Most commercial skylights need robotic cleaning every 4-8 weeks in urban or coastal air, and every 8-12 weeks in low-dust areas. The trigger is light loss and bonded dust, not the calendar.
Facility managers who clean twice a year are usually reacting, not planning. By the time a roof looks dirty from the floor, transmission has already dropped 12-18 percent and the film has bonded. Bonded film needs more water, more passes and sometimes a mild acidic agent – all of which raise cost per square metre.
What actually sets the cleaning interval?
Three inputs decide it: local dust fall, rainfall pattern, and the type of dirt. Desert and industrial sites see 3-5 g/m2 of dust per month. A coastal site adds salt, which etches if left for a season. Sites under aircraft approach paths get kerosene residue that nothing removes once polymerised.
- Urban, traffic-heavy: 4 weeks
- Coastal or salt air: 4-6 weeks
- Suburban, tree pollen season: 6 weeks in spring, 10 weeks rest of year
- Rural, low dust: 8-12 weeks
- Post-rain dust storms: clean within 72 hours before the mud bakes
Is dirty glass really a cost?
Yes, and it is measurable. A 3,000 m2 atrium losing 15 percent of daylight pushes lighting load up roughly 8-11 percent in a retail building. Over a 2,000-hour trading year that is real money on the utility bill. Retail and office sites justify 4-week cycles on energy alone; warehouses with low daylight dependency can stretch it.
The Lingyun Y3 cleans about 180-220 m2 per hour at cruise, so a 3,000 m2 roof is a two-day job including water refills. If your cycle feels expensive, check the m2/h rate first – often the machine is fine and the route plan is bad.
When is cleaning too often?
Every 2 weeks is overkill for most sites and wears the suction cups faster. Cups are the consumable that fails first – 6-12 months in daily use, longer with light schedules. If you are replacing cup sets more than twice a year, reduce frequency before blaming the machine.
Also watch water use. Cleaning a dry, dusty roof too frequently without a filtration loop wastes water and leaves streaks that are worse than the dust. The filtration setup matters as much as the schedule.
A simple rule for 2026
Clean when the camera inspection shows more than 20 percent visible haze, or on a fixed 6-week default. Set the default, then trim it based on your first six months of photos. Sites that measure daylight before and after cleaning tend to land on 5-7 week cycles regardless of climate.
Who should skip fixed schedules entirely? Small roofs under 400 m2 in clean air. There, one or two cleanings a year genuinely is enough, and a robot pays back slower. Renting or a shared service is the better call.
Does the cleaning cycle change the machine you need?
It does more than most buyers expect. A 4-week cycle on a dirty urban roof needs a machine with strong filtration and a bigger tank, because it will spend more time on heavy film. An 8-12 week cycle on a clean rural roof can run with a smaller tank and lighter water use. Pick the cycle first, then the specification – not the other way round.
Frequent light cleaning also keeps suction cups cleaner, which extends their life. Cups last longest when they meet glass, not grit. A denser schedule with gentle passes often costs less per year than a sparse schedule with aggressive scrubbing, because the aggressive passes wear both glass coating and rubber.
Key Takeaways
- Default to a 6-week cycle, then adjust with photo evidence.
- Urban and coastal sites: 4-6 weeks. Rural low-dust: 8-12 weeks.
- Light loss of 12-18 percent is invisible from the ground but hits energy bills.
- Post-storm, clean within 72 hours before mud bonds.
- If cups fail twice a year, you are cleaning too often.
Need a schedule plan for your building? Send the roof size and location and we will map a cycle.
How do you set a schedule without over-cleaning?
Start with photos. Shoot the same three panes from the same spot on day one, then every month. Compare haze against a printed grey card. When the card stops being readable through the glass, that is your trigger. This beats guessing and beats any climate table, because your roof’s dirt mix is unique.
Shift the cycle with the season. Spring pollen in temperate cities adds a burst around weeks 10-14. Harvest dust in agricultural regions spikes in late summer. Sites near construction run dirtier for the whole build period and then drop back. A fixed calendar misses all of it.
What does a badly timed clean cost?
Cleaning too late means bonded film that needs extra passes, more water and sometimes a mild agent, which raises the cost per square metre by 20-40 percent. Cleaning too early wears consumables and lands you in the same week as rain that undoes the pass. Find the sweet spot in the first season and hold it.
Rain is not a free clean. Light rain turns dust into a mud film that is harder to remove than dry dust. Heavy rain on clean glass is fine. So if a storm passes, run the robot within 72 hours rather than waiting for the next scheduled slot.