Spare Parts Management: Stocking the Right Parts Without Drowning Capital

Spare parts management is the discipline of deciding which parts to hold, how many to hold, and how to keep that stock accurate, so a breakdown never waits on a part you should have owned and your storeroom does not drown capital in parts you will never use. Every stocking decision trades the cost of downtime during a stockout against the cost of carrying inventory, and good practice makes that trade deliberately, by consequence, rather than by habit. This guide covers criticality-driven stocking, min/max levels in plain terms, insurance spares, data hygiene and the dead-stock problem.

Why spares are always a trade-off

Every part you put on a shelf is a bet. Hold too little and a breakdown that should have been a two-hour repair becomes a multi-day wait for freight, with the line down the whole time and someone paying courier rates at midnight. Hold too much and you have turned working capital into boxes that gather dust, occupy racking, get counted every year and quietly lose value as seals harden, grease separates and electronics go obsolete. Both failure modes cost real money, but only one of them shows up on a downtime report. The other hides inside the inventory valuation, which is why storerooms drift towards bloat: nobody gets shouted at for stock that sits, but everybody remembers the night the packing line waited on a bearing.

The practical consequence is that spare parts management is not an inventory problem, it is a risk decision. The question is never simply 'how do we cut stock' or 'how do we stop stockouts'; it is 'which failures can this organisation afford to wait on, and which can it not'. Purchasing tends to see only the carrying cost and maintenance tends to see only the stockout, so the stocking policy has to be agreed between them, in writing, class by class. If you have never priced an hour of downtime on your constraint line, do that first: the whole trade-off is meaningless without that number.

Stock by consequence, not usage frequency

The most common stocking mistake is ranking parts by how often they move. Usage frequency feels objective, but it optimises for exactly the wrong parts: fast movers are usually cheap, generic and available from several suppliers by tomorrow. The parts that hurt are the slow movers on critical assets, the drive card that fails once in a blue moon and stops the whole plant when it does. Rank by consequence instead. For each candidate part ask three questions: if this fails and we do not have one, does production stop or degrade; is there a safety or environmental exposure while we wait; and how long is the real replenishment lead time, door to door, including approvals and goods-in?

That gives you a simple classing scheme. Critical spares: failure stops a constraint or creates a safety exposure and the lead time is longer than you can tolerate; stock them, always, regardless of how rarely they move. Important spares: failure hurts but a workaround exists, or the supplier can genuinely deliver within your tolerance; hold a smaller quantity or a firm supplier agreement. Commodity spares: buy on demand. Crucially, tie the classing to your asset criticality ranking, not to the part in isolation. The identical gearbox can be critical on the moulding line and irrelevant on a redundant transfer pump; the asset context decides, and the where-used record (covered below) is what makes that context visible.

Min/max and reorder points in plain terms

Min/max is the workhorse policy for stocked spares, and it is simpler than most explanations make it. Max is the most you ever want on the shelf. Min is the level that triggers a reorder. Set the min so that what remains covers expected use while the replacement is on its way, plus a buffer for bad luck: that is your reorder point. In plain arithmetic, if you use about two of a part per month and the true lead time is two months, you expect to consume four during replenishment, so a min of five or six gives you a cushion. The max then follows from sensible order quantities and how much shelf risk you accept.

Two habits make min/max actually work. First, use real lead times, measured from requisition raised to part on shelf, not the supplier's optimistic quote: approvals, shipping, goods-in and inspection all live inside that window. Second, review the levels on a schedule, because they rot. A min/max that was right three years ago is wrong today in one direction or the other. A short quarterly pass over the critical list and the top movers is enough; the aim is a living policy, not a parameter someone set once and buried in the system. At each review, check four things:

• Re-measure lead times: requisition to part-on-shelf, not the quoted figure.
• Check demand: compare actual usage over the period against what the min assumes.
• Check where-used: if the assets that consumed the part are gone, the stock level should go too.
• Check criticality: has the asset's role, redundancy or duty changed since the level was set?

Insurance spares: the long-lead exception

Some parts break every rule above: they may never be used at all, and stocking them still makes sense. These are insurance spares, held purely because the lead time is intolerable, not because usage justifies them. Typical candidates are a main drive motor with a long rebuild lead time, a custom gearbox, a transformer, inserts for a critical moulding tool, or processor racks for a control system the manufacturer no longer supports. The justification is pure consequence arithmetic: the probability of failure may be low, but the lead time multiplied by the hourly cost of downtime on that asset dwarfs the purchase price, and no supplier agreement can conjure a custom casting in a week.

Treat insurance spares as a separate population. They will show zero turns by design, so exclude them from any slow-mover or dead-stock report before someone helpfully disposes of them. Store them properly: motor shafts rotated on a schedule, electronics kept dry and static-safe, elastomers cool and dark. An insurance spare that has perished on the shelf is worse than no spare at all, because it gives you confidence you have not earned. And review the list whenever assets change: when the machine that justified the spare is retired, the spare should either follow it out of the door or be re-justified against something else.

Data hygiene and the dead-stock problem

None of the above survives bad data, and three disciplines carry most of the weight. One part, one number: duplicate part numbers are how you end up with six of a seal held under three different codes and a stockout recorded on all of them at once; standardise the numbering and naming convention and merge duplicates ruthlessly. BOMs linked to assets: a technician standing at a stopped machine should be able to go from the asset to its bill of materials to a bin location in one lookup, at three in the morning, without phoning anyone. Where-used records: every part should know which assets it fits, because that is what lets you class it by consequence, and what tells you it is safe to dispose of once the last machine that used it leaves the plant.

Dead stock is rarely created by one bad decision; it accumulates. Assets get retired and their spares stay. A painful breakdown triggers a 'never again' purchase that is never consumed either. An engineering change supersedes a part but nobody tells the storeroom. Duplicate numbers breed quietly, each with its own min/max. Because inventory is booked as an asset, the pile looks like value rather than loss, so it survives audit after audit. The fix is periodic and boring: report every part with no movement over a long window and no live where-used link, have maintenance (not finance) confirm each line, then disposition it: return it, sell it, scrap it or write it down. Done yearly, the exercise stays small; skipped for five years, it becomes a project nobody wants to own.

Underneath all of it sits one input: honest failure data. If you know which machines actually stop, how often and why, then criticality classes, reorder points and the insurance list all follow from evidence instead of memory, and MTBF for each asset stops being a guess. The partner we recommend, Fabrico, reads stops directly from the machines and routes the resulting work orders to the right people, so every part you fit is recorded against a real, timestamped failure. Fabrico is a partner we recommend; the tools here are free regardless.

FAQ

Related: cost of downtime · how to calculate MTBF · preventive vs predictive maintenance · planned vs unplanned downtime