Setting Up a Kanban Pull System Between Cutting and Sewing

Cutting rooms push. Sewing floors pull. When the two are connected by a production schedule rather than by a signal, the result is a bundle store that grows all week and a sewing floor that is simultaneously starved of the style it needs.

Why push fails between these two departments

Cutting is a batch process with a strong economic incentive to cut large lays. Sewing is a flow process that consumes bundles at a rate set by line efficiency and style change. The two rates are structurally different, so scheduling both against the same plan guarantees a mismatch that shows up as work-in-progress.

Calculating the number of kanban cards

Number of cards = (D x L x (1 + S)) / C

  D = average demand per period (bundles/hour)
  L = replenishment lead time (hours)
  S = safety factor (0.1 - 0.3)
  C = container size (bundles per trolley)

For a line consuming 14 bundles per hour, a cutting replenishment lead time of 3 hours, safety factor 0.2, and a trolley holding 12 bundles: (14 x 3 x 1.2) / 12 = 4.2, so five cards. Total WIP between the departments is then capped at 60 bundles, by design, rather than by whatever the schedule produced.

The first three weeks

Three failures showed up consistently during implementation and are worth anticipating.

• Cards going missing. Within a week two cards had been used as note paper. Laminate them, number them, and audit the count at every shift change.
• Cutting overproducing anyway. The cutting supervisor was still measured on pieces cut per day, so the system fought the metric. Change the metric to on-time replenishment against cards before changing the process.
• Style changeovers breaking the loop. A pull system assumes reasonably stable consumption. On the first style change the cards were all in the wrong place. The fix was a defined changeover procedure that pulls all cards back and reissues them for the new style.

Results after eight weeks

Note the last row. Cutting output fell slightly, which is the correct outcome and the hardest one to defend internally: the cutting room was previously producing inventory, not product.

What I would do differently

I would start with a single line rather than the whole floor, and I would change the cutting room metric before installing a single card. The technical part of a pull system takes a week. The measurement and incentive part is the whole project.