Skip to main content

Assessing the Factors Underpinning PV Degradation Through Data Analysis

This DuraMAT project focuses on analyzing photovoltaic (PV) system degradation with production data and investigating the correlation between bill of materials (BOM) and module durability.

The PVPro method enables the estimation of single-diode model parameters at the string level using production and weather data and predicts future degradation. This allows operators to not only understand overall degradation rates, but also identify specific causes of power loss. PVPro builds precise physical models for degradation analysis and power prediction.

Additionally, the project includes a comprehensive analysis of how BOM features correlate with module thermomechanical durability using machine learning techniques. The analysis here uncovers the predominant design factors that influence module durability among entangled BOM features, and it also quantitatively demonstrates the trend of this influence. Through this dual approach, the research addresses the challenges of monitoring existing PV system health, identifying root causes of degradation and improving future module designs for enhanced durability.

Core Objective

Central Data Resource

Team Members

Anubhav Jain, Baojie Li, and Xin Chen at Lawrence Berkeley National Laboratory (LBNL)

Impact

The PVPro method enables PV system operators to perform detailed degradation analysis and obtain accurate power predictions using commonly available production and weather data. This research directly supports the DuraMAT goal of analyzing and predicting the long-term degradation of PV systems. The BOM project identifies key BOM features that influence module durability and provides insights into improving future module designs. The findings support DuraMAT’s goal of understanding the influence of materials on module durability.

Learn More

Publications

Chen, Xin, Todd Karin, and Anubhav Jain. 2025. “Analyzing the Impact of Design Factors on Solar Module Thermomechanical Durability Using Interpretable Machine Learning Techniques.” Applied Energy 377: 124462.

Li, Baojie, Xin Chen, and Anubhav Jain. 2024. “Power Modeling of Degraded PV Systems: Case Studies Using a Dynamically Updated Physical Model (PV-Pro).” Renewable Energy 236: 121493.

Li, Baojie, Todd Karin, Bennet Meyers, Xin Chen, Dirk Jordan, Clifford Hansen, Bruce King, Michael Deceglie, and Anubhav Jain. 2023. “Determining Circuit Model Parameters From Operation Data for PV System Degradation Analysis: PVPRO.” Solar Energy 254: 168–181.

Tools and Other Information

PVPro (GitHub)

BOM Analysis (GitHub)

DuraMAT DataHub: BOM Dataset

Contact

To learn more about this project, contact Anubhav Jain, LBNL.

Production data (voltage & current) and weather data (irradiance & temperature) are the physical model parameters that produce charts showing past degradation analysis data and future power prediction data

PVPro workflow, from production data to degradation analysis and power prediction.

Method includes data collection (BOM files, dataset, modules, thermal cycling) that leads to data preprocessing, ML fitting and selection, SHAP analysis, and statistical testing that result in Inference (bar and line charts).

SHAP analysis reveals key BOM factors affecting module thermomechanical durability.