A maintenance window that overruns teaches a plant to stop granting windows. Scoping, pre-staging and a hard fallback rule are what keep the window credible.
Every plant has a story about the eight-hour maintenance window that ran into its third shift. The damage outlasts the incident: production planners remember, and the next window request meets resistance, so maintenance gets deferred, so the next intervention is bigger and less predictable. A window that holds is not a scheduling detail — it is the asset that keeps preventive work possible at all.
Windows overrun for reasons visible in advance. Scope that was never surveyed, so the job discovered on site is larger than the job that was quoted. Parts that arrive with the technician instead of before. Access nobody arranged — the locked-out adjacent line, the missing permit, the scaffold that was not booked. Almost none of the classic overruns begin during the window; they begin in the weeks before it.
Our planning rule is that a window contains no discovery. The machine is surveyed beforehand, the work scoped against that survey, and every part, tool and consumable staged on site before the line stops. If the survey cannot be done without stopping the line, that survey is its own short window — a planned hour now against an unplanned shift later.
Inside the window, sequence for early truth. Front-load the diagnostic and disassembly steps that could reveal surprises, so bad news arrives in hour two, while options remain, rather than hour seven, when the only option is overrun. Define a go/no-go checkpoint partway through: past it, the work completes within the window; at it with open findings, the fallback executes.
The fallback is what makes the schedule credible. For every job we ask in planning: if this cannot finish, how does the machine run tomorrow anyway? Sometimes the answer is refitting the old part, sometimes deferring one sub-task to the next window. A window with a rehearsed fallback can be granted with confidence. A window without one is a gamble the plant is making, whether or not anyone said so.
Then close the loop. Actual times against planned, per task, reviewed after every window — because plants that measure this quote the next window from evidence, and their windows get shorter and more trusted in the same year. The plants that do not measure it keep telling the story about the third shift.
— GANI Lifecycle engineering team