Batch production: running a fixed quantity, then changing everything over

What a lot size commits you to before anything is made

Choosing to run in batches means accepting that the plant can only make one thing at a time on a given asset, so every product waits its turn. The lot size sets the length of that queue and therefore the responsiveness of the whole factory. It also creates cycle stock by construction: a batch built to cover the interval until the next run sits as inventory the moment it is finished. Longer runs spread setup cost across more units and starve the other products of access; shorter runs improve responsiveness and consume capacity in changeovers. The right answer moves with demand, and plants that set lot sizes once rarely revisit an assumption that has quietly gone stale.

A lot is judged as one thing, and can be lost as one thing

Quality here is defined at lot level. Sampling, in-process checks and release testing all speak about the batch rather than the unit, which is efficient until something goes wrong, because the exposure is the whole quantity. Contamination or carryover from the preceding product, an ingredient added out of sequence, a control parameter that drifted mid-run: any of these can condemn everything made since the last verified point. That is why cleaning verification between runs, and unambiguous identification of material in quarantine, are not paperwork but the mechanism that limits the size of a loss. Sequencing products to reduce the consequence of carryover is a scheduling decision with a direct quality payoff.

Shared assets, and the products that can tolerate waiting their turn

The capital argument for batching is that one set of equipment serves many products, which suits mid-volume ranges where no single item justifies a dedicated line. Vessels, mixers, ovens, presses, coaters and packing lines are sized for the family rather than for one member of it. Products must be able to wait, which rules out anything with an immediate service promise unless finished stock covers the gap. The constraint that bites is rarely nominal capacity. It is the cumulative time lost to changeovers and cleaning, which converts an apparently under-loaded plant into one that cannot fit the schedule.

Genealogy, lot sizing rules and buying that matches them

Systems must carry lot identity end to end: which raw material lots went into which production lot, which finished lot shipped to which customer, and the status of each at any moment. Without that, a recall or investigation becomes a search through paper. Planning software needs sensible lot-sizing logic and a scheduler that treats setup and cleaning as real time rather than as an adjustment factor. Purchasing follows the same grain. Buying in quantities that do not divide into production lots creates part-used containers, which are the usual source of both waste and identity errors. Agreeing supplier pack sizes against actual batch sizes removes a persistent, low-grade cost.

Adding campaigns, and the suppliers who have to keep up

Growth comes first from longer campaigns of the same product, which improves apparent efficiency while lengthening the cycle every other item waits through. That trade eventually becomes unacceptable to customers, and the next step is either dedicated equipment for the highest runners or a second shared line. Upstream, suppliers must deliver against a lumpy call-off pattern rather than a smooth one, and materials with limited shelf life complicate that further, since a large purchase to match a large run can expire before the following campaign. Suppliers who can deliver in step with the campaign plan are worth more than a lower price paid for material that ages in a store.

Frequently asked questions

What should really drive batch size?

The cost of changing over, the cost of holding what the run produces, and the service commitment the plant has made, weighed together for each product rather than set as a plant-wide default. Physical limits often decide it anyway, since vessel volume or oven capacity fixes a practical minimum. The mistake worth avoiding is treating a historic lot size as a constant. Demand shifts, changeover practice improves, and material costs move, so the figure that was right at commissioning rarely stays right.

How do you limit the damage when a lot fails release?

By narrowing what a lot contains and by having intermediate verification points inside the run. If a control check partway through a long production run is recorded and passed, everything before that point may be separable from what came after. Clear physical segregation matters just as much, because a rejected lot that has already been mixed into a bulk store or stacked with released stock forces you to condemn far more than actually failed. Decide the segregation rules before you need them.

Is batch production compatible with short customer lead times?

Only with finished stock in front of it, or with very short cycles between runs of the same item. The customer experiences the interval between production opportunities, not the run itself, so a product made rarely has a long effective lead time however fast the line is. Plants that need both usually split the range: frequently ordered items get short, regular runs and a small stock buffer, while slower items are campaigned and quoted honestly with a longer delivery.