Soiling from dust, pollen, pollution, and bird droppings can reduce PV output by 215% depending on climate and tilt. Cleaning restores yield but adds O&M cost and operational complexity. At Energy Solutions, analysis combines field studies, SCADA data, and O&M contracts to quantify when cleaning improves lifecycle returns and when rain and self-cleaning coatings are sufficient.
PV module output declines when surfaces accumulate dust, organic material, and other contaminants. Losses depend on climate, tilt angle, module framing, and nearby activities such as agriculture, mining, or traffic. Soiling typically develops gradually and can be missed in routine performance-ratio checks unless specifically monitored.
Energy Solutions analysis distinguishes between light uniform soiling (e.g., thin dust layer), non-uniform shading (e.g., bird droppings on cells), and event-driven soiling such as construction dust storms. Each requires different cleaning strategies and thresholds.
| Climate / Site Type | Example Regions | Typical Annual Soiling Loss | Loss Range Observed |
|---|---|---|---|
| Temperate, moderate rainfall | Northern Europe, US Pacific Northwest | 23% | 15% |
| Subtropical urban/industrial | Southeast US, Southern Europe | 36% | 28% |
| Arid / semi-arid high-dust | MENA, India, US Southwest | 812% | 620% |
| Agricultural proximity (dust & pollen) | Rural Europe, Latin America | 47% | 310% |
Source: Energy Solutions portfolio analysis (20242025); fixed-tilt systems, 1-axis trackers show slightly lower average loss.
Cleaning economics hinge on the interaction of soiling rate, PPA or tariff price, and cleaning cost per visit. The following table summarises stylised benchmarks for a 1 MWdc plant under different site categories.
| Site Type | Baseline Soiling Loss | Cleaning Strategy | Cleaning Cost (USD/year) | Extra Revenue (USD/year) | Net Gain (USD/year) |
|---|---|---|---|---|---|
| Temperate rooftop C&I | 3% | 2 manual cleanings/year | 6,000 | 10,500 | 4,500 |
| Arid utility-scale (desert) | 10% | 6 mechanical cleanings/year | 18,000 | 36,500 | 18,500 |
| Agricultural fringe | 6% | 4 manual cleanings/year | 10,000 | 19,500 | 9,500 |
Source: Energy Solutions modelling assuming PPA USD 55/MWh, degradation 0.6%/year, real discount rate 6%.
The most common mistake is to adopt fixed calendar-based cleaning (for example, monthly) without reference to soiling rates or market prices. More advanced operators employ soiling stations, satellite data, or comparative string analysis to trigger cleaning when marginal value exceeds marginal cost.
Methods range from manual washing with water and soft brushes, to tractor-mounted systems and fully automated robotic cleaners on trackers or rooftops. Each has different CAPEX/OPEX profiles and labour requirements.
Source: Energy Solutions O&M benchmarks for utility-scale PV in OECD markets (excludes land lease and insurance).
A logistics owner with 24 rooftop systems across three countries adopted data-driven cleaning schedules using inverter data and reference modules.
A single-axis tracking plant in a high-dust environment implemented robotic cleaning with limited water use.
Cleaning practices vary widely. In parts of Europe and North America, portfolios rely heavily on rainfall and occasional manual washing. In MENA and parts of India, cleaning is treated as a central design parameter with dedicated water logistics and robotics. Asia-Pacific markets span both extremes, from coastal, rain-rich climates to dusty inland regions.
Source: Energy Solutions surveys of O&M providers (20242025); typical number of cleanings per year.
While under-cleaning erodes revenues, over-cleaning introduces its own risks:
Many institutional investors therefore favour trigger-based cleaning policies anchored in soiling data and clear yield thresholds rather than fixed schedules.
By 2030, several trends are likely to reshape cleaning economics:
In Energy Solutions scenarios, cleaning and soiling management remain a small share of total PV O&M spend but a material driver of performance, especially in high-irradiance markets where each recovered percentage point of yield is valuable.
There is no universal rule. In many temperate urban settings, one to two cleanings per year combined with rain is sufficient. In dustier or industrial areas, three to four targeted cleanings triggered by soiling data are often justified.
Light to moderate rain can remove some loose dust but often leaves behind sticky residues, pollution films, or mineral deposits. Sites with frequent rain still benefit from occasional cleaning, especially near traffic or industrial activity.
Most module warranties require that cleaning methods do not damage glass, coatings, or frames. Abrasive tools, high-pressure jets, or incompatible chemicals can jeopardise warranties. O&M contracts typically reference manufacturer-approved cleaning guidelines.
At minimum, high-resolution production data and reliable irradiance proxies are required. Dedicated soiling sensors, reference strings, or nearby reference plants significantly improve accuracy and allow robust estimation of soiling-induced loss.