PVCracks Documentation
Overview
PVCracks is the DuraMAT project that investigates the effects of cracks on power loss in photovoltaic (PV) solar cells and tracks crack progression over time. This effort was lead by the Photovoltaic System Evaluation Lab at Sandia National Laboratories. We provide:
Open‐source cell‐level imaging and electrical datasets
MultiSolSegment: crack, busbar, and dark‐area segmentation in EL images
Variational Autoencoder (VAE): unsupervised parameterization and clustering
XGBoost model: per‐cell power‐loss estimation (ΔPMPP)
pvspice_lite: lightweight SPICE tools for I–V curve simulation
Schematic
This is how the workflow is set up. In addition to the SPICE simulation capabilities that are being added.
Data & Models
All datasets, trained model weights, and additional resources are hosted on DuraMAT DataHub: https://datahub.duramat.org/project/pv-crack-stress-and-power-loss
Cell‐level EL & electrical data (publication pending)
MultiSolSegment training images & masks (DOI: 10.21948/2587738)
MultiSolSegment model weights (DOI: 10.21948/2997859)
VAE model weights (DOI: 10.21948/2997860)
Publications
MultiSolSegment In revision for Solar Energy (Elsevier). Preprint (10.2139/ssrn.5506469)
Variational Autoencoder (VAE) EUPVSEC 2024 conference proceeding (10.4229/EUPVSEC2024/3BO.15.6)
Power-Loss Model (XGBoost) 2025 IEEE 53rd Photovoltaic Specialists Conference (PVSC) proceeding (10.1109/PVSC59419.2025.11132966)
Updates
Dec. 9 2025: DuraMAT webinar is uploaded here: www.duramat.org/news-and-events/webinars Direct Link to YouTube
Acknowledgments
Funding was provided as part of the Durable Modules Consortium (DuraMAT), an Energy Materials Network Consortium funded by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Solar Energy Technologies Office under agreement number 32509. The views expressed in this article do not necessarily represent the views of the DOE or the U.S. Government. The U.S. Government retains—and the publisher, by accepting the article for publication, acknowledges—that it retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this work, or to allow others to do so, for U.S. Government purposes.
Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.
Installation
Clone and install:
git clone git@github.com:NormanJost/pvcracks.git
cd pvcracks
# 1. Upgrade tooling
pip install --upgrade pip setuptools wheel
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install in editable mode
pip install -e . # enables intra‐project imports