Commissioning a skylight cleaning robot takes 1-2 days and must include a wet suction test on the actual glass, a tether or rail layout drawing, and a signed handover sheet listing brush pressure, water dosing and battery health. Skip the wet test and you will discover suction failure in month three.
Most buyers spend weeks comparing specs and then rush the handover. That order is backwards. The robot’s real limits show up on your roof, on your glass, with your water. A Lingkong K3 that grips a clean sample panel in a showroom can still slip on a 22-degree atrium pane carrying three weeks of diesel film.
What should be tested during commissioning?
Four things matter more than everything else. First, wet suction on the dirtiest pane you own, not a wiped sample. Second, edge behaviour: watch the robot cross a 40 mm mullion and a raised gasket. Third, water recovery, because streak complaints usually trace back to a full or misaligned recovery tank, not the brush. Fourth, the emergency stop and tether anchor load test.
Ask the supplier to run the Lingyun Y3 through a complete cycle, not a 3-metre demo. A full cycle on a 600 m2 roof takes 2-4 hours and exposes heat drift in the suction motor. Sample runs hide it.
How long does handover actually take?
Budget one full day for a single robot and two days for a fleet of three or more. Of that, roughly 3 hours go to documentation: anchor point drawings, IP rating certificates (IP54 minimum for outdoor skylight decks), and the CE file. The rest is operator practice. Train two operators, not one. Fleets stall when the single trained person takes leave.
Which numbers belong on the handover sheet?
| Parameter | Typical value to record | Why it matters |
|---|---|---|
| Suction holding force | record measured value, compare to 1.4x spec | safety margin on slope |
| Brush pressure | set to supplier figure, log it | protects low-E coatings |
| Water dosing | ml per m2, e.g. 30-45 | streaks and tank runtime |
| Battery health | state of health % at delivery | warranty baseline |
| Cycle time | measured m2 per hour | feeds your cleaning schedule |
Write the measured numbers down. When a dispute appears at month eight, the commissioning sheet is the only baseline you have.
Who should not commission a robot this way?
If your roof has fewer than 200 m2 of glass, a full commissioning program is overkill. A Lingfeng S1 on a simple flat skylight with a single operator and a one-page checklist is enough. The heavy process earns its cost on buildings above 800 m2, where downtime on one robot breaks a crew’s whole week.
One risk to name plainly: some integrators treat commissioning as a sales demo and leave without a written sheet. Insist on the document before you sign acceptance. For scheduling follow-up service, our team handles it through the contact page.
Does the robot need a dedicated roof anchor?
Every tether-based skylight robot needs at least one certified anchor point rated well above the robot’s weight plus dynamic load. On a flat skylight deck a single 500 kg-rated eye bolt is usually enough; sloped glass often needs two so the tether angle stays shallow. The trap is reusing an old window-washing anchor without a load certificate. Those anchors were rated for a cradle, not a tugging robot, and re-rating them is not free.
If the building has no anchor at all, budget the engineering. Drilling into a waterproofed roof and re-sealing costs USD 300-900 per point in most markets. Skip the seal detail and you trade a clean roof for a leak above a meeting room.
How do you handle glass that is not uniform?
Mixed glazing breaks clean assumptions. A lobby might combine low-E coated panes, laminated glass and a few polycarbonate sections. Each behaves differently under suction and brush pressure. Map the roof into zones before commissioning: coated glass gets low brush pressure and slower speed, laminated glass tolerates more, and polycarbonate is best cleaned by hand because it scratches. Write the zoning on the handover sheet so operators do not re-learn it every shift.
One measured fact worth noting: low-E coatings can be damaged by abrasive brushes at pressures above roughly 2-3 N per cm of brush contact. Most robot defaults sit under that, but heavy dirt tempts operators to press harder. Training on the coating limit matters more than training on the buttons.
What does a bad commissioning look like in hindsight?
The pattern is consistent. A demo runs on a cleaned pane in mild weather. The robot grips, the squeegee leaves no trail, everyone signs. Three months later the operator reports slipping on the north slope after rain. The suction cups were never tested wet; the margin was assumed from the showroom. By the time the fault surfaces, the supplier considers commissioning closed.
The correction is cheap and boring: test in the worst realistic condition on day one. Wet glass, dirt present, the steepest section the robot will ever meet. If it holds a 1.4x margin there, you have a baseline worth defending. If it does not, better to find that out during handover than during a live cleaning window.
How do you keep commissioning valid over five years?
Re-test annually. Suction cups age, anchors loosen, tether hardware wears. An annual re-commission is a two-hour job for one operator and a checklist, and it catches drift before it becomes an incident. Keep the original handover sheet so you can compare. A number that has fallen 15% year on year is a buying decision, not a surprise.
Add firmware and brush-setting changes to the re-test trigger. A software update that alters speed or dosing can quietly move the robot outside the conditions you commissioned it for.
Key Takeaways
- Always run a wet suction test on the dirtiest real pane before acceptance.
- Log suction force, brush pressure, water dosing and battery health on a signed sheet.
- Train at least two operators; single-operator fleets stall on leave days.
- Below 200 m2 of glass, a lightweight checklist beats a two-day program.
- Demand the anchor drawing and CE/IP paperwork as part of handover.