What Is SMED (Single-Minute Exchange of Die)
SMED, or Single-Minute Exchange of Die, is a methodology for reducing the time it takes to switch a production line from one product to another. The name comes from the original goal: die changes on stamping presses completed in single-digit minutes, meaning under 10. The methodology was developed by Shigeo Shingo at Toyota between 1950 and 1969 and is now a core tool in lean manufacturing worldwide.
On a packaging line, changeover means the entire sequence from last good bag of product A to first good bag of product B. This includes stopping the line, cleaning the hopper and contact parts, swapping film rolls, swapping or adjusting seal jaws, changing the forming collar or tube, swapping the auger or filler tooling, reprogramming the HMI, re-threading the new film, running trial bags, and adjusting until the seal, weight, and cut-off meet spec.
Without SMED, packaging line changeovers routinely run 45 to 120 minutes. With SMED, the same changeovers run 10 to 25 minutes. The 30-70% reduction range in the title is conservative and well-documented across hundreds of implementations.
Why changeover matters: every minute a line is down for changeover is a minute of lost production. A line running 40 bags per minute loses 40 bags per changeover minute. At a contribution margin of $0.30 per bag, every changeover minute costs $12. A 60-minute changeover run twice per day costs $1,440 per day, or $360,000 per year on a single line. SMED is the cheapest way to recover that capacity without buying another line.
For the broader production-line picture, see the Production Line pillar. This article covers SMED specifically. For integrated line operations see filling, sealing, and labeling integration, and for downstream automation see end-of-line automation.
The 4 SMED Principles
SMED rests on four principles. Understand them before applying the steps because the principles guide every decision.
Principle 1: Separate internal from external work. Internal work requires the machine stopped. External work can be done while the machine is still running. Most changeovers in non-SMED plants are 40-60% external work being done as internal work because nobody prepared in advance.
Principle 2: Convert internal to external wherever possible. Once you separate, look at every internal task and ask whether it can be moved external. Pre-heating seal jaws, pre-staging film rolls, pre-assembling tooling, pre-mixing cleaning solution, pre-filling the new product hopper on a cart. Most internal tasks can be moved external with thought.
Principle 3: Streamline remaining internal work. What cannot be converted must be made faster. Use quick-change fasteners (hand knobs, cam locks, spring-loaded pins) instead of bolts. Standardize height and position so the same setup works across products. Use parallel operations where two or three operators work on the line simultaneously instead of one operator doing everything in series.
Principle 4: Streamline external work. External work also takes time and can bottleneck. Stage carts with everything pre-staged. Use kanban to ensure consumables (film, glue, labels) are always available. A line that has fast internal work but slow external work is not actually fast because external delays push the next changeover start.
Internal vs External Work: The Key Distinction
The internal vs external distinction is the entire engine of SMED. Without it, you cannot make meaningful progress.
Internal work examples on a VFFS line: removing the old seal jaw assembly and installing the new one. Swapping the forming collar. Adjusting the film spindle width. Re-threading the new film through the forming tube. Cleaning the auger filler between flavors (food safety requirement). Running trial bags and tuning fill weight.
External work examples on the same line: rolling the new film roll to the machine on a cart. Pre-filling the cleaning solution bucket. Pre-assembling the new seal jaws on a spare change plate. Pre-heating the new jaws on a benchtop warmer. Pulling the next product batch from the warehouse and staging at the line. Pulling up the next recipe in the HMI preview screen.
A typical pre-SMED changeover has the operator stop the line, then walk to the warehouse to get the film. That is 8 minutes of internal time spent on a task that is entirely external. The SMED fix is trivial: stage the film roll at the line before stopping.
Audit your current changeover by video recording one full changeover, then categorizing every minute as internal or external. Most plants find 30-50% of their changeover time is actually external work being done internally. This is the lowest-hanging fruit and the source of the 30-40% reductions in the first month of a SMED program.
Step-by-Step SMED Implementation
Implementation follows a six-step sequence. Steps 1-3 are analysis. Steps 4-6 are execution.
Step 1: Video a current changeover. Set up a phone on a tripod and record the entire changeover from last good bag of product A to first good bag of product B. Time-stamp everything. Most plants have never actually watched their own changeover in detail.
Step 2: Build a task list with times. From the video, list every task and its duration. Example: walk to warehouse 8 min, retrieve film 3 min, clean hopper 12 min, swap seal jaws 9 min, swap forming collar 6 min, re-thread film 5 min, swap auger 7 min, program HMI 4 min, trial bags 10 min, fine tune 8 min. Total 72 minutes.
Step 3: Classify each task as internal or external. Mark each task as I (internal, machine stopped) or E (external, can be done while running). In our example, walk to warehouse is E, retrieve film is E, clean hopper is I, swap seal jaws is I, swap forming collar is I, re-thread film is I, swap auger is I, program HMI is I (preview recipe), trial bags is I, fine tune is I.
Step 4: Convert external work being done internally. Walk to warehouse and retrieve film: 11 minutes that can be done external. Savings 11 minutes. Recipe programming: do this in HMI preview mode while line still running. Savings 4 minutes. New changeover: 72 - 15 = 57 minutes.
Step 5: Streamline remaining internal work. Quick-change seal jaw plate (hand knobs instead of 8 bolts): reduces 9 minutes to 3. Pre-warmed jaws (no warm-up time): saves 0 in our case but saves 4-5 minutes on lines where jaw warm-up is in the sequence. Quick-change forming collar (single lever): reduces 6 minutes to 2. Pre-staged cleaning cart with all tools and chemicals: reduces 12 minutes to 7. New changeover: 57 - 4 - 4 - 5 = 44 minutes.
Step 6: Use parallel operations. Two operators now: operator 1 cleans hopper and swaps auger (14 minutes parallel), operator 2 swaps jaws and forming collar and re-threads film (10 minutes parallel). Trial bags and fine tune: 18 minutes. New changeover: 14 + 10 + 18 = 42 minutes... but with parallel work it is max(14,10) + 18 = 32 minutes. Then trial bag and fine tune reduce with recipe tuning: 32 - 10 = 22 minutes.
Final changeover: 22 minutes from a starting 72 minutes. That is a 70% reduction.
Worked Example: 45-Minute Changeover Reduced to 12
A real-world example from a snack food plant running potato chips in three flavors on a single VFFS line. Original changeover between flavors: 45 minutes, four changeovers per day. Video audit revealed operators were skipping cleaning to hit 45 minutes, causing cross-contamination and consumer complaints.
Five SMED rounds brought the true changeover (cleaning included) from 55 minutes to 12.
Round 1 (convert external): film staged at line before stopping. Recipe loaded in HMI preview. Saved 8 minutes. New time: 47 minutes.
Round 2 (streamline internal): quick-change seal jaw plate with hand knobs replaced eight bolts, reducing jaw swap from 10 to 3 minutes. Quick-change forming collar with single lever, 8 to 2 minutes. Pre-staged cleaning cart, 14 to 9 minutes. New time: 29 minutes.
Round 3 (parallel operations): added a second operator during changeover. Operator 1 cleans hopper and auger while operator 2 swaps jaws, collar, and re-threads film. Sequential time compressed to max(9, 10) plus trial tuning. New time: 22 minutes.
Round 4 (recipe discipline): each flavor has its own recipe with auger rotation, jaw temperature, dwell, film pull, and registration stored in HMI. Trial bags dropped from 15 to 4, tuning from 12 to 4 minutes. New time: 14 minutes.
Round 5 (further conversion): swappable auger cartridge per flavor eliminated cleaning between flavors. Swap is 2 minutes instead of 9 minutes cleaning. New time: 12 minutes.
Result: 45 minutes reduced to 12 minutes, a 73% reduction. The line recovered 132 minutes per day of capacity, equivalent to 5,280 additional bags per day at 40 BPM. At $0.40 contribution margin per bag, that is $2,112 per day or $660,000 annually. Total SMED investment: $18,000 in tooling and cleaning cart. Payback: under two weeks.
Common Pitfalls
SMED implementations fail in predictable ways. Here are the six we see most often.
Pitfall 1: Skipping the video audit. Teams think they know their changeover. They do not. Memory compresses and smooths over delays. The video is the only objective record. Skip it and you optimize based on myth.
Pitfall 2: Buying hardware before fixing process. Plants spend $30,000 on quick-change jaws without first staging film rolls at the line. Process fixes are free or cheap and often deliver 30-40% of the gain. Hardware fixes deliver the next 30-40%. Do them in order.
Pitfall 3: Single-operator optimization only. Most SMED analysis assumes one operator. Adding a second operator for the changeover window is often the highest-leverage move and the cheapest. The cost is 15-30 minutes of additional labor per changeover. The benefit is parallel work that compresses the sequence.
Pitfall 4: Ignoring the trial-and-tune phase. Trial bags and tuning are often 25-40% of changeover time. Recipe discipline, accurate recipes, and a formal first-article inspection process are the cure. Without recipe discipline, you cannot get below 20 minutes on a typical packaging line.
Pitfall 5: No standardized changeover sequence. Each operator does it differently. Cycle time varies from 25 to 70 minutes depending on shift. Standardize on one sequence, document on a visual job aid posted at the machine, and audit compliance weekly.
Pitfall 6: No measurement after initial improvement. The team celebrates the win, then stops tracking changeover time. Within 6 months, changeover creeps back up because shortcuts accumulate. Track changeover time weekly, review monthly, and address regression immediately.
Sustaining Gains
Sustaining SMED gains requires three institutional commitments. Without all three, regression is inevitable.
Standard work. Every changeover has a written, visual standard work document posted at the machine. Photos of each step. Expected time for each step. Operator name and sign-off. The document is updated whenever the changeover sequence changes. Standard work is the single most important sustainment mechanism.
Daily and weekly measurement. Changeover time is recorded for every changeover event. The data goes into a trend chart visible to operators, supervisors, and managers. Regression is visible within a week, not after a quarter.
Monthly review and quarterly kaizen. Once per month, the site leadership reviews changeover trends across all lines. Once per quarter, the team runs a half-day kaizen to attack the worst-performing changeover. Continuous improvement is not optional. Without it, the next improvement cycle never starts.
The plants that hold SMED gains for 5+ years do all three. The plants that lose them within 18 months typically do only the initial workshop and skip the institutional sustainment.
SMED compounds with other lean tools. 5S reduces search time during changeover. Total Productive Maintenance ensures the machine does not break during the next run. Kanban ensures materials are staged. Visual management surfaces problems immediately. The compounding effect is why lean plants run changeovers under 15 minutes as standard while non-lean plants run 60+ minutes.
If you take one thing from this article: video your next changeover. The rest follows.