Capital planning & simulation · Case 05
Proving a line change in simulation before the spend
A capacity decision made on a spreadsheet is a guess with decimal places. This engagement replaced the guess with a discrete-event model — the line, rebuilt in software and loaded with real timing and variance — so the throughput number was defended before any capital moved.
The problem
The plant had a proposed change and a number attached to it, but the number came from a static average. Static averages hide the thing that actually governs a line: variability. Two stations with the same mean cycle time behave completely differently if one of them has a long tail, and a spreadsheet can't see that. Committing capital on the strength of an average is how plants buy a second machine and discover the constraint was somewhere else.
The approach
I built the line as a discrete-event simulation in Arena, loaded with element-level timing and its measured distribution rather than a single mean, and let it run. The model surfaced the true constraint under load, showed what the proposed change actually did to throughput once queueing and blocking were accounted for, and let alternative changes be tested in an afternoon instead of a quarter. The full 65-page model and its analysis live in the vault.
The result
A throughput figure with a defensible basis, and a clear read on where the money should and shouldn't go — validated before a dollar of capital was committed, not after.
Client identity withheld. Drawings, configurations, and deliverables are available under mutual NDA. The simulation portfolio is in the vault.
Have a version of this problem?
Tell me the number that isn't moving, or the system that isn't trusted, and I'll tell you whether it's the kind of thing I fix.