The Team That Fixed the Wrong Line
A site runs two assembly cells, A and B, making the same product family. Cell A produces 900 units a day with 4% rework and a 15-minute average wait between stations. Cell B produces 700 units a day with 1% rework and almost no waiting, but operators walk an average of 6 km a shift fetching components. A Yellow Belt is asked to lead a two-week kaizen on Cell A because 'its numbers look worst'. He notes that total demand is 1,500 units a day, Cell A's rework consumes two full-time reworkers, and Cell B's output is capped by the walking. The site manager wants the kaizen to focus on Cell A's rework.
1. Using Lean flow and waste logic, assess whether Cell A is the right target and what the system-level constraint is.
Read the worked answer
Cell A looks worst on rework, but the system constraint is total output against 1,500/day demand. Current output is 900 + 700 = 1,600, only 100 above demand - thin cover, and Cell A's 4% rework means its good output is about 864. Effective good output is roughly 864 + 693 = 1,557, barely above demand.
Cell A's waiting (15 minutes between stations) signals a flow problem - imbalance or batch build-up - and its rework consumes two full-time people, which is a large, visible cost. Cell B's walking is motion waste that caps its throughput: it is the reason B cannot simply absorb more.
So both cells have real waste, but the question is which one limits the system. If demand rises or A's rework spikes, the site misses. The kaizen should treat A as the priority only if A's rework and waiting are the binding constraint; the Belt should verify by checking where work-in-progress accumulates and which cell has spare capacity. Choosing A because 'its numbers look worst' is targeting a symptom.
2. The site manager insists the kaizen focus only on rework. How should the Belt respond, and what would a two-week kaizen realistically deliver?
Read the worked answer
The Belt should accept the rework focus as the entry point but widen the scope with evidence, not argue from principle. A practical response: 'Rework is a good start - two full-time reworkers is a real cost. Can we spend the first two days mapping where the rework is generated and how the 15-minute waits relate to it? If they are linked, fixing rework alone may not hold.' That respects the sponsor's priority while keeping the flow question open.
Realistic two-week deliverable: a mapped current state for Cell A, the top two or three rework causes verified with data, one or two countermeasures trialled (e.g. standard work at the station generating most defects, a small buffer change to cut the 15-minute wait), and a documented new standard with an owner. Expect a measurable reduction in rework on the targeted defect, not a transformation of the cell.
What to avoid: promising Cell B's walking problem will be solved in the same two weeks, or claiming a system-level fix from a single-cell kaizen. The Belt should log Cell B's motion waste as a separate, evidenced opportunity for the next cycle.