Soiling losses

In this document

We will introduce you to the Solargis soiling model, which is essential for estimating soiling losses in photovoltaic (PV) systems, particularly in regions prone to dust accumulation. While the model's low spatial resolution of inputs and limited validation data may introduce some uncertainty, it provides valuable insights into the impact of dust on solar modules. This understanding enhances loss estimation and supports effective system performance planning.

Overview

Soiling in photovoltaic (PV) systems refers to the buildup of external materials, such as dust, pollen, pollutants, and sea salt, on solar module surfaces. This accumulation obstructs solar energy transmission to photovoltaic cells, leading to decreased efficiency and power output, a phenomenon known as soiling loss.

Soiling losses poses a significant challenge for solar PV projects, resulting in billions of dollars in economic losses globally each year and increasing planning and maintenance costs. Accurately modeling soiling is crucial for analyzing expected energy production, optimizing maintenance strategies, and forecasting outputs in operational solar plants.

The complexity of soiling arises from various interdependent factors that vary across time and space. Key influences include particulate concentration, chemical properties, and atmospheric conditions such as temperature, humidity, and wind. Rainfall can both contribute to soiling through wet deposition and mitigate it by washing away particles; its effectiveness depends on factors like quantity and distribution. Wind can either enhance particle interaction with module surfaces or resuspend particles, complicating the establishment of specific thresholds for these factors. Consequently, measuring and modeling soiling loss introduces significant uncertainty in solar project assessments.

The Solaris soiling model quantifies the impact of soiling by estimating the soiling ratio (SR) or transmission loss (TL), which correlates with the mass of deposited particles. This relationship is derived from advanced physics-based models tailored for PV plants. Despite uncertainties in input data, this model provides a reliable basis for estimating performance losses worldwide. It incorporates meteorological data (e.g., PM2.5, PM10, rainfall, wind speed, temperature) along with variables such as air density and module effective tilt to evaluate soiling effects. Cleaning events—whether natural (like rain) or manual—reset the accumulation process.

Modeling soiling losses in PV systems

Soiling loss in photovoltaic (PV) systems is a critical factor impacting energy production, particularly in areas prone to dust accumulation. This section outlines the methodologies used to model soiling losses, providing insights into the calculations and factors influencing the soiling ratio (SR).

Quantifying Soiling Losses

Soiling loss is quantified by the soiling ratio (SR), which compares the energy production of a PV module under soiled conditions (Epv-soiled) to its production under clean conditions (Epv-clean). The formula for SR is defined as SR=Epv-soiled/Epv-clean.

The SR is a dimensionless value ranging from 0 (complete obstruction by soiling) to 1 (fully clean panel). The relationship between SR and transmission loss (TL) is often assumed to be linear: TL=1-SR

This relationship accounts for the reduction in solar energy received by PV cells after passing through the module's glass and encapsulant. Given that uncertainties in model inputs can exceed the effects of nonlinearity between SR and TL, this assumption is practical for assessing soiling-related losses.

The soiling model

The Solargis soiling model (fig.1), based on research by Coello and Boyle (C&B), consists of two main components:

  • Accumulation of particles

  • Soling ratio calculation

The model calculates the mass (m) of suspended particles deposited on PV surfaces, considering both natural and scheduled cleaning events. The formula used is:

m=∫(v2.5C2.5+v10-2.5C10-2.5)cosβtdt

where:

  • C represents the concentration of particles in different size ranges,

  • v denotes deposition velocities,

  • t is accumulation time,

  • and β is the module inclination angle.

Solargis enhances this model by dynamically estimating deposition velocities using advanced physics-based formulations.

Once the accumulated mass (m) is determined, an empirical relationship estimates the soiling ratio:

SR=1-0.3437erf(0.17m0.8473)

Soiling losses calculation time during the simulation depends on the number of segments in the energy system. More segments result in longer estimation computations.

Figure 1: Scheme of the soiling model implemented by Solargis. Deposition velocities are highlighted as the component of the scheme where innovation work has been carried out.

Data sources and resolution

The required meteorological parameters, including temperature, atmospheric pressure, wind speed, and rain, are obtained from ERA5 and ERA5-Land. The spatial resolution of ERA5 data is approximately 31 km globally, while ERA5-Land provides finer detail at 9 km resolution.

Two key input parameters for the soiling model—PM2.5 and PM10 particulate matter concentrations—are derived through a harmonization process involving multiple sources. These particulate matter data adhere to the spatial and temporal resolutions of CAMS reanalysis, which are approximately 80 km (0.75º x 0.75º lat-lon regular grid) and 3 hours, respectively. This coarser resolution may limit the precision of localized soiling predictions for PV systems.

Factors influencing soiling

Several factors affect particle deposition effectiveness:

  • Atmospheric variables: Wind speed, temperature, relative humidity, rainfall, and atmospheric pressure significantly influence deposition rates.

  • PV system configuration: The inclination angle of modules, tracking systems, and materials used can enhance or reduce particle accumulation.

Dry deposition process

Dry deposition occurs in three steps:

  1. Aerodynamic Transport: Particles are transported from the atmosphere to a stagnant air layer at the panel surface.

  2. Molecular Transport: Particles move through this layer to the panel surface.

  3. Uptake at the Surface: Particles adhere to the surface.

Cleaning Events

Cleaning processes—both natural (e.g., rainfall) and manual—interrupt particle accumulation on PV panels. In the Solargis model, these events are treated independently and affect the mass accumulation variable directly (fig.2). The cleaning efficiency factor ranges from 0 (no effect) to 1 (completely clean), determining how effectively cleaning removes accumulated soiling.

Figure 2: Time series of daily mean values of SR observed and estimated by the model. The daily cumulative rainfall values are also shown.

Figure 3: Global map of long-term annual average values of daily SR estimated with the soiling model of Solargis for the period 2003-2021 with a precipitation cleaning threshold of 3.8mm/day.

Conclusion

Accurately calculating soiling losses is essential for improving the reliability of PV system performance predictions by reflecting real-world environmental conditions. It enables better energy yield forecasts, helps optimize cleaning schedules, and minimizes economic losses caused by reduced energy production. By integrating local factors such as particle concentration, rainfall, and wind, advanced soiling models provide globally adaptable and consistent results, enhancing both system design and operational efficiency.

This document provides an overview of the Solargis soiling model, highlighting an innovative configuration for estimating soiling losses based on the work of Coello and Boyle (2019). The key advancement lies in the improved parameterization for calculating deposition velocities, along with adjustments tailored to specific PV installations.

Evaluations show that this new approach enhances dynamic deposition velocity calculations compared to previous models. Additionally, it offers flexibility to adapt to the unique environmental conditions of various locations worldwide.

While the soiling model provides significant advancements in estimating energy losses, it is important to recognize its limitations. The model incorporates simplifications, such as constant or parameterized deposition velocities, which may not fully capture local pollution sources or the diverse physicochemical properties of particles. Additionally, uncertainties arise from factors like partial cleaning by rain or wind and the spatial resolution of input data. Since soiling is a highly localized and complex phenomenon, further research is required to improve the model’s accuracy and adaptability across varying environmental conditions. Current efforts focus on refining rain and wind cleaning efficiencies, but the model still does not account for wet deposition mechanisms or the specific impact of contaminating particles like black carbon or sulfates, which are more effective at blocking solar irradiance than mineral dust. These gaps highlight areas for future investigation to enhance the model’s reliability under diverse scenarios.

Comparison with other software

Soiling losses are accounted for similarly in other solar simulation software:

Software

Parameter name

Notes

Solargis Prospect

Dirt, dust, and soiling

Accounted as an irradiance loss (GTI).

Solargis Evaluate

Pollution losses

Accounted as an irradiance loss (GTI).

PVsyst

Soiling loss factor

Accounted as an irradiance loss.

SAM (NREL)

Soiling

Applies to both the beam and diffuse components of the POA irradiance (page 33).

SolarFarmer (DNV)

Soiling

Irradiance reduction.

Further reading