Key takeaways
In most plants, the production schedule and the maintenance schedule are built independently and only meet when they collide. Production publishes a shift plan based on orders, capacity and changeovers. Maintenance has its preventive tasks coming due based on hours, cycles and failure history.
The two plans are made in different systems, by different people, on different cadences. When they meet, usually three days before they collide, someone has to lose.
What usually happens: production wins. The PM gets pushed. Three weeks later the same PM is overdue. Six months later the asset class shows up as a recurring failure because the PMs that would have caught early degradation never ran. The piece on the preventive maintenance schedule covers the compounding failure cost.
The structural problem is not that production is unreasonable. It is that production has visibility into the order book months in advance, while maintenance has visibility into PM due dates only a few weeks out. The two planning horizons do not align. The fix is to make them align.
Protected windows are fixed slots, planned at least 90 days out, that production cannot reschedule without a formal exception. They cover the big PM tasks, annual overhauls, multi-hour mechanical work, anything that needs a clean break. There should be roughly one to two per quarter per critical line.
Protected windows are negotiated jointly between production and maintenance when the quarter is planned, and they are published in the production schedule. Once published, they are immovable except by a defined exception process (plant manager approval, documented reason, replacement window scheduled within 30 days).
The point of protection is that the maintenance team can actually plan around these windows, order parts, schedule technicians, contract specialists, without the work disappearing in week 11 because of a sudden order. The article on manufacturing KPIs covers the PM-completion metric this defends.
Flexible windows are 1-4 hour slots that have a target date but can move within a 5-day band. They cover medium PM tasks and condition-based interventions: a bearing inspection that needs to happen this week but could be Monday or Thursday.
Production owns the move within the band. Maintenance owns showing up when production picks the slot. This is the compromise that resolves most week-to-week conflict: production gets the flexibility to optimise around orders, maintenance gets a guarantee that the work happens inside the band.
The implementation rule: by Thursday of each week, the next week's flexible windows are locked in specific time slots. Before Thursday, production can shuffle; after Thursday, the slots are fixed.
Opportunistic windows are short tasks (under 30 minutes) that maintenance executes whenever production hits a natural gap, between batches, during a changeover, after an unplanned stop is resolved. There is no scheduled time; there is a backlog of opportunistic tasks ready to run, and the line supervisor calls them when the moment opens up.
For this to work, the opportunistic backlog needs to be visible at the asset, prioritised, and pre-staged with the parts needed. Most plants underuse this tier because the backlog is not organised; the work exists but the team cannot grab it in five minutes. The work order management system needs to surface this list at the line, not bury it in a queue.
Protected windows are immovable in normal weeks. They are not immovable in absolute terms. Sometimes a critical order genuinely cannot wait. The compromise lives or dies on the exception process.
A working exception process has four properties:
The structure works because the cost of breaking a protected window is visible and slightly inconvenient. Without that friction, every protected window becomes a flexible one.
The three tiers need to be visible to everyone at the right time.
The connection to root cause analysis matters here: when an asset shows up repeatedly in opportunistic tasks, it usually means the protected window is too far apart and the PM cadence needs tightening.
A typical mid-market plant with 50 critical asset classes ends up with:
The before-and-after picture: pre-compromise, PM completion is often stuck in the 60-70% range with high week-to-week variance. Post-compromise, it typically stabilises into the high-80s to around 90% (the SMRP world-class level) with much lower variance. Production output usually does not drop, because the opportunistic tier captures slack that was previously unused.
The three-tier calendar can be run in any plant with a CMMS and a production schedule.
Where it works best is when the OEE event stream feeds the opportunistic tier directly, a stop on Line 3 surfaces the asset's pending opportunistic tasks, the line supervisor sees them on a tablet, the technician grabs one before the line restarts.
Fabrico is built so the maintenance windows, the OEE events and the work orders share the same calendar. To see how the three-tier compromise looks against your scheduling reality, book a demo .
The protected and flexible tiers can be in place in a month. The opportunistic tier takes a quarter to mature because the backlog needs to be organised and the on-floor surfacing needs to be working. Plants that try to ship all three tiers on day one usually drop the opportunistic tier within a month.
That is a plant-manager conversation, not a maintenance one. The cost case is straightforward: PM completion below 70% drives recurring failures, which cost more production hours than the windows would have. Show the failure-cost number alongside the window proposal.
Shutdown weeks are an extreme case of protected windows. The three-tier calendar reduces the volume of work that needs to happen during the shutdown, because more of it has been caught in the regular windows. Plants that run the compromise well find their shutdown weeks shrink noticeably.
Emergency reactive work is a separate stream and gets handled in real time. The calendar does not apply to emergencies. What the calendar does is reduce the rate of emergencies by keeping PM compliance high, which is where most of the avoidable emergencies come from.
Making the protected tier negotiable in practice. If "protected" really means "we will move it when production asks," the whole system collapses to a single tier of negotiated slots, which is what plants have today. The plant manager has to enforce the exception process or the framework does not stick.
Login
Blog
