SMED — Single-Minute Exchange of Die — is Shigeo Shingo's method for cutting equipment changeover time down to a single-digit number of minutes: under ten, not literally sixty seconds. Developed at Toyota and Mazda between the 1950s and 1970s, the method matters because changeover time isn't just downtime — it's the reason batch sizes stay large. A two-hour changeover forces a plant to run big batches to amortize it, which builds inventory, hides quality problems inside large lots, and works directly against a Lean flow.
SMED's central insight is deceptively simple: most of what makes a changeover slow isn't the mechanical work itself, it's how much of that work is done while the machine is stopped that didn't need to be. Separating what can be prepared in advance from what truly requires a stopped machine, and only then improving what's left, typically cuts changeover time by 50–90% with little or no capital investment.
Why SMED Matters
Unlocks Smaller Batch Sizes
A fast changeover removes the economic pressure to run large batches, which is the single biggest lever for reducing work-in-process inventory.
Recovers Capacity Without Capital
Most SMED gains come from reorganizing existing work and simple fixtures, not new equipment — changeover time is often the cheapest capacity a plant has.
Improves Quality at Startup
Standardized, positioned setups reduce the startup adjustment period that produces first-piece scrap after every changeover.
Feeds Directly Into OEE
Setup and adjustment time is one of the Six Big Losses that reduces Availability — see the OEE Guide.
Core Terms
| Term | Meaning |
|---|---|
| Internal Setup | Changeover work that can only happen while the machine is stopped, such as removing a die or bolting on a new fixture. |
| External Setup | Changeover work that can happen while the machine is still running the previous job, such as staging the next tool or pre-heating a mold. |
| Changeover Time | The elapsed time from the last good part of the previous run to the first good part of the next run. |
| Functional Clamping | A fastening method (cam, one-turn bolt, spring clip) that secures a part in less than one full turn, replacing multi-turn bolts. |
| Adjustment Elimination | Replacing trial-and-error positioning with fixed stops, jigs, or numbered settings so a part locates correctly the first time. |
| Parallel Operations | Two people performing internal setup steps simultaneously on opposite sides of a machine instead of one person working sequentially. |
Shingo's Four Stages
SMED is applied in a fixed order because each stage builds on the one before it, and the earliest stages are also the cheapest — separating work costs nothing but attention, while redesigning tooling costs engineering time and, sometimes, money.
| Stage | Name | What Happens |
|---|---|---|
| 0 | Undifferentiated Setup | Internal and external work are mixed together and typically performed in whatever order habit dictates. This is the untouched starting point, not a target. |
| 1 | Separate Internal from External | Every step is classified as internal or external, usually by filming or timing an actual changeover. Simply moving genuinely external steps — staging tools, retrieving the next die — to before or after the stop is often the single largest reduction in the whole project. |
| 2 | Convert Internal to External | Steps currently done while stopped are redesigned so they can be done while running: pre-heating a mold on a separate stand, pre-assembling a fixture off-machine, standardizing die heights so no re-shimming is needed. |
| 3 | Streamline Remaining Internal Setup | What's left is improved directly: functional clamps replace bolts, positioning stops eliminate adjustment, and parallel operations split remaining work across two people. |
Practical Techniques by Stage
Each stage has a small set of techniques that do most of the work. None of them require new capital equipment — they're almost entirely about reorganizing existing work, standardizing existing tooling, and replacing generic fasteners with ones designed for speed.
Checklist the Current Changeover
Film or time an actual changeover step by step before changing anything — teams routinely misremember which steps are truly internal.
Pre-Stage Everything External
Tools, dies, gaskets, and fixtures for the next job are retrieved and positioned at the machine before the current run ends.
Standardize Functions, Not Parts
Standardizing die height or clamping location — not the dies themselves — lets one set of fixtures serve many different jobs.
Replace Bolts With Functional Clamps
A one-turn cam or spring clip does the same clamping job as a multi-turn bolt in a fraction of the time.
Use Intermediary Jigs
Pre-assembling a die or tool on a jig away from the machine, then swapping the whole jig, converts what was internal work into external work.
Eliminate Adjustment With Positioning Stops
Fixed stops, guide pins, or numbered settings mean a part locates correctly on the first try, removing trial-and-error entirely.
How to Run a SMED Event
SMED is usually run as a focused, short project rather than a background initiative — a single piece of equipment, one to three days, and a small cross-functional team.
- Select and scope. Pick one machine with frequent changeovers and a clear owner, not the whole plant at once.
- Time the baseline. Film an actual changeover, start to finish, and log every step with its duration.
- Classify every step. Mark each one internal or external as a team, discussing any step that's ambiguous.
- Apply Stage 1. Move every genuinely external step off the downtime clock before touching anything else.
- Apply Stages 2 and 3. Convert what can be converted, then streamline what remains with clamps, jigs, and positioning aids.
- Document the new standard. Write down the new sequence and train every operator who performs the changeover.
- Re-time on a schedule. Internal work creeps back without periodic verification against the documented standard.
Worked Example: Meridian Stamping's Press Changeover
Meridian Stamping runs a 400-ton press that changes dies six times per shift. The current changeover takes 42 minutes, all of it logged as downtime because nobody had separated internal from external work.
- The team times the current changeover and classifies each step: retrieving the next die from the rack (6 min), disconnecting the old die (5 min), removing it (4 min), cleaning the bolster (3 min), positioning and bolting the new die (14 min, including re-shimming for height), connecting utilities (4 min), and first-piece adjustment (6 min).
- Stage 1 — separate: retrieving the die (6 min) is purely external and is moved to before the press stops. New changeover time: 36 minutes.
- Stage 2 — convert: dies are standardized to a common shank height using shim kits pre-fitted on a jig away from the press, eliminating the 14-minute positioning-and-bolting step's re-shimming portion (9 minutes recovered). New changeover time: 27 minutes.
- Stage 3 — streamline: multi-turn bolts on the now-5-minute position/bolt step are replaced with functional cam clamps (drops to 1 min), a second technician performs utility connection (4 min) in parallel with cleaning (3 min) instead of after it (the pair now takes 4 min instead of 7), and fixed locating pins remove the need for first-piece trial adjustment (6 min drops to a 1-minute confirmation check).
- Final changeover time: 5 (disconnect) + 4 (remove) + 4 (clean and utility connect, run in parallel) + 1 (position/bolt) + 1 (confirm) = 15 minutes.
A 42-minute changeover done six times a shift consumed 252 minutes of downtime — over four hours. At 15 minutes, the same six changeovers consume 90 minutes, recovering roughly 2.7 hours of press capacity per shift without buying any new equipment.
Self-Assessment Questions
Before calling a changeover "optimized," a team should be able to answer yes to each of these:
- Has the current changeover actually been timed step by step, not estimated from memory?
- Is every step classified as genuinely internal or external, with no "we've always done it this way" steps left unquestioned?
- Has every purely external step been moved off the downtime clock?
- Is there a specific plan — not just an aspiration — to convert the largest remaining internal step to external?
- Is the new changeover sequence written down and trained to every operator who performs it, not just the person who improved it?
- Has batch size actually been reduced to capture the benefit of the faster changeover?
Common Mistakes
Jumping to Stage 3 First
Redesigning tooling before separating internal from external work usually improves steps that shouldn't have been on the machine's downtime at all.
Skipping the Baseline Timing
Without an actual timed changeover, teams reliably underestimate how much "internal" work is really external habit.
Treating SMED as a One-Time Event
The biggest gain often comes from the first pass, but changeover time creeps back up without a standard and periodic re-timing.
Ignoring the Adjustment Step
First-piece adjustment after a changeover is frequently longer than the mechanical swap itself, and it responds directly to positioning aids.
Quick Reference
Setup Checklist
- Time or film an actual changeover before changing anything.
- Classify every step as internal or external.
- Move genuinely external steps off the downtime clock first.
- Only redesign tooling after separation and conversion are done.
Using the Result
- Track changeover time as a standard, not a one-time project result.
- Re-time periodically — internal work creeps back without a standard.
- A faster changeover only helps if batch sizes are actually reduced afterward.
- Feed the recovered time into the Availability factor of OEE.
Sources and Further Reading
- Shigeo Shingo, A Revolution in Manufacturing: The SMED System.
- Shigeo Shingo, Non-Stock Production: The Shingo System for Continuous Improvement.
- Taiichi Ohno, Toyota Production System: Beyond Large-Scale Production.
- ASQ Certified Quality Engineer Body of Knowledge.