Production batching: choosing how much to run before you change over

The batch size argument is a changeover argument in disguise

Large lots exist to spread setup time and setup scrap across more units. Shrink the changeover and the case for large lots weakens immediately, which is why the setup-reduction methods that belong to lean practice change this calculation more than any planning rule. Before touching lot sizes, establish what a changeover really consumes: not just the machine downtime, but the material scrapped while dialling in, the labour tied up, the first-off inspection wait, and the output lost if the equipment is a constraint. Plants that have never measured this almost always underestimate it on constrained equipment and overestimate it elsewhere.

Campaigning and the bill it sends downstream

Grouping similar work — all one colour, one grade, one allergen status, one tool family — reduces changeovers and looks efficient on the machine running the campaign. The cost lands elsewhere: the products not in the current campaign wait a full cycle, finished stock builds for items made early in the run, and the plant becomes unable to respond to a late order without breaking the campaign it just committed to. Campaign length is therefore a service-level decision as much as an efficiency one, and it belongs to whoever answers for both, which usually means it should not be settled by the machine supervisor alone.

Splitting the transfer batch from the process batch

The quantity a machine runs and the quantity that moves to the next operation need not be the same. Moving material forward in smaller transfer quantities lets downstream operations start before the upstream batch finishes, which cuts throughput time substantially without changing setup frequency at all. It costs more handling moves and needs somewhere to put partial containers, and it exposes any quality problem to the next process earlier — usually an advantage, occasionally a nuisance when it triggers premature holds. For a plant that cannot yet cut its setups, this is often the least costly lead-time gain on offer.

Quality exposure scales with the size of the lot

If a process drifts, the quantity at risk is whatever was made between the last known-good check and the point of detection. Big lots concentrate that risk: one contaminated batch, one tool wearing out of tolerance, one wrong material issue, and the whole run is in question. Smaller lots with in-process verification bound the exposure. In regulated production the lot is also the recall unit, so lot size determines how much stock must be held or retrieved when something is found. Quantifying that exposure changes the economics considerably in sectors where a single bad lot can dwarf the setup savings.

Who is allowed to change the run quantity, and on what basis

In practice, run quantities drift because people on the floor change them for sensible local reasons — finishing a coil, avoiding a late-shift setup, using up a mixed material lot. The problem is not the judgement but the invisibility: the extra units become stock nobody planned, and the planner's parameters no longer describe reality. Set an authority level and a tolerance for floor-level deviation, require the reason to be recorded on the order, and review the recorded reasons periodically. That review is usually where a plant discovers its planning parameters were wrong all along.

Frequently asked questions

Does a smaller batch always mean a shorter lead time?

Usually but not automatically. Smaller lots shorten queue and processing time at each operation, which is why they cut throughput time. But if the equipment is already the constraint and setups are long, cutting lot sizes adds setups the plant cannot afford and lead time rises as queues grow behind the machine. The order of operations matters: reduce the changeover first on constrained equipment, then reduce the lot size, and measure throughput time rather than assuming the improvement.

How do we handle minimum batch sizes imposed by the process itself?

Some processes have a genuine floor — a vessel that will not mix below a fill level, a furnace load, a coating bath, a print run with fixed make-ready. Treat that as a constraint on the equipment, not on the order, and decide separately what happens to the surplus: planned stock for a known repeat, offered to another customer, or scrapped as a cost of the process. Recording that surplus honestly is what eventually justifies investing in smaller-capacity equipment.

Should scrap allowance be built into the batch quantity?

Build in what the process reliably loses, based on recorded yield for that product and route, and keep the allowance visible as a separate figure rather than hidden in the order quantity. Two things then stay honest: the planner knows how much good product to expect, and any improvement in yield shows up as a reducible allowance rather than disappearing into surplus stock. Allowances that nobody reviews tend to ratchet upward and never come back down.