Dirt costs output and cleaning costs money, and the right schedule is whichever makes those two together smallest. For most domestic arrays in a rainy climate, that is no schedule.
What do kWp and kWh mean?
kWh — kilowatt-hour
A unit of energy, and the thing your electricity bill charges you for. A
1,000-watt heater running for one hour uses one kWh. A typical home uses somewhere between
2,000 and 10,000 kWh a year depending on where it is and how it heats.
kWp — kilowatt-peak
A unit of capacity — how big the solar array is, not how much it makes.
It is the output the panels would produce under standard test conditions: bright, cold and
perfectly aimed. Real roofs rarely see those conditions, which is why a 4 kWp array does not
generate 4 kW for most of the day. One modern panel is roughly 0.4 kWp, so 4 kWp is about ten
panels.
Putting them together
kWp is the size of the system; kWh is what it produces over time. The ratio between them
— kWh generated per kWp installed, per year — is the honest way to compare
locations, because it strips out how big the system happens to be. It runs from roughly 700 in
cloudy high latitudes to over 1,800 in sunny deserts.
Never — let the rain do itbest cleaning schedule
+£0a year, against never cleaning
£2dirt costs if you never clean
0.17%average soiling, rain only
Leave it to the rain. Cleaning this array costs more than the dirt does — never cleaning loses £2 a year, and the cheapest schedule that improves on that does not exist.
Rain is the reason. It rinses the array roughly every 8 days here, which
resets the dirt before it has time to build up — and a clean booked for a week it has
already rained on buys nothing at all. The drier the site, the more that changes.
Every schedule, costed both ways
Every frequency costed the same way: the energy dirt takes, plus the money cleaning costs. The
best row is the one where those two together come to least — not the one that keeps the glass
cleanest.
How often
Cleans a year
Average soiling
Energy lost
Worth
Cleaning cost
Total a year
Against never
Never — let the rain do it — best
0
0.17%
10 kWh
£2
£0
£2
+£0
Twice a year
2
0.17%
10 kWh
£2
£240
£242
−£240
Three times a year
3
0.17%
10 kWh
£2
£360
£362
−£360
Quarterly
4
0.17%
9 kWh
£2
£480
£482
−£480
Every two months
6
0.17%
9 kWh
£2
£720
£722
−£720
Monthly
12
0.16%
9 kWh
£2
£1,440
£1,442
−£1,440
Fortnightly
26
0.14%
8 kWh
£2
£3,120
£3,122
−£3,120
Weekly
52
0.10%
6 kWh
£1
£6,240
£6,241
−£6,239
Notice what happens at the frequent end. Cleaning weekly does keep the glass cleanest — the
average soiling figure keeps falling all the way down the table — and it is still the worst
option on the page, because each visit costs more than the dirt it removes was ever going to take.
A cleaning schedule optimises money, not cleanliness, and those have different answers.
How the soiling is modelled
Dirt accumulates at the rate you set, day by day, until it reaches the ceiling you set —
glass does not keep getting dirtier for ever, and a tilted panel sheds more than a flat one. Two
things reset it: rain, on its own cycle, and cleaning, on yours. Soiling on any given day is
measured from whichever happened most recently.
The loss is then weighted by that day's own generation. This matters more than it
sounds. A 5.40 kWp array in London, United Kingdom makes
several times as much in June as in December, so dirt that accumulates through a bright dry spell
costs far more than the same dirt in midwinter. Multiplying an annual average soiling figure by an
annual generation total — the usual shortcut — misses that entirely.
What this does not model
The soiling rate is your figure, not a prediction. Nothing here knows about
the pollen, the construction site, the farm track or the gulls near your roof. Published
soiling rates span more than a factor of ten between sites, and the named options above place
familiar conditions on that range rather than measuring anyone's.
Rain is an average interval, not weather. Real rain arrives in clusters and
dry spells, and a dry spell in high summer costs disproportionately; an average smooths
exactly the variation that matters most. Every figure here is averaged across all the ways
the rain could line up with your cleaning dates, because otherwise the answer depends on
whether the two intervals happen to share a factor — which is arithmetic, not weather.
Light rain can make soiling worse. Drizzle on a dusty panel leaves streaks
and deposits. Only rain heavy enough to rinse is modelled, and only as a full reset.
No snow, and no bird droppings. Snow is a total loss while it lies and then
cleans the glass as it goes; localised soiling on one cell can cost far more than its area
through the module's bypass diodes. Neither is a uniform derate and neither is here.
No damage risk, and no cost of access. Walking a roof has a cost that is not
the cleaner's invoice.