A white goods maker moulded more than 200 components from a monthly forecast, held large stocks and still ran short. A kanban pull system between moulding and assembly cut component inventory by 52%, stock-outs from nine to two a month and storage space by half.
Key numbers
A white goods manufacturer had a warehouse full of moulded components and an assembly line that still stopped for missing parts. Both problems had the same cause: moulding was scheduled from a monthly forecast. When the forecast was wrong, which it usually was, the plant made too much of some parts and too little of others.
The situation
The plant moulded more than 200 different components for its assembly lines. Component inventory was worth around $1.2 million and occupied 620 m² of storage. Despite that, the assembly line averaged nine stock-outs a month, each causing delays, expediting and overtime.
What the data showed
An analysis of demand showed a classic split. A relatively small number of high-running components accounted for most of the volume and were used every day. Many low-runners were used occasionally. Both groups were being scheduled the same way, from the same forecast, in large batches driven by long changeovers.
The value stream map showed the inventory piling up between moulding and assembly, with no link between what assembly actually used and what moulding made next.
What changed
- Components split into high-runners, managed by kanban, and low-runners, made to order.
- Kanban loops sized from actual demand and replenishment lead times, including realistic changeover times. Changeover reduction work using SMED made smaller batches practical.
- A supermarket created next to assembly, holding a defined quantity of each high-runner.
- Moulding scheduled by kanban signals rather than the forecast: when assembly used a container, the card went back to moulding as the signal to replace it.
- Kanban quantities reviewed monthly and reduced as reliability improved.
Results
Component inventory fell from $1.2 million to $0.58 million over seven months, a 52% reduction. Stock-outs dropped from nine a month to two, because moulding was now making what assembly actually used. Storage space halved, from 620 to 310 m².
Why it worked
Kanban connected moulding directly to real consumption, removing the dependence on an inaccurate forecast. Separating high-runners from low-runners meant each was managed in the way that suited it. And shorter changeovers made the smaller, more frequent batches economic.
Lessons for your operation
- High inventory and frequent shortages usually point to forecast-driven push scheduling.
- Not every part suits kanban. Start with high-running, stable-demand items.
- Size kanbans from real demand and real lead times, then reduce them over time.
- Changeover reduction and kanban work best together.
