TOC in Operations – Advanced · Lesson 6 of 12
Decoupling the feeder shops
Choking the release flushed out the work in process, and then the feeder shops drowned in set-ups. Stock buffers decouple them, until a long disruption breaks the decoupling and the shop is left with two rules. Use the release dates you already know: a coverage report batches the backlog with the near-future releases, under one rule.
The situation
A material-handling equipment maker builds a special-purpose truck under contract for an overseas brand. Demand is strong: the line is overbooked, with a backlog of two months. The flow is gas cutting, CNC milling and drilling, fabrication and chassis assembly, painting, final assembly, then inspection and dispatch. Before TOC it was scheduled monthly, with an internal lead time of four weeks.
The standard solution went in: work in process quarantined, material released by the rope, focus on completing orders, existing work in process flushed out. The production buffer was cut by half, to two weeks.
Then a new problem appeared. The line makes about one truck a day. Released order by order, every order is a set-up at gas cutting and at the CNC machines, every day, for every part. The feeder shops, which had looked comfortable, ran out of capacity.
First extension: stock buffers
The natural step: decouple the pre-fabrication shops. Keep a stock buffer of ready-for-fabrication components between the CNC and fabrication. Fabrication onwards is make-to-order with a two-week time buffer; gas cutting and CNC work to replenish the stock, in economic batches.
It worked, until a disruption at gas cutting or CNC lasted long. Stock ran out; the decoupling broke; the MTO buffer report showed a growing backlog of orders waiting for parts. The feeder shops switched to making what the backlog needed, part by part, and set-ups exploded again. Someone proposed a fix: when making for the backlog, add the stock buffer quantity to the batch.
The feeder managers objected, and not without reason.
Coverage instead of stock
The answer is to decouple with information instead of stock: a coverage report.
- For each feeder item, list what the orders already in the plant need and haven’t got, and what the orders due for release in the next few days will need.
- Negative coverage: quantity for orders that are past their release date and still waiting at pre-fabrication. This is the backlog.
- Positive coverage: quantity for orders whose release date hasn’t come yet.
- Batch = backlog + future releases within a restricted window.
Set-ups fall, because each batch covers several days of orders. Capacity stealing doesn’t return, because nothing is made for an order outside the window. And the feeder shops have one rule, whether they are ahead or behind.
Pause and think: why not make the window as wide as possible, say a month, and save even more set-ups?
- Parts for orders a month away take capacity from parts needed now: the stealing the window prevents
- Because the report can’t look further than the orders due for release in the next few days of the plan
- Because orders a month away can still be cancelled by sales
The window trades set-ups against stealing. A short, fixed window keeps the trade under control; a wide one quietly reintroduces the old monthly batching.
The negative branch: shared feeders
Gas cutting and CNC also fed the company’s other assembly lines, and those lines worked on stock buffers. Now the feeders served two masters with two signals: coverage for one line, stock for the others. Which goes first? And the overbooked line could tolerate no loss of output at fabrication.
The injection: soft-dedicate machines in each feeder department to that line. Not a physical rebuild, just a standing assignment, so the priority question doesn’t arise every hour.
The rule
In make-to-order plants whose feeder shops have set-up problems:
| Decouple with | |
|---|---|
| Release dates of future orders are known | A coverage report: backlog + restricted future releases |
| Future orders are not known | Stock buffers |
In both cases, keep one rule for the feeder shop, and soft-dedicate machines where several lines share it.
Check your understanding
Answer all five. Get four right to complete the lesson.
1. After the work in process was flushed out, the feeder shops ran short of capacity because…
- The flush-out took too long
- At one truck a day, releasing order by order meant a set-up for every part, every day
- The CNC machines had grown slower
- Fabrication had become the constraint once the work in process was flushed out
Small release quantities turned set-ups into the feeder shops’ main load.
2. Stock buffers for the feeders failed when…
- Fabrication released too many orders in a single week
- The parts were too heavy to store
- The stock was counted wrongly
- A long disruption emptied them, and set-ups returned
Then the shops had two rules and set-ups returned.
3. Negative coverage is…
- Quantity for orders past their release date still waiting at pre-fabrication
- Stock above the norm
- Parts made for orders that sales has not yet booked or confirmed with the customer
- Rejected parts
It is the backlog; positive coverage is for releases still to come.
4. A coverage report can be used only when…
- The feeder shop has no set-ups at all
- All parts are bought out
- The release dates of future orders are known
- There is one product
Without future information, decouple with stock buffers instead.
5. Shared feeders serving lines with different rules were handled by…
- Giving the coverage line absolute priority in every department
- Soft-dedicating machines in each department to the line
- Moving all lines to coverage
- Adding a shift
A standing assignment keeps the priorities simple.