WIP Control

Cap work-in-progress with a fixed number of job cards; new work only enters when finished work leaves.

Time40 min
FormatSmall group
StageMake

WIP Control: what it is and why it works

WIP control caps the amount of work-in-progress allowed in a segment of the process and releases new work only when finished work leaves. In its simplest form, a fixed number of job cards circulates: each job needs a card to enter the segment, and the card returns to the start when the job exits. This is the logic of CONWIP (constant work-in-process), described by Spearman and colleagues in 1990. Unlike product-specific kanban, a CONWIP card is not tied to a part number, which makes it suitable for job shops and high-mix flows. The team counts current WIP, sets a cap, enforces release rules and tunes the cap over time.

The method rests on Little's Law, from John Little's 1961 proof: average WIP equals throughput multiplied by average lead time. For a given throughput, cutting WIP must cut lead time. Most plants release work as soon as orders arrive, believing an early start means early delivery, but extra jobs simply queue, compete for resources and hide priorities. Capping WIP makes queues short, priorities visible and lead times predictable. If the cap is set too low, the bottleneck may starve; too high, and queues return. WIP control works with level scheduling to steady release, supports lead-time compression, and protects the constraint identified with Theory of Constraints.

What you need

A count of current work-in-progress between each pair of stations or in each segment
Throughput data for the segment over recent weeks
A release point where new jobs enter, and authority to hold jobs there
Physical or digital job cards and a place to display them

What you get

A WIP cap for each controlled segment
A release rule linking new work to completed work
Lead time tracking before and after the cap
A monthly review of caps against throughput and bottleneck starvation

When to use it

When shop-floor inventories grow invisibly and due dates slip.

How to do it, step by step

Count current work-in-progress between each pair of stations.
Set a cap per segment: no more job cards than the cap allows.
Release new work only when finished work exits the segment.
Watch lead time fall as queues shrink.
Tune the caps monthly — too tight starves, too loose inflates.

Worked example: Capping WIP in a pipe spool fabrication shop

Illustrative scenario — figures are realistic but not from a real company.

A pipe fabrication shop produced carbon steel and stainless spools for capital projects in chemicals and water treatment. It shipped about 40 spools a day, but roughly 1,200 spools sat between cutting, fit-up and welding, NDT and hydrotest, and painting. Average lead time was about 30 working days, and project managers constantly expedited.

The team counted WIP by segment and confirmed Little's Law: 1,200 spools divided by 40 per day gave the 30-day lead time customers were experiencing.
They set a cap of 480 spools for the whole shop, using laminated cards attached to each spool's traveler. Cutting could start a new spool only when a card came back from painting.
New orders waited at release in a visible queue ordered by required ship date, rather than being cut immediately.
The weld and inspection teams watched for starvation. When welders ran short of work in the second week, the cap was raised slightly for that segment.

Result. Within six weeks WIP settled near 500 spools, throughput held at about 41 per day, and lead time dropped to around 12 to 13 working days. Expediting fell sharply because priorities were set at release. The team learned that the queue at release was far easier to manage than a queue spread across the shop floor.

Open it in the FieldKit Supply Chain app → Get the Supply Chain template pack — join the waitlist

Common pitfalls and how to avoid them

Setting the cap so tight that the bottleneck runs out of work.Start with a moderate reduction, watch the constraint for starvation, and tighten gradually.
Allowing exceptions for urgent jobs that bypass the cards.Handle urgency by resequencing at the release point, not by adding work outside the cap.
Capping WIP but still releasing orders as soon as they arrive.Hold new orders at release and sequence them by due date or priority.
Never revisiting the cap after the initial setting.Review the cap monthly against throughput, lead time and product mix changes.

Frequently asked questions

What is Little's Law?

Little's Law states that in a stable system, average work-in-process equals average throughput multiplied by average lead time, often written as WIP = TH × LT. It holds regardless of arrival patterns or process details, as long as averages are measured over a long enough period. In practice it means that for a given throughput, reducing WIP reduces lead time proportionally.

What is the difference between CONWIP and kanban?

Kanban uses cards tied to specific parts, each triggering replenishment of that part between two stations. CONWIP uses generic cards that cap total WIP across a line or segment regardless of product; when a job exits, a card releases the next job from the queue. CONWIP is simpler to set up and suits high-mix environments, while kanban gives finer control for repetitive parts.

How do you set a WIP limit?

Start from current throughput and a target lead time; Little's Law gives the WIP that would achieve it. Set the initial cap somewhat above that figure to avoid starving resources, then reduce it in steps while watching the bottleneck. If the constraint begins to starve or throughput drops, the cap is too tight; if queues build, it is too loose.

Origin

CONWIP & Little’s Law — John Little, 1961; Spearman et al., 1990.

Used in these playbooks

Inventory cash release 2–3 days

Free cash from stock in days: classify, kill the obsolete, right-size buffers, cap WIP — with service protected.

Lead time halving week 1 week

One week to find where time hides: map the flow, walk it, compress queues and lots — measure before and after.

Related methods

More in “Make”

Turn materials into products at the constraint’s rhythm.