. ....-_..:
\
I ..
• 0
>J::l
-10% ••-. o;.,.
.. .. . •
0 I i
I
�
-30% !
I
!
I
i
-50% I I i I
0 200 400 600 800 1000 1200

Irradiance (W /m2)

Module Energy Rating Methodology 108

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 3 Current at 14.4V (nominal)
'
2.0
H • Anderson I 1 I I
i�-- I
I
I
I
I
1.8 I
---.
..c:
. I I I I 1 ---- 1
i i
<t:
-...
I

..... ...-
1.6
1 .4 I I
I •
I I . l-----J
•
• I
I

•
�
Q)
>-<
I I 00 I I I
� � I I I I •

>-<
;:l
1.2
I I Ii .I � r 0 �I . :I I •
•
•
•
•
I
0
I
1.0 .
I 0 I 0.. 0 t 0 -" II
I
u
Q)
"'d
I
I � . . I 0 II i •
..... 0.8
·
.1 0 l
I

� . I I •
I

I0 1 0 l�
ro I. I • I I
.§...,
(fJ
0.6
1 .
0

j· · 0 I 0 I 1 0 I I I I I I
I
I
0.4
!
�
\ • I �i .• I •
"' I I I
0.2
� .. - 1 ·� • "
I I I i I I
I ! I
I

I
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

2.0 ' i
I 1 --
1 .8 H • Blaesser I I I I i! I I

---.
..c:
. I I I I I
I1
---- I' I I
1.6 I

I I i I I� I I
I

$ 1.4 I I
..., I I : •
I� I I
0
I
I
� I I I "'
Q)
>-<
>-<
;:l
1.2
IV I 0 I I I �� . •
II
0
0 0. 0 I :I
•- I
I
I
ol
0
I

u 1.0
I· I
I
I
I
1..--- -:-: 0 • II I I
• I·
I•
•
o
••
•
• '.
I
!
�
I

...,
Q) I . 0
"'d 0.8
. 0 · I
0 Il/<iI� 0 t .
ro I . •
I i •
·

•
• I

.' .0 ,.. .- I • I
0.6
.§..... I'
(fJ '! I 0 •I • I i
I . • I I I
0.4 • • I
.... .. �� I • I I
�
I i I I I I
0.2 • I I
•
I
I

0.0 ....... ... . �- I I j I I

I !
'

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

I I !
2.0
1.8 1-! • Interpolation I I I I I
I
I I
I
1I --
:2 1.6 I I I � ---- !
I
I I !
I

$ 1.4
..., I I
I 0 i
I I 1----
j i ! •

� [ • 0 i
I I
I
. �� I 0 i
i
I

Q) .
>-< 1.2 I I
• .. •
>-<
;:l I
I
I I
0 �I � 0.. ·, j .I
0 . I I o o o o
• !
u
"'d
Q)
1.0
!
I
v 0 II �� 0 00 �- 8 · -" I
--• i
o
! o
I
I
�
0.8
i' · i 1� 0.. . 1I .
• · ! ! I oo! • I
•

.§..... 0.6 0I
•
�- !
•
I · I i I
!
I I

·• II � ' • 0
•

(fJ
�
0.4
I 0 o
I

.. .. I' I 0 i
I I
' '
i I

0.2 1

0 I : I

0.0 �� .., I I i I I '

i I

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Actual Current (A-h)

Module Energy Rating Methodology 109

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 3 Current at 14.4V (nominal)
50%
•
•
J •
I o Anderson
40%
•• •
30% • -
•
I
I
I
!
20% • •
•
• I
I
I I
0 r 0 I
. I,
•
... .
•
� -
•
0 I
10%
•
I

I · 0 • o •• •• •vwJl· •
• •
�
•. , • I
•
� • • • •
0 0 • ;_)
0%
r:...:l
0
� -10% • • • • I � 0 • •
• • • • • I
· · • •
..

-20%
. . ..
. .. ,
.. ..!.....
w
.. -1 .. • "
#
•
, �
•
,
0 I
•• �
.
••
.
•
• 0 •
II 0 0
•
0 0 I 0
-30% • �-�-..... 4\t�· 0
i
•.
-40% -"-• • v ! I
I
• •
I
-50%
...
. I ! I 0

0 200 400 600 800 1000 1200

50%
I
i I I
I

40%
••
.
.
I
JI I I I
I o Blaesser �
30% I
I I
I !
I
i I
•
• 0 • •• • I
•
I
I
20% • •
�
..
. I I
0
�
�
10%
0% •
•o 8
O o. • I
I
I

I
" . 0
r:...:l
• _.. .o..
• •
I . .JI
•
0 .
. ........
•
•
�
,
-10%
• 0• ,
•
. •
...
•
I 0
lo 0 • • •
• I.'.
.
• • .• · · .• • •
I I
• 'b .t·
-20%
_ II .. • II ·· •.
• • . : • •l 0 •• • ..
. •. •. . •
,
•
.
•
-30% a• •• •.,.,.... • •
I ' • • • •
•
••• : ·· 0 .v· .
•
•
-40% .
-50% ... •
.
I I •

0 200 400 600 800 1000 1200

50% • I
40% 0
• I
I
I I I • Interpolation I
30%
•

0
I
I
Ii
0 I
20%
0 !
� I -··
. i I

• • • • •• .. . •.8 0
0
10%
0 .
�
� •• I • • • •
• • • • 0 0 • ·· · --
0% i o
I I
r:...:l ••• •
0 • •
.
II I
olII
cf2. -10%
•• . !' 1
I
... • • . • • i • •
-20% . II (f . .,. • • . . ' � .. ..
•
I i
i
.

II .
I

I
•
-30%
• • I• •
•• 0 •
.' I I i
i
•
• • i •
!
-40% 0 .. . !
I 'I
. .. I
'
iI
-50% ..
. I • ! • I I

0 200 400 600 800 1000 1200

Module Energy Rating Methodology

llO
 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 4
60
� 50 U • Anderson 1 J I
I I •
0
I�
1-<
Q)
� 40
I
'

I
I ....... · -r-� I
0
P... I I
I ... ���, .::::---- I I
I
30
I I
· • • ��'I�� I
.....
Q)
"'d

· ... . · I I
•.f#�' ·�
<1:l 20
.....§. I
I
I
I
10
CJl
u::l
I i
0 - ! !

0 10 20 30 40 50 60

60
�1-<
50
Q)
� 40
0
P... 30
.....
Q)
"'d
<1:l 20
.§..... 10
CJl
u::l
0
0 10 20 30 40 50 60

,..-._
60
� 50 • Interpolation
1-<
Q)
� 40
0
P... 30
.....
Q)
"'d
<1:l 20
.....
.§
CJl
10
u::l
0
0 10 20 30 40 50 60

60
�1-<
50
Q)
� 40
0
P... 30
.....
Q)
"'d
<1:l 20
.§..... 10
CJl
u::l
0
0 10 20 30 40 50 60

Actual Power (W)

Module Energy Rating Methodology 111

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 4
50%
• •
.��� I
I

0 • •• •• q, I ! I • Anderson I
1-<
30% :
• •• •
"'
· . . .... 00 '.
• 0. #o
... . ""...
.. .. f#·. ""' J
••• • • "' 0 • elf'#• •"- Jl>ool •r, d' J> 3' /IIIII�
0 0 • . .. •
0
1-<
10%
1-<

I
�
0
-10% I I
I I
;:,g
-30% I
I I
I

I
-50% I

0 200 400 600 800 1000 1200

·: • • • 0
50%
I

I
I

: 0 ."!. • I I • Blaesser I
I
30%
1-<
0
·�· . ·,. 0 · ·
q,""........
· . • I l I
�
1-<
1-<
10%
. ��.. -. - •
• • • •. t·. . 0
.
; ·: � � 11,\,.-
•• . . .
o • -'I.!J> o II: (> I
G
.... , · · ·· �� .
V'
�·

0
;:,g -10% . ,. I
I
I
i I
-30%

-50%
I
i I
I
I
0 200 400 600 800 1000 1200

... .. ,. 0 • • I
I!
50%
I
I

. .. . . 0 I• I • Interpolation I
"'· ., i 0 0 l
30% 0 ' ' .. •. ••
1-<
0
1-<
1-<
10% • 0�
..;0)}:.
. . ...
ol
.
;,I0I 0
.0 I
0. .. . 1
I
,
· ·II ·
\ r,·· '""'0 • "' ..-.��-
•
• �
., '- ·· • ·11 ' •• ,. ��··� . I1... 0
I
I
0
I I
�
� I !
0 I
-10%
I I
;:,g
!
I
I
I
I
'
-30% I
I
I
I
-50% i i I

0 200 400 600 800 1000 1200

. !
I
50%
0 .
, .. 0 •• i I I o Myers I
. .. . .
30%
1-<
0
1-<
1-<
10%
·: .,. . . 0 0
...
to.;:,,.-.... I •
I. ..
...
I
J... o_
I
I
i I
Il
• 0"'"Jt;, .: "'� • ,. :" � · I ;:•,.•• .'!"'o • .. \d"'>41P'I''�> 0 •• -• ]"IF,P- · - · ·· 0 i.
I
� 0
0
-10% ••ll•
II I
;:,g
i I
I I
I
-30%
I
I
I
I I II
-50%
0 200 400 600 800 1000 1200

Irradiance (W /m2)

Module Energy Rating Methodology 1 12

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 4 Current at 14.4V (nominal)
I
1 .0 I I
I
I
� • Anderson I I I I I
I I. I �--
,... . 1 ---- !
I
I
0.9 0
..c:I
......_
ol I I • I i I j•
<!; ...
....
0.8
I .
!
I I
0
I I .
0 I
I
I 0
II •• : .: 1.:..--:- I 0 i
I I
0.7
! I I • "· 0• vr-
0
i I
....,
t:::: . 0
Q) I • !
'-< I 0
'-<
0.6
· r� 1 .J
:::s
0.5 I 0
II 0 0
j •• I I
•
I iI
u
! �� I• 0 • ! •
I II I
Q) 0.4 ""L•
.
"'d
I
I . r:.;..r 1 II iI I I
I

!
....,
ro I

• I
0.3
.....,§ I� I • I I !I II I I
I
!
"' 0.2
j:.I:l
. :.�.-...
. � .� I I I I II I I
I
I
I
0.1
II I i'
'r.. I I
!
BIP'W" I
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

1.0 I I
I I I I I
'
0.9 � • Blaesser I
I
I , __
I
I
..c:I
......_

0.8 I I I I
I
I i I
I
I ____ I I
<!;
.......
II I
!
• I
I
·I I1�1 0
i I
I
i
. I� ! I
0.7
t:::: 0 0
I 0
...., 0
Q)
'-< 0.6 I
I
• • I I I
'-<
I I I
.
1� •��� -
•
I • I
•
:::s o
0.5 II
i 1"'0 I
: I
I
u
Q)
"'d I • I .• • • I •
I
I
I •
•
i

� I ' I
•
0.4 I
I ••
i i • I I
..
....,
ro � I '
.
I I

.§ 0.3
I
I I� I
• i
0
I
II I ! !
II

#I �
....,
j:.I:l
"' 0.2
I I [ I I
•
I I
I
I

i
0.1
· ·r· 1 ! i
' I I I
!
0.0 '
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

I I I I
I
1 .0
-j_ • Interpolation � I I
I
! I ![ _...-
io
I' I I
0.9
:::2 0.8
.
i
I•
!
.;,... I ____II 0

.-:-
I •
I •• • • �.:.;. • !
.
!
<!; ... I . 0 :
I

0
I
•
�� II
....
...., 0.7
t::::
Q)
I . I L1:
I
• 0
I i .
. ·
.
'-< . I
'-<
:::s
0.6
0
II I 0
I � ' 10
, I '
I
� 0
I
I
I
! I i
0
. ��
0.5
u
Q)
"'d I
I

"'L
i.
i
. I
• •
I I
I
I
i
I
• •
0.4
II I
....,
ro
0.3 • l:.;..r I
I
I
I
.§"' 0 i �� • I . I I
j:.I:l
0.2
0/1
·
�
. ._-.,. �I � I I I ! !
0.1
0.0 llillii8'""4'.
Jo'¥' I
I

I i 'I ! I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Actual Current (A-h)

Module Energy Rating Methodology 1 13

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 4 Current at 14.4V (nominal)
50% 0.... . I
I I I •
1 "I • Anderson
40%
• • • i
o'!
.....
• I
I

I
I
i

•• • ..•• I
I
30%
• • • ! • I
I

•
20%
•
.
i • • • • r ..• •
I

I• · I·
�
.:
I
.. : , 0 : • •.
....
0 ... •;. .. . • :
,.
• .
·-
l· • • .
10%

• .• • • I
v

.... ,. ·. . . .
.... .
• • t • •

I •
0%
;r
�
0
� -10% • :. • • i
I
I
•• • • • I •
-20%

•• • I
I
-30%
-40% • ' I I I
l I I I
-50%
0 200 400 600 800 1000 1200

50%
• • •• ! • •
• • • I
I I I • Blaesser �
40%
•
.... .. I
I I
I I
• •
•• .• . I1. •
30%
20%
,
..... • . • I
....
• • • • •0 • • l .. I

I •
• •
• I
0
I•
10%
.... •
I ••
• •• • • •• • 0 •0 • • • oo •
.... .
� 0%
.· ·· to· •• ••
..
.
·. .. ...
•. • • 0 0 •• • ..
. ·· ........ .
0
� -10% ••••• • • • i ·v
0 I
• ••
0
I • I
-20%
• 0 0 0

-30% • •• • I • I
I
0 • I ••
I
!
I I
-40% • 0 I 0
I
I
I
I
I I
-50%
0 200 400 600 800 1000 1200

50% . '
0
II
! I •
I • Interpolation I
I I
40%
.•
.
...
.
! I
30%
. ... I • . I
I
I

•• • •0••• I l I
0 •
I i 0

I • . 0 0 0 II oo. • . .: •
0

....
20%
• I · · • • . . . .
. ..
: q, : 0 r ... .
....

.. •·•
0 .� .
• � ... •
i • • I.
10% I

.... •
.... ! .
00
. : .. . . .
0.

�
0
�
0%
··
ell •• !
I
I .
• •• I
-10%
•. • I I
I
.
I
-20%
•• 0

.
:
.
'
I .
I
I
!
I
I
-30% 0. .
•• •. . !
• '
!
I
-40%
-50%
.
i I i I I
i

0 200 400 600 800 1000 1200

Module Energy Rating Methodology 1 14

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module S
.......
35 I
I I

� 30 H • Anderson ! I I II I --
1-< I I I�....� .. I ---- I
Q.)
I
-· j
25
;s: I . , ,. .I I
0
I
20
tl.. I i 0:. .,�.. I� I o.S.
I
"'d
Q.)
.r. r::---- I
15
.... .. -·
(';) I. • I I
10
• •• •
•• O O ! � I I I
.§
...... . •.- "4 I I I I
Ul
!
5
!:I:! I
I
0 � ·I I I I
0 5 10 15 20 25 30 35

�1-<
35
30 H
I
•
I
Blaesser
I
�� I I i
i
I
I 1 ---
1 .-.:----- 1
Q.) 25
;s: I I 1 I � II I i
0 20
�I �
tl..
"'d I I I
i
I
Q.)
15
.... ..
(';)
10 I �� �-- I
.§
...... .. �� I
!• I I I
Ul 5
!:I:! i,
0 ��I I
I I
I
0 5 10 15 20 25 30 35

35 I
I '
� • i I
I I
--
� 1-<
30 Interpolation •
I
I
i
I
I

... ... I ----

I
I
Q.)
l
;s:
25
I I
0 I ... ... J� I
' I
l
20
tl.. I I ·· "' 111"!� ! I
i I
"'d
Q.)
-�.:�...- � l
I
15
..... I I II
(';) A 'r: ;;;;-
10 I
• ' i
.§...... •• !•
· � I i I I I
Ul I I
.J. :.,.,.--
5
!:I:! I I
0 ...liii/l1fW" ! I I
I
I
I

0 5 10 15 20 25 30 35

35
� 1-<
30
Q.) 25
;s:
0 20
tl..
"'d
Q.)
.... .
15
(';)
10
.§
......
Ul 5
!:I:!
0
0 5 10 15 20 25 30 35

Actual Power (W)

Module Energy Rating Methodology 1 15

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module S
50% •• • o 0'l. ••
I '• • •

_'I. . 0
I
-�I o0!'o ;Oo
I • Anderson I
.. . •• •
... .
• • ,...• ..... ,...�·· ,. , _ .,..�
30%
....
0
....
....
10% II I 0
I .,
I
,

I I
"' •
� I
0
-10%
I
�

I
-30%

-50% I
0 200 400 600 800 1000 1200

50% ·•�• • • I
.,�. I I I I • Blaesser I
� --.; ••• • ·
30%
.... · : !:. · .
0
00
I I
I
0 . 0.
.
• ..•. ..
....
....
10%
0 o;)o\. 0 0
, : .. 0 .
I

� • •
• . tiiJ.� :.•• • • • .. o •• • 0
.. . o ..... �r.... .._..-
I
!" I of r • n .---
I
�- •
0
-10% ... . -
I
�
-30% I
I
I
I I
-50%
I
I
I
! I I
0 200 400 600 800 1000 1200

50% • I • Interpolation I
I •. I I I
30% •
• ••• ' .J. o • I

·· �I · •
• • • L• I •
• •o .,I •
.... .. 3> 0 • 0 - .-.��..a �-
0
....
....
10%
all&&
....
0
•' , . O o .._... I
....._
••

I
,�·-- _! ! ! '
•
�
• "' v '
0 II
-10%
I
�
• i
. .. . , I
!' I i
-30%

-50% ...
.. I I I I
0 ' 200 400 600 800 1000 1200

50% . ... .). • I

\ --..
ll· o ·.._ .
: . .
I
I "'
·
I
I
i I I · Myers I
• I
30%
....
0
....
....
10%
· ':t'i.
"··
•
I
I
•.__1 . 0
•
' ·I • � .... .
+ • 0 •
I
I
i
1 . •••
I
II
II
• • �-
� .... 1
• .. . I• .
.. .
.
...�:-,, 0
.
. .. ... .. If"·: ·, • • 0' ---
!II
I
.
0
-10%
I
�
II I
I
I
I ! I
-30% I
I
I I I
I
II i !
-50%
0 200 400 600 800 1000 1200

Irradiance (yY /m2)

1 16
Module Energy Rating Methodology
 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 5 Current at 14.4V (nominal)
2.5
I • Anderson
0
I
:? 2.0
$
......
c
Q)
'-< 1.5
'-<
::l
u
"0
....
Q).. 1.0
ell

:§
<ll
J:r.l
0.5

0.0
0.0 0.5 1.0 1 .5 2.0 2.5

2.5
I • Blaesser I I I
I

::2 2.0
I
I
$
�
Q)
'-<
'-<
1.5
0
.II • •
•
0
0
0 0
::l I •
u ·
"0 0 II
I
1.0
0
.l:l •
('(l •
•
:§
<ll
•
J:r.l
0.5 0
0

0.0
0.0 0.5 1.0 1.5 2.0 2.5

..c:I
,....._
2.5
I • Interpola�on II
2.0
$ o iI • • 0
�
• •
Q) 0 •
'-< 1.5
'-< ·: i •
::l I

u 0 •
"0
2
1.0
('(l \ ..
•
-E<ll 0.5
0
� 0

0.0
0.0 0.5 1.0 1.5 2.0 2.5

Actual Current (A-h)

Module Energy Rating Methodology 1 17

 Module Energy Rating Validation Data
NREL Outdoor Test Facility
Module 5 Current at 14.4V (nominal)
• "' • I
"
50%
40% • ••
• • •
I I •
•
• I I • Anderson
• .,
I [I o I 0
.• • I • •• •
30% • •
. I •• I I·
I • • .. . . ...,
20% • •
.1> •
• 0 • •0 . 0
I 0 • •• 0 • •
• 0
•
1-< .
0 0 ••• • • • ••0
I· • . I
10% •
1-< ••
1-<
0% •
• I u

j:I.l
0
� -10% 0
I . r
I

-20% I • I
-30% • .
0
I
0 I
I
I I
I

I
i i I
II
-40%
-50%
• I
0 200 400 600 800 1000 1200

50% .

40% . .•.
I I • i i
I
I o I • Blaesser �
..0. . • I
• •0 • i
I I
!
I
I
iI
30%
20% •
0 I•
t
•
I 0
I
•
• I
II i Ii •
1-<
0 •Z I • I • •
I
10%
1-< � •
1-< ! •.. I I .
0% .2
•. • • .. • . •I · ... 0 . 0 • • •..• 0 0 0 • 8•
j:I.l
.... I •• ..
II • •
lO

• I ··•.l I
0
• cJ . 0 . • • III • • 0 •
� -10% •
•
-20% 0
oe:. •
• •• I .
! • 4 8
•• •
! !I
•• • I
-30% • ,. . <I' •
<I' I I !
I
I
• f •
I
I
I
I I I •
• I I
I • I
-40% •
-50%
• •
I • I I ! I

0 200 400 600 800 1000 1200

50o/o .��-�
.---, . ••.--
• .-------------,
• • 01-------------.- 1 ------------,---------- . --�-------------.
40% +--· ....
__ .
,.___ ____ --+- 0 -------------+
0 1----------·----+1-------------+--------------jl o Interpolation I
0• I 0 • •I 0
30o/o +---.--,�------+1--------------+I�--------------+,---0-----y- . ---+-------------+--------------+
0 1
20o/o +---A---ro.----+1---·�·------�--+--------------+I��L-----�-+�----
!0
·--
••·� �- . +� --------------+
0 1 • ! 0 • i. • • 0 ·0 ,
-
10o/o + �·�-J·� ·----+-----� +
------- 1� ------------
· !--,---------+
-+ 1 --� .--·�� v
!� - +--------------+
�
�-< ·: ---- i • I0 .,! . : ,. • • • o • l o o•
Ja Oo/o �--��� � -+1-----------8�7- o o· r.
, ��----------+�----��-L��--+-----+--+------------
' ··· . � ·· · · � • --+
� -10o/o +-�·�0� - ----+ -- +� -- ----·
----�-, � -- --� ·
�-
·0 ·- -·---·-+ :_ _�l
• -+ 1------------
o••oo• --+
___________ ____________
I •
-20o;;0 +----� •��--�-+ • I--------------+'----------------------�---+-------------+ • ------------
. . . --+
� 1
-30o/o +-� . -.--JL------+1---------.•.-----------------+-*�·---------+-------------+1--------------+
-40o/o +----
· · ,.�,-------+:-----o--------,---------·----�------------�--------------+l------------
!, . .
: �
--+
o/o �------------�-------------'--------------+--------------+1--------------+------------�
-50
0 200 400 600 800 1000 1200

1 18
Module Energy Rating Methodology
 Fonn Approved
REPORT DOC U M ENTATION PAG E OMB NO. 0704-0188

�
fi
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewin instructions, searching existing data sources,

d9 ( ) g
gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regardin t is burden estimate or any other aspect of this
collection of information, mcluding suggestions for reauci this burden, to Washington HeadCJuarters Services, Directorate for In o ration O� �tions and Re orts, 1 2 1 5 Jefferson
Davis Highway, Suite 1204, Arlington, VA 22202-4302 , an to the Office of Management and Budget , Paperwork Reduction Proiect 0704-0188 , Washinpton, C 20503.

1 . AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED
January 1 998 Final Report

4. TITLE AND S U BTITLE 5. FUNDING N U M BERS

Photovoltaic Module Energy Rating P rocedure; Final Subcontract Report C: AAI-4- 1 4 1 92-01
TA: PV806301

6. AUTHOR(S)

C.M. Whitaker and J . D . N ewmiller

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGAN IZATION

REPORT NUMBER

Endecon Engineering
2500 Old Crow Canyon Road
San Ram on, California 94583

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 1 0. SPONSORI NG/MONITORING

AGENCY REPORT N U MBER

National Renewable Energy Laboratory

1 6 1 7 Cole Blvd. SR-520-23942
Golden, CO 80401 -3393

.
11. SUPPLEMENTARY NOTES

NREL Technical M onitor: B. Kroposki

1 2a. DISTRIBUTION/AVAILABILITY STATEMENT 1 2b. DISTRIBUTION CODE

; UC-1 270

1 3 . ABSTRACT (Maximum 200 words)

This docum ent describes testing and com putatio n procedu res used to generate a photovoltaic Module E nergy Rating (MER). The MER
consists of 10 estimates of the amount of energy a single module of a particular type (make and model) will produce in one day. Module
en ergy values are calculated for each of five different sets of weather conditions (defined by location and date) and two load types.
Because reproduction of these exact testing conditions in the field or laboratory is not feasible, limited testing and modeling procedures and
assumptions are specified.

1 4. S U BJECT TERMS 15. NUMBER O F PAGES

ph otovoltaics ; m odule energy rating ; MER ; module characterization ; performance 1 26

1 6. PRICE CODE

17. SECU RITY CLASSIFICATION 1 8. SECU RITY CLASSIFICATION 1 9. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT
OF REPORT OF THIS PAGE OF ABSTRACT
Unclassified U nclassified U nclassified UL

\JSN 7540-0 1 -280-5500 Standard Form 298 (Rev. 2-89)

Prescribed by ANSI Std. 239-18
298-102