Most robots should not run on single-glazed skylights older than about 30 years unless the glazing bars and panes are confirmed to carry the load. Untempered annealed glass fails without warning, and a tracked machine concentrates 22-28 kg on two small contact patches. Get a structural check before anything goes up.
Why is single glazing risky for robots?
Three reasons stack up. The glass is thin, often 4-6 mm, with little reserve strength. It may be annealed rather than tempered, so it cracks instead of holding together. And old putty or silicone bedding has hardened, so the pane is not fully supported at the edges where a load concentrates.
Add thermal stress on a sunny day and the margin shrinks further. A machine that is fine on modern laminated low-E glass can be the straw that breaks an 1980s atrium pane.
What load can the roof take?
You need two numbers: the allowable maintenance load in kN/m ², and the point-load limit for the glazing bars. Ask the structural engineer, not the cleaner. As a rough screen, roofs rated below 0.4 kN/m ² for maintenance access are poor candidates for a tracked robot.
- Tempered laminated glass, modern bars: generally robot-compatible.
- Tempered single pane, untested bars: require engineer sign-off.
- Annealed/old single glazing: assume not suitable without certification.
- Leaded, wired or heritage glazing: use non-contact methods only.
What are the alternatives?
For fragile roofs, options are rope access with a light touch (cleaners stand on ropes, not glass), telescopic pole systems from walkways, or a suspended platform. None is as cheap per visit as a robot on a modern roof, but all keep the load off the glass. If the roof has walkable sections, clean from those and leave the glass untouched by tracks.
What happens if a pane cracks?
Cost and safety. Replacing one coated insulated unit in a heritage atrium can run into thousands and take six to twelve weeks. If the pane is overhead and untempered, fragments fall. That is why the rule is simple: if you cannot prove the glass carries the load, do not put a machine on it.
Who can safely run a robot on old roofs?
Buildings that recently re-glazed with laminated glass, that have documented structural capacity, and that can afford a light machine run at low speed. Everyone else should stay with rope access or pole systems until the glazing is upgraded. Upgrading the glazing is often the real project; cleaning is the trigger.
Can you retrofit a fragile roof instead?
Often that is the real answer. Replacing old single glazing with laminated units of known capacity both improves safety and opens the door to machine cleaning later. If the roof is being worked on anyway, ask for the maintenance load rating in writing while it is accessible, and keep it for the cleaning plan.
What is the safe fallback?
Rope access with a strict light-touch method, or a pole system from a walkway, and never a tracked machine. Cleaners work from ropes so their weight never lands on the glass. It costs more per visit, but it removes the crack risk entirely, and on fragile heritage roofs that is the only number that matters.
How do you check capacity without a full survey?
A walkover with the original drawings is a start, but only an engineer can confirm capacity. Look for signs of movement: cracked putty, sagging panes, corroded glazing bars, water staining at the frame. Any of these means assume the roof is fragile until proven otherwise, and plan for rope access or poles in the meantime.
What about partial robots on walkable sections?
A hybrid works on some roofs. Clean the glass from adjacent walkable areas with poles, and use a robot only where there is a firm, rated platform. That keeps the load off the fragile panes while shortening the rope work. It needs a clear route plan so the machine never crosses an unsupported span.
How often should fragile glazing be inspected?
At least once a year, and always before a cleaning campaign. Look at putty, seals, glazing bars, fixings and any pane that moves under light hand pressure. Catching a failing seal before water gets behind it is far cheaper than replacing a unit later. The inspection is quick; the consequence of skipping it is not.
What is the correct decision posture?
Assume fragile until proven capable. The burden of proof is on the roof, not on the machine. A structural note with a load figure turns a guess into a decision, and it is cheap compared with a cracked atrium pane. When the capacity is documented, machine cleaning is safe and fast; without it, the cautious route is the only defensible one.
If the roof is unlisted but old, the same caution applies. Age matters less than glass type and support. A 1970s tempered laminated roof with solid bars may be perfectly fine; a 2005 build with annealed panes and a failing seal may not be. Judge the build-up, not the date, and get it in writing either way.
If the answer is no, say no. Walking away from a fragile roof that cannot be certified is a professional decision, not a failure. The building keeps its glass, you keep your record clean, and the cleaning can wait for a re-glaze that resolves the problem properly.
Key Takeaways
- Single glazing over 30 years old usually fails a robot load check.
- Get kN/m ² capacity and bar point-load limits from an engineer.
- Below 0.4 kN/m ² maintenance rating, do not use a tracked machine.
- Rope access and pole systems keep loads off fragile glass.
- One cracked heritage pane can cost more than years of cleaning.
Send glazing build-up details through the contact page. Load and retrofit notes sit in the news archive.