Authors: | Presented by Tomas Cebecauer, PVPMC 2025 (Albuquerque) |
Tuesday, May 13, 2025 |
Overview
Soiling—accumulation of particulate matter on PV module surfaces—is a significant loss factor in photovoltaic system performance, contributing to substantial global energy yield reductions and financial impacts (estimated at €4–7 billion in 2023). This presentation outlines the latest advancements in soiling loss modeling, now fully integrated within Solargis Evaluate 2.0, which enables precise, site-specific simulation of soiling effects across multi-year, high-resolution time series.
The improved soiling model leverages environmental and meteorological data to capture local and seasonal dynamics of soiling, supports operational planning (such as manual cleaning), and significantly reduces yield uncertainty in both pre-feasibility and operational project stages.
Key elements of the model and simulation chain
Solargis Evaluate 2.0 features
Advanced 3D scene modeling, multi-year time series (1- to 15-minute resolution), new component and environmental models (soiling, snow, albedo), and cloud-based execution with validated outputs.
Soiling model
Based on the Coello & Boyle/HSU approach with Solargis's physically-based parametrization. It integrates:
Atmospheric data: PM2.5, PM10 (MERRA-2, CAMS reanalyses), harmonized for long-term, global, and consistent coverage.
Weather data: Rainfall (key for cleaning), wind speed, air temperature, humidity, atmospheric pressure.
Technical parameters: Module tilt, mounting type, manual cleaning events.
Model enhancements: Physical deposition velocity, PV-mounting adjustments, calibrated rainfall cleaning efficiency.
Soiling loss quantification
Supports simulations with or without cleaning, and flexible output aggregation (daily, monthly, long-term).
High spatial, seasonal, and interannual soiling loss variability.
Uncertainty of yearly soiling estimation is typically 5–10% but can be higher in some regions.
Validation and practical application
Model validation
Calibration performed using 39 ground measurement sites from diverse geographies.
After quality checking (QC) and corrections for cleaning and offsets, the model achieves bias of –0.5% (stdev 1.7%), improving to +0.1% (stdev 0.9%) without outliers.
Analysis covers daily and monthly soiling losses, showing strong agreement between measured and modeled results.
Integration into PV simulation chain
Soiling losses are applied directly to the solar irradiance step within the full PV simulation workflow.
Accessible in Solargis Evaluate under “Environmental losses.”
Supports TMY and time series simulations, evaluating performance under both default (no cleaning) and defined manual cleaning strategies.
Case studies demonstrate strong impact of cleaning strategy on yearly energy yield—highlighting potential to optimize OPEX and reduce financial risk.
Limitations and considerations
Coarse resolution of PM datasets and missing local emission sources can lead to underestimation of soiling losses at micro scales.
Global model harmonization and continuous data improvements address many of these challenges.
Users may further reduce uncertainty by integrating measured site-specific soiling data if available.
Conclusions
The Solargis soiling model (released in Evaluate v2.3.0, June 2025) offers global, site-specific, and highly validated simulations of soiling losses, eliminating the need for purely local measurement data. This new model enables realistic energy yield estimates, helps quantify operational cleaning impacts, and improves investment-grade bankability by replacing subjective “expert guess” values with science-based environmental data. As a result, PV yield uncertainty is reduced, expected OPEX is better understood, and project assessments are more reliable across development and operational phases.
Further reading
Soiling Losses – Impact on the Performance of Photovoltaic Power Plants by Tomas Cebecauer, Vicente Lara Fanego, Simon Stassel, Martin Mihal, Viktor Sklencar, Tomas Sasko, Lukas Dvonc, Branislav Schnierer