This article is for plant managers, production managers and continuous improvement, methods and maintenance leaders who are preparing a SMED project and need to decide where to start: which line, which product, which transition. Before shortening a single changeover, it pays to know what all changeovers really cost over a month, and to use that measurement to set the baseline the project will be judged against.
- The longest changeover is not always the one that costs the most hours per month.
- Measuring every changeover for weeks reveals frequency, critical transitions and gaps between shifts.
- Time to first good part is part of the changeover.
- Freeze the baseline before the first SMED workshop, with written rules.
Induscor Partners’ point: apply SMED where it really limits you
The editorial team at Induscor Partners, a Barcelona based industrial operations consultancy specialising in lean and SMED, has published SMED: Qué es, Cómo Implementarlo y Resultados Reales en Planta Industrial (in Spanish). The article covers the four steps of the method, the most common mistakes and a 90 day implementation plan. One of those mistakes is picking the wrong machine, and the conclusion is clear: “Hay que aplicar SMED donde el cambio limita realmente la flexibilidad o la capacidad.” (“SMED should be applied where the changeover really limits flexibility or capacity.”)
The Induscor Partners article provides the method to run SMED: observe the changeover, separate internal and external activities, convert, streamline. This article sits just before that: how to get, machine by machine, the data that lets you choose the pilot machine on facts and set a reliable starting point. The consultant runs the project; measurement gives it an objective basis.
Why a one off stopwatch study is not enough to choose
The usual approach is to ask the team which changeover is the longest, then time or film a few changeovers. That is essential for task analysis, but it answers a different question poorly: which changeover costs the most production hours per month? Three biases come up again and again:
- People remember the longest changeover, not the most expensive one. A long, rare changeover makes more of an impression than a short one repeated every shift. For example, a line that changes over twice per shift with 40 minute changeovers can lose more hours per month than a press with a single two hour changeover per week.
- The observed changeover is not the everyday changeover. When someone holds a stopwatch, the team gets organised. Night shift changeovers, a different crew or the absence of the lead setter rarely make it into the sample.
- The changeover does not end when the machine restarts. Measuring only the stop ignores adjustment parts and the time to the first good part, which often weigh as much as the teardown.
We have already covered how to measure a changeover properly, ramp-up included: Changeover time: how to really measure it, ramp-up included. The question here is different: how to use the measurement of every changeover to decide where to start.
What to measure on every changeover for a month
Ideally, record every machine changeover for four representative weeks, across all shifts, with the outgoing and incoming product. Each measurement supports a decision:
| Measurement | What it reveals | Decision it supports |
|---|---|---|
| Number of changeovers per line per month | Where frequency is concentrated | Which line gains most from every minute saved |
| Average duration and monthly total | The real cost in production hours | Rank lines by hours lost, not by gut feeling |
| Duration by transition (from one product to another) | Critical transitions: material, colour, tooling | Which product or transition to tackle first |
| Gap between shifts or crews | Variability of the current method | Whether to standardise before redesigning |
| Time to first good part | Adjustments, scrap and slow restart after the changeover | Whether effort belongs on settings rather than teardown |
Crossing these measurements with each line’s load answers the question Induscor Partners raises: a changeover only deserves priority if it limits capacity or flexibility. On a line with spare capacity, the first goal may be smaller batches; on a bottleneck, every minute recovered is output, and OEE points.
A five step method to choose where to start and set the baseline
- Define what a changeover is. Write the rule down: from the last good part of the outgoing product to the first good part of the incoming one. The same rule for every line and every shift.
- Measure without acting for four weeks. Automatically record every stop qualified as a changeover, with product, shift and duration. Nothing in the method changes during this period.
- Rank by hours lost per month. A Pareto by line, then by transition. Flag the bottleneck lines.
- Analyse the spread. If the same changeover takes twice as long depending on the shift, the first workstream is a shared standard; if every shift is slow, the tasks need redesigning.
- Freeze the baseline. For the pilot machine, record average duration, spread, time to first good part and starting OEE, with the calculation rules. That is the reference SMED will be measured against.
With this starting point, choosing the pilot machine stops being a matter of opinion, and the filming and timing in the following steps focus on the changeover that really matters.
Who does what: the SMED project and the measurement
The SMED project is run by the plant team or its consultant: task analysis, converting internal activities to external ones, standardisation, training. TEEPTRAK provides the measurement. A module wired into the electrical cabinet and a sensor at the line side automatically record each machine’s state, with no PLC connection; installation takes about one hour per machine and OEE is available within 48 hours. The operator qualifies the stop as a changeover, and its duration, shift and the restart that follows are recorded. TEEPTRAK is not an MES: it measures OEE, downtime and micro-stops in more than 700 plants across more than 30 countries.
Further reading
- SMED: Qué es, Cómo Implementarlo y Resultados Reales en Planta Industrial (Induscor Partners editorial team, in Spanish)
- Changeover time: how to really measure it, ramp-up included
- Performance tracking: 5 measurement pitfalls that distort decisions
FAQ
How long should you measure before choosing the pilot machine? Long enough to cover every shift and a representative product mix, usually a few weeks. A measurement that is too short describes one particular week, not the reality of the shop floor.
Does automatic measurement replace filming the changeover? No. Measurement shows which changeover costs the most and how much it varies; filming and task timing explain why. Choose with data first, then analyse in detail.
Which indicators should a SMED baseline include? At minimum, average changeover duration, its spread across shifts, time to first good part and the pilot machine’s starting OEE, measured with written rules that will not change during the project.
0 Comments