Many items, one order

The situation

A home-textile exporter ships about 40 containers a month against 50 to 60 sales orders. Each order is an assortment: sheets, pillow covers and sets in many sizes, thread counts and colours. More than 3,000 finished items are possible, made from about 800 dyed fabrics, from 80 to 90 grey fabrics, from 10 to 12 yarns. Nothing ships until the whole assortment of an order is ready.

The plant diverges, from a few yarns to thousands of items, then converges at dispatch. Many components are common across orders, so there is always the option to club the work of several orders together.

The standard solution was in place: release choked, buffer colours on the floor, a central buffer meeting for the black and red orders. Yet these effects stayed:

• Material stealing: common fabric meant for one order went into another, to make a clubbed batch.
• Capacity stealing: common resources made components that weren’t needed yet, and delayed the ones that were.
• De-synchronised arrival: at the integration points, the components of one order arrived days apart.
• High work in process at intermediate stocking points, most of it not needed soon.
• Late full kits: late dispatch and low on-time-in-full.

The gap: to club or not to club

The dye house can dye three shades a day. In front of it are two orders, both black: order 1 needs shades 1, 2 and 3; order 2 needs shades 1, 4 and 5. Both have the same colour, so buffer management can’t choose between them. Choose any three shades across the two orders, and you open both without finishing either. Every day of that adds elapsed time to every order. Yet dyeing strictly order by order loses the saving of dyeing shade 1 once for both, and can turn the dye house into a bottleneck.

The flow meeting

The extension to the standard solution: in addition to the choked release and colour priority, run a flow meeting at each integration point every day.

• Each open sales order is listed with its buffer colour, the number of its items (here, shades), the date its input was complete and the date work on it started.
• Every day the elapsed days since the start are updated, with the progress made.
• The people who control the feeder lines attend, so they see each order’s elapsed time as they decide.
• Clubbing decisions are taken on the elapsed time, within each buffer colour.

Read the chart in two directions. Progress down the list, many orders each a little further on, means the plant is opening up orders. Progress across, a few orders moving to completion, means it is flushing them out. The meeting steers towards flushing out.

The assembly version: a bank of full kits

Assembly plants meet the same problem in another form. A maker of material-handling equipment starts final assembly as soon as a chassis and a few sub-assemblies are ready. Lots of incomplete trucks fill the floor and then a staging area. When the missing part finally arrives, fitting it takes four to ten times the normal work content, because access in a half-built truck is restricted. Some trucks develop quality problems. Incomplete trucks can’t be posted as finished goods, so customers chase, and orders are diverted from the trucks already built.

The rule: no assembly starts without a full kit. Where the orders allow it, go further and separate two kinds of order:

• Type 1 orders, completed entirely with standard parts. Keep a bank of full kits always ready in front of assembly, replenished by consumption (order to the supplier daily = norm − stock − pending orders; norm = maximum consumption during the supply lead time). Assembly lead time falls below a week, its variation collapses, and manufacturing starts as late as possible, after all confirmations.
• Type 2 orders, with one or more unique parts. Their lead time is the assembly buffer plus the time to procure the unique items. Quote them with the planned load plus a supply buffer, and use the management time freed by type 1 orders to follow them up.

A home-linen supplier that applied this went from about 30% on time and a 60–80-day lead time to 95% on time and 30–45 days in eight months. Its month-end spike, 80% of dispatches in the last days with a carry-over into the next ten, became a month that closed by the 29th.

Check your understanding

Answer all five. Get four right to complete the lesson.

1. Two orders are both black in front of the dye house. Colour priority…

• Says to finish the one with the larger value first
• Says to dye whichever shades have the biggest lots
• Can’t decide between them; another signal is needed
• Is suspended for black orders

Within one colour, the elapsed time of each open order decides.

2. The weak assumption in the clubbing conflict is that…

• Clubbing must be decided without seeing how long each order has been open
• Each lot of a shade costs a set-up and a full dyeing cycle in the dye house
• An order ships only complete
• The dye house is close to the constraint

With elapsed time visible, clubbing can be chosen to complete orders.

3. On the open-order chart, progress across many orders at once means the plant is…

• Flushing orders out
• Balancing its load across the departments
• Working to the healthy mix
• Opening up orders without completing them

The aim is to move a few orders to completion each day.

4. Why is assembling with an incomplete kit so damaging in truck assembly?

• Missing parts are always more expensive when bought late
• The late part takes four to ten times the work to fit
• Operators prefer full kits
• It changes the colour of the order

Restricted access multiplies the work, and incomplete trucks can’t be shipped.

5. A type 2 order is quoted as…

• The same as a type 1 order, since assembly is the same
• Whatever the customer needs
• The planned load plus a supply buffer for its unique parts
• Twice the type 1 lead time

Its unique items must first be procured; type 1 orders draw on the kit bank.