A skylight cleaning robot manages roughly 600 to 900 m2 per hour on open, uniform glass in 2026. On real roofs with frames, ridges and access breaks, expect 300 to 500 m2 per hour. Battery and tank size set the ceiling, not the brush.
Manufacturers quote the clean-lab number. Buyers live with the real one. Understanding why the gap exists is how you plan labour and schedules so you are not surprised on the first shift.
What sets the real throughput?
Four things, in order of impact. Panel layout, battery runtime, water capacity, and how often the operator has to stop to move the machine or refill. On a wide open skylight with no frames, the robot barely pauses. Add framing every 1.5 metres and the brush is constantly hitting a joint, slowing the pass.
| Roof type | Realistic m2/h | Limiting factor |
|---|---|---|
| Open laminated glass | 600-900 | Battery |
| Framed panels | 400-600 | Joint transitions |
| Sloped 20-35 deg | 300-500 | Grip and climb speed |
| Mixed solar-glass | 250-450 | Obstacle avoidance |
How long does the battery actually last?
Most commercial units run 2 to 4 hours per charge under load. Climbing a slope drains faster than crossing flat glass, because the motors fight gravity. A 35 degree roof can cut runtime by 20 to 30 percent versus flat work.
If a full pass of your roof takes longer than one charge, you need a spare battery or a second robot. Changing a battery mid-cycle costs 10 to 15 minutes, and on a big roof that adds up to a real dent in daily output.
Does tank size limit m2 more than battery?
Sometimes, yes. A 10 litre tank covers roughly 400 to 600 m2 of glass depending on how dirty it is. Dry, dusty glass drinks water. If your roof is 1,200 m2, that is two or three refills per pass.
The honest fix is not a bigger tank, it is a water supply point closer to the roof. Hauling 20 litre jerry cans up a ladder three times a shift is the hidden labour cost nobody quotes. Plan the water logistics and the robot’s rated speed becomes reachable.
Can you raise throughput without buying a bigger robot?
Yes. Clean more often and the dirt load per pass drops, so the robot moves faster and uses less water. A monthly cycle runs noticeably quicker than a twice-yearly heroic clean. Run a first pass and a rinse pass rather than three heavy scrubs, and schedule the work at night when the roof is cooler and water evaporates less.
Where do quoted figures mislead?
Watch for m2/h figures measured with no stops, over a 30 metre straight run, on clean glass. Extrapolating that to a 2,000 m2 site is how owners end up promising a client a one-day clean that takes three. Always ask for the figure with stops included, or ask the vendor to state battery, tank and refill assumptions alongside the number.
How do you turn m2/h into a schedule?
Multiply realistic m2/h by usable hours, then subtract refills and repositioning. A 1,000 m2 roof at 500 m2/h with a one hour net loss to refills and setup lands at roughly three hours per pass. That is the figure to give a client, not the marketing number.
Does cleaning frequency change throughput?
Yes, and it is the most useful lever you have. Monthly light cleaning runs faster and uses less water than a quarterly heavy clean of the same roof. If a client complains about cost, the answer is often more frequent, lighter passes, not fewer heavy ones.
What limits output on large roofs?
- Battery runtime between swaps
- Tank capacity and refill logistics
- Panel joints that slow the brush
- Access and repositioning time between roof sections
How do you measure and improve throughput over time?
Log each pass. Record the area covered, the time taken, the water used and any stops. After three or four cleans a pattern appears, and it usually shows one dominant loss, whether that is refills, battery swaps or a difficult roof section. Fixing that one item typically lifts output more than any specification change.
Operators get faster too. The same machine on the same roof is measurably quicker by the fourth clean, purely from familiarity with the route and the refill points. Do not judge output from the first visit.
What is the single biggest throughput killer?
Refills and swaps, almost every time. A robot that stops three times to refill and twice for a battery loses more than an hour a shift. Moving the water point closer and carrying a charged spare battery is cheaper than buying a faster machine, and it closes most of the gap between the rated speed and the real one.
How do tank and battery interact in planning?
They rarely run out at the same time, and that is the trick. A machine might empty its tank before the battery drops, or the reverse, depending on dirt load. Log which one limits each roof. Then you can decide whether a second battery or a second tank is the better investment, instead of guessing.
Key Takeaways
- Expect 600-900 m2/h on open glass, 300-500 m2/h on real framed or sloped roofs.
- Battery runtime is 2-4 hours; sloped work cuts it by 20-30 percent.
- A 10 litre tank covers roughly 400-600 m2, so large roofs need refills.
- Frequent light cleaning is faster per pass than occasional heavy cleaning.
- Quoted speeds usually exclude stops, refills and repositioning.
Want a throughput estimate for your roof? Send us the dimensions.