Cement dust on a ±15.6 MWp solar plant in North West Province

The plant is a ±15.6 MWp solar PV plant of about 25,000 modules in North West Province, next to a cement works. Dust from the works settles on the modules, and cement dust doesn’t stay loose: moisture dissolves part of it, and it recrystallises into a film bonded to the glass that dry cleaning can’t break (IEA PVPS, 2022). On this plant the dust had cured onto the glass, and water-only cleaning could not remove it.

We looked at the glass, the cleaning water and the soiling data.

The water matters as much as the dust. At ±250 ppm as CaCO₃ it is hard by any measure: the USGS starts its “very hard” class at 180 mg/L. Used untreated, every rinse would leave scale on the glass.

The pilot, and how we measured it. On 24 June 2026 we cleaned one string with Chemitek Inorganic Removal Agent (CRA), working alongside the plant’s O&M team. We applied no protective coating, so the pilot tested restoration only.

We measured string current with a clamp meter. Soiling cuts the light reaching the cells and the current falls with it, so string current is a quick, direct reading of what the glass is costing. We read the cleaned string against comparable uncleaned strings in the same irradiance window, so both readings saw the same sun.

This is one CRA-cleaned string, read on one day, with no coating applied. It shows that CRA removes this plant’s cured cement, and that the cleaned string read 7.2% more current than comparable uncleaned strings in the same irradiance window. It is not an annual energy figure. That depends on how fast the dust returns, on rain and on how well the coating holds.

We also photographed the modules with a thermal camera on the day of the pilot. At the crosshair, cleaned modules read 26.9 °C and 28.3 °C, and an uncleaned module read 32.5 °C.

That matters because power is voltage times current. The clamp meter measured current, and a cooler module also holds a little more voltage: crystalline modules typically lose about 0.3–0.4% of their power for every degree they warm up (module datasheets). Field studies have found dusty modules running warmer than clean ones (Rashid et al., 2023). Taking the camera’s readings at face value, the cleaned modules ran about 5 °C cooler, which is worth roughly another 1.5–2% of power. Added to the 7.2% more current, that puts the power gain nearer 9% than 7%.

That is an estimate, not a measurement. A thermal camera reads the front glass rather than the cells, and clean and dusty glass reflect the sky’s heat differently, which shifts the readings (IEC TS 62446-3). Strings that share an inverter input also run at the same voltage. Measuring the temperature effect properly takes sensors on the backs of cleaned and uncleaned modules, logged alongside string power.

Heat also ages modules. NREL’s review of more than 11,000 degradation rates found that hotter climates, and mountings that keep modules hotter, can raise degradation in some products (Jordan et al., 2016). Clean glass helps a module run cooler, which is one more way soiling costs more than the lost current.

The pilot shaped a three-phase programme on a 24-month calendar, written as a washing procedure the plant’s O&M team can run.

Every wet step uses water softened with Chemitek Water Softening Agent.

A three-person relay. The first cleaner sprays diluted CRA from a backpack sprayer, staying about 15 seconds ahead. After that ±15-second dwell, the second agitates with a battery-powered dual rotating brush. The third follows with a water-fed rotating brush on softened water, rinsing before anything dries. When the row is done, a water-powered doser meters ASA 2.0 into the feed, and the water-fed brush makes one quick pass to apply the coating, which cures dry with no rinse.

Rules that protect the modules. Start on dry glass. Work in the early morning, late afternoon or under cloud, when the glass is cool. Never let product dry on the module. Use soft bristles. Never dry-brush cement-covered glass, and keep high-pressure jets off frames, seals and junction boxes.

The calendar. A restoration phase restores and coats the whole plant, followed by dry washes, softened-water washes and an SWP 50 protective clean. A repeating annual cycle then takes over: an inspection every January, SWP 50 protective cleans in May, September and November, softened-water washes in March and July, and dry washes between them where possible.

Equipment. A battery-powered dual-brush kit, a water-fed rotating brush, a water-powered dosing pump, a backpack sprayer, graduated mixing containers, a non-rotating telescopic brush for dry washes, and a clamp meter.

The +7.2% is the pilot’s result, not the programme’s. To measure the full gain, the procedure sets up monitored strings: crews log string current before and after each section and at every wet clean, plus weekly energy yield on strings agreed with us. With the site’s optical soiling sensor, those readings can show how fast soiling returns after each phase and whether the coating is holding.

The aim is that the plant never needs a full CRA wash again. Each January’s inspection and the monitoring data decide the year: with no severe contamination, it starts with ASA 2.0 and SWP 50; if cement has re-bonded, with CRA and ASA 2.0.

Want the same evidence from your own strings? Book a free pilot, or see how Restore → Protect → Maintain works.