What is SMED: how to reduce changeover time

In one line

SMED (Single-Minute Exchange of Die) reduces the time it takes to change over a product or tooling by separating what can be prepared while the machine is running from what truly requires stopping it — and then attacking each of those two parts separately.

What it is

The method was developed by industrial engineer Shigeo Shingo in the 1950s and refined by him over more than twenty years inside Toyota, as part of the Toyota Production System. "Single-Minute" does not literally mean one minute: the original definition is to reduce the changeover to less than 10 minutes, that is, to a single-digit number.

The distinction that organizes the whole method is this:

• Internal activity: it can only be done with the machine stopped (removing the old mold, mounting the new one).
• External activity: it can be done while the machine is running (gathering tools, preparing the next batch of supplies).

SMED is not limited to "changing a part quickly": the Kaizen Institute makes clear that it covers all the setup activities — adjustments, cleaning, preparation and verification —, not just the physical change of the tooling.

What it is for

A slow product changeover is machine downtime that does not show up as a "failure" in any report, but it weighs just as much on availability. In fact, in the article on OEE this same loss is named explicitly among the Six Big Losses, in the Availability category: "setup and adjustments". SMED is the specific tool for attacking that particular loss.

The benefit goes beyond availability. If changing over a product stops taking hours, you can produce in smaller batches without losing productivity — which means less work-in-process inventory, more flexibility to respond to varied orders, and less capital tied up in stock waiting for its turn in production.

How it is applied: the 5 steps of SMED

The Kaizen Institute organizes the implementation into a starting point (an unoptimized setup, with internal and external activities mixed together without any criterion) and 5 steps:

Step 1 — Study the current situation. You observe and film the complete changeover, from start to finish, where it happens (gemba), with the operator who does it every day. You map the movements with a spaghetti diagram, time each task separately with a stopwatch, classify each one as internal or external, and identify waste: waiting, unnecessary movements, searching, rework.

Step 2 — Separate internal from external activities. This is the step with the greatest impact for the least effort: nothing physical is changed yet, you only reorganize what is done before stopping, what is done with the machine stopped and what is done after restarting. The Kaizen Institute illustrates this with three very different sectors: in healthcare, cleaning and disinfecting the operating room is internal, but preparing the surgical instruments and bringing in the next surgery kit is external; in industry, removing the old mold and installing the new one is internal, but gathering tools and screws or preheating the new mold is external; in banking, activating a new service in the system is internal, but preparing forms and training the staff is external.

Step 3 — Convert internal activities into external ones. You preheat tooling or materials, standardize interfaces so that you do not have to adjust with the machine stopped, assemble kits in advance, and use duplicated or precalibrated parts. A real example documented in the guide: on a pasta production line, a second holder was prepared for the packaging film roll with all the supplies already ready, so that the operator replaces the roll without having to stop the line for so long to prepare it.

Step 4 — Optimize what remains internal. What could not be converted to external is made faster: quick clamps (levers, jaws, pneumatic or hydraulic clamping) instead of screws that have to be loosened one at a time, parallel work with two operators at the same time, precalibrated or duplicated tooling to eliminate manual adjustment, and color codes or positioning guides to mount quickly and with no room for error.

Step 5 — Reduce the external work. Even though it does not stop the machine, external work still consumes time and resources from the support areas: tools and materials are brought closer to the point of use, dedicated cabinets and shadow boards are organized, and 5S practices are applied at the workstations and support areas so that this order is sustained.

Finally, the guide insists on standardizing and training: document the optimized sequence (what is done, when, who does it and with what tool) and leave visual checklists at the workstation, so that the result does not depend on which shift or operator is on duty that day — without this, the time gained tends to be lost again as the months go by.

Real example

A thesis from the Universidad Nacional Abierta y a Distancia (UNAD) in Colombia documents the application of SMED on a soft capsule encapsulation machine, at a pharmaceutical plant in Barranquilla with a capacity of 25,000,000 capsules per month.

Starting situation: the setup for a product changeover had a standard time of 240 minutes (4 hours). With approximately 20 product changeovers per month, that represented 4,800 minutes (80 hours) per month of stopped machine due to setups alone — without counting any failure.

Method applied: the work team filmed the complete setup, timed each activity separately with a stopwatch, made a spaghetti diagram of the operator's movements (dismantling tanks and parts, moving them to the washing area, mounting the new tooling, cleaning, filling in documentation) and put together a "3W" type action plan — what anomaly was detected, how it is solved and who is responsible for closing it.

Result: of the activities that used to be done with the machine stopped, 9 could be converted into external ones (91 minutes in total): mainly transfers and washing of parts — tanks, pumps, baskets, piping — that came to be done while the machine kept producing the previous batch. The internal setup time dropped from 240 to 150 minutes (2.5 hours), a reduction of 37.5%. With the same 20 monthly changeovers, that is equivalent to freeing up about 30 hours of machine capacity per month, without buying any new equipment.

Benefits

• More equipment availability without investing in additional capacity — the encapsulation machine case freed up the equivalent of more than three full production shifts per month just by reorganizing the setup
• It allows smaller production batches without losing productivity, which reduces work-in-process inventory and gives more flexibility for variable orders
• Most of the improvement comes from method and organization (separating and converting activities), not from buying machinery — the implementation cost is usually low compared to the savings
• Less pressure on the operator during the changeover: with the standardized sequence and visual checklists, the temptation to rush steps while the machine is moving is reduced

Limitations to keep in mind

• SMED does not fix a changeover that was badly designed from the start: if the tooling requires complex mechanical disassembly by design, the real gain also requires a redesign of the mold or fixture, not just reorganizing the method
• Most of the gain comes from step 2 (separating internal from external); if the work team does not have the authority to change the existing procedure, the project stalls at the diagnosis and never gets implemented
• As with 5S, the order achieved tends to get mixed up again over time if there is no written standard or visual checklist at the workstation — sustaining it demands the same discipline as step 5 of the 5S
• The encapsulation machine case is the result of a one-off project with monthly follow-up documented in a thesis; not all product changeovers have the same margin for converting internal to external — simpler machines may have less room for improvement than the 37.5% of this case

In summary

SMED attacks product changeover time by first separating what requires a stopped machine from what does not, and then converting and optimizing each part separately — with no need to buy new equipment. The case of the pharmaceutical encapsulation machine in Barranquilla shows it in concrete numbers: from 240 to 150 minutes per setup, 37.5% less, using only video, a stopwatch and reorganization of the work. The connection with OEE is direct: every minute of setup that is eliminated is availability gained without touching any other variable.