Guide · Simulation
Line Simulation Explained: Little's Law, Kingman's Equation & Factory Physics
ngaide · 28 June 2026
Most of the money spent “fixing” a production line is spent moving the wrong constraint. A line simulation tells you where flow actually breaks — and why — before you touch anything. This guide walks through what a line simulation does and the queueing science it runs on.
What line simulation is (and why a spreadsheet can’t do it)
A spreadsheet computes averages. A line does not behave like its average. Stations starve and block each other, work arrives in bursts, machines go down — and it is those interactions, not the averages, that decide your throughput and your queues. A line simulation models the line as it behaves over time, so you can watch queues form, locate the constraint, and test a change before committing capital.
Little’s Law: the identity behind flow
Little’s Law is the starting point. It states a simple, exact relationship that holds for any stable process, regardless of the underlying distributions:
WIP = throughput × flow time (L = λ × W)
Because it is an identity, knowing any two terms gives you the third. Its real value is diagnostic: if WIP keeps climbing while throughput is flat, flow time is growing — work is piling up, not moving faster.
Variability is the real enemy: Kingman’s equation
Throughput is not lost only to slow machines; it is lost to variability. Kingman’s equation (the “VUT” form) approximates the average time a job waits in queue at a station:
CTₙ ≈ V × U × T — variability × utilization × process time
- V (variability) = (cₐ² + cₛ²) / 2 — the squared coefficients of variation of arrivals and service.
- U (utilization) = ρ / (1 − ρ) — and this term explodes as ρ approaches 1.
- T = the effective process time of the station.
The utilization term is the punchline. Push a variable station to 95% utilization and the queue does not grow a little — it grows enormously. That is why “just keep every machine busy” quietly destroys flow, and why cutting variability often beats buying capacity.
Factory Physics: the best-case line and critical WIP
Factory Physics (Hopp & Spearman) gives the benchmarks to judge a line against. The bottleneck rate rₛ and the raw process time T₀ define the critical WIP:
W₀ = rₛ × T₀
Critical WIP is the amount of work-in-process at which a best-case line reaches full throughput with the shortest possible flow time. Run below it and you starve the bottleneck; run far above it and you only add flow time without buying throughput. Real lines sit somewhere between the best case and the worst case — simulation is how you find out where yours actually is.
Discrete-event simulation, in practice
A discrete-event simulation (DES) steps through events — an arrival, a service completion, a breakdown — advancing the clock event by event, so the model captures starving, blocking and downtime exactly as they interact. Run it long enough, across several replications, and you get distributions rather than single averages: throughput, WIP, queue lengths, and the station that constrains them all.
How to use this on your line
- Map the stations, their rates, their variability and their downtime.
- Find the constraint — the station that sets the pace — not the loudest complaint.
- Size buffers and WIP around W₀, not by habit.
- Test a change in the model before you spend a cent on it.
ngaide’s line simulator runs exactly this — discrete-event simulation with the Kingman/VUT relationship and Factory Physics references — in your browser, so you can find the real bottleneck before you move it.