Shop Floor Control

Every factory has a plan and every factory has a floor, and shop floor control is the machinery that connects them. It covers four things: releasing work orders to the floor with the materials and instructions attached; dispatching them to specific machines, lines and operators; tracking their physical progress through the routing; and reporting actual output, scrap, downtime and time consumed back to the planning and costing systems.

In most textile factories, three of those four are handled well and the third one is not handled at all. Work gets released. It gets dispatched. It gets reported at the end. But between release and completion, the order disappears into the plant, and the only way to find out where it is is to walk out and look.

The WIP blindness problem

A garment order does not move through a factory as a block. It is cut into bundles — a size, a colour, twenty or thirty pieces, tied together with a bundle ticket — and those bundles crawl through cutting, sewing, washing, finishing and packing at different speeds. At any moment a 20,000-piece order might have 4,000 pieces packed, 9,000 somewhere inside the sewing line spread across thirty operations, 5,000 waiting to be washed, and 2,000 still in the cutting room.

Ask a factory without floor control where that order stands and you will get a number that is an aggregate of guesses. The cutting supervisor knows his part. The line supervisor knows roughly how many bundles are on his line. Nobody knows the whole. So the merchandiser calls the floor, the floor estimates, and the estimate goes to the buyer — and this is precisely how a shipment gets confirmed for the 30th and then slips on the 28th, with two days to react instead of two weeks.

The delay was not created on the 28th. It was created three weeks earlier, when operation 14 started falling behind and nothing in the system noticed. Shop floor control is, more than anything else, an early-warning system.

What gets tracked, and at what granularity

The core decision in any floor-control implementation is granularity, and it is a real trade-off. Track too coarsely and you learn nothing you did not already know. Track too finely and you burden operators with scanning until they stop doing it accurately.

Operation-level bundle tracking is the practical standard for garment plants. The bundle already exists as a physical object with a ticket on it, so the marginal cost of putting a barcode on that ticket and scanning it at each operation is small. See [barcode and RFID tracking](/glossary/barcode-rfid) for how the ticket itself is constructed.

In a mill rather than a garment plant, the equivalent object is the [fabric roll](/glossary/fabric-roll) or the dye batch, and the equivalent scan points are loom doff, batch start, batch completion and inspection.

Downtime reasons are the other half

Tracking where work is tells you that you are behind. Tracking why tells you what to do about it. A shop-floor system that captures production but not downtime produces a factory that knows its output is low and has to hold a meeting to speculate about the cause.

Reason codes have to be captured at the moment of the stop, by the person who saw it, from a short list they can actually pick from. A list of forty reason codes will be answered with whichever one is at the top. A list of eight — no material, machine fault, no operator, power, quality hold, changeover, planned break, waiting for previous operation — gets answered honestly and rolls up into the availability component of [OEE](/glossary/oee-textile).

The most valuable of those eight in a sewing line is usually "waiting for previous operation", because it is the fingerprint of a line-balance problem rather than a machine problem, and it is fixed by moving an operator rather than by calling a mechanic.

Feeding the loop back

Floor control is worth doing only if the data returns somewhere. Actual output per operation feeds line balancing and [SAM](/glossary/sam-smv) validation. Actual downtime feeds OEE and maintenance priorities. Actual completion dates feed the schedule, so that the next order's promise date is based on what the plant does rather than what the plan assumed. Actual material issued and consumed feeds costing, and closes the variance between what the [BOM](/glossary/bom) said and what was really used.

A floor-control system whose output is a report nobody opens is an expensive way to make operators scan things. The test of whether it is working is whether next month's plan is different because of what last month's floor data showed.

Vastra ERP's [shop-floor module](/features/shop-floor) tracks bundles and rolls at operation level against the routing issued by [production planning](/features/production-planning), captures downtime with reason codes at the machine, and returns actual times and quantities to costing and scheduling in the same system, so WIP position is visible without a floor walk.