Quality
SMED: When Your Organization Discovers That the Fastest Changeover Is
the One Nobody Thinks Is Possible — and the Nine Minutes You Shaved Off
Setup Became the Most Profitable Nine Minutes in Your Factory’s
History
The Four-Hour Tombstone
The press line at Erlau Metalworks had a rhythm. Three shifts, five
days a week, forty-seven weeks a year. Stamping body panels for three
different automotive OEMs. The line ran well — when it ran. The problem
was what happened between runs.
Every time the production schedule called for a die change, the line
went dark. Four hours. Sometimes five. A team of four technicians would
descend on the 800-ton press with wrenches, hoists, and a collective
memory of how the last changeover went wrong. They’d unbolt the outgoing
die, wrestle it onto a flatbed cart, wheel in the next one, align it by
eye and feel, bolt it down, feed through trial blanks, measure the first
parts against the go/no-go gauges, adjust, try again, measure again, and
— if everything cooperated — declare the line ready.
Four hours of a capital asset worth more than the building it sat in
producing absolutely nothing. Four hours of operators standing around.
Four hours of the schedule bleeding into the next shift’s time. And
because every changeover was a four-hour penalty, the schedulers ran
massive batches — 5,000, 10,000, sometimes 20,000 parts at a stretch —
to amortize the pain. Which meant mountains of work-in-process
inventory, long lead times, and a factory that could respond to a
customer change request about as fast as a cargo ship could change
direction.
The plant manager, a pragmatic German named Kessler, had accepted
this as physics. “You cannot change the laws of manufacturing,” he told
his team during the quarterly review. “Setup takes what it takes.”
He was wrong.
A Man Who Refused to Accept
Physics
Shigeo Shingo had a habit that made factory managers uncomfortable.
He would stand next to a machine during a changeover and watch. Not for
ten minutes. Not for an hour. He would watch the entire setup, start to
finish, with a stopwatch and a clipboard, writing down everything he
saw.
What he saw — at hundreds of factories across Japan in the 1960s and
1970s — was not a manufacturing problem. It was a classification
problem.
Shingo noticed that every activity during a changeover fell into one
of two categories. There were tasks that could only be done when the
machine was stopped — removing the old die, positioning the new one,
aligning it. And there were tasks that could be done while the machine
was still running — gathering tools, pre-positioning fixtures,
preheating dies, transporting materials.
The first category he called internal setup. The
second he called external setup.
His observation was devastating in its simplicity: most factories
were treating external setup as if it were internal. Technicians were
searching for tools after the machine had already stopped. They were
walking to the tool crib mid-changeover. They were waiting for parts to
be delivered while the press sat idle. They were doing things that
required no stopped machine at all — but they were doing them while the
machine was stopped.
The math was brutal. At a typical factory Shingo visited, the
four-hour changeover contained roughly forty-five minutes of actual
internal setup. The other three hours and fifteen minutes was external
work being performed during downtime — or pure waste, like searching for
a wrench that should have been in the technician’s hand before the
machine stopped.
Shingo’s method — which he eventually named Single-Minute
Exchange of Die, or SMED — was built on a single radical
premise: if you move every possible task to external setup, the
remaining internal setup can be reduced to under ten minutes.
Single digit. Hence the name.
The Three Stages That
Change Everything
SMED is not a trick. It is not a hack. It is a systematic method that
unfolds in three stages, each one building on the last. Understanding
these stages is the difference between a factory that shaves twenty
minutes off changeover time and one that turns four hours into nine
minutes.
Stage One:
Separate Internal and External Setup
This is the foundation, and it is shockingly difficult for
organizations that have never examined their changeovers critically.
The process is simple in concept: film or observe a complete
changeover. For every task, ask one question: “Does this require
the machine to be stopped?” If the answer is no, it is external
setup. If the answer is yes, it is internal setup.
At Erlau Metalworks, when the team finally conducted this analysis,
they discovered that 68% of their changeover time was external work
being done while the press sat idle. Technicians were locating dies in
the storage area after the line stopped. They were walking across the
plant floor to retrieve specialized bolts. They were checking quality
measurement tools that could have been calibrated an hour earlier.
The technicians weren’t lazy. They were following the procedure that
had been in place for years — a procedure that had never been designed,
only accumulated. Nobody had ever asked the fundamental question:
what actually requires the machine to be stopped?
When Erlau moved all external tasks to the period before the line
stopped, their four-hour changeover dropped to one hour and fifteen
minutes. Stage one alone recovered nearly three hours of productive
time.
Stage Two:
Convert Internal Setup to External Setup
Stage one stops the bleeding. Stage two goes deeper. The question
changes: “Of the tasks that currently require the machine to be
stopped, which ones could be redesigned so they don’t?”
This is where engineering creativity enters the picture. Shingo
documented dozens of conversion techniques:
Pre-heating. If a die needs to reach operating
temperature before it can produce good parts, don’t heat it on the
machine. Heat it on a pre-heating station while the previous job is
still running. When the old die comes out, the new die is already at
temperature.
Intermediate fixtures. Instead of bolting a die
directly to the press bed — a slow, precise process that requires the
machine to be stopped — bolt it to a standardized adapter plate. The
adapter plate can be pre-aligned and pre-bolted externally. When it’s
time to change, the entire plate-and-die assembly drops into the press
with a single clamping motion.
Parallel operations. If a changeover requires work
on both the top and bottom of a press, don’t do them sequentially with
one technician. Assign two. Internal setup time is not the sum of all
tasks — it is the critical path through all tasks.
At Erlau, converting internal to external setup brought their
changeover time from one hour and fifteen minutes to twenty-two minutes.
They did it with intermediate adapter plates, a die pre-heating station,
and two technicians working in parallel instead of one.
Stage Three:
Streamline Remaining Internal Setup
Now you’re in the final minutes. The remaining internal setup is
genuinely tasks that require a stopped machine. The question becomes:
“How can we make these tasks faster, simpler, and more
reliable?”
Shingo’s techniques here are elegant:
Eliminate adjustments. The single biggest time sink
in most changeovers is not removal or installation — it’s adjustment.
Technicians aligning dies by eye, running trial parts, measuring,
adjusting, running more trials. Shingo’s solution: design the die and
press interface so that alignment is automatic. Precision locating pins.
Guide rails. Stops. The die should drop into the correct position and
stay there — no adjustment needed.
Replace bolts with clamps. Shingo observed that most
bolts in changeovers are used to fix dies in position. But bolts are
slow — you have to align them, thread them, tighten them, check them.
One-point clamping mechanisms, toggle clamps, and hydraulic quick-lock
systems can secure a die in seconds instead of minutes.
Standardize. Every die should use the same clamping
points, the same bolt pattern, the same connection heights. When every
changeover follows the same physical interface, the technician doesn’t
need to think about geometry — only sequence.
Erlau’s final stage brought their changeover from twenty-two minutes
to eight minutes and thirty seconds. Eight minutes and thirty seconds.
From four hours. On a press that had been running four-hour changeovers
for over a decade.
The Hidden Payoff: Batch
Size Reduction
Here is where SMED’s true impact reveals itself — and it has nothing
to do with the changeover time itself.
When a changeover costs four hours, the economics of manufacturing
force you into massive batches. You need to produce enough parts in each
run to spread that four-hour penalty across enough units that the
per-part cost remains acceptable. The math is relentless: if your
changeover costs $2,000 in lost productive time and your part margin is
$0.50, you need to run at least 4,000 parts before the changeover cost
is amortized below your margin.
But when your changeover costs eight minutes — when the penalty drops
from $2,000 to $67 — the entire economic model inverts. You can now run
500-part batches. Or 200-part batches. Or fifty.
This is not a marginal improvement. This is a fundamental change in
what your factory can do.
Lead times collapse. Instead of waiting for a
10,000-part batch to finish before your 500-part order gets scheduled,
your order runs next. Customer lead time goes from weeks to days.
Inventory evaporates. You don’t need safety stock
when you can change over in eight minutes. You produce what’s needed,
when it’s needed, in the quantity needed.
Flexibility becomes real. When changeovers are
cheap, you can respond to demand changes, engineering changes, and rush
orders without destroying your schedule. The factory that was a cargo
ship becomes something closer to a speedboat.
Quality improves. Smaller batches mean defects are
caught faster. Instead of producing 10,000 parts with a misaligned die,
you produce 200 before the next changeover and the next first-article
inspection. The cost of a quality escape drops by a factor of fifty.
Why SMED Fails in Most
Organizations
If SMED is so powerful — and it is — why doesn’t every factory
implement it?
The Myth of “We Don’t Have
Time”
The most common objection: “We can’t afford to stop production to
study our changeovers.” This is the manufacturing equivalent of saying
you can’t afford to pull over for gas because you’re in a hurry. You’re
going to stop anyway. The question is whether you stop deliberately and
learn, or stop randomly and suffer.
SMED does not require a massive capital investment. Stage one —
separating internal and external setup — requires observation, analysis,
and reorganization of existing tasks. It costs nothing but time. And the
time it recovers pays for itself within the first week.
The Expertise Trap
Changeovers are often performed by the most experienced technicians
on the floor — the people who have been doing them for years and have
developed an intuitive feel for the process. These same technicians are
often the most resistant to SMED, because SMED externalizes and
standardizes the knowledge they carry in their hands and their
heads.
This is not a criticism. It is a natural human response to having
your expertise codified. But it must be addressed directly. SMED does
not replace expert technicians — it frees them. When the routine
elements of a changeover are standardized and externalized, the expert’s
attention is available for the genuinely difficult problems that
remain.
The “Good Enough” Ceiling
Many organizations implement Stage One of SMED, see a 50% reduction
in changeover time, declare victory, and stop. Stage One feels like
enough. It isn’t.
The difference between a two-hour changeover and an eight-minute
changeover is not a difference of degree — it is a difference of kind.
At two hours, you still run large batches. At eight minutes, you run
what the customer needs. The economic inflection point — the point where
batch size reduction becomes not just possible but economically
irrational to ignore — typically falls somewhere below twenty minutes.
If your SMED implementation stops above that threshold, you’ve captured
the easy gains but missed the transformation.
The SMED Implementation
Roadmap
For organizations ready to move beyond theory, here is the practical
sequence:
Week One: Baseline. Film three complete changeovers
on your critical constraint. Time every task. Classify each as internal
or external. Calculate your current ratio. If you’re like most
factories, you’ll find that 60-70% of your changeover is external work
performed during downtime.
Week Two: Externalize. Create a setup preparation
checklist. Every external task should be completed before the machine
stops. Assign responsibility. Create shadow boards for tools. Pre-stage
dies and materials. This single step typically cuts changeover time by
40-60%.
Weeks Three Through Six: Convert. Identify the
highest-impact internal tasks and engineer conversions. Start with the
cheapest, highest-return opportunities — pre-heating, parallel
operations, intermediate fixtures. Each conversion should be measured
and validated before moving to the next.
Weeks Seven Through Twelve: Streamline. Attack the
remaining internal setup with clamping conversions, alignment
standardization, and adjustment elimination. Target single-digit
minutes.
Ongoing: Sustain and Expand. Document the new
standard. Train every technician. Apply SMED to every changeover on
every critical machine. Review quarterly.
The Deeper Truth
SMED is nominally about changeover time. But its real subject is
something more fundamental: the relationship between fixed costs
and flexibility.
Every manufacturing process carries fixed costs — the cost of
changing from one product to another. When those fixed costs are high,
the organization is forced into behaviors that optimize around them:
large batches, long lead times, high inventory, rigid schedules. The
factory becomes a machine that produces what it was set up to produce,
and changing direction is expensive.
When those fixed costs are low — when changeover is measured in
minutes instead of hours — the organization gains the freedom to produce
what the customer actually wants, when they want it, in the quantity
they need. The factory becomes responsive. Agile. Customer-centered.
Shigeo Shingo understood this. SMED was never really about dies or
presses or bolts. It was about removing the economic barrier between
what your factory can do and what your customer needs. Every minute of
changeover time you eliminate is a minute of freedom — freedom to
respond, to adapt, to serve.
Erlau Metalworks discovered this the hard way. After their SMED
implementation, they reduced their average batch size from 8,000 parts
to 600. Their work-in-process inventory dropped by 73%. Their lead time
fell from eighteen days to four. And their on-time delivery — which had
been hovering around 82% for years — climbed to 97.6%.
It started with a stopwatch, a camera, and a question that nobody had
thought to ask: what are we actually doing while the machine is
stopped, and how much of it requires the machine to be stopped?
The answer, at almost every factory, is humbling. And the opportunity
that answer reveals is extraordinary.
Peter Stasko is a Quality Architect with 25+ years
of experience transforming organizations across automotive, aerospace,
and pharmaceutical industries. He has led SMED implementations that
reduced changeover times by over 90% and unlocked batch size reductions
that transformed factory economics from rigid to responsive. His
approach combines Shingo’s systematic method with modern constraint
theory and real-world production realities — because the best changeover
strategy is the one your technicians will actually follow.
