Ask an O&M contractor how often panels should be cleaned and you will usually get an answer in cleans per year. Ask why that number, and the reasoning is thin.
The honest answer is that it depends entirely on where the plant sits, and it is measurable.
| Environment | Typical annual soiling loss |
|---|---|
| Urban rooftop, regular rainfall, no local sources | 1 - 2% |
| Industrial estate, moderate traffic | 2 - 4% |
| Near cement, quarrying or bulk materials handling | 6 - 15% |
| Agricultural, harvest season dust | 4 - 10% seasonally |
| Coastal, salt film | 3 - 6% |
| Arid region, long dry season | 8 - 20% by end of dry period |
The spread is enormous, which is precisely why a generic schedule is a poor instrument.
Rain cleans tilted modules reasonably well. It cleans low-tilt modules poorly, and it barely cleans flat ones at all – water sits, evaporates, and leaves the dissolved solids behind as a film.
Below about 10° tilt, natural cleaning is weak and a dirt band forms along the lower frame edge where runoff collects. Above 20°, rainfall does much of the work in any climate with regular precipitation.
Commercial flat roofs frequently use 5-10° tilt to reduce wind load and row spacing. That choice, made for good structural reasons, quietly commits you to a higher cleaning requirement for twenty-five years.
Two practical methods:
Clean a control section. Pick two comparable strings with matching orientation and tilt. Clean one, leave the other. Compare normalised output over the following weeks. The divergence is your actual soiling rate, and the recovery immediately after cleaning is your current soiling loss.
Soiling sensors. Dedicated instruments compare a maintained reference cell against a soiled one. More expensive, and generally only justified on larger plants or where a soiling clause sits in a performance guarantee.
The control-section method costs almost nothing and answers the question well enough for a cleaning decision.
The calculation is simple once you have a soiling rate.
Take a 700 kW plant producing 800 MWh a year, displacing power worth 0.14 per kWh. Each percentage point of soiling loss costs about 8 MWh, or roughly 1,120 per year.
If cleaning costs 2,800 per visit and recovers 3.5 percentage points, that visit is worth about 3,900 in the year following – assuming soiling rebuilds gradually rather than instantly. Worth doing. If the same visit recovers 1.2 points, it returns about 1,340 against a 2,800 cost. Not worth doing.
The variables that decide it are your soiling rate, your energy value, and local cleaning cost. All three are knowable for your specific site, which means this should never be a matter of opinion.
One well-timed clean often outperforms three poorly timed ones. Clean at the start of your highest-yield period, and after whatever local event drives the worst accumulation – harvest, a dry season, a nearby construction phase.
Cleaning in late autumn in northern Europe, just as irradiance collapses and winter rain arrives, is close to wasted money. It happens routinely because it fits a maintenance calendar.
Aggressive cleaning that damages the AR coating causes permanent loss that no amount of subsequent washing recovers.