SMED for food production: reducing allergen changeover time

Why allergen changeovers are different

Standard SMED methodology, developed by Shigeo Shingo in the 1950s for automotive die changes, assumes that changeover tasks can be reorganised freely as long as the line produces good parts at the end. In food manufacturing with allergen transitions, that assumption breaks down. The sequence of cleaning steps is not a matter of preference — it is a food safety control measure with legal and regulatory weight.

A peanut-to-tree-nut changeover where the cleaning sequence is reorganised incorrectly does not produce a late first good part. It produces an undeclared allergen in a finished product, which is a consumer safety incident and a potential recall. This is why many food plants have historically treated allergen changeovers as fixed-duration events immune to improvement, and why SMED has been applied unevenly across food production lines.

The error in that thinking is conflating the food safety requirements — which are fixed — with the total changeover duration, which contains significant compressible time. Allergen changeovers in food manufacturing typically run 90 to 240 minutes. In most cases, 30 to 50% of that time is recoverable without touching any food safety control.

The internal/external distinction in allergen changeovers

SMED divides changeover tasks into internal tasks, which can only be done while the line is stopped, and external tasks, which can be done while the line is still running. The core SMED move is converting as many internal tasks to external as possible.

For allergen changeovers, the internal/external split looks like this in practice. Internal tasks that cannot be moved include all cleaning and sanitation steps, allergen swabbing and verification, line clearance and physical inspection, first article production and quality release, and any regulatory documentation that requires the line to be in a known state before signing off. These tasks define the minimum changeover duration and cannot be compressed by reorganisation alone.

External tasks that can be prepared before the line stops include staging all cleaning chemicals, tools, and equipment needed for the changeover, pre-warming CIP supply tanks to target temperature, preparing the next product's raw materials and packaging, programming recipe parameters on the HMI, briefing the incoming crew on the changeover plan, and completing all paperwork that does not require line inspection.

In a typical allergen changeover, external task preparation can recover 20 to 40 minutes of line downtime without any change to the cleaning protocol or food safety controls.

Compressing the internal tasks

Once external tasks are maximised, attention turns to the internal tasks themselves. The question is not whether cleaning can be skipped or shortened — it cannot. The question is whether each internal task is being performed at its minimum validated duration, with no wasted time between steps.

The most common sources of recoverable time inside the internal changeover window are waiting time between steps, crew coordination failures, equipment that is not staged correctly before the line stops, and verification procedures that are not designed for speed.

Waiting time between CIP steps is often the largest single source of waste. If the acid wash tank takes 15 minutes to reach temperature after the caustic step completes, that 15 minutes is dead time. The solution is pre-heating the acid tank during the caustic step, so it is ready when the intermediate rinse finishes. This requires no change to the cleaning protocol — it only requires that the CIP skid controls are programmed to pre-heat in parallel rather than sequentially.

Crew coordination failures during changeovers are common and expensive. If the cleaning crew finishes the CIP and then waits for the QA team to arrive for swabbing, that wait time is internal downtime. The fix is a defined hand-off protocol: the cleaning crew completes CIP, immediately notifies QA, and begins the physical line clearance while QA prepares for swabbing. No dead time between steps.

Allergen swabbing and verification: designing for speed

Allergen verification is a fixed internal task, but its duration is not fixed. Many plants run lateral flow immunoassay (LFIA) strip tests at the line, which give results in 5 to 10 minutes per sample. The number of samples, their location, and the decision protocol for borderline results all affect how long verification takes.

A verification protocol designed for speed collects the minimum number of samples at the highest-risk locations — typically the most difficult-to-clean equipment in the allergen contact zone — runs all samples simultaneously, and has a clear decision rule for pass/fail that does not require a supervisor to review each result individually. A verification protocol designed for audit compliance often does the opposite: samples every surface, runs samples serially, and requires sign-off at each stage.

The right protocol design balances speed with risk control. For most allergen transitions, three to five samples at defined high-risk locations, run simultaneously, with results reviewed and signed off in a single step, is both faster and more defensible than a lengthy protocol that covers low-risk surfaces exhaustively.

Standardisation across shifts

Research on manufacturing changeovers consistently finds that shift-to-shift variation in changeover duration is often larger than the gap between the average and the target. A changeover that averages 120 minutes may range from 90 to 180 minutes depending on which crew performs it. The 90-minute example already exists — the problem is that it is not the standard.

Standardising allergen changeovers means documenting the fastest validated sequence in enough detail that any trained operator can replicate it, training all shifts to that standard, and tracking changeover duration by shift to identify where variation occurs. In food manufacturing, this documentation often already exists in the form of the allergen cleaning procedure — the gap is that the procedure describes what to do, not how to do it at minimum time.

A changeover work instruction that includes the sequence of tasks, the time target for each task, the handoff point between crews, and the pre-start checklist of external tasks gives operators the information they need to hit the target consistently. Without that level of detail, changeover duration depends on individual knowledge and habit, and improvement is not sustainable.

What SMED cannot compress

It is worth being explicit about the limits. The validated cleaning contact times — caustic wash duration, acid wash duration, sanitizer contact time — are not compressible by SMED. They are set by cleaning validation data and regulatory requirements. Reducing them without revalidation is a food safety violation, not a lean improvement.

Similarly, the allergen verification steps, the line clearance sign-off, and the first article quality release are not compressible below their minimum validated duration. SMED works around these fixed elements, not through them.

The realistic target for a well-executed SMED programme on allergen changeovers is a 20 to 35% reduction in total changeover duration, achieved entirely through external task conversion, dead time elimination, and standardisation. That is typically 20 to 60 minutes on a 90 to 180 minute changeover — significant production recovery without compromising a single food safety control.