Run spot passes weekly on visible hot spots and a full wash cycle every 8 to 12 weeks for most commercial glass roofs. Skipping the full cycle turns the roof grimy; running it weekly wears brushes, wastes water and doubles your cost with no visible gain.
What counts as a spot pass?
A spot pass is a targeted run over the areas the public actually sees: the entrance atrium band, the strip above reception, and any panel facing a street or a camera. On a 2,000 m² glass roof that is often under 300 m². You run the robot at normal speed and full flow, but only on that band. Ten minutes of setup, twenty minutes of work. Spot passes handle bird droppings, leaf litter and the pollen film that builds on the south-facing panels first. They do not remove the gradual mineral haze that covers everything after a dry month.
| Roof type | Spot passes | Full cycle | Annual water |
|---|---|---|---|
| Urban office atrium | Weekly | Every 10 weeks | Under 18,000 L |
| Desert / near cement plant | Twice weekly | Every 4 weeks | Under 25,000 L |
| Coastal salt air | Weekly | Every 12 weeks | Under 16,000 L |
| Under deciduous trees | Seasonal heavy | Every 8 weeks | Under 20,000 L |
Why does the full cycle still matter?
Full cycles exist to reset the whole surface, not to chase visible dirt. Rain carries dissolved minerals, jet fuel residue in city centres, and fine construction dust that lands evenly and dries into a uniform film. You will not notice it panel by panel, but it dulls the daylight transmission of the whole roof by a measurable amount. The full wash removes that layer and lets you inspect every panel, frame and seal. It also flushes the brush and the water path, which spot passes never do. Eight to twelve weeks is the sweet spot for most climates.
How does interval choice affect wear and water?
This is where operators waste money. A full cycle on a 2,000 m² roof uses roughly 900 to 1,400 litres depending on flow setting. Do that weekly and you burn around 60,000 litres a year, plus brush replacement every few months. Move to a spot-plus-quarterly mix and the annual water figure drops under 15,000 litres and a brush set lasts 18 months. The savings are not in the machine, they are in the calendar. I have seen sites cut cleaning cost per m² by a third just by stopping the weekly full wash nobody needed.
Which roofs need a different rhythm?
Deserts and industrial estates near cement plants need shorter full cycles, sometimes every four weeks, because dust is relentless. Coastal sites with salt air need more frequent rinsing to stop salt etching but can stretch the full cycle since rain keeps the surface wet. Roofs under trees get a heavy autumn spot programme and a light rest of year. One rule holds everywhere: if you can see a difference between panels from the inside, you waited too long, whatever the schedule says.
How do you decide the interval with no fixed rule?
Start with a baseline interval, run three cycles, and look at the glass before the third. If the roof still looks clean at ten weeks, stretch to twelve. If it looks dull at eight, pull back to six. The interval is a measurement, not a policy, and it will change with season and weather. A dry windy month loads the glass faster than a wet one. Log what you see and let the schedule follow the evidence. Rigid annual calendars are how buildings end up either over-cleaning or under-cleaning and paying for both mistakes.
Watch the north and south faces separately. The sunny face films over faster and may need its own shorter interval, while the shaded face can wait. One schedule for the whole roof usually means one half is wrong.
Does weather change a cycle’s success?
It does more than most operators admit. Cleaning in strong sun dries the water before the brush lifts the film, leaving streaks that look worse than the dirt. Cleaning just after rain is the opposite problem: the surface is already wet, so you spend water chasing runoff and the brush does less work. The sweet spot is a dry morning with the glass still cool, before the sun climbs onto the panes. Wind above about 40 km per hour should stop the run, because suction margins fall and debris blows back onto freshly cleaned glass.
Track the conditions against the result for a few months and you will find the window that suits your building. It is usually a two-hour slot, and hitting it reliably cuts your rework almost to zero.
Related reading: 2026 cost and payback breakdown, and the square metres per hour and battery figures.
Key Takeaways
- Weekly spot passes plus a full wash every 8 to 12 weeks suits most commercial glass roofs.
- A full cycle on 2,000 m² uses 900-1,400 litres; running it weekly wastes 60,000 litres a year.
- Mixing spot and quarterly schedules can cut cleaning cost per m² by about a third and extend brush life to 18 months.
- Desert and industrial sites need 4-week full cycles; coastal sites can stretch to 12 weeks.
What about water recycling during long cycles?
Recycling sounds efficient and often is, but only with filtration. A robot that reuses water without a fine filter sends concentrated minerals back onto the glass, and over a long cycle the last panels get the dirtiest rinse. If your site recycles, check the filter and replace it on run hours. If you cannot maintain filtration, use fresh water and accept the higher draw. Clean glass from fresh water beats rewashed glass every time, and the volume difference is smaller than most people assume.
Track litres per cycle for a month. Once you know the real number you can judge whether recycling pays on your roof, instead of adopting it because it sounds greener.