2014 PV Performance Modeling Workshop: Optimizing PV Designs with HelioScope: Paul Gibbs, Folsom Labs

1. NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.
Improvements to PV
System Modeling in SAM
Sandia/EPRI PV Modeling Workshop
Aron P. Dobos
5 May 2014

2. 2
What’s new in SAM for Photovoltaics?
We’ve been working on a number of projects that will help
optimize the design of PV systems and more accurately model
their performance.
In 2013, we held a technical review committee that
recommended several areas for improvement.
• New “11 parameter” module model using IEC-61853 test data
• Much more flexible inverter models
• Validation work shows SAM and other models are able to predict
performance well
• SAM Shade Calculator for 3D scenes
• Historical NSRDB database update for inter-annual variability (P50/P90)
analysis
We are thankful for the support of DOE SunShot Systems
Integration subprogram that facilitates these efforts.

3. 3
IEC-61853 Module Model (1)
Problem:
• Sandia module model is accurate but requires complex testing, and predicts only 5
points on the IV curve
• 5/6 parameter single diode models can be inaccurate at low irradiances and for
some module technologies
Proposed solution:
• Create a backwards-compatible extension of the single diode model whose
additional parameters can be automatically calculated from IEC-61853 test data.
• Parameters measured at each test condition : Pmp, Vmp Voc, Isc
IRRADIANCE Spectrum Module Temperature
W-m-2 15 C 25 C 50 C 75 C
1100 AM1.5 NA
1000 AM1.5
800 AM1.5
600 AM1.5
400 AM1.5 NA
200 AM1.5 NA
100 AM1.5 NA NA

4. 4
IEC-61853 Module Model (2)
Recall: single diode models consist of two parts:
o The nonlinear I-V curve equation defined by parameters: a, IL, Io, Rs, Rsh
o These five parameters are STC values (@ 1000 W/m2 & 25 C)
o A set of auxiliary equations that translate the five parameters from STC to
operating conditions (temperature & irradiance)
Rs is assumed constant!
OK
OK
OK
?
NOT OK
NOT OK
 Egref is fixed at 1.121 eV

5. 5
IEC-61853 Module Model (3)
Observation: the I-V curve equation can fit module
performance data quite well for many technologies. Auxiliary
translation equations for parameters however are not adequate.
Approach: define new auxiliary equations that fit the data better.
Procedure: at each IEC-61853 test condition:
IRRADIANCE Module Temperature
W-m-2 15 C 25 C 50 C 75 C
1100 NA
1000
800
600
400 NA
200 NA
100 NA NA
1. Estimate the diode factor a from Voc &
temperature coefficient
2. Numerically solve for the remaining 4
parameters IL, Io, Rs, Rsh given Pmp, Vmp, Voc, Isc
3. Fit the behavior of these parameters as a
function of temperature and irradiance

6. 6
IEC-61853 Module Model (4)
Result: an 11 parameter model.
For a thin film module, these
parameter values were
obtained from the automated
solution procedure:
Improved auxiliary equations for Rs and Rsh:

7. 7
IEC-61853 Module Model (5)
• For a First Solar thin film module, the
improved model reduces power
prediction error on average from 6.1%
with the 5 parameter model to 1.3%
• Next version of SAM (Autumn 2014)
will include an implementation of the
module model and solver
• This model can help you optimize
system design by having a more
accurate representation of module
performance based on measured test
data

8. 8
Inverter Model Improvements (1)
By default, SAM uses the Sandia
Model for Grid-connected Inverters
in conjunction with the CEC inverter
database.
We’ve added two options:
• Enter inverter weighted efficiency
(European or CEC)
• Enter part-load inverter efficiency
curve
Model uses operating range inputs
to calculate clipping losses.
Voltage ranges are not enforced, but warnings
are issued during simulation.

9. 9
Model Validation and Intercomparison (1)
• Validation of SAM
model against 9
systems with measured
data
• Annual agreement*
within ± 3%
• Hourly agreement*:
o RMSE within 5.1%
o MBE within ± 1.0%
• Download report:
http://www.nrel.gov/do
cs/fy14osti/60204.pdf
* Mesa Top and DeSoto excluded from these
results. Annual numbers based on quality-
controlled data, with system downtime, outages,
snow hours, etc removed.

10. 10
Model Validation and Intercomparison (2)
Using same measured
datasets, compare SAM,
PVsyst, PV*SOL, and
PVWatts
Notes:
• PVWatts V1
underpredicts on
average by 14% with
default inputs (not
shown)
• PV*SOL does not
properly model 1 axis
tracking systems
• Many thanks to PVsyst
and PV*SOL teams for
helping review this work
Full report forthcoming this
year
Tool Error ranges for 6 systems
SAM -5.0% to 4.1%
PVsyst -1.7% to 5.5%
PV*SOL -5.5% to 1.4%
PVWatts -16.2% to -8.9%

11. 11
Updated Weather Datasets/Formats (1)
The NREL TMY2 dataset is based on 30
years of measured and modeled hourly
solar data from 1961-1990 at 239
locations in the United States.
The underlying actual year data hasn’t
been freely available until recently, and
there were significant gaps in
temperature or other meteorological
values which caused simulations to fail.
We’ve applied backfilling algorithms to
fill missing data with representative
values to enable simulations to run.
https://sam.nrel.gov/NSRDB
We’ve also developed a flexible
Excel-editable CSV-based
standard file format that will be
used in the next version of SAM.
• SAM will include 1619 locations
• Much easier to import/edit your own weather files

12. 12
New SAM Desktop Application (1)
In Autumn 2014, we will
release a completely
revised SAM desktop
application.
• Significantly streamlines
the user experience
• Much better access to
hundreds of outputs
• Simulations like
parametric and Monte
Carlo analysis will use
parallel processing
• Libraries and weather
data in Excel-editable
CSV files
• Much more powerful
scripting language and
SAM SDK integration

13. 13
Tonight’s Presentation
Upcoming major changes
to PVWatts algorithms
http://pvwatts.nrel.gov
Live demo of SAM Shade
Calculator Beta
http://sam.nrel.gov/shade

14. NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.
Major Updates to PVWatts
Sandia/EPRI PV Modeling Workshop
Aron P. Dobos
5 May 2014

15. 15
Why Update PVWatts?
• PVWatts V1/V2:
o Originally developed in the
late 90s based on algorithms
derived from performance
data measured in the 80s
• Turns out modern systems
perform better
• Our own validation work
shows that PVWatts
underpredicts
• Pretty consistent feedback
from users that PVWatts
underpredicts
The web tool also gets a
new modern look and feel.
http://pvwatts.nrel.gov

16. 16
Key Model Changes
• Module type options: “Standard”, “Premium”, or “Thin-film”
• Option to specify a DC-to-AC nameplate sizing ratio
• System losses are specified as a percentage, default of 14%.
• Inverter performance curve updated
• One axis tracking systems: linear self-shaded or backtracked

17. 17
Module Types and System Losses
• Default losses are roughly
equivalent to a V1 derate of 0.825,
which is about 7% higher
• Energy prediction is actually about
8-9% higher due to the revised
inverter performance curve.

18. 18
Inverter
• Based on
statistically most
representative
actual inverter in
the CEC database
since 2010
• Nominal efficiency
can be set by the
user, default is 0.96.

19. 19
Performance Summary
Old PVWatts is on average about 11.8% low.
New PVWatts is on average 1.8% low.
Notes: The 9 systems are well maintained and are generally considered
“unshaded”. The default shading loss of 3% in the new PVWatts was set to zero
to match the old PVWatts default.
-20
-15
-10
-5
0
5
10
1 2 3 4 5 6 7 8 9
AnnualEnergyPredictionError(%)
PVWatts Annual Error for 9 Systems
New PVWatts
Old PVWatts

20. 20
What’s Next for PVWatts
• Draft technical manual available online (see survey link
below)
• Online survey to get feedback on these changes
https://www.surveymonkey.com/s/pvwattsv5
• Changes will be finalized and new algorithms will be
rolled out in the next couple of months

21. NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.
SAM Shade Calculator
Sandia/EPRI PV Modeling Workshop
Aron P. Dobos
5 May 2014

22. 22
SAM Shade Calculator Beta
Based on consistent and strong feedback from the user
community and our Technical Review Committee in 2013, the
Department of Energy funded work to develop a 3D shading loss
calculation tool.
Download from: http://sam.nrel.gov/shade

23. 23
SAM Shade Calculator Overview
Our tool will:
o Enable basic analysis of shading impacts for a particular scene
o Provide a relatively crude representation of a 3D scene with
preprogrammed shapes
o Enable intercomparison of many shading tools
o Be able to read and write common file formats and ideally provide
an open standard for PV shading geometry data
Our tool will not:
o Render specific modules, racking systems, wiring conduits, or any
details of the balance of system
o Be able to display a realistic rendering or visualization of a
particular PV installation
Full roadmap available online.