For 10,000 m2 of skylight, plan on two robots if you clean in a single shift, or one robot over four to five nights. A single machine realistically covers 600 to 900 m2 per shift once refills, repositioning and setup are counted.
Vendors quote pass rates. Pass rates are not coverage. The gap between the two is where fleet plans fall apart, and it is a gap of roughly 40 percent on a real roof.
How do you convert pass rate into real coverage?
Start with the pass rate, then subtract everything. A machine running 8 m/min with a 600 mm head theoretically lays down 288 m2/hour. On a factory roof with raised seams, two refill stops, one cable reposition and a 20-minute setup, the honest figure lands near 120 to 150 m2/hour. Run a six-hour productive shift and you have your 600 to 900 m2.
| Roof condition | Theoretical m2/h | Realistic m2/h | Shift coverage |
|---|---|---|---|
| Flat, unbroken panels | 288 | 170-200 | 1,000-1,200 m2 |
| Seamed factory strips | 288 | 120-150 | 700-900 m2 |
| Atrium, many obstructions | 288 | 80-110 | 480-660 m2 |
Why does cleaning frequency change the answer?
If you clean four times a year, one robot handles 10,000 m2 over five nights each cycle. That is fine. If you clean monthly, the same roof now needs eight to twelve nights per year of shutdown access, and shutdown windows are usually two nights long. That is where a second machine pays for itself, not on hourly throughput but on access windows.
The trap: buying two machines to solve an access problem you could solve by cleaning more often in better weather. Glass gets dirtiest in spring pollen and autumn rain. Clean twice in those months and skip the mid-summer pass; you will save more than a second robot costs.
What about water and power as a constraint?
Water is the constraint nobody budgets for. A 10,000 m2 roof at 40 ml/m2 needs about 400 litres per full cycle. A machine with a 25-litre tank and a 60-metre hose run means either a lot of refills or a proper water point on the roof. Two robots do not halve that — they double the demand and you will be filling both at the same tap.
Sizing a fleet without sizing water is the single most common planning error we see. The Lingkong K3 handles larger tanks better for this reason; on very large roofs a K3 plus a Y3 split between tank capacity and manoeuvrability often beats two identical units.
Who should run more than two robots?
Only sites with sustained, year-round glazing and a full-time cleaning crew. Airports and large transport hubs qualify. A single shopping centre almost never does. Above two machines you also need a person whose job is robot maintenance, because pads, brushes and filters will wear roughly three times faster and unscheduled downtime multiplies.
Sizing checklist
- Measure net cleanable glass, not the footprint of the building.
- Count obstructions per 100 m2 — every fin or vent costs you 8-15 seconds.
- Confirm water volume per cycle: roughly 40 ml per m2.
- Check access windows: how many nights per year can you actually get on the roof?
- Budget one spare pad set and brush set per machine per quarter.
If you have drawings, the sizing gets much sharper. Send the plan through our contact page and we will work the numbers against a real pass rate rather than a brochure one. The coverage rate article explains where the pass-rate figures come from.
How do you phase the rollout?
Do not buy the fleet on day one. Run one machine for two full cycles, record real coverage per night, and size the second unit against those numbers. Almost every site that skipped this step either overbought or discovered its access windows were shorter than the drawings suggested. One cycle of data is worth more than any sizing model.
What is the biggest planning mistake?
Assuming the roof is available when you want it. Production, tenant access and weather all take the roof back at short notice. A fleet plan built on perfect access will miss its cycle. Build in two contingency nights per cycle and treat any extra as a bonus rather than the plan.
How do you track whether the machine is earning its keep?
Log three things every cycle: hours run, square metres covered, and water used. After six cycles the trend line tells you more than any brochure. If coverage per hour is drifting down, a consumable is failing. If water use is rising, your filtration has degraded. Owners who log these three numbers spot problems in weeks rather than discovering a damaged roof at year end.
What does a typical cleaning cycle look like?
A full cycle is more than the time the machine spends cleaning. It is the walk to the roof, the water fill, the boundary setup, the run itself, and the pack-down. On a well-mapped roof the setup is 20 minutes and the run dominates. On a difficult roof with restricted access it can be 45 minutes before the machine lays down its first pass. Budget the whole cycle, not the pass, when you plan a shift.
How does glass type change the brush and pad choice?
Coated glass, laminated units and textured or fritted panels all behave differently under a rotating brush. Fritted glass wears pads faster and needs lower brush pressure. Coated low-E glass scratches easily, so a clean pad and a fine brush are not optional. Textured or prismatic glazing traps dirt in the pattern and often needs more passes, which quietly doubles your clean time on those sections.
Key Takeaways
- One robot covers 600-900 m2 per shift on a real roof, not 1,700 m2.
- 10,000 m2 suits one robot over five nights or two robots in one.
- Water demand is about 400 litres per full 10,000 m2 cycle.
- Access windows, not throughput, usually decide whether a second machine pays.
- Above two machines you need a dedicated maintenance role.