In non-regulated manufacturing, SMED often starts with video analysis at the changeover. On regulated lines, that isn't enough. The visible stoppage is only part of the changeover. Cleaning, documentation, sign-offs, material status, ambient conditions, line clearance and possible revalidation determine the real changeover.
A practitioner who only optimises the mechanical tasks can save time in the short term and create compliance risk in the long term.
The right question
Not: how do we make the changeover faster?
Better: which tasks are genuinely required for safety, quality and compliance, which can be prepared in advance, which are waiting needlessly, and which standards prevent variation?
Bring QA in early
QA shouldn't only check at the end whether the new sequence is acceptable. They belong in the analysis. Which documents are affected? Which change needs change control? Which tests or validations are required? What risks appear if tasks are parallelised or moved?
Variation often matters more than best time
Many lines have an impressive best time and a poor average. In regulated environments, variation is dangerous, because it destabilises planning, sign-offs and shift handovers. SMED should therefore improve repeatability, not just target time.
A practical start
Pick a changeover with high volume or high pain. Observe the entire end-to-end changeover, not just equipment downtime. Separate the mechanical, documentary, quality-related and logistical tasks. Then build the new standard with production, QA, engineering and planning.
The takeaway
SMED works on regulated lines. But it has to be understood as process- and quality-system work, not as speeding things up on the shop floor.