A skylight cleaning robot cleans 80 to 120 m² per hour on flat or gently sloped glass, and 50 to 70 m² per hour on steeper curtain walls. A typical battery gives 3 to 5 hours of runtime, and a 12 L tank covers 100 to 140 m². Real-world output is usually 60 to 75 percent of the rated figure once setup and refills are included.
Vendors quote the best number they ever saw on a clean, flat, indoor roof. Your roof is dusty, has obstacles, and sits outdoors. Here is how to build a figure you can actually schedule against.
What actually slows a skylight robot down?
Five things, in order of impact: setup time, cup cleaning, obstacles, refills and slope. Setup is the biggest — 30 to 45 minutes before the first square metre is touched. On a 1,000 m² roof that is already 8 percent of a full day gone.
Obstacles are the second tax. Every vent, pipe or mullion forces a reverse and re-approach, and each one costs 20 to 40 seconds. A roof with 40 obstacles spaced across the field can lose an hour across a full cycle.
| Surface | Rated (m²/h) | Realistic (m²/h) |
|---|---|---|
| Flat indoor atrium | 120 | 85-95 |
| Sloped 15° outdoor | 100 | 70-80 |
| Curtain wall vertical | 70 | 50-60 |
| First clean, heavy dust | — | 40-55 |
How far does one battery charge go?
A 48 V, 20 Ah pack is about 960 Wh. At a draw of 200 to 300 W including pumps, that is 3 to 5 hours of run time. The Lingyun Y3 and Lingfeng S1 sit in that band. In cold weather, expect 25 to 35 percent less, because lithium cells lose capacity below 5 °C.
Charging takes 2 to 4 hours, so most teams run two batteries and swap. A second pack at $400 to $700 is the cheapest productivity upgrade you can buy — far better value than a faster machine.
How much water do you need per shift?
Assume 10 litres per 100 m² on a maintenance clean and 20 litres on a first clean. A 1,200 m² roof therefore needs 120 to 240 litres per visit. A 12 L on-board tank means 10 to 20 refills. That refill cycle, not the motor, is often what limits a shift.
Sites with a nearby tap and a hose reel finish far faster than sites where someone carries jerry cans up a ladder. If you can plumb a permanent water point to the roof, do it before you order the robot.
Who should expect lower numbers than the brochure?
Anyone with steep glass, heavy dust, cold weather or awkward access. Coastal sites with salt film are slow because the first pass barely cuts through. Post-construction sites need two passes and half the throughput. If your roof matches any of those, plan using the bottom of the range and you will rarely be disappointed.
Measure your own output on the first three cycles and treat that as truth. Our notes on cycle time and battery versus tether power go deeper on both. Ready to size a machine? Start at the contact page.
How do you schedule a job from a throughput number?
Take the number of square metres, divide by your realistic rate, then add the fixed costs. A 1,500 m² roof at 80 m²/h is 19 hours of cleaning, plus 45 minutes of setup, plus refills. Spread over two days at 10 working hours each, that is a comfortable schedule with margin. Trying to force it into one day is how crews cut corners and leave streaks.
Always add 15 percent contingency for a first visit to any roof, and keep the buffer even on repeat cycles. Wind stops, a stuck sensor and a low battery will eat an hour before you notice.
Does machine weight affect throughput?
Less than most people think. A 20 kg machine and a 28 kg machine on the same roof run at similar speeds. Weight matters for point load and for how far the machine can push water, not for raw speed. Do not pay a premium for a light machine if your roof is already rated for the load.
What does affect throughput is maintenance state. Dull squeegees, clogged filters and hard cups all slow the machine in ways the spec sheet never shows.
Why do two identical roofs run at different speeds?
Microclimate is the usual reason. A roof that faces the prevailing wind collects dust on one side faster and stays wetter longer, so the machine slows where grip is marginal. Altitude and reflected heat from a nearby wall also change how quickly water flashes off.
This is why comparing your numbers with a neighbouring building rarely helps. Log your own cycle times for three months and you will have a baseline better than any brochure figure. Then improvements become measurable.
Key Takeaways
- Real output is 60-75 percent of rated: 80-120 m²/h flat, 50-70 m²/h steep.
- Setup eats 30-45 minutes; batch bays to spread that cost.
- One battery gives 3-5 hours; buy a second pack before a faster machine.
- Budget 10 L of water per 100 m², double for a first clean.
- Steep, dusty, cold or coastal sites should plan on the bottom of the range.