The Impact of Setup Time

Measure setup without equating it with waste

Changeover time can be defined from the previous product's last good unit to the next product's first good unit. State whether stabilization is included and which quality criterion applies. Separately record external tasks performed before or afterward without stopping equipment.

External setup means work that can occur while equipment runs, not work that is inherently removable. Cleaning, safety, verification, and permits can be necessary. SMED supports separating tasks and investigating preparation outside downtime; it does not authorize omitted controls.

Two measurement questions

Start-and-finish records for each changeover describe event duration and variability between references. State sampling estimates the observed share in setup, waiting, or other activities. They are not interchangeable: longer changeovers are more likely to appear in snapshots.

Distinguish the person's activity from equipment state. Preparing tools while a machine produces does not imply machine downtime.

Scenario example, not a customer outcome

Assume a 480-minute shift with thirty minutes excluded from planned production, leaving 450. Five eighteen-minute changes total ninety minutes. With no other stops, availability is (450−90)/450=80%.

If a trial achieved six-minute changes, total setup would be thirty minutes and availability (450−30)/450=93.33%. The difference is 13.33 percentage points under these assumptions. It does not show that SMED will achieve six minutes or that all added capacity will be sold.

With performance and quality unchanged, an initial 65% OEE would scale to approximately 75.83%. Recalculate if those factors change. The OEE formulation treats changeovers as availability losses when they occur during planned production time.

Design categories and sample

Separate disassembly, cleaning, installation, adjustment, verification, and confirmed waiting with non-overlap rules. Categories describe activity; they do not establish causes or prove that all time is reducible.

Planning example for setup share: p₀=0.20, z=1.96, and e=0.03 yield n₀=682.951… and n=683. NIST explains the calculation. Coverage across shifts, families, and days is a separate design task. Scheduling seven hundred readings does not guarantee independence or freedom from bias.

Prioritize using your own data

Use records from your process to identify tasks, change counts and total duration. A percentage from another factory does not replace that measurement; differences in product, method and change frequency limit comparisons across sectors.

Trial a concrete measure such as staging tooling before a change. Assess safety, first-piece quality, preparation workload, and complete duration. Moving work outside downtime does not make the work disappear.

Sustain the result

Document method, conditions, and responsibility. Compare compatible references and avoid attributing every difference to the intervention. Saved time may represent potential capacity rather than cash savings; value it without double counting.