FacebookXLinkedIn

Solar PV energy is one of the fastest-growing renewable technologies, with projects now deployed across nearly every climate zone contributing to the global energy transition. New solar projects begin with projections based on detailed models that estimate how much energy a plant will generate over the next 20 or 25 years. These models become the foundation for investor presentations, power purchase agreements and loan approvals, and as such, they are what make or break a project’s financial future. But as the industry has matured, project stakeholders have learned that seemingly small technical assumptions can have large financial implications.

Among the consequences of these assumptions, soiling loss, which is the reduction in energy yield caused by the accumulation of dust, sand, pollen, snow or other particulates on PV modules, remains one of the most underestimated. Soiling, which varies widely depending on geography, climate and season, reduces irradiance reaching solar cells, thereby lowering power generation. While soiling may seem like a minor maintenance detail, in reality it is a performance driver that can substantially impact a project’s economics.

In some regions, gentle rainfall takes care of the problem naturally, washing panels clean every few weeks. In others, like the Sonoran Desert of Mexico, soiling is a relentless force. There, annual losses can average 3.1%, far above the industry’s “default” assumption. If that higher rate is ignored during the early engineering phase, the project will promise investors more energy than it can actually deliver. Year after year, the gap between forecasted and real production grows wider, eventually costing millions of dollars in unrealized revenue. Even regions with moderate soiling face hidden risks. In central Chile, for example, soiling may stay near two percent on an annual basis, but spike dramatically during March, just when sunlight and market power prices are at their highest. That one month of underperformance can have an outsized effect on the plant’s bottom line, creating a problem that no simple annual average can capture.

As a result, soiling has major implications across the project lifecycle. In the EPC (Engineering, Procurement and Construction) phase, even a 1% error in soiling assumptions can ripple through a project’s financial model, shifting internal rate of return calculations, distorting debt sizing, and leading to design decisions that fail to optimize performance. In the O&M (Operations and Maintenance) phase, poor soiling management reduces performance ratios (PR), increases costs and accelerates equipment degradation. Tilt angles, tracker selection and ground clearance are all influenced by expected soiling. Miss the mark, and the project may forego anti-soiling coatings, robotic cleaners or other solutions that could have paid for themselves. It may even fall short of contractual performance guarantees, setting up disputes between contractors, owners and financiers.

Soiling losses in the EPC phase: The role of assumptions in energy yield estimates

Energy yield estimates are the cornerstone of project financing. During EPC, developers typically use simulation software such as PVSyst or Helioscope to forecast annual energy production. To account for soiling, many EPC teams adopt a flat assumption, most often 2% annual loss. This shortcut is common because detailed soiling data are rarely available in early project stages.

Once the project goes live, soiling is no longer a number: it is a dynamic, moving target. Whether driven by dry-wet cycles, dust storms, pollen seasons or industrial activity, the temporal distribution of soiling matters more than the annual average. Some months, wind and rain keep panels clear and production high. Other months, dust settles faster than crews can clean it off. Irradiance sensors sit gleaming and spotless, so performance reports don’t always tell the whole story. Hidden underperformance creeps in. Operations teams face a delicate balancing act: cleaning panels too often wastes money, labor, water and equipment life; cleaning too infrequently lets energy slip away, unrecovered. Over time, uneven soiling can create hot spots on modules, accelerate material degradation and even risk voiding manufacturer warranties. What started as a simple 2% assumption during the design stage has now become one of the most significant controllable costs in the plant’s operating budget.

Soiling thus affects both short-term production and long-term asset reliability. By proactively managing soiling, O&M teams protect asset value, extend module lifespans and ensure compliance with warranty and insurance requirements. To address soiling correctly, the optimal strategy lies in data-driven cleaning schedules. O&M providers increasingly rely on soiling sensors, meteorological data and predictive modeling to identify when cleaning provides the highest economic return. For example, cleaning may be scheduled just before peak summer irradiance in desert climates or delayed until after pollen season in temperate regions.

Bridging EPC and O&M: An integrated approach

Based on my experience, I envisioned a new approach: one that could turn raw data into a living, adaptive model of soiling behavior. This is what inspired me to develop the PV Soiling Loss Calculator, a tool designed to replace static assumptions with real, site-specific intelligence. Built with a Python-based backend using the Streamlit framework, the calculator integrates multiple data sources and modeling techniques and employs regression and time-series machine learning (ML) models trained on historical soiling data to predict accumulation rates. It employs regression and time-series Machine Learning Models trained on historical soiling data to predict accumulation rates, pulls real-time meteorological data such as precipitation and wind speed through API integrations, and allows users to input site-specific parameters such as panel orientation, tilt angle and cleaning frequency. Dedicated functions automatically pull real-world data from NOAA, including monthly rainfall, snow days and average temperature, while another function collects monthly humidity data from CityRating. The results simulate yield losses and estimate revenue impacts under different cleaning strategies, enabling sensitivity analysis and cost-benefit comparisons.

The calculator’s output is simple but powerful: a month-by-month table and an easy-to-read graph that visualizes the expected performance impact. In just seconds, users can explore what happens if they change tilt angle, cleaning frequency or even the timing of a cleaning cycle. There is no need to open a spreadsheet because everything is calculated and displayed interactively. Behind the scenes, Python libraries like pandas, NumPy and Plotly handle the data processing and visualization, while machine learning models built with scikit-learn allow the calculator to improve as more real-world data are collected.

For EPC teams, this means more accurate pre-construction energy modeling and better financial projections. For O&M teams, the PV Soiling Loss Calculator drives cleaning schedules that are dynamically aligned with seasonal soiling patterns, weather forecasts and even market price signals, optimizing a value-based, data-driven approach.

By serving as a shared and continuously improving reference point, the calculator bridges the long-standing gap between EPC assumptions and O&M realities. This alignment improves the accuracy of performance guarantees, reduces disputes between stakeholders, strengthens investor confidence, and ultimately drives better profitability and bankability for solar projects.

Soiling losses are often underestimated, yet they influence nearly every stage of a solar PV project from financing and design decisions in EPC to cleaning strategies and long-term reliability in O&M. Treating soiling as a dynamic, site-specific variable rather than a flat assumption is essential to protecting performance and profitability. Tools like the PV Soiling Loss Calculator show how data-driven approaches can align EPC forecasts with actual operating environments and ensure that solar assets deliver their full value over their lifetime.