What Is Work in Process (WIP) in Manufacturing?

Work in process (WIP) is inventory that has started production but is not yet a finished, saleable good. It is the middle state between raw material and finished goods: the parts that have cleared at least one routing operation but still have operations left to run. A tray of shafts that has been cut and drilled, yet still needs grinding and inspection, is work in process. WIP represents money and labor already committed that cannot be shipped or invoiced until the job is complete, so keeping it to the minimum the schedule needs is a core goal of finite capacity planning.

This entry defines work in process and shows how it behaves inside EDGEBIC by User Solutions. For the wider index of planning terms, see the manufacturing glossary, and for the rate that WIP eventually becomes, read what is throughput in manufacturing.

How it works

The mental picture is a wool shop full of half-knitted scarves. The wool on the shelf is raw material. The finished scarves in the display case are finished goods. Everything in between, the scarves that are started but not yet done, is work in process. They have absorbed time and material, but none of them can be sold until the last row is knitted.

On a real routing, a job becomes work in process the moment its first operation is recorded as started, and it stays WIP until its final operation is recorded as complete. In between, each unit spends part of its life being worked and part of its life waiting in a queue for the next machine. In most plants the waiting dwarfs the working, which is why WIP and lead time rise and fall together: cut the queues and both shrink.

Three forces push WIP up. Large batch sizes mean the whole batch is in process at once. Long queue and buffer times between steps mean parts sit rather than move. And an overloaded constraint means work piles up in front of the slowest machine. A scheduler that respects finite capacity attacks all three by placing work only where real hours exist and by overlapping steps where it can.

A concrete example

A job runs 200 machined parts through cutting, then drilling, then inspection. Cutting takes ten hours for the full batch. If drilling cannot start until every part is cut, then for those ten hours all 200 parts are work in process and none can move on. Only when cutting finishes does the batch of 200 shift to the drilling queue, where it may wait again.

Now overlap the steps by transferring parts in trays of 25 as they are cut. Drilling starts after roughly the first tray is ready, about an hour and a quarter in, and from then on the two operations run largely in parallel. Far fewer parts sit idle between the two stations at any moment, and the job clears the floor sooner. The finished parts leave the WIP pile earlier, so the average amount of work in process the plant carries drops even though the same 200 parts were made.

How EDGEBIC uses it

EDGEBIC represents work in process through the actual start and end dates on each scheduled operation. An operation with an actual start and no actual end is live work in process, and the job as a whole carries an in-progress status until every step is closed. Because completed steps are immutable to the rescheduler, WIP that has already been recorded stays exactly as it happened while the remaining, unstarted work is replanned around it.

The product also lets you attack the causes of WIP directly. Lot streaming, set per routing step, overlaps sequential operations so partial batches move downstream before the upstream step finishes the whole lot, which shrinks both the peak WIP and the lead time. Queue time and transit time are modeled explicitly, so the buffers that turn into waiting WIP are visible and tunable rather than hidden. And the backlog view on the resource calendar can show, for each work center and day, the forward sum of remaining committed hours, which is the honest measure of how much work in process is queued ahead of that station.

Because the engine schedules against real, finite capacity rather than assuming infinite machines, it will not stack more hours onto a work center than it can run, so WIP does not silently balloon in front of an overloaded resource the way it does under infinite-capacity planning. To see how overlapping operations shrinks WIP, continue with what is lot streaming. For the buffer time that turns into queued WIP, read what is queue time in manufacturing scheduling. And for how load stacks up against real capacity, see what is load vs capacity in scheduling.