This article is for maintenance managers, plant managers, production managers and continuous improvement leads who already plan their maintenance work properly but still struggle with one question: when should the work actually happen? If your preventive tasks keep getting pushed back because “the line cannot stop this week”, or if your planned windows land right in the middle of a busy run, the issue is often not the plan. It is the picture of machine time the plan is built on.
- Most maintenance windows are chosen from the schedule, not from real machine time.
- Automatic stop data reveals real idle periods, recurring micro-stops and changeovers.
- Four weeks of stop data is usually enough to map weekly patterns on a line.
- Ranking the backlog by lost production time complements criticality analysis.
Why maintenance windows are often picked from the calendar, not the shop floor
A solid starting point is Maintenance Planning: 7 Best Practices That Cut Downtime, written by the FastMaint Team at SMGlobal. The article separates planning, which answers the “what” and “how”, from scheduling, which answers the “when” and “who”, and walks through asset records, criticality analysis, job templates and planner roles. On prioritization, the FastMaint Team puts it simply: “Criticality-based prioritization ensures your planning effort and resources go where the impact is greatest.”
That article gives you the method. This one focuses on a single input to that method: knowing, with real data, when your machines are actually stopped, why, and for how long. In most plants, maintenance windows are chosen from the production schedule, the shift pattern or a fixed weekly slot. The schedule says the press runs Monday to Friday on two shifts, so maintenance gets Saturday morning. The problem is that the schedule describes intended machine time, not real machine time.
On the shop floor, lines wait for material, for an operator, for a quality release or for the next order. Changeovers take longer than planned. Short stops of a few seconds to a few minutes add up without anyone writing them down. Each of these is a hidden window, or a hidden warning, that a calendar-based approach never sees.
What automatic stop data reveals that the production plan does not
When stops are captured automatically at the machine, rather than typed into a log at the end of the shift, three patterns usually become visible within a few weeks.
- Real idle periods. Recurring slots where the machine is stopped but not because of a fault: waiting for upstream parts, end of a production campaign, a lunch break that is not covered, a night shift that ramps up slowly. These are natural candidates for short maintenance tasks.
- Recurring micro-stops. Short, repeated stops that rarely trigger a work order but often signal wear, misalignment or a sensor drifting out of tolerance. Their frequency and timing tell you which asset is degrading before it breaks.
- Changeovers. Planned stops that already take the machine out of production. If a changeover regularly lasts 40 minutes, for example, some inspection or lubrication tasks can be done in parallel instead of requiring a separate stop.
None of this replaces criticality analysis or a good job plan. It gives the scheduler a factual map of machine time, so the “when” is based on what the line actually does rather than what it is supposed to do.
A step-by-step method to choose maintenance windows from stop data
Here is a practical sequence you can run on one line before rolling it out more widely.
- Collect at least four weeks of automatic stop data. You need start time, duration and a reason category for every stop. Four weeks is usually enough to see weekly patterns and separate one-off events from recurring ones.
- Map stops by hour and by day. Build a simple heat map: days of the week in columns, hours in rows, total stop minutes in each cell. Recurring low-activity slots stand out immediately.
- Separate the stop types. Split the map into unplanned breakdowns, micro-stops, changeovers and organizational waits (material, staffing, quality). Only the last two are good candidates for planned maintenance windows.
- Match task duration to window length. Use your job plans to sort tasks by duration. A recurring 20 minute wait suits a visual inspection or a filter change; a full overhaul still needs a dedicated stop, but you now know which shift or day costs the least output.
- Book the window and measure what happens. After each planned window, check that the following runs show fewer breakdowns and fewer micro-stops on the treated asset. If they do not, the task, the frequency or the window needs adjusting.
The aim is not to squeeze every task into a gap. It is to stop guessing, and to give production and maintenance a shared, factual basis for negotiating machine time.
Prioritizing the maintenance backlog with real stop patterns
The same data helps decide what goes first. The FastMaint Team’s maintenance planning best practices recommend refining estimates from history: “Use your CMMS work order history to compare planned versus actual times, parts consumption, and labor requirements.” Automatic stop data adds the production side of that history: how much output each open issue is costing, and how often it strikes.
The table below shows one way to combine the two views. The example values are illustrative, not measured results.
| Signal in stop data | What it suggests | Backlog decision |
|---|---|---|
| Micro-stops on one station increasing week after week | Progressive wear or drift | Move the related inspection up the backlog and schedule it in the next real idle slot |
| Breakdowns clustered after long runs | Task frequency too low for actual operating hours | Switch from calendar based to usage based triggering for that task |
| Changeovers that regularly overrun | Mechanical adjustment issues or missing standards | Combine adjustment work with the changeover and review the procedure |
| Recurring organizational waits at the same hour | Predictable idle time | Use the slot for short preventive tasks with no output loss |
| Rare but long stops on a critical asset | High impact failure mode | Keep at the top of the backlog regardless of frequency |
For example, a packaging line where one labeler shows a steady rise in short stops may not look urgent in the work order system, because nobody raised a ticket. In the stop data, that labeler can be the largest single source of lost minutes on the line. Ranking the backlog by lost production time, then cross-checking with criticality, keeps the team working on what matters for output and for risk.
If you want to go further on the maintenance side, our article on maintenance downtime tracking explains how MTBF and MTTR are built from stoppage data, and our guide on how to reduce unplanned downtime in manufacturing covers the wider improvement loop.
Where TEEPTRAK fits in this approach
TEEPTRAK captures machine stops automatically and turns them into OEE, downtime and micro-stop analysis in real time. The hardware is a module wired into the electrical cabinet plus an edge-mounted sensor, installed in about an hour per machine, without any connection to the PLC, and OEE is available within 48 hours. Operators qualify stop reasons on a tablet, so each stop carries a cause as well as a duration. TEEPTRAK is used in more than 700 plants across more than 30 countries, where customers see on average a gain of 10 to 15 OEE points and 30% fewer unplanned stops. It does not replace your CMMS or your planning process: it gives them a reliable picture of real machine time.
Further reading
- Maintenance Planning: 7 Best Practices That Cut Downtime (the FastMaint Team, SMGlobal)
- Maintenance Downtime Tracking: Using Equipment Stoppage Data to Optimise Your Maintenance Strategy
- How to Reduce Unplanned Downtime in Manufacturing: A Data-Driven Approach
FAQ
What is a maintenance window in manufacturing? It is a period during which a machine or line is deliberately made available for maintenance work. It can be a dedicated stop, a weekend slot, or time that is already idle, such as a changeover or a recurring wait for material.
How much data do I need before choosing windows from stop patterns? Around four weeks of automatic stop records is usually enough to see weekly patterns on a line. Plants with seasonal or campaign based production should check the data again when the product mix changes.
Can micro-stops really help prioritize the maintenance backlog? Yes. A rising number of short stops on the same station is often an early sign of wear or drift. Tracking their frequency and lost minutes lets you schedule the right inspection before the issue becomes a breakdown.
0 Comments