Which Spare Parts to Stock

Most spare parts inventories are a record of past failures. Something broke, the plant waited for it, and afterwards somebody bought two. Repeat for fifteen years and you have a store containing a great deal of capital, an unknown fraction of which is for equipment that no longer exists.

A defensible answer starts from three variables rather than one. The first is consequence: what happens to production, safety and compliance when this item fails. The second is likelihood, which for most industrial components means a wear pattern rather than a random draw. The third is lead time, and it is the one most often left out of the conversation entirely.

Lead time deserves its own paragraph because it inverts intuitions built on price. A fifty-euro proximity sensor with a twenty-six week lead time can stop a line for longer than a fifty-thousand-euro drive that a distributor holds regionally and ships overnight. Value tells you what the item costs to stock; lead time tells you what not stocking it costs. Any criticality assessment that sorts by price is answering the wrong question, and it is the reason so many stores are full of expensive things while the line waits for a cheap one.

Then there is the category beyond long lead time, which is no lead time at all. Industrial plants last twenty to thirty years and their electronics do not: a controller family reaches end of production in seven or eight, and the last-time-buy notice arrives with a window measured in months. Vendors publish those notices and most plants do not read them. Anything under an obsolescence notice needs a decision — buy the final stock now, qualify a replacement, or accept that the next failure is a migration project — and the decision is much cheaper made deliberately than discovered.

With those three axes you can sort the store into behaviours rather than a single list. Insurance spares are the items you hold one of, that cost a great deal, that you hope never to touch, and whose justification is entirely about the consequence of not having them: a spare transformer, a main drive, a critical bearing set. Consumables are governed by reorder points and consumption rates, and the mistake there is managing them by attention rather than by rule. Commodity items — fasteners, general cable, standard fuses — should mostly not be stocked at all, because a local supplier holds them better than you do.

The largest lever, though, is not stocking policy. It is commonality. A plant with three drive families, four PLC vintages and a dozen sensor brands needs a store several times larger than a plant that standardised, because coverage is per distinct part number rather than per function. That standardisation is decided during projects, by purchasing decisions made one machine at a time, usually against a capital budget that does not carry the spares consequence. The engineering fix is a plant standards document that constrains what new equipment may specify — and it only works if someone has the authority to enforce it against a project that found a cheaper drive.

None of this is possible without knowing what is installed, and that is where most programmes actually fail. An accurate asset register with model numbers, firmware versions and installed locations is the input to every decision above, and it is usually somewhere between incomplete and fictional. Capturing it at commissioning is the cheap moment; reconstructing it later means walking the plant with a clipboard.

Two operational details matter more than their profile suggests. First, spares age on the shelf. Electrolytic capacitors dry out, elastomeric seals harden, greases separate, bearings can brinell under static load and vibration, and boards need proper ESD packaging and dry storage. A spare that fails on installation is worse than no spare at all, because it consumes the outage window and sends the diagnosis in the wrong direction. Second, test them. Rotate stock so the oldest is used first, power up electronic spares periodically, and run spare motors and pumps rather than leaving them idle for a decade.

Measure three things. The stockout rate on items classified critical tells you whether the classification is right. Inventory turns tell you whether the store is working capital or a museum. And the count of emergency freight shipments per year is the most honest metric of the three, because every one of them is a spare that should have been on the shelf and a decision that was made by default rather than deliberately.