Internal vs External Setup in SMED: Changeover Optimization

Separating internal setup, tasks that can only happen while a machine is stopped, from external setup, tasks that can happen while it's running, is the foundation of [SMED in JIT Manufacturing: Single-Minute Exchange of Die], but making that split actually reduce changeover time requires more than sorting tasks into two lists. It requires a specific sequence: documenting every task as it's currently done, classifying each one honestly, converting as many internal tasks to external as genuinely possible, and only then working on streamlining whatever remains internal. Getting this sequence right matters directly to [Heijunka in JIT: Leveling Production for Mixed Models], which depends on changeovers short enough to support frequent switching, and to [Cellular Manufacturing in JIT: Benefits and Implementation Guide], where a cell serving several product variants needs genuinely fast, not just reclassified, changeovers to stay viable as a single shared cell.

Documenting the Changeover Before Classifying Anything

Why Documentation Has to Come First

Before any task can be correctly classified as internal or external, it has to actually be observed and recorded, step by step, exactly as it happens today, not as it's supposed to happen according to a procedure that may no longer match current reality on the floor. A changeover documented from memory or from an outdated work instruction routinely misses steps that have crept in informally, tool searches, extra inspections, interruptions, that are consuming real time but aren't written down anywhere official.

What Good Documentation Actually Captures

Useful documentation records every single action with its duration, in the order it actually occurs, typically gathered by timing an actual changeover with a stopwatch or filming it for later review and discussion with the operator involved.

Key Insight: The documentation step feels like overhead before any improvement has happened, but skipping it means classifying tasks based on assumption rather than observation, which is exactly how obviously-external tasks end up staying internal for years without anyone noticing.

Classifying Each Task Honestly

The One Question That Decides Classification

Classification asks one question of every documented task: does this specific task require the machine to be physically stopped, or could it happen while the machine is running the previous job or already running the new one? A task like removing mounting bolts clearly requires the machine to be stopped. A task like retrieving a wrench from a tool crib doesn't, even though it's currently being done while the machine sits idle and nobody has questioned why.

Habit Is the Enemy of Honest Classification

The honesty required here is the step most changeover documentation gets wrong, since habit tends to classify a task as internal simply because that's when it's always been done, rather than asking whether it genuinely has to be. A task performed during the stopped window out of long-standing habit isn't automatically an internal task; it only becomes genuinely internal once there's a real physical reason it can't happen at another time, something a team has to consciously check for rather than assume. A useful discipline here is to ask, for every single task, what would physically have to be different for this to happen while the machine is running: if the honest answer is nothing, the task has been misclassified as internal purely out of habit, and belongs in the external category starting with the very next changeover.

Converting Internal Tasks to External Wherever Possible

Moving Tasks in Time, Not Changing the Task

Once tasks are classified, the real work of SMED actually begins: looking at every task still marked internal and asking what would need to change, in staging, in preparation, or in sequencing, to make it genuinely external instead of staying locked into the stopped window by default. A die retrieval that's internal only because the die isn't pre-staged nearby becomes external the moment a staging cart places it within reach before the changeover starts; the retrieval task itself doesn't change, only when and how it happens.

Common Conversion Techniques

A handful of techniques account for most successful internal-to-external conversions: pre-staging tools and materials at the point of use before the changeover begins, pre-heating or pre-conditioning a die or mold so it's ready the moment it's installed, using checklists prepared in advance so verification during the stopped window is confirmation rather than discovery, assigning a second person to handle external tasks in parallel with the machine operator's internal work, and standardizing fasteners and connectors across different dies so the same gathered toolkit works for every changeover rather than needing a different one searched out each time.

Key Insight: Most of these techniques cost very little to implement; the barrier to converting a task is almost always organizational, deciding who does the preparation and when, rather than requiring new equipment or significant capital investment.

A Worked Example of Classifying and Converting a Changeover

The Documented, Unclassified Changeover

Consider a plastic injection molding machine changing over between two different mold sets used for related but distinct parts. Filming an actual changeover reveals the operator spends the first twelve minutes locating the correct mold in a crowded storage area, eight minutes gathering the specific bolts and alignment pins this mold uses from a shared parts bin, thirty-five minutes removing the current mold and setting it aside, fifteen minutes cleaning mold-release residue from the machine platen, fifty minutes installing and aligning the new mold by hand, twenty minutes connecting cooling lines and verifying temperature settings, and finally runs a ten-minute warm-up cycle before producing an acceptable first part. The entire sequence, over two and a half hours, happens with the machine stopped.

Classifying and Converting Each Step

Classifying each step honestly shows that locating the mold, gathering bolts and pins, and even the ten-minute warm-up cycle don't actually require the machine to be stopped at all; the warm-up in particular could run on a separate heating station entirely independent of the press. Converting these three tasks alone, by pre-staging the mold and hardware kit together on a cart before the changeover and running the warm-up externally, removes roughly thirty minutes from the stopped window immediately. A second pass at the alignment step, adding locating dowels to the mold so it seats correctly without the fifty minutes of manual adjustment, cuts that step to under fifteen minutes. The changeover that started at two and a half hours now runs under ninety minutes, with further conversion opportunities still available in the cleaning and cooling-line steps for a future pass. The cleaning step, in particular, is a strong candidate for a third pass, since dedicating a second set of pre-cleaned platen surfaces to swap in could eliminate most of that fifteen-minute task entirely rather than merely shortening it.

Common Mistakes in the Classification and Conversion Process

A handful of recurring mistakes undermine this process even when the basic internal/external concept is well understood:

• Classifying tasks from memory instead of direct observation. A changeover that hasn't been watched and timed recently often has outdated assumptions baked into its classification, missing newer habits or workarounds that have crept into the process since it was last reviewed.
• Converting a task to external without assigning clear ownership of it. A task reclassified as external but with nobody specifically responsible for doing it beforehand tends to simply happen during the stopped window anyway, out of habit or convenience, erasing the improvement in practice.
• Stopping the conversion effort too early. The first pass through a changeover usually converts the most obvious tasks; a second and third pass, done after living with the first round of changes, frequently finds additional tasks that can also move once the obvious ones are gone.
• Treating the classification as permanent once completed. A new die design, a new material, or a new piece of tooling can change which tasks are genuinely internal, so classification needs to be revisited whenever the underlying equipment or process changes meaningfully, not left as a one-time exercise.
• Converting tasks without verifying the conversion actually holds under real conditions. A pre-staged cart that works well in a trial run can still fail in practice if the cart isn't restocked consistently or ends up parked somewhere inconvenient, quietly reintroducing the search time the conversion was meant to eliminate.
• Assuming every changeover on similar equipment needs the identical classification. Two machines running superficially similar processes can still have genuinely different internal requirements depending on their specific tooling, so a classification copied wholesale from one machine to another risks either missing real constraints or leaving easy conversions undiscovered.
• Documenting only the operator's actions and ignoring supporting roles. A changeover often involves a maintenance technician, a material handler, or a quality inspector at some point, and classification that only tracks the primary operator's steps can miss opportunities to convert tasks these supporting roles perform as well.
• Discarding the original documentation once a conversion is implemented. Keeping the before-and-after record makes it possible to retrain a new operator quickly, to spot when a converted task has quietly drifted back internal, and to compare this changeover's progress against similar ones elsewhere in the plant.

When This Process Delivers the Most Value

Undocumented Changeovers Are the Best Candidates

Good to Know: The internal/external conversion process delivers the most value on changeovers that currently have a lot of informal, undocumented work happening during the stopped window; a changeover that's already tightly standardized and documented may have little left to convert, since most of the obvious opportunities have likely already been captured in an earlier effort.

A changeover that hasn't been formally reviewed in years, by contrast, especially one run primarily by muscle memory and informal habit, is usually the best candidate for this process, since undocumented changeovers accumulate unnecessary internal time the way unused storage space accumulates clutter, gradually and without anyone noticing until it's measured directly against a clear baseline.

Prioritizing Which Changeover to Tackle First

A plant with several candidate changeovers across different machines is usually better served starting with whichever one combines high changeover frequency with the least prior documentation, since that specific combination offers both the largest potential time recovery and the most undiscovered opportunities still sitting in the process waiting to be found. A rarely-run changeover, even a poorly documented one, simply won't generate enough cumulative benefit to justify the documentation effort the same way a frequent one will.

Measuring Whether the Conversion Actually Worked

Comparing Before and After Consistently

The clearest evidence that internal-to-external conversion is working is a direct before-and-after comparison of stopped time, measured the same way both times, using the same documentation method that identified the original tasks. A reduction that holds up when checked weeks later, not just immediately after the improvement effort, confirms the converted tasks are actually being done externally in practice and haven't quietly drifted back into the stopped window out of habit once attention moved elsewhere.

Checking That Converted Tasks Stay Converted

Beyond the headline time reduction, it's worth periodically re-observing the changeover itself to confirm each specific task that was converted is still actually happening externally rather than silently sliding back into the stopped window. A task can appear converted on a checklist while an operator, under time pressure or simply out of old habit, ends up doing it during the stoppage anyway, which the aggregate time measurement alone might not reveal until the pattern has already become entrenched again. Pairing the aggregate number with an occasional spot check of individual tasks gives a far more complete picture of whether the conversion effort genuinely stuck.

Within the Lean System

Connection to Lean Principles

[SMED in JIT Manufacturing: Single-Minute Exchange of Die] is the broader method this classification and conversion process actually executes; the internal/external split is SMED's foundational technique, not a separate concept.

Connection to Lean Tools

[Heijunka in JIT: Leveling Production for Mixed Models] depends on exactly this kind of disciplined changeover reduction to make frequent switching between product variants practical without losing production time.

Connection to Continuous Improvement

[Cellular Manufacturing in JIT: Benefits and Implementation Guide] often surfaces this classification work directly, since a cell serving multiple product families needs genuinely converted, not just reclassified, changeover tasks to stay viable.

Frequently Asked Questions

Classifying sorts each task into internal or external based on whether it genuinely requires the machine stopped; converting then changes staging, preparation, or sequencing to move a task that's currently internal into the external category without changing the underlying task itself in any way.

Because classifying from memory or an outdated procedure misses tasks that have crept in informally, like extra tool searches or unplanned inspections, which are consuming real stopped time but wouldn't be caught without directly observing and timing an actual changeover first, start to finish.

Pre-staging tools, materials, or dies at the point of use before the changeover begins, which converts a search-and-retrieve task from internal to external without requiring any new equipment, simply by changing when and where the preparation work actually happens beforehand.

Yes. A new die design, material, or piece of tooling can change which tasks genuinely require the machine to be stopped, so classification done once isn't permanent and needs revisiting whenever the underlying equipment or process changes in any meaningful way at all going forward.

Usually less so. A changeover that's already tightly standardized has likely captured most of the obvious conversion opportunities already; this process delivers the most value on changeovers still running largely on informal habits and undocumented, outdated workarounds nobody has questioned recently.