A last-time-buy is a one-shot forecast with no correction possible. Sizing it from consumption data, failure history and remaining machine life beats sizing it from fear.
A last-time-buy is the only purchase order you cannot correct. Order too few and the shortfall arrives years later, unfixable at any price. Order too many and the surplus is capital parked on a shelf until scrap day. Because both errors are invisible on the day you sign, most LTBs are sized by fear — a round number that felt safe. There is a better method.
Start with the population and the horizon: how many units of the component are installed, in which machines, and how long must each machine keep running? The horizon question belongs to plant management, not purchasing, and forcing it into the open is half the value of the exercise — an LTB is a bet on machine life, and someone should be consciously placing it.
Then replace opinion with records. Actual consumption of the part over the past years — failures, preventive replacements, commissioning damage — projected over the horizon, with an allowance for the failure rate rising as the fleet ages. Where service history is thin, the manufacturer’s failure data or our fleet-wide experience with the same component substitutes. The forecast will not be exact. It will beat the round number.
Storage is part of the buy, not an afterthought. Electrolytic capacitors age on the shelf; batteries in stored HMIs and processors discharge and leak; firmware on a shelved spare drifts out of step with the installed base. Serious LTB stock is stored under controlled conditions, exercised or checked on a schedule, and firmware-managed against the machines it protects. A dead spare discovered during a breakdown is the worst outcome of all — you paid for cover you did not have.
Consider the alternatives before signing, because an LTB is not always the answer. A form-fit-function substitute qualified now may cover the tail more cheaply than a decade of shelf stock. Sometimes a partial buy bridges five years to an already-planned retrofit. The full-horizon buy is the right answer only when the machine truly must run unchanged to the end.
We run this calculation with customers whenever a discontinuation notice lands on a component we supply or have designed in. The output is one page: population, horizon, projected consumption, recommended quantity with its assumptions stated. The assumptions are the point — a number you can argue with is worth ten you cannot.
— GANI Contract Manufacturing engineering team