Capacity Planning

Selling is easy and capacity is finite, and capacity planning is where those two facts meet. It asks, over a defined horizon, whether the load implied by the order book fits inside the capacity available on each resource — and where it does not, what to do: move the date, move the work to another machine, subcontract it, add a shift, or decline the order.

The answer is only useful if the resources are modelled at the level at which they actually constrain. "We have 200 looms" is not a capacity statement. It is a machine count.

Capacity in textiles is not a single number

A weaving shed's capacity depends on which fabric you intend to weave. A 200-loom shed running a coarse construction and a 200-loom shed running a fine one have different capacities measured in metres, because pick density changes the metres per hour. And not every loom can weave every construction — width, loom type (rapier, airjet, waterjet, projectile), reed and beam all constrain which orders can go where. See [loom scheduling](/glossary/loom-scheduling) for how that assignment problem plays out.

So the honest unit of capacity varies by process, and the unit determines what you can plan.

Notice that the sewing line converts to pieces only through SAM. A line of 30 operators working an 8-hour shift has 14,400 available operator minutes; at 80 percent efficiency that is 11,520 productive minutes; divided by a garment SAM of 18 minutes, it is 640 pieces. Change the style and the capacity of the same line, measured in pieces, changes completely. This is why sewing capacity must always be planned in minutes and only converted to pieces per style.

Finite versus infinite capacity planning

[MRP](/glossary/mrp) plans materials assuming capacity is infinite. That is a useful simplification for the materials question and a dangerous one if it is where planning stops, because it produces schedules that are materially feasible and physically impossible.

Finite capacity scheduling does the opposite: it loads work onto specific resources, respects the capacity of each, and pushes work out in time when a resource is full. The output is later and less pleasant than the infinite-capacity plan, and it is the only one that will actually happen.

Most factories need both, in sequence: a rough-cut capacity check on the master schedule to catch gross overload early, an MRP run for materials, and then finite scheduling on the genuinely constrained resources. Trying to finitely schedule every machine in the plant is usually wasted effort — only the bottleneck resources determine throughput.

Changeover is capacity

In textiles, setup and changeover time is not a rounding error, and treating it as one is the most common way capacity plans go wrong.

A beam change on a loom takes hours during which the loom produces nothing. A dye vat that switches from a dark shade to a pale one must be cleaned. A sewing line changing style loses time to re-layout, re-training on new operations, and the efficiency ramp as operators learn the new work — a line that runs at high efficiency on a style it has made for two weeks will run well below it on the first day of a new one.

So real capacity is a function of the *sequence*, not just the volume. Five orders of 10,000 metres each on one loom consume far less changeover than fifty orders of 1,000 metres. A capacity model that multiplies machine hours by a rated speed and ignores sequencing will systematically overstate what the plant can do, and the overstatement grows precisely when the order mix fragments — which is exactly when the planner most needs to be told the truth.

The bottleneck moves

Every plant has a constraint, and the temptation is to identify it once and plan around it forever. In textiles the constraint moves with the product mix. A run of heavy, dark, simple fabrics may bottleneck at dyeing. A run of light, complex, printed styles may bottleneck at finishing or at cutting. A month of intricate garments with high SAM may bottleneck at sewing while the fabric side idles.

This is why capacity planning has to be re-run against the actual order book rather than settled once from experience. The plant's capacity is not a property of the plant; it is a property of the plant and the mix together.

Capacity planning and the promise date

The commercial payoff of capacity planning is not efficiency, it is the ability to answer a customer honestly at the moment they ask. A merchandiser who can see that dyeing is committed through week 9 can quote week 12 and mean it. A merchandiser who cannot will quote week 8 because that is what the buyer wants to hear, and the factory will discover the problem in week 7.

Every capacity overcommitment eventually gets paid for, and it is usually paid for in air freight, discounts, or an order that arrives after the season it was made for. The purpose of the plan is to move that discovery from week 7 to the day the order was quoted.

Vastra ERP models looms, knitting machines, dye vats, cutting tables and sewing lines as distinct resources with their own capacity units and changeover rules, loads the order book against them with finite capacity, and exposes the resulting available-to-promise position to [production planning](/features/production-planning) and [order management](/features/order-management) so quoted dates reflect the plant's real position.