Reducing Garment Style Changeover Time with a Measured Trial

Measure the return to acceptable production

A useful style-changeover measure ends with acceptable output from the next style. Threaded machines and cleared tables show progress. The factory also needs proof that the new method produces the right garment.

The measurement becomes more useful when it separates the first accepted piece from steady production. A team may make one good piece quickly, then spend another hour on adjustments. Waiting for parts or teaching a new operation can also delay the run. Recording only the first piece can hide that later loss.

This guide uses a hypothetical changeover trial in one defined sewing module. The aim is to shorten the transition. The approved method, safety controls and quality requirements must still hold. All times and results are illustrative. They demonstrate how to build a study, not a performance target for every factory.

The same approach can support a wider line change. Start with a clear boundary, record what happens, then change the work that causes delay. A broad target such as “cut changeover by half” gives little direction. An observed twenty-minute wait for a missing folder points to a specific task.

Agree on the clocks before observing the work

NIST’s quick-changeover guidance defines two events: the last good piece of the current run and the first good piece of the next. The time between them is the changeover interval. This provides a clear starting measure when both events refer to the same process boundary.

For the pilot, record three separate results:

Define “accepted” using the current product criteria. A piece waiting for required approval is not yet an accepted first piece. Record when the check ends, along with any delay in obtaining that check.

The stable-output rule also needs agreement before the trial. In this example, the module must stay within an agreed output band for thirty continuous minutes. The output must also pass the approved quality checks. This is a local trial rule, not an industry standard. Another process may need a different observation window.

Keep the event timestamps as well as the final durations. If a later review changes the interpretation of one event, the original times remain available. The two acceptance times make a “sixty-minute changeover” claim easier to check. Without them, a reviewer cannot tell which events the observer used.

Choose a transition that can teach you something

Select a recurring style transition with a visible cost or delivery effect. It should be important enough to matter and narrow enough for the team to observe fully.

A repeat change between related styles may reveal missing preparation, poor tool location or unclear roles. A move to a new construction may also involve untested methods and operator learning. Those are different studies. Label the transition so later results are compared with similar work.

Identify the two styles and their method versions. Record the machine group, attachments and material conditions as well. Note who performs the work and whether the new operators have already practised the critical tasks. Include the availability of approved cut parts, trims and quality references.

The trial team needs people who perform and support the change. An operator can explain why a setting is adjusted twice. A mechanic can identify which tasks require a stopped machine. A quality representative can define the evidence needed for restart. Planning can check that preparation does not quietly delay another order.

Use the existing approved process as the baseline. First correct any unsafe practice or known quality failure. The baseline must be a method that is suitable to repeat. A time study should not preserve a harmful action just to create a convenient comparison.

Record the whole changeover, including waiting

Observe the work from the agreed start to the agreed end. Record each activity, its start and finish, the people involved and the reason it occurs. Include searching, walking, waiting, repeated adjustments and checks that fail.

Avoid vague labels such as “machine problem.” A useful record might say that the folder could not be fitted because the mounting screws were missing. That description points to a preparation fault, while the vague label encourages the team to blame the machine.

Where tasks overlap, draw them on the same time scale. Adding their durations gives total task time, but it does not necessarily give elapsed changeover time. Two people working for ten minutes in parallel use twenty person-minutes and occupy ten minutes on the clock.

Explain the observation to the team and follow workplace rules for recording or filming. The purpose is to understand the method and support needs. A rushed demonstration staged for the observer may be faster than normal work yet impossible to sustain.

Oregon Manufacturing Extension Partnership recommends observing the full transition, including waits and adjustments. It then calls for testing the revised method over several changeovers. That sequence prevents a tidy checklist from replacing evidence about what actually happens.

A baseline that exposes the lost time

In the hypothetical module, the last accepted old-style piece leaves at 10:00. The activities below occur in sequence. The first accepted new-style piece leaves at 11:42, giving a 102-minute changeover.

The record does not show that all 102 minutes are waste. The module needs a safe transition, correct settings and a valid quality check. It does show that twenty minutes are spent finding items after the old work has stopped.

The two-person fixture change also deserves attention. Why does it take twenty-five minutes? The answer may involve access, a missing adjustment reference or a difficult fastening method. Those causes call for different changes, so the team should inspect the work before proposing a faster target.

First acceptance is only the first endpoint. In this example, another fifteen minutes of adjustment follow. Only then does the thirty-minute steady-output window begin. The window ends at 12:27. Time to confirmed steady output is therefore 147 minutes from the original 10:00 start.

Separate preparation from work that needs a stopped process

SMED stands for Single-Minute Exchange of Die. The Lean Enterprise Institute defines two types of setup. Internal setup needs stopped equipment. External setup can occur while the current job runs. The name refers to a single-digit-minute ambition. It does not set a time limit that every garment line must meet.

Apply the distinction to each observed task. The question is whether the task can be completed safely and correctly before the affected process stops. Simply moving its timestamp earlier is not enough.

External preparation still uses time, people and space. The person collecting a kit may have other duties. The kit may occupy a shared bench or borrow a tool from another line. Include those needs in the revised method. Otherwise, preparation may delay other work even as this module restarts sooner.

External setup does not override energy-control rules

Preparing an attachment at a separate bench avoids work on the running machine. Adjusting equipment that can start or release stored energy raises different risks. The task classification must follow the approved safe method.

In the United States, OSHA 29 CFR 1910.147 addresses hazardous-energy control during covered servicing and maintenance. Its scope includes risks from unexpected starting and stored energy. Other jurisdictions have their own requirements. Use the rules that apply to the site and the machine manufacturer’s instructions.

Ask the responsible technical staff to review changes to access, adjustment or isolation. Their review should cover the proposed method before the trial. A shorter time achieved by skipping a required control is not an acceptable trial result.

Prepare the next style as a verified kit

A kit is useful when someone can confirm that it is ready for the specific change. A trolley containing assorted tools is only another place to search.

For the pilot, the kit record names the style and method version. It lists the folders, gauges, tools, thread and trial material needed. It also shows what has been checked and who is responsible for a missing item. A clear physical location prevents prepared parts from being borrowed without notice.

Trial material should represent the approved production conditions. A convenient scrap may behave differently from the actual fabric layers. Settings developed on a substitute remain provisional. State which checks are still needed when the actual material arrives.

Where a preset or reference setting is proposed, test that it can be repeated. Markings and recorded values can reduce search and adjustment. Worn tools or a change in material may still require a controlled check. Keep the first-piece acceptance step even when the setting was successful on the last order.

Find the task that controls the restart time

Several tasks can be improved without changing the time at which production restarts. This happens when a longer, dependent task still sets the finish time.

In a separate simplified example, two independent setup paths take 28 and 35 minutes. Both must finish before a twelve-minute joint check can start. Assume separate resources are available and the paths can run safely in parallel. Total elapsed time is then 35 + 12 = 47 minutes. Adding all task durations would give 75 minutes and overstate the stopped period.

Reducing the shorter path from 28 to twenty minutes does not change the 47-minute finish. Reducing the longer path from 35 to thirty minutes would shorten the finish to 42 minutes. That assumes no other task becomes the constraint.

Use this dependency view when choosing the next improvement. It does not make the shorter task unimportant; that task still uses labour and may create risk. The claimed restart saving must follow the full sequence. Adding minutes removed from unrelated tasks can overstate the gain.

Run the revised method and preserve the same endpoints

The team prepares and verifies the kit before the old style ends. That preparation uses twenty-four person-minutes. The approved fixture method and setting references also shorten some tasks during the stopped period.

The revised sequence is:

The first accepted piece arrives sixty minutes after the last accepted old-style piece. The same quality criteria apply to both runs.

Elapsed time saved = 102 − 60 = 42 minutes.

Elapsed reduction = 42 ÷ 102 × 100 = 41.2%.

After first acceptance, the trial needs five more minutes of adjustment. The thirty-minute steady-output window begins after that. That window finishes 95 minutes after the starting event. Compared with the baseline’s 147 minutes, confirmed steady output arrives 52 minutes earlier.

Report both results. They answer different questions, and neither should be relabelled as the other. The first-good-piece improvement is 42 minutes; the improvement in the full confirmation window is 52 minutes.

Count labour as well as elapsed time

The baseline table uses 154 person-minutes:

12 × 1 + 20 × 1 + 25 × 2 + 18 × 1 + 12 × 2 + 15 × 2 = 154.

The revised stopped-period work uses 98 person-minutes. Adding the twenty-four minutes of advance preparation gives 122 person-minutes for the measured changeover tasks.

Total labour reduction = 154 − 122 = 32 person-minutes.

Recorded changeover labour falls by 20.8%. Elapsed time falls by 41.2%. The difference is expected: moving work outside the stopped period changes the clock without making that work vanish.

These labour totals end at first acceptance. They exclude the later ramp-up period. Measure that period separately if its labour cost is needed for the business case. Keeping that boundary explicit prevents a partial labour total from appearing to cover the entire transition.

Parallel work can further reduce elapsed time. The tasks must be safe to overlap, and the required people must be available. Adding a second mechanic may speed one change while delaying another line. Check the shared schedule before treating that extra person as free capacity.

Protect the quality decision during restart

The first-piece record should identify the style, size and method version. Add the relevant settings and acceptance result. Record any failed checks and the changes made before acceptance.

A first piece from one operation may not prove that the whole module is ready. Some faults appear only when parts join or when the garment reaches a later check. Define which checks must occur before restart and which continue during the early run.

Keep old-style and trial pieces clearly identified. A fast change can mix components or lose the status of a rejected sample. Both faults create more work later. The outgoing style needs a known final quantity and a clear location for any unfinished or held items.

The trial result is acceptable only if the agreed quality conditions hold. Review rework, scrap and any defect that escaped the restart check. A lower setup time accompanied by a larger repair queue may simply move the loss downstream.

ApparelDiary’s guide to rework loops and inspection capacity can help assess that later workload. The changeover study should capture it as an outcome, even though the detailed capacity analysis belongs elsewhere.

Compare similar transitions and repeat the trial

One improved run is useful evidence about the proposed method. It is not enough to promise the same saving across every style family, shift or operator team.

For a small follow-up, suppose three comparable baseline changes took 102, 98 and 110 minutes. Three revised changes took 60, 64 and 62 minutes. The median falls from 102 to 62 minutes, a reduction of 39.2%. The observed ranges are 98–110 and 60–64 minutes.

These six invented observations describe the example only. They do not establish a precise long-term average. Nor do they prove that preparation alone caused the difference. The revised method also changed fixture work and setting references.

Compare like transitions where possible. Record material, attachments, staffing, prior experience and any outside interruption. NIST’s guidance on blocking explains how other factors can affect a comparison. Grouping similar conditions can make a trial more useful. A formal experiment needs its own design and analysis.

Keep unusual events visible. A power interruption should not disappear because it makes the trial look worse. Show the full elapsed result and, where useful, a separate explained interruption measure. Use the same reporting rule for the baseline and revised runs.

Distinguish machine readiness from operator learning

Some delays end when tools, settings and material are ready. Others reflect the time needed to learn a new task or coordinate a changed sequence. Treating all of that as mechanical setup can lead to the wrong remedy.

If a difficult operation needs practice, plan that practice with suitable material and a qualified trainer. Record the method learned and the conditions under which the operator can perform it. A brief explanation beside a stopped line may not provide enough preparation.

Coats Digital’s planning framework separates learning profiles from changeover time. Both are capacity inputs. For the factory, the practical point is to keep the observed restart delay separate from the later output curve.

The thirty-minute confirmation window in this pilot is only an early check. Follow the new style through the relevant production period to see whether output and quality remain acceptable. A good first hour cannot establish that the method will hold across the whole order.

Use the recovered time for a real production decision

An earlier restart can be valuable when useful work is ready. Downstream stages must also be able to accept that work. It may support more output, an earlier handover or a smaller batch. The result depends on the rest of the route.

A shorter change may make smaller transfer batches practical. Review the release rule with the same care. ApparelDiary’s Kanban cutting-to-sewing guide explains the link between replenishment signals and WIP limits.

Assume the process makes sixty accepted pieces per hour. On time alone, forty-two recovered minutes offer room for forty-two pieces. This is an upper-bound illustration before ramp-up, material limits and downstream constraints. It is not a forecast of extra sales or profit.

Do not value the same benefit twice. Suppose extra output is valued through its contribution. Adding the full wage value of the same recovered minutes can overstate the gain. If paid staffing stays unchanged, lower person-minutes may release time for other work without reducing cash payroll.

Include the cost of the improvement: duplicate fixtures, kit preparation, training, validation and upkeep. A modest recurring saving may still justify a simple change. The case becomes more credible when it shows those costs instead of treating preparation as free.

Apply the method to lines that change progressively

A complete sewing line may change in stages. The first operations can begin the new style while old-style pieces remain further downstream. In that case, “last old piece” and “first new piece” need a defined observation point.

The new style may reach the output point before the final old-style piece. Subtracting those timestamps can then produce a negative interval. That does not mean the changeover created time. It means the two styles overlapped and the chosen clock does not describe a simple stopped transition.

Use module or station windows to measure the affected resources, then track the overall transition separately. Record when the new route is ready and its first complete accepted garment emerges. Then record when the planned output becomes stable. Preserve the old-style balance so leftover work does not vanish from the study.

This approach also reveals whether a local gain matters to the line. A faster folder change has limited value if the next operation still waits for a specialist machine. Follow the dependency through to the actual output constraint before reporting the benefit.

Make the tested method available for the next change

Once the trial meets its agreed conditions, record the tasks and who performs them. Include preparation timing, kit contents and restart checks. Link the instruction to the correct machine and style family. Keep the evidence behind any setting reference or fixture change.

The owner should review the next few uses with the people doing the work. Missing items, repeated adjustments and unclear handovers are reasons to improve the method again. They should not become accepted exceptions hidden inside an ambitious target time.

Finish the changeover record with actual elapsed time and labour used. Include the quality outcome and unresolved issues. That gives planning a grounded estimate for the next transition and gives the team a specific problem to solve. The shorter time then comes from a repeatable method whose conditions are understood.

References

1. NIST Manufacturing Extension Partnership. Lean and process improvement: quick changeover.

2. Lean Enterprise Institute. Single-Minute Exchange of Die.

3. Oregon Manufacturing Extension Partnership. Setup reduction for manufacturers.

4. US Occupational Safety and Health Administration. 29 CFR 1910.147: hazardous-energy control.

5. NIST/SEMATECH. Randomized block designs.

6. Coats Digital. Apparel production planning framework.