This article is for plant managers, production managers, maintenance managers and continuous improvement leads who notice that Monday morning, or the restart after a planned shutdown, never looks like any other shift. It breaks down restart losses (warm-up, adjustments, ramp-up, start-up scrap), explains how to measure them separately and how to build a restart standard.
- Warm-up, adjustments, ramp-up and scrap get scattered across OEE.
- Monday mornings and restarts after planned shutdowns cost more than a normal shift.
- A defined restart window lets you measure these losses separately.
- A restart standard mainly reduces variability from one restart to the next.
A loss nobody measures separately
In most plants, the restart does not exist as a category. Warm-up is logged as downtime, or not at all. Adjustments get mixed up with breakdowns. Ramp-up shows up as diffuse slow running. Start-up parts go to scrap without anyone linking them to the restart. The result: the loss is there in Monday’s OEE, but it is scattered across three components and never becomes a topic.
That is a problem, because restart losses are among the easiest to standardise. They repeat at every restart, under the same conditions, and depend largely on preparation. You just have to see them.
Where restart losses come from
A restart combines several losses of different kinds, each hitting one OEE component.
| Source of loss | OEE component affected | Sign on the shop floor |
|---|---|---|
| Warm-up, bringing to temperature or pressure | Availability | Machine powered but not yet producing |
| Restart adjustments and settings | Availability | Repeated short stops in the first hour |
| Ramp-up | Performance | Speed deliberately reduced or unstable |
| Start-up scrap | Quality | First parts scrapped or reworked |
| Waiting at start-up (material, people, sign-off) | Availability | Line ready, but nothing arrives |
Taken one by one, these losses look normal: you have to warm up, you have to adjust. It is their sum, and above all their variability from one restart to the next, that deserves attention.
Monday morning: the weekend effect
Monday stacks several factors. Machines are cold, some fluids have settled, tools may have been removed for cleaning. The morning shift is not always the one that stopped the line on Friday, and information on the machine’s condition got lost in between. Material is not always at the line on time. And settings that worked on Friday evening on a hot machine no longer work on a cold one.
In many shops, this shows up as a Monday OEE that is structurally lower than on other days, something everyone knows and nobody quantifies. Comparing Monday’s first shift with the average first shift of the other days is an excellent starting point.
After a planned shutdown: maintenance, holidays, stocktaking
Restarting after a planned shutdown adds its own risks. After maintenance work, the machine may have been dismantled, parts replaced, settings changed. After summer or year end holidays, the shutdown is long enough for routines to be lost and latent defects to appear. These restarts are rarer, so less practised, and they often cost far more than an ordinary Monday.
The classic mistake is to treat the planned shutdown as over when the work is finished. In reality, it ends when the line is back at nominal speed with good parts. In between lies a grey period nobody owns: maintenance considers the machine handed back, production considers it not yet received.
How to measure restart losses separately
Measuring the restart separately requires a clear definition, applied the same way everywhere.
- Define the restart event. For example: any restart after a stop of more than eight hours, plus any restart after planned maintenance. The rule matters less than its consistency.
- Define the start and end of the window. Start: the planned restart time, not the actual time of the first part. End: the moment nominal speed is reached and held for a set duration, with conforming parts.
- Create a dedicated category. Stops and slow running inside the window go to a restart family, with sub-reasons (warm-up, adjustment, waiting for material, waiting for sign-off).
- Count start-up scrap separately, with a dedicated bin or code, so it is not mixed with defects from normal running.
- Compare. Window duration, losses by component, scrap: by line, by day of the week, by shift and by restart type. Variability often says more than the average.
A word of caution: starting the window at the actual time of the first part erases exactly the losses you are looking for, especially waiting and warm-up.
Worked example: an extrusion line on Monday
For example, an extrusion line shut down every weekend (fictitious figures, for illustration only). The restart is planned for 6 am. After four weeks of separate measurement, the team sees that the restart window lasts 2 h 10 on average, but varies from 1 h 20 to 3 h 30 depending on the week. Warm-up accounts for 50 minutes and is stable. The variation comes from die adjustments and from waiting for material when the morning forklift driver does not have the picking list. Automatic heating before the team arrives and a picking list left on Friday shorten the window and, above all, narrow its spread.
The lesson is general: warm-up is rarely the priority target, because it is known and stable. Variable losses are the expensive ones, and they only show up when every restart is measured.
Building a restart standard
Once losses are visible, the restart standard addresses them one by one. It fits on one page per line and contains at least:
- Preparation the day before: material, tooling, documents, handover notes from the shift that stops to the shift that restarts.
- The start-up sequence, with planned warm-up times and what can be launched before the team arrives.
- Cold reference settings, distinct from production settings, and when to switch from one to the other.
- The first good part criterion and who signs it off, with a response time.
- The ramp-up profile: steps and durations, rather than an improvised climb.
- The target window duration, tracked at every restart and discussed at the production meeting when exceeded.
For restarts after maintenance, the standard adds an explicit handover between maintenance and production, with the end of the planned shutdown defined by production recovered, not by work completed.
What TEEPTRAK brings to restart measurement
TEEPTRAK automatically measures each machine’s activity, stops, micro-stops and speed, with precise timestamps. A module wired into the electrical cabinet and a sensor at the line are all it takes, with no PLC connection, about one hour of installation per machine and OEE available within 48 hours. The restart window becomes visible: restart time, short adjustment stops, ramp-up profile. Operators qualify reasons, so a restart family can be tracked week after week. Equipped plants see on average a 10 to 15 point OEE gain and 30% fewer unplanned stops.
Further reading
- Changeover time: how to really measure it, ramp-up included
- The 6 big losses: what your manual log doesn’t see
- Quality Rate OEE: How to Calculate, Measure and Improve This Critical Manufacturing Metric
FAQ
Are restart losses planned downtime? Warm-up defined in the standard can be treated as planned time, provided its duration is fixed. Anything beyond the standard is a loss and must show up in OEE.
Should restarts be separated from changeovers? Yes. Both include ramp-up, but their causes differ: a cold machine and a returning team on one side, a product and tooling change on the other. Mixing them blurs both analyses.
How do you set the end of the restart window? Use a measurable criterion: nominal speed reached and held for a set duration, with conforming parts. The exact duration depends on the process; what matters is that it is the same for every restart compared.
0 Comments