Most skylight cleaning robots in 2026 should stop above wind speeds of 10 to 12 m/s (about 22-27 mph), and must stop in rain, hail or on any glass showing frost. Light passing cloud is fine; gusty conditions are not.
Weather planning is where roof work goes wrong. Not because the limits are hard to find, but because the pressure to finish before a forecast change pushes crews to keep going just a little longer.
Why is wind more dangerous than rain?
Rain just ruins the result; you clean and it streaks. Wind destabilises the machine. A glass roof on a tall building channels and accelerates airflow, so the 8 m/s reading at ground level can be 13 m/s at roof edge. Above roughly 10 m/s, a light machine starts to be pushed laterally, and that lateral force works directly against the suction grip. The safety margin on a skylight robot is smaller than people assume.
- Under 8 m/s: normal operation.
- 8-12 m/s: run only with tethering and a spotter, reduce speed.
- Above 12 m/s: stop and secure the machine to the roof.
Can you clean in light rain?
No. Rain carries dust and pollutants onto the glass and re-deposits them as it dries, so you are paying for a clean that leaves a dirtier panel. There is also a practical suction issue: a wet, contaminated seal loses grip faster than a dry one. Wait for the rain to pass and the glass to dry. The exception is a post-rain clean done deliberately to remove the residue the rain left behind, which is different work entirely.
What about heat and condensation?
Two limits. Above roughly 40 C surface temperature, rinse water flashes off before it does its job and the machine’s own electronics get stressed. At the other end, condensation after a cold night leaves a film that defeats suction and smears on contact. The safe window on most glass roofs is mid-morning to late afternoon, after dew has cleared and before peak heat.
One more thing worth saying plainly: a wet glass roof with a 15-degree pitch is a slip hazard for the person setting up the machine, not just the machine. Wind, rain and slope compound. If two of the three are marginal, stop.
Who sets these limits on a real site?
They should be written into your roof access plan and your method statement, not left to the operator’s judgement on the day. Track local wind at roof height with an anemometer, not from a phone app for the city centre. Sites that log conditions alongside each clean build a useful record for insurance and for arguing about weather-related delays.
How do you read wind forecasts properly for roof work?
Use a roof-level source, and treat gust figures, not averages, as the number that matters. A forecast saying 9 m/s average with gusts to 15 m/s should stop work, because the gust is what pushes the machine. On tall buildings, wind is also funnelled around corners and can change direction sharply within a few metres of an edge, so the safe side of the roof at 9 am may be the exposed side at noon.
A cheap handheld anemometer kept at the roof access point is more useful than any app. Fifteen seconds of measurement before you start tells you more than an hour of guessing.
What is the recovery plan after a weather stop?
When the wind rises mid-clean, the machine comes off the roof and gets secured, not left in place. A half-tonne gust can move a light unit and a tether is only as good as its anchor. Before you start any clean, decide where the machine will go if the weather turns, and make sure one person can execute that plan alone if needed.
- Stop, don’t pause, when gusts exceed the limit.
- Secure the machine below the roof line if a longer stop is likely.
- Log the stop and the reason; it protects you in delay disputes.
How does weather risk change between regions?
A roof in a windy coastal city loses far more days than one inland. If wind typically rises by mid-afternoon, schedule cleans for the morning. If your site sits in a corridor that funnels wind, budget fewer, more thorough cleans rather than many marginal ones, and build weather contingency into the contract so you are not blamed for days the roof simply could not be worked.
What is the cost of a lost weather day?
More than the clean it displaces. A booked roof visit that gets cancelled loses the crew’s day, the access equipment hire and the parking of the next job in the queue. On a tight programme, one lost day can push a whole sequence back a week. That is why experienced operators build a weather buffer into every multi-day roof job rather than promising exact dates.
If you are contracting the work, ask how weather days are handled. A supplier who pretends weather will not interfere is either new or optimistic. A supplier who quotes a window and a buffer is telling you they have done this before.
Key Takeaways
- Stop above 10-12 m/s wind; roof-edge wind is far higher than ground readings.
- Never clean in rain; it re-deposits dirt and weakens suction grip.
- Safe window is usually mid-morning to late afternoon on most glass roofs.
- Write limits into the method statement, not the operator’s mood.
- Log weather per clean for insurance and delay claims. See our safety guide.