Work in Progress Control on the PCB Floor

Work in progress is the material that has been started and not yet finished, and it is the most expensive inventory in a board shop. It has already consumed material, labour and machine time, it occupies space, it hides defects until they are found late, and it lengthens the time between an order and a shipment. Wip control is therefore not a housekeeping exercise; it is one of the few levers that improves delivery, quality and cash flow at the same time, and it is almost entirely a matter of the habits on the floor rather than of software.

Why Wip Matters

Wip is expensive for four reasons at once. It ties up cash in material and labour, it occupies floor space and racks, it delays the discovery of defects because a faulty panel that would have been found in an hour waits a week behind other work, and it lengthens the lead time quoted to the next customer. A shop with three times the wip it needs is therefore slower, poorer and less accurate at the same time, and the three effects come from one cause.

The relationship between wip and lead time is direct: for a given output rate, the time a job spends in the plant is the wip divided by the throughput. Doubling the wip doubles the lead time without changing the output by a single panel. This is why wip control is often the cheapest lead time improvement available, and why it is usually overlooked in favour of equipment.

Setting a Limit

Wip control works best with an explicit limit rather than with a general intention to reduce it. The limit is a maximum number of lots or a maximum quantity in a defined area, and when it is reached no further work is released into that area until something leaves. A limit that is never enforced is worse than none, because it teaches the floor that the rule is decorative.

The limit should be set from the constraint rather than from the available floor space. A useful starting point is a small multiple of the work that the constraint can process in a shift, which keeps the constraint fed without burying it. Once the limit is in place, the pressure to exceed it becomes visible, and that pressure is itself useful information about where the capacity is short.

Lot Tracking That Works

Lot tracking fails when it requires more effort than the value it produces. A traveller that is signed at every step is workable, a system that requires a terminal entry at every step often is not, and the result is data that is entered late and trusted by nobody. The practical rule is to record the movement of a lot at the few points where a decision depends on its location.

The record should carry the lot identity, the current step, the quantity and the date, and it should be readable from the floor rather than from an office. Where the step sequence is long, the date at each step is enough to show where a lot has been waiting, and waiting is the condition that wip control is trying to detect. A simple board with a card per lot is often more accurate than an elaborate system, because it is updated at the moment the lot moves.

Managing Queues

A queue in front of a step is not an accident; it is the result of variability meeting high utilisation. The larger the queue, the longer the wait for everything in it, and the greater the chance that a defect made much earlier is discovered much later. Queue management therefore consists of keeping the queues small and visible rather than of adding capacity, at least until the queues have been made honest.

The first step is to count. A shop that knows how many lots are waiting in front of drilling, plating and test can see which queue is growing and when. Painting a target area on the floor for each queue with a marked maximum is a simple and surprisingly effective control, because an overflowing area is visible to everybody including the person who is about to add to it.

Shop Floor Control Habits

The habits that hold wip down are small and consistent. Start fewer jobs and finish more of them, which usually means resisting the temptation to release a job because a machine is momentarily free. Move a lot only when the next step can accept it, which keeps material from accumulating in transit. And treat the completion of a lot as the event worth celebrating, rather than the start of one.

The most useful habit is a short daily review of what is in progress against what was planned. Ten minutes with the current lot list, the constraints of the day and the work that must finish, settled by the person who owns the schedule, will do more for wip control than any weekly report. The process flow provides the structure for that conversation, and the fabrication notes provide the record.

Wip and Quality

Wip hides defects. A panel with a plating fault that is discovered at the next step costs one panel, while the same fault discovered after the panel has been through five more steps costs the panel plus everything that was done to it, and it may also mean that a whole batch shares the fault. Reducing wip shortens the discovery loop and therefore reduces the cost of every defect, without changing the defect rate at all.

This effect is why wip reduction often shows up first as a quality improvement rather than as a delivery improvement. The scrap figure falls, the rework hours fall, and the reason is not that the process improved but that the feedback arrived sooner. It is worth pointing this out, because the quality and production teams will otherwise argue about which of them earned the gain.

Practical Rules

Set an explicit limit, keep the queues visible, and release work only when the next step can accept it. Record the lot’s location at the few points where a decision depends on it, and measure the time each lot spends waiting rather than only the time it spends being processed.

Review the wip position daily and the trend weekly, with the quality data beside the delivery figures so that the connection between the two is visible. The objective is not the smallest possible wip but the smallest wip that keeps the constraint fed and the delivery dates honest, and finding that level is a matter of experiment rather than of theory.

FAQ

How much wip is too much? When lead time grows without output growing, wip is too high. A useful start is a small multiple of what the constraint can process in a shift.

Does wip control need software? No. A visible board with a card per lot and an enforced limit works well, and it is often more accurate than a system that is updated late.

Why does less wip improve quality? Because defects are found sooner. Shorter queues mean the distance between making a fault and discovering it is smaller, so less work is wasted on the affected board.