Flowline › Learn

Learn production flow by running a factory

Flowline is an idle factory game whose simulation is checked against the operations-management literature in CI. These pages explain the laws it implements — and every one of them ends in a puzzle plant where you can watch the law bite.

The concepts

• Theory of Constraints One station caps your whole line. Fix it and the cap jumps somewhere else. What Goldratt’s Theory of Constraints says, and a factory sim you can test it in.
• Little’s Law Little’s Law says WIP = throughput × cycle time. What it means on a factory floor, why extra WIP buys time and not output, and how to see it measured.
• OEE OEE multiplies Availability, Performance and Quality. Why the three losses need different fixes — and why the intuitive one so often fails.
• Variability and queues Kingman’s VUT equation explains why queues explode near full utilization, why a perfectly balanced line still stalls, and what actually fixes it.

Why trust the numbers

Every expected value in the test suite is recomputed independently from the same primitives the simulation consumes — station cycle-time distribution, MTBF/MTTR, yield — never by asking the game’s own readout whether it agrees with itself. And the suite runs against real game tuning rather than a stripped-down mock, because a law proved with the mechanic switched off is not proof of anything.

The measured headline: throughput versus work-in-process sits 17–24% above Hopp & Spearman’s Practical Worst Case curve, which is what the theory predicts for this line’s own measured variability — and a deterministic control run settles exactly on the Best Case curve, so the gap is attributable to variability rather than to a modelling error.

The curriculum

Challenge Mode is 18 fixed-seed puzzle factories across 8 chapters, each broken in one specific, textbook way. By default the game withholds its own diagnosis — no spotlight naming the bottleneck — so the instrument panel is the only assistance.

• Reading the numbers — The overlay is not decoration — it is the same diagnosis the factory floor already shows you, spelled out in numbers.
• Find the bottleneck — The slowest step caps the whole line — and it does not stay put.
• Variability and buffers — A cushion of work-in-progress buys flow. Too little of it and you run dry; too much and it just sits there, costing you.
• The hidden losses — Uptime, speed and scrap (OEE = A × P × Q) waste time in three different ways — and each wants a different fix.
• Flow discipline — How work is released matters as much as how much capacity you own.
• Market and supply — The bottleneck is not always inside your factory.
• Putting it together — Every lesson so far, one at a time. These need more than one at once.
• Time and the customer — Everything so far judged the plant by what it made. These judge it by when the customer got it — a different, and harder, question.