DuraMat CO1 Central Data Resource: How it started, how it’s going …
DuraMat CO1 Central Data Resource:
How it started, how it’s going …
Anubhav Jain*
LBNL
Aug 30, 2022
2016 proposal Now
*w/ contributions from: Todd Karin (LBL), Xin Chen (LBL), Robert White (NREL), Thushara Gunda (Sandia), Cliff Hansen (Sandia), Bennet Meyers (Stanford), Brittany Smith (NREL) and their respective teams
slides (already) posted to
https://hackingmaterials.lbl.gov
A variety of projects with some common philosophies
Core functions and
data common to many
PV analyses
Operation and
degradation in the
field
data hub
pvanalytics
pv climate zones
pv-pro
pvOps
pvARC
pv-vision
simple LCOE calculator
vocmax
Planning and
reduction of LCOE
Open-
source
licenses
Python /
pydata
stack
Maintain
via
Github
Suitable
for large
data sets
DuraMat Data Hub (R. White)
Current Consortium Stats
239 86 271 9653
Users Projects Datasets Resources & Files
Soiling Data Analysis Optical Mapping EL Crack detection Fleet degradation
Summary Impact
A secure data portal was
developed and deployed at:
https://datahub.duramat.org
… and populated with
DuraMat 1.0 project data!
~10 months usage
Citation Indexes and Registries
Science Communities and Public
Data Hub
DOI request applied
to archived data file
• Increased data
visibility via DOIs
• Infrastructure
synergy with other
EMNs
• Continues as data
platform for
DuraMat 2.0
• See also:
Energy Material Network Data Hubs Software Platforms for Advancing
Collaborative Energy Materials Research
https://doi.org/10.14569/ijacsa.2021.0120677
downstream effects of DOI
pvanalytics (Cliff Hansen, Will Vining, Kirsten Perry, Matt Muller)
Summary Impact
The pvanalytics project implements a suite
of data cleaning and data filtering tools that
are available at:
https://github.com/pvlib/pvanalytics
Automated tests ensure code correctness
• Advance state of the art for various
data filtering algorithms
• Updates functions in pvlib
• Provides workflow-independent tools
for various downstream analyses
Updated clear sky detection submitted for publication (C. Hansen, D. Jordan)
pvOps (Thushara Gunda, Michael Hopwood, Hector Mendoza)
Summary Impact
The pvOps projects contains modules for
working with O&M text data:
• Consistent labels for various types of
O&M procedures
• Routines for aligning O&M logs to
production data
https://github.com/sandialabs/pvOps
• Allows for large-scale
analysis of O&M logs
• Better estimates of
actual production
losses from various
mechanisms
• Software repository
can be expanded as
methods improve
See also:
pvOps: Improving Operational
Assessments through Data Fusion.
https://doi.org/10.1109/PVSC43889.202
1.9518439
Analysis of PVROM
data set (800+ sites)
Affected hours
(inverter vs tracker vs module)
Total production loss
(inverter vs tracker vs module)
Production loss distribution
(inverter issues)
pv-vision (Xin Chen, Todd Karin, Anubhav Jain)
Summary Impact
The pv-vision project contains routines for analyzing
module EL images and extracting:
• Defect classes at the cell level
• Pixel-level segmentation of cracks, busbars, and
power-loss areas
• Feature extraction such as crack length, isolated
area as defined by crack/busbar pattern
https://github.com/hackingmaterials/pv-vision
• Allows for large-scale
analysis of EL data,
being adoped by
industry
• Can be used to
determine how
environment affects
crack features which in
turn affects power
(ongoing PREDICTS2
collaboration)
• See also: Automated defect identification in electroluminescence
images of solar modules.
https://doi.org/10.1016/j.solener.2022.06.031
Effect of fire damage on defect count
(>18,000 modules)
Correlating power loss w/crack length
Detection and classification of defective cells (cell-level)
Labeling of crack, dark regions, and busbars (pixel-level)
Vectorizing cracks for
feature analysis
vocmax (Todd Karin, Anubhav Jain)
Summary Impact
The vocmax project allows planners to accurately
estimate string lengths to improve overall LCOE
• Conventional methods evaluates Voc at minimum
historical ambient temperature AND irradiance of
1000 W/m2 (extremely conservative)
• Vocmax models the Voc and Voperating at the desired
location to produce a distribution of expected
voltages – leading to longer strings
https://github.com/toddkarin/vocmax
Web tool at:
https://pvtools.lbl.gov/string-length-calculator
• In the US, modeled
string lengths
increased by 10%
using site—specific
modeling, leading to
potentially 1.2%
reduction in LCOE !
• Article in Solar
Power World online,
continued interest by
IEs
• See also: Photovoltaic String Sizing Using Site-Specific Modeling.
https://doi.org/10.1109/JPHOTOV.2020.2969788
Distribution of increased string sizes
for various criteria vs standard
Distribution of simulated voltages as compared
to various conventional thresholds
Other DuraMat 1 Central Data Resource Projects
PV Climate Zones
Comparative
LCOE calculator
pvarc pvpro
https://github.com/DuraMAT/pvarc https://github.com/DuraMAT/pvpro
www.github.com/NREL/PVLCOE
Perform instantaneous
technoeconomic analyses –
e.g., break-even cost
analysis of new technologies
based on performance and
system specs.
Given DC voltage/current
& environmental data as
input, PVPro determines
time-dependent values of
single diode parameters
to understand root causes
of power degradation.
Method for field estimation
of ARC thickness using an
integrating sphere, dSLR, or
even phone camera
Spectrum / color of reflected
light indicates ARC condition
Defines climate zones
more relevant to PV
degradation than Koppen-
Geiger
Used by others to analyze
PVFleets & PVROM data
Photovoltaic degradation climate zones.
https://doi.org/10.1109/PVSC40753.2019.8980831
https://github.com/toddkarin/pvcz
Photovoltaic module antireflection coating degradation
survey using color microscopy and spectral reflectance
https://doi.org/10.1002/pip.3575
Nondestructive Characterization of Antireflective Coatings on
PV Modules.
https://doi.org/10.1109/JPHOTOV.2021.3053482
• DuraMat 1.0 funded many projects that advance how
PV degradation is analyzed with data and software
• These projects were not one-off studies, rather they all
have associated public code
• Many of these new software implementations will be
leveraged going into DuraMat 2.0
Conclusions