Control components announce their death years in advance. Plants that treat those announcements as calendar entries retrofit on their own terms; plants that ignore them retrofit during a breakdown.
No PLC platform dies suddenly. The manufacturer announces the phase-out years ahead, publishes a last-order date, then a service end date, and the aftermarket price curve does the rest. Obsolescence only becomes a crisis in plants where nobody put those dates in a calendar — and then the retrofit happens at breakdown prices, on breakdown timescales, with production waiting.
The lifecycle has recognisable stages, and each has a correct response. Active: no action. Announced phase-out: assess your installed base and decide. Last-order window: execute a last-time-buy of spares if you are bridging. Service end: you are now self-insured, running on your shelf stock and the used market. The plants that suffer are the ones that discover stage four without ever having acknowledged stages two and three.
The decision at the phase-out announcement is genuinely two-sided, and we argue it case by case. A machine with five years of planned life left justifies a last-time-buy of processors, I/O and HMI spares — a few percent of retrofit cost buys the right to do nothing. A machine expected to run fifteen more years justifies a planned migration, because no shelf of spares bridges fifteen years and the engineering knowledge around the old platform is retiring too.
That last point is underrated: obsolescence applies to people and software, not just parts. The programming tool that only runs on an operating system nobody supports, the serial protocol the new SCADA cannot speak, the engineer who knew the machine retiring next spring. A spares shelf addresses none of these. A migration plan addresses all of them.
The mechanics of staying ahead are undramatic. An installed-base register — which controllers, drives and HMIs, on which machines, at which firmware — checked against manufacturer lifecycle notices once a year. Two days of work annually for a mid-size plant. Every entry that turns red produces a scheduled decision instead of an emergency, and scheduled decisions are the cheap kind.
We hold this register for customers as part of service agreements, but the tooling is not the point; ownership is. Someone in the organisation must own the calendar. Where nobody does, the calendar still runs — it just delivers its notifications through a machine that will not start on a Monday morning.
— GANI Engineering engineering team