Reliability-Centered Spares: Stocking by Risk

Reliability-centered spares (RCS) sets stocking levels from risk, not habit. You stock a part when the cost of not having it, its failure probability, times the downtime it would cause, times its lead time, outweighs the cost of holding it on the shelf. Critical, long-lead, hard-to-predict parts get stocked; cheap, fast, plentiful parts do not.

Most storerooms are built the opposite way: on gut feel, on "we got burned once," and on parts someone ordered a decade ago for a machine that is long gone. The result is the worst of both worlds, shelves full of dead stock and stockouts on the parts that actually stop the line. Reliability-centered spares replaces that with a defensible rule for every part.

What are reliability-centered spares?

Reliability-centered spares is a method for deciding what to stock, and how much, based on each part's risk to production rather than tradition. It applies the same logic as reliability-centered maintenance to the storeroom: identify the failure modes that matter, then hold exactly the inventory that protects against the consequences worth protecting against. It is the storeroom half of a reliability program, and it connects directly to broader spare parts inventory management.

The goal is not minimum inventory or maximum inventory. It is right inventory: enough of the parts whose absence would be expensive, and none of the parts whose absence would not. That sounds obvious, and almost no storeroom actually does it, because the default is to keep buying and never review.

Getting spares wrong is expensive in two directions at once, which is why gut feel fails. Over-stock and you tie up cash in parts that may never turn, pay to store and count them, and eventually write them off when the asset they fit is scrapped. Under-stock the wrong part and a routine failure becomes a multi-day outage while you wait on a supplier, at a cost that dwarfs a year of carrying the part. A storeroom run on instinct usually manages to do both, bulging shelves and empty bins on the same day, because instinct remembers the last painful stockout but never remembers the slow bleed of dead inventory. Reliability-centered spares fixes that by forcing every part to earn its place against the same yardstick.

How do you decide whether to stock a part?

You decide by weighing the risk of a stockout against the cost of holding the part. The risk of a stockout is roughly the part's failure probability, multiplied by the downtime that failure would cause, multiplied by how long it takes to get the part when you need it. When that expected cost is larger than the annual cost of keeping the part on the shelf, you stock it. Four factors drive the decision:

• Criticality of the parent asset. A part for a machine that can stop the plant carries far more stockout risk than one for a redundant or trivial asset. Pull this straight from your equipment criticality analysis.
• Failure probability of the part. How likely is this part to fail in a given period? Parts on your RCM analysis and failure history tell you which parts actually fail, versus which just feel risky.
• Lead time. A part you can get overnight barely needs stocking; a custom casting with a sixteen-week lead is a different conversation entirely. Lead time is often the single biggest driver.
• Cost of holding. Purchase price plus the ongoing carrying cost, storage, capital tied up, obsolescence, and spoilage. Carrying cost is commonly planned at roughly 15–30% of a part's value per year, which is why hoarding "just in case" is not free.

How do you set min/max and reorder points on that basis?

Once a part earns a place on the shelf, its min/max levels come from lead time and usage, not a round number someone liked. The reorder point (the min) should cover expected usage during the lead time plus a safety buffer for variability; the max sets how much you replenish to. The steps:

1. Confirm the part is worth stocking. Run it through the risk test above. If it fails the test, do not set a min/max, set it to buy-on-demand.
2. Estimate usage during lead time. How many will you likely consume between placing an order and receiving it? Use failure history, not guesses.
3. Add safety stock for variability. Cover the uncertainty in both usage and lead time. Higher consequence and less reliable suppliers justify more safety stock.
4. Set the reorder point (min). Reorder point equals expected lead-time usage plus safety stock, the level that triggers replenishment before you run out.
5. Set the max. The reorder point plus a sensible order quantity, balanced against holding cost. For a critical insurance spare, the max may simply be one.
6. Review on a cadence. Usage and criticality change. Re-check min/max at least annually and whenever an asset is added or retired, using cycle counting to keep counts honest.

Which parts should you never stock, and which always?

Never stock cheap, fast-moving, easily sourced parts for non-critical assets, the holding cost outweighs a stockout risk you can cover with a phone call. Always stock the parts that are simultaneously critical, long-lead, and prone to failure: the custom seal, the sole-source bearing, the long-lead gearbox for the machine that runs the plant. These "insurance spares" may sit for years, but the one time you need one, its absence costs many times its price in lost production.

The hardest category is the critical part that almost never fails. Failure probability is low, so a naive cost model says do not stock it, but the consequence is catastrophic and the lead time is long. Here the right answer is usually to hold a single insurance spare and accept the carrying cost as cheap protection. This is exactly the judgment that separates reliability-centered spares from a spreadsheet.

How do RCM and spares connect?

RCM feeds the spares list directly. When a reliability-centered maintenance analysis identifies which failure modes matter and which get on-condition or scheduled tasks, it also tells you which parts those tasks and failures will consume. The RCM output is, in effect, a shopping list weighted by consequence. Running RCM and then setting spares from its findings is far more defensible than stocking from memory. The same criticality ranking that drives predictive maintenance and condition-based maintenance decisions drives the storeroom.

Condition monitoring also changes the spares math in your favor. When you can see a bearing degrading weeks out on the MTBF trend, you can order the part on that warning instead of holding it year-round, turning some shelf spares into just-in-time orders. The better your warning time, the less insurance inventory you need.

How do you avoid dead stock and stockouts at once?

You avoid both by reviewing inventory against current reality, not the reality of five years ago. Dead stock accumulates because parts are added and never removed; stockouts happen because the parts that matter were never identified. A disciplined RCS review kills both:

• Purge parts tied to retired assets and parts with no usage in years and no criticality justification.
• Re-rank remaining parts by current asset criticality and failure history.
• Reset min/max from current lead times and usage, not legacy numbers.
• Track the outcomes, stockout events and inventory value, to prove the policy is working.

The public data backs the reliability logic underneath all of this: age-related failures are under 20% of the total, with the rest random (Nowlan & Heap, 1978), and a well-run predictive program saves 8–12% over preventive alone (PNNL, Maintenance Approaches), savings that only land if the right spare is on the shelf when the warning comes.

Harmony's agents connect the two ends of this loop: they watch the failure signals across your assets, flag the parts a coming failure will consume, and stage the work and the spare together so a predicted failure becomes a planned change-out. See it on a real floor in the CLS case study or the wider view under equipment reliability.