4-Hour Changeover Trimmed to 40 Minutes by a SMED Method on a Bosch Packaging Line

A Bosch SVE 2520 flow-wrapper changeover fell from 240 minutes to 40 after a single-minute-exchange-of-die project separated stopped-machine work from staging work. The line runs three product formats per shift, and the main recovery came from moving 14 of 22 setup tasks off the critical path.

4-Hour Changeover Trimmed to 40 Minutes by a SMED Method on a Bosch Packaging Line

Twenty-two tasks on film

The project began with one complete changeover on a Bosch SVE 2520 horizontal flow-wrapper, recorded from the last run of chocolate bars through the first verified biscuit sleeves. Two technicians and a line operator stayed in view across the full 240-minute stop. During playback, the team broke the work into 22 separate tasks and logged each task against the point where the last good pack of format A left the outfeed and the first good pack of format B was accepted.

The largest block was reel splicing and film tension setting around the Bosch forming shoulder. Fetching the correct forming tube also absorbed a noticeable part of the stoppage because the tool crib sat 70 metres from the line and the format B tube was stored on an unlabelled shelf. Cutter jaw temperature stabilisation created another idle stretch, with the crew waiting at the machine while the jaws reached operating temperature.

That waiting time became an early target. Once a controller can bring a component to temperature without a person standing beside it, the heating period can be placed outside the stopped-machine window. The same logic applied to the tube, the reel, and the basic kit of changeover parts.

Internal work, external work

SMED, formalised by Shigeo Shingo at Toyota and documented in A Revolution in Manufacturing, uses a simple split. Internal tasks require the machine to be stopped. External tasks can be completed during the previous order, after restart, or away from the line while production continues.

On the SVE 2520, the 22 tasks divided into 8 genuinely internal jobs and 14 that had been pulled into the stoppage through habit. The forming tube for format B could be brought to the line during the final part of the format A run. A second, dedicated set of sealing jaws could be pre-heated on a bench-mounted controller wired into the same motor control centre. The format B film reel could be mounted on a spare mandrel, torqued, and tension-checked before the wrapper stopped.

Moving those tasks changed the shape of the changeover before any technician worked faster. The critical path lost waiting, walking, and searching. Mechanical adjustment still remained, but the machine no longer carried every preparation task inside the stop.

The jaw-swap proposal drew resistance at first. A second jaw set tied up money and would sit idle for much of the week. The plant accepted the purchase after comparing repeated downtime on three daily changeovers with the costed value of line-hours recovered by pre-heating.

The tube was a storage problem

The forming tube delay pointed straight at storage discipline. The crib was reorganised with shadow boards and QR-coded shelf locations, and each Bosch forming tube received a labelled position. The changeover instruction now names the exact location, so retrieval is handled as external staging before the machine stops.

This change also removed a small ambiguity from the job. A technician no longer had to know where a rarely used format part had drifted after the previous run; the storage system carried that information.

Clamps, reels, and parallel hands

The forming shoulder on the SVE 2520 had been held by six M8 hex bolts. Removing and refitting them with a socket took several minutes, and cross-threading had caused two scrapped shoulders in the preceding year. Shingo’s functional-clamp principle says a bolt clamps on its last turn; the earlier turns add motion without adding clamping force.

The six bolts were replaced by two quick-release toggle clamps and a pair of pear-shaped slotted holes. The shoulder drops onto locating dowels and the clamps snap shut. Fitting became a short, repeatable motion, and the dowels removed the small position-finding routine that had previously depended on feel.

Repeatable positioning mattered as much as speed. Film tracking on the sealed pack depends on shoulder alignment within a fraction of a millimetre. With the dowel scheme, the shoulder seats in the same place each time, so the technician is no longer judging final location by eye.

The film reel mandrel received the same treatment. The old M10 centre bolt was replaced with a cam-lock hub, turning the reel change into a lock-and-release action. Standardising the fasteners on the changeover path to two clamp types also reduced the tool load at the line.

Before the change, the crew carried a socket set, a screwdriver, and two Allen keys for different parts of the sequence. After the conversion, one tool covered the remaining adjustment work. The tool-hunting motion visible in the original video largely disappeared because the standard no longer sent technicians back and forth for small items.

Parallel work supplied the last major improvement. With external staging in place, one technician prepared the outfeed and cartoner interface while the other worked at the forming section. The original video showed both people waiting on the same serial tasks at several points; the new dual-column instruction assigns separate work streams and posts them side by side at the line.

The numbered instruction matters because parallel work can collapse without a visible standard. Each person can see the current task, the next task, and the handoff point. The line operator also has a defined role in staging and confirmation, so the technicians are not using stop time to ask which parts are ready.

Where the time moved

The stopped window fell from 240 minutes to 40 minutes because activity moved out of the stop, because fixtures became quicker to locate, and because the two technicians stopped queuing behind each other. The 14 external tasks did not disappear. They shifted into the running window before the last format A packs and into the period after the first good format B packs were verified.

The big movement was easy to see on the recording. Pre-heated jaws removed the idle wait at the wrapper. The staged forming tube removed the walk to the crib. A pre-kitted reel on a spare mandrel reduced live splicing and tension work. Quick clamps cut wrenching and helped hold alignment. The dual-column standard took serial waiting out of a two-person job.

No servo speeds changed, and no operator was asked to rush. The machine did the same basic work with a shorter stoppage because the stop was reserved for tasks that truly needed access to a non-running wrapper.

Holding the 40-minute standard

A SMED gain can fade when the checklist becomes optional. Four weeks after implementation, the plant filmed another changeover and confirmed the 40-minute figure, while also finding that two tasks had slipped back onto the critical path because the pre-heat controller was being started late. The fix was a checklist trigger tied to the pack counter: at 500 packs remaining on format A, the display prompts the operator to begin external staging.

The dual-column work instruction is laminated at the line and revision-controlled in the plant’s document system. New operators run their first three changeovers with a trained technician against the timed standard. Where the plant had accepted 240 minutes as the cost of variety, it now schedules four short runs a day where it once ran two long ones, reducing work-in-progress inventory of semi-finished product between the wrapper and the cartoner.

The project left an open operational question: how much of the remaining changeover cost sits outside the wrapper, in upstream and downstream equipment still planned around the old assumption?