This article is for production managers, manufacturing engineers, continuous improvement leaders and plant managers who run high-mix lines: several formats, materials or tool sets on the same equipment. On these lines a single SMED standard does not hold, because a format change has little in common with a material change. The question addressed here is specific: how to group products into families based on changeovers that were actually measured, then write and maintain one standard per type of transition.
- On a high-mix line, a single SMED standard does not hold.
- A from/to matrix reveals the real technical families.
- Write one standard per family and per transition type.
- Tracking the gap per family targets audits and protects OEE.
What the SMED method reminds us, and where high-mix lines push back
On the Organilog blog, Serago P. Ubifavirutu published SMED : réduire le temps de changement de série en production (in French). The article goes back to the origins of the method at Toyota with Shigeo Shingo, details the four steps (observe, separate internal and external tasks, convert, streamline), illustrates them with a bottling line running four bottle formats, and lists the expected benefits, including smaller batch sizes and less scrap at restart. It also stresses durability: “Les temps de changement dérivent naturellement si les standards ne sont pas maintenus.” (in other words: changeover times naturally drift if standards are not maintained). The author recommends auditing the standards regularly.
That warning matters even more on a high-mix line. When a line chains very different transitions, drift does not show in an overall average: a format change can stay under control while a material change slips week after week. Organilog lays out the method; this article suggests a way to organize it by product family, with measured rather than estimated times.
Why one standard is not enough
On a high-mix line, changeover time depends mostly on what separates the outgoing product from the incoming one. Moving to a neighboring format sometimes takes a simple guide adjustment. Changing material, color or a full tool set means cleaning, swapping parts and a longer ramp-up. A single standard ends up too heavy for small transitions and too light for big ones, and operators stop following it.
The answer is to think in transition types, often captured in a from/to matrix: for each pair of outgoing and incoming products, look at the real changeover time, ramp-up included. Pairs with similar times form a family. Standards are then written per family and per transition type, not per product.
Building families from measured changeovers
Commercial families (by customer or product range) do not always match the technical families of the line. The most reliable starting point is the history of measured changeovers. The table below is a starting grid, to adapt to each shop.
| Transition type | What changes on the machine | What the standard covers | What to measure |
|---|---|---|---|
| Within the same family | Fine settings, labels, parameters | Settings list with target values, preparation outside the stop | Stop duration and parts made before nominal rate |
| Format change | Guides, change parts, stops | Labeled change parts, fixed-value settings | Stop duration, micro-stops in the following hour |
| Material or color change | Cleaning, purge, start-up check | Cleaning sequence, first-part approval criteria | Stop duration, scrap at restart |
| Full tool change | Tool, clamping, program | Complete external preparation, dedicated cart, task order | Stop duration, ramp-up, gap between shifts |
Once the matrix is filled in, two findings come up often. Some transitions believed to be simple are actually long, because ramp-up drags on. And the sequence of orders weighs as much as technique: for example, a line that alternates freely between two families goes through more big changeovers than a line that groups its orders by family.
Writing and holding a standard per family, step by step
- Extract the history. Pull several weeks of timestamped changeovers with outgoing product, incoming product, stop duration and return to nominal rate.
- Fill in the from/to matrix. For each pair, compute the median duration and the spread. A wide spread signals a transition with no real standard.
- Define technical families. Group products whose internal transitions are short and consistent. Validate the families with setters, who know the exceptions.
- Write one standard per transition type. Apply the four SMED steps to each type: within family, format change, material change, tool change.
- Sequence orders. With planning, group orders by family when lead times allow, to limit the number of big transitions.
- Track the gap per family. Every week, compare the measured time of each transition type with its standard. Drift on a single family triggers a targeted audit rather than a general review.
What family-level tracking changes for OEE
Changeover time weighs on availability, restart scrap on quality and a slow ramp-up on performance. Tracked by family, OEE becomes readable: you know which transition type costs the most and which family a SMED project should target first. It also avoids a common trap, judging a project on the average of all changeovers, when the weekly product mix alone is enough to move that average.
This tracking does not replace shop floor observation. It tells you where to look; task analysis, video or timing a full changeover tell you what to change.
What TEEPTRAK brings
TEEPTRAK measures OEE, stops and micro-stops in real time. A module wired into the electrical cabinet and a sensor at the line record machine activity automatically, with no connection to the PLC. Installation takes about one hour per machine and OEE is available within 48 hours. Each changeover is timestamped and qualified, so the from/to matrix can be filled with measured durations, ramp-up included, and the gap to standard tracked family by family. The solution is deployed in more than 700 plants in over 30 countries.
Further reading
- SMED : réduire le temps de changement de série en production (Organilog, in French)
- Before SMED: measure every changeover to decide where to start
- Changeover time: how to really measure it, ramp-up included
FAQ
How many families should a line have? As few as possible, as long as transitions within a family stay short and consistent. If a family shows a wide spread of times, it probably hides two technical families.
Should orders be grouped by family even if it makes batches longer? Not always. Grouping is decided with planning, weighing the changeover time saved against customer lead times and inventory. Measured times per transition type make that trade-off factual.
How often should family standards be reviewed? Whenever the measured gap for a transition type stays above its standard, and at least during a regular audit. Tracking the gap weekly points to the family that is drifting.
0 Comments