Dust costs more than most EPCs think, and cleaning costs less

Every generation estimate you issue carries a soiling assumption. Most carry it silently, and most carry it wrong.

It is worth getting right, because soiling is the largest recoverable loss on an Indian rooftop and the only one the customer can fix with a bucket.

The numbers

Reported soiling losses in India run from around 6 percent in coastal regions to about 15 percent in arid zones. In the dusty parts of Rajasthan and Gujarat, performance ratio can fall by 0.5 to 1.0 percent per day between cleanings during dust season.

Cleaning frequency drives the average directly. Studies of cleaning intervals put monthly average soiling loss at roughly 1 to 1.5 percent with weekly cleaning, rising to around 5 percent when cleaning drops to monthly. For urban rooftops in high pollution areas the commonly recommended interval is every ten to fifteen days.

Two things follow that are worth saying out loud to a customer:

• Soiling is reversible. Clean the modules and the output comes back.
• Degradation is not. Modules lose output permanently at roughly half a percent a year. Customers conflate the two constantly, and a dirty array in year three gets blamed on "panel quality".

The arithmetic that makes the case

Take a 100 kW commercial rooftop generating around 1.4 lakh units a year, on a tariff of Rs 8 per unit.

Move from monthly cleaning to fortnightly and you are recovering something in the order of three percentage points of annual output. That is roughly 4,200 units, about Rs 34,000 a year, on one 100 kW system.

Run the same calculation on the customer's own numbers and it usually answers the question of whether cleaning is worth paying for. It generally is, by a comfortable margin, and it is one of the few conversations where the honest answer also happens to be the one that sells an annual maintenance contract.

What to put in the generation estimate

State the soiling assumption as a number, and make it match the cleaning regime you are recommending.

A proposal that models 2 percent soiling while the customer intends to clean twice a year is not an estimate, it is a future complaint. If they will not commit to a cleaning schedule, model the loss that matches what they will actually do, and show them both cases. The second column is the most persuasive maintenance pitch available.

Where this fits in the wider document is in what actually goes in a solar proposal, and the related honesty problem of shading is in shading analysis without guesswork.

Practical cleaning, without wrecking the asset

A few things worth standardising, because bad cleaning costs more than no cleaning.

• Water quality matters. Hard water leaves mineral deposits that are worse than the dust they replaced.
• Early morning or evening. Cold water on a hot module risks thermal stress, and midday cleaning evaporates before it does anything.
• Soft brushes and no abrasives. The anti reflective coating is the thing being destroyed when someone uses a scrubber.
• Never stand on modules. Micro cracks do not show until the output does, two years later.
• Access first. If the design left no walkway, the system will not be cleaned properly, whatever the contract says.

That last point is a design decision, not a maintenance one, and it is made on the day you lay out the array.

Who should be doing it

Three models, and you should have a view rather than leaving it vague at handover.

The customer cleans. Cheapest and least reliable. Works for a motivated residential owner with terrace access and a tap. Give them a written schedule and a short demonstration at handover, or it will not happen.

You clean, under an AMC. The right answer for most commercial systems. It gives you a recurring revenue line, a reason to be on site regularly, and early sight of faults before they become warranty claims. It also protects your generation estimate.

Automated systems. Robotic cleaners and waterless systems are now common on larger plants, and reported to cut per MW operations cost meaningfully. On a rooftop they need to be designed in, not retrofitted, and the economics usually need scale or severe water scarcity to work.

Where the water comes from

Ask at the site visit. A terrace with no tap means every cleaning visit carries water up a staircase, which is what turns a fortnightly schedule into a quarterly one in practice. In water scarce districts this question decides the cleaning model, and it belongs in the design conversation rather than in a footnote after handover.

The short version

• Soiling in India runs from about 6 percent on the coast to 15 percent in arid zones, and can cost 0.5 to 1.0 percent of performance ratio per day in dust season.
• Weekly cleaning holds average loss near 1 to 1.5 percent. Monthly cleaning lets it reach about 5 percent.
• Soiling is reversible. Degradation, at roughly half a percent a year, is not. Customers confuse the two.
• On a 100 kW system, moving from monthly to fortnightly cleaning is worth tens of thousands of rupees a year. Do the sum with the customer.
• State the soiling assumption in the generation estimate, and make it match the cleaning regime you recommend.
• Design for access and water at layout stage, or the cleaning schedule will not survive handover.

Sources

• Power Loss Due to Soiling in Solar PV Modules, Green AI Energy, trade coverage
• Machine learning based prediction of soiling losses under different cleaning frequencies, Scientific Reports
• Assessing the role of soiling mitigation in boosting solar plant generation efficiency, Journal of The Institution of Engineers (India)