Batch Sizing

Setting Economic Batch Sizes That Hold Up in Production

Batch size is a money decision disguised as a scheduling habit. Here is how to set it with real setup and holding costs, and when to override the formula.

Batch size quietly sets two costs at once: every setup you run and every pallet you store. Run batches too small and setup cost eats margin; a $250 changeover spread over 200 units adds $1.25 each, but over 2,000 units only $0.13. Run them too big and cash sits on the floor; carrying inventory costs most plants 18 to 30 percent of its value per year in capital, space, damage, and obsolescence. Most shops set batch sizes by habit, usually whatever the scheduler picked five years ago, and the habit is worth real money either way it errs.

The economic batch quantity is the square root of two times annual demand times setup cost, divided by annual holding cost per unit. Worked example: demand of 48,000 units per year, setup cost of $250, holding cost of $4 per unit per year. Two times 48,000 times 250 is 24,000,000; divided by 4 is 6,000,000; the square root is about 2,449. Round to 2,450 units, roughly a two and a half week supply. The Batch Size calculator does the arithmetic; your job is getting the three inputs honest.

Setup cost is machine time plus labor plus scrapped startup parts. A 90 minute changeover on a machine with a $120 per hour rate is $180, plus one operator at $32 loaded for 1.5 hours is $48, plus 10 startup scrap parts at $3 each is $30, total $258. Holding cost per unit is unit cost times your annual carrying rate: a $16 part at a 25 percent rate carries $4 per year. Plants that guess these inputs typically land 30 to 50 percent off, and the batch size lands equally wrong.

The good news is the square root forgives moderate input error: a 20 percent error in setup cost moves the answer only about 10 percent. The bad news is habits err by 300 percent, not 20. A shop running 8,000 unit batches when the economics say 2,450 is carrying more than three times the optimal cycle stock. At $16 per unit that is roughly $44,000 of average inventory that should be $14,000, costing about $7,500 a year in carrying charges on one part number.

The single biggest lever is setup reduction, because it shrinks the optimal batch itself. Cut that $258 setup to $65 through SMED, meaning externalizing prep work, standardizing tooling, and staging the next job, and the economic batch drops from 2,450 to about 1,250 units. Halving setup cost cuts the optimal batch by 29 percent every time, which halves cycle stock over time. Plants running serious SMED programs routinely take 90 minute changeovers to under 20 minutes within 6 months, and batch sizes and lead times fall with them.

Know when to override the formula. Shelf life caps the batch: never make more than the dated life allows minus transit. Capacity caps it too; if the machine is loaded above 85 percent, extra setups may not fit and batches must grow temporarily. Demand lumpiness matters: for a part ordered 4 times a year in fixed quantities, match the order, do not run EBQ. And rerun the number whenever demand shifts more than 25 percent or a setup improvement lands, because a batch size computed on last year's inputs is just a new habit.

Deploy it on a cadence. Monthly, recalculate batch sizes for your top 20 part numbers by inventory value; that usually covers 70 to 80 percent of cycle stock. Quarterly, audit actual run quantities against the calculated batches and make the scheduler explain gaps over 30 percent. Track two numbers on the board: setups per week and average batch size versus economic batch size. World-class plants run changeovers under 20 minutes, batch sizes within 20 percent of economic on A items, and revisit the math every quarter instead of every regime change.

Published 2026-07-02.