Beginner

How to do capacity planning in manufacturing

A practical method for working out whether you can take on the work in front of you — define your workstations, measure real capacity, lay demand against it, and decide what to do where the two do not fit.

Published

Outcome

You can answer, with a number rather than a feeling, whether the work you have committed to fits the capacity you have — and where exactly it does not.

This guide describes a method, not a product. It works with pen and paper, with the free template, or with software. The method is the part that matters; the tool only changes how much of it you do by hand.

Step 1 — Decide what you are counting capacity of

Capacity planning fails at this step more often than at any other, and it fails quietly.

The unit should be the place where work happens: the mill, the lathe, the deburr bench, the assembly bay, the paint booth. Not the department, which is too coarse to schedule against, and not the individual asset separated from its location, which produces two lists that disagree.

A machine stands at a workstation. It is described by that workstation — what it is, what it can hold, what tolerances it holds — rather than tracked as a separate thing that needs reconciling with a floorplan. If you keep one list, the plan stays consistent; if you keep two, you will eventually schedule a machine into a bay that is already occupied.

Start with the workstations that constrain you. A shop with thirty workstations usually has four or five that decide its output. Planning all thirty in month one is how capacity-planning projects die.

Step 2 — Establish real capacity, not nameplate capacity

For each workstation, capacity is the work it can absorb in a period — usually hours per day.

Take the available hours and subtract what never gets used for jobs: setup and changeover, planned maintenance, the first-off inspection, breaks, the shift that is only half-staffed on Fridays. What remains is the number to plan against.

The temptation is to use the theoretical figure because it looks better. Resist it. A plan built on nameplate capacity is a plan that is quietly 20% overbooked from the first day, and the overbooking surfaces as lateness rather than as a number anyone can point at.

If you do not know the real figure, start with an estimate and correct it after a month of comparing plan to actual. An honest estimate beats a precise fiction.

Step 3 — Express demand in the same unit

Every committed job needs to say how much of a workstation it will consume, and when. Hours per day on a named workstation, between a start and an end date, is enough to be useful.

Two practical notes. Use durations your shop can actually estimate — if quoting says “about three days on the mill”, plan in days rather than inventing an hourly precision nobody believes. And include everything that consumes the workstation, not just the machining: if the job sits in the bay for two days waiting on inspection, the bay is not free.

Step 4 — Lay demand against capacity

This is the whole exercise: for each workstation, for each day, compare booked hours against capacity.

Three results matter.

Under capacity — there is room. Worth knowing, because it is where the next job should go, and because chronic underuse on an expensive workstation is a real cost.

At capacity — full, with no absorption left. Treat this as a warning rather than a success. A shop where every workstation is planned to exactly 100% has no capacity for the rush job, and rush jobs are not optional.

Over capacity — two or more jobs want the same place at the same time. This is a decision you have not made yet, and it does not become less true by staying invisible.

Step 5 — Decide, at the overbooked days

Overbooking is resolved in one of four ways, and it is worth naming them so the choice is deliberate rather than reflexive: move the job, extend the capacity (overtime, a second shift), move the work elsewhere (another workstation or another site), or move the commitment — tell the customer a real date.

The last option is the one shops avoid and the one that costs least when taken early. A date renegotiated three weeks out is a conversation; the same date missed on the day is a relationship.

Step 6 — Choose a horizon and a cadence

Plan far enough ahead to be able to act, which for most shops is four to twelve weeks. Beyond that, demand is too speculative to schedule and the effort is wasted.

Review on a fixed rhythm — weekly is typical — and after any event that invalidates the plan: a rush order, a breakdown, a late delivery. A capacity plan that is updated only when someone remembers is worse than none, because people trust it while it is wrong.

The failure mode to watch for

Capacity planning collapses when maintaining the plan costs more than the plan saves. That is usually not a method problem; it is a mechanics problem. If every change means rebuilding a grid by hand, the plan will fall behind reality, and once it is behind, the shop stops trusting it and goes back to asking the shop floor.

So keep the plan as coarse as you can get away with, automate the comparison in step 4 if you can, and be honest when the manual version stops keeping up — which has a recognisable set of symptoms.