How to Cut Changeover Time in Half With SMED
SMED sounds like a black belt project. It is mostly a camera and a checklist. Here is the internal-vs-external split and a worked example that found 30 percent fast.
I have watched a press operator spend forty-five minutes changing over a die, and for about thirty of those minutes the press was sitting stone cold while he walked to the tool crib, hunted for the right wrench, realized the next die was still on a rack across the building, found a forklift, waited for the forklift, and finally got back to the machine. The actual work of swapping the die took maybe fifteen minutes. The other thirty was a scavenger hunt with the machine held hostage.
That gap is the entire opportunity in changeover reduction, and you do not need a black belt or a kaizen consultant to grab it. SMED, which stands for single-minute exchange of die, sounds like a methodology you have to be certified to run. It is not. At its core it is a camera and one question asked over and over: did the machine have to be stopped for this step, or could it have been done while the machine was still running? Answer that honestly for every step of a changeover and you will find your first 30 percent without spending a dollar.
SMED Is One Question: Internal or External?
Strip away the lean vocabulary and SMED rests on a single distinction.
- Internal setup is work that can only be done while the machine is stopped. Actually unbolting the old die and bolting in the new one. The machine has to be off for this.
- External setup is work that can be done while the machine is still running the previous job. Staging the next die at the machine. Gathering the tools. Pre-heating the mold. Filling out the paperwork. Finding the forklift.
Here is the whole insight: in most shops, half or more of what happens during a changeover is external work being done internally. The operator is doing setup tasks with the machine off that he could have done with the machine still running. Every one of those minutes is pure waste, and you reclaim it just by moving the task earlier in time. You do not change the work. You change when the work happens.
That is the easy 30 percent, and it usually costs nothing but attention. The harder gains come later, from making internal steps faster and eliminating adjustment. But you start with the sort: internal or external. Sort the changeover and the waste announces itself.
Step 1: Video the Changeover
Do not try to analyze a changeover from memory or from the operator’s description. Video it. Phone on a tripod, or just propped on a toolbox, recording the whole changeover from last good part to first good part of the next job.
The camera is the most important tool in SMED, and it is the one people skip because it feels like overkill. It is not. When you watch the playback, you see things nobody remembers in the moment: the three trips to the tool crib, the ninety seconds of standing and waiting, the hunt for the right fastener, the operator reading the setup sheet because the steps are not memorized. People cannot narrate their own waste accurately. The camera does not lie.
Watch the video with the operator, not at them. They are not the problem, the layout and the process are. Watch it once straight through to feel the rhythm. Then watch it again with a stopwatch and a pen, and log every step.
Step 2: Sort Every Step Internal or External
Take your logged steps and put each one in a bucket. Here is the worksheet, with a press die change filled in so you can see the shape of it.
CHANGEOVER STUDY: Press die change, Part A -> Part B
Baseline (last good part to first good part): 45:00
STEP TIME I/E CAN MOVE TO EXTERNAL?
----------------------------------------------------------------------------
Walk to crib, get wrench set 3:30 INT YES - stage at machine
Find/stage next die (forklift hunt) 9:00 INT YES - stage before stop
Unbolt old die 6:00 INT no (true internal)
Remove old die w/ hoist 4:00 INT no
Clean die area 2:00 INT YES - some pre-clean
Set & bolt new die 7:00 INT no (speed up later)
Hunt for correct shim pack 2:30 INT YES - kit it
Adjust shut height (trial & error) 6:00 INT partial - set stop later
First-piece check & sign 4:30 INT YES - prep paperwork
----------------------------------------------------------------------------
TOTAL INTERNAL (machine stopped): 45:00
MOVED TO EXTERNAL (done while running): 3:30 + 9:00 + 2:30 + ~1:00 = 16:00
NEW INTERNAL TIME (machine stopped): 45:00 - 16:00 = 29:00
FIRST-PASS REDUCTION: 45 -> 29 min = 36% faster, $0 spentRead what happened. Just by moving four steps to before the machine stops (staging the wrench kit, pre-staging the die, kitting the shim pack, prepping the paperwork), the machine-down time dropped from 45 minutes to 29. Thirty-six percent, no capital, no new equipment. The work did not get faster. It got moved.
Step 3: Attack What Is Left
Once you have moved everything you can to external, the remaining internal time is the real target. Now you make those true-internal steps faster and more repeatable. A few standard moves, in rough order of payoff:
- Kit everything. Build a changeover cart for each job or family: every tool, fastener, shim, and gauge that changeover needs, staged together. No more trips to the crib. A $200 cart and a shadow board can save more minutes than a $40,000 machine option.
- Kill the adjustment. Trial-and-error adjustment (the “set shut height by feel, run a part, measure, repeat” loop) is the biggest hidden time sink in internal setup. Replace it with a stop, a setting block, a mark, a torque value. Make the right setting the only setting. An adjustment you make by hand once is fine. An adjustment you hunt for by trial and error every changeover is a project.
- Standardize the bolts. If your changeover uses four different fastener sizes, it needs four different tools and four chances to grab the wrong one. Standardize fasteners and you cut tool changes and errors at once.
- Quick clamps over threaded bolts. Where it is safe and the load allows, a clamp that throws in one motion beats a bolt you run down six turns. This is real money over hundreds of changeovers.
- Write the new standard. A faster changeover that lives in one operator’s head is not a gain, it is a liability. The minute you find a better sequence, it becomes a standard work instruction with the kit list, the settings, and the sequence on one photo-first page. Otherwise the next operator reinvents the scavenger hunt.
That last point is where shops leak their gains. They run a great SMED event, cut the changeover in half, and never write it down. Six months later it has crept back to 40 minutes because the new hire never learned the new way. The improvement has to be captured in the standard or it evaporates.
Why This Is Worth Your Week
Changeover time is not just lost machine time. It is the hidden reason your batches are too big. When a changeover costs 45 minutes, the operator and the scheduler quite rationally avoid changing over, so they run giant lots to amortize the setup. Giant lots mean more inventory, longer lead times, and more cash tied up in work-in-process sitting on shelves. Cut the changeover and small lots suddenly become affordable, which is exactly the queue problem that shows up on a value stream map. Setup reduction is often the lever that unlocks everything downstream of it.
So the 30 percent you find is not even the whole prize. The bigger prize is that cheap changeovers let you run the smaller batches that shrink your lead time.
Keep the Gain From Slipping Back
A short discipline so a SMED win sticks:
- Re-video after the change. The first video proves the waste. The second proves the gain and catches the steps creeping back internal.
- Track changeover time on every change, on the production tracking sheet. A number nobody watches drifts back up. A logged number stays honest.
- The kit cart is part of the standard, not a nice-to-have. If the cart is incomplete, the scavenger hunt returns.
- Eliminate adjustment before you try to speed up bolting. Adjustment hides more time than fastening.
- New best time becomes the new standard the same day. Undocumented gains evaporate.
Start tomorrow with the cheapest possible version: video one changeover on your worst machine, watch it with the operator, and circle every step that could have been done while the machine was still running. Just doing that sort, with no other change, usually surfaces a fast 20 to 30 percent. Move those steps to external next time and you have your first win, for the price of a phone video.
If you want the changeover study worksheet, the internal-versus-external sort sheet, and a kit-cart checklist as clean, printable pages, they are part of my systems series, where I cover the same camera-first approach across setup reduction, standard work, and flow. The worksheet will not change the die for you. It just turns your changeover into a list of steps you can move, shorten, or kill.
Video it. Sort it. Move every external step out of the machine-down window. That is the half you can cut this month.
Liked this? Get the free SOP Writing Cheat Sheet.
The 6-step SOP format on one page. Free.