Key takeaways
- Planned downtime is scheduled (maintenance, changeovers, no demand). Unplanned is the surprise (breakdowns, micro-stops).
- Unplanned is the expensive kind: no warning, secondary damage, rush costs.
- In OEE, changeovers and breakdowns inside planned production time are availability losses; no-demand time is excluded.
- Both are reducible: SMED for changeovers, predictive maintenance and fast diagnosis for breakdowns.
Not all downtime is equal. Telling planned from unplanned is the first step to attacking it, because each has a different cause and a different fix, and they show up differently in your numbers.
What counts as each
| Planned downtime | Unplanned downtime |
|---|---|
| Preventive maintenance | Breakdowns and equipment failures |
| Changeovers and setup | Micro-stops and minor stops |
| Scheduled cleaning / CIP | Material or supply starvation |
| No demand / no orders | Quality holds and unplanned adjustments |
How they hit OEE
OEE is measured against planned production time. So time when you are not scheduled to produce (no demand, planned shutdown) sits outside OEE entirely. But downtime that happens during planned production time, whether a breakdown or a changeover, is an availability loss and pulls OEE down.
The nuance that trips people up: a changeover is "planned", but it still costs availability. So reducing changeover time (with SMED) directly improves OEE, even though the changeover was scheduled.
Turn downtime hours into the real annual figure: lost throughput, idle labour and restart.
Why unplanned downtime hurts more
Planned downtime is budgeted: you have parts, people and a window. Unplanned downtime arrives without any of that, which is why it costs far more per hour: scrambled labour, secondary damage, restart scrap, expediting, and missed shipments. The goal is not zero downtime, it is converting unplanned into planned, and shrinking both.
The levers to cut each
- Unplanned breakdowns: move from reactive to predictive maintenance, and cut diagnosis time so repairs start faster (lower MTTR).
- Micro-stops: capture and attribute them, because they are often the biggest unplanned loss nobody logs. See the hidden factory.
- Changeovers (planned): SMED and standard work to shrink the window.
- Everything: trend it. If you do not measure stops accurately, you cannot tell planned from unplanned, let alone reduce them.
What a neutral study found
In a survey of US discrete manufacturers for 2016, the US National Institute of Standards and Technology found that unplanned downtime took 7.8% of planned production time, and that 31.7% of that downtime was due to reactive maintenance. The plants relying most on reactive maintenance lost 13.0% of planned time to downtime, against 4.0% for the plants relying on it least (source: NIST, AMS 100-34, 2020). Planned work is not free, but it is the downtime you choose. More sourced figures are on the manufacturing downtime statistics page.
How to classify a stop
Agree the rules before you log, or the split will mean different things on each shift:
- A stop is planned if it was on the schedule before the shift started: planned maintenance, changeovers in the plan, cleaning, trials.
- A stop is unplanned if it was not: breakdowns, jams, waiting for material, quality holds.
- A planned stop that runs over counts as planned up to its planned length and unplanned after that.
- Log the reason as well as the class, so the downtime log shows where to act.
The measurement problem
Most plants undercount unplanned downtime because the short, frequent stops never get logged, so they look better on paper than reality. Fixing that starts with capturing every stop automatically. The partner we recommend, Fabrico, reads stops straight from the PLC, shows the stop on video with a suggested cause, where cameras are fitted, and lets a technician open the work order from the stop, so unplanned losses become visible and fixable. It is EU-built, with data hosted in the AWS EU region and holds ISO 27001 / 20000-1 / 9001 (supports audit-readiness). Fabrico is a partner we recommend; the tools here are free regardless.
Planned stops that are long by design
Some planned stops are long by design: cleaning, washing and sterilising. The guides for the CIP system, HTST pasteuriser, bottle washer, crate washer and freeze dryer cover the faults that make these planned stops run over. Industry context: the food and beverage and pharmaceutical OEE and downtime pages.
Is a changeover planned or unplanned downtime?
Planned, because it is scheduled. But it still counts as an availability loss in OEE during planned production time, so reducing it improves OEE.
Should no-demand time count as downtime?
It is planned non-production and sits outside OEE (which uses planned production time). It does show up in TEEP, which measures against all calendar time. See OEE vs TEEP.
What is a realistic split?
It varies by plant, but unplanned downtime is usually the bigger and more variable cost. Measure your own split first, then attack the largest reducible category.