Condition Evidence Doesn't Decide Spares

No slide sells a monitoring programme faster than the storeroom slide.

If the machine tells us weeks in advance, why is that seal sitting on a shelf earning nothing? Release the capital. Order on alert. The arithmetic fits in one line and the saving lands in the quarter you present it.

It is the most seductive arithmetic in the business, and the bill arrives later and somewhere else. The event was outside the monitored mode. The warning was shorter than the model assumed. The part turned out not to be interchangeable. The repair loop stalled at a vendor’s door. Or the spare that was there had spent four years quietly going out of specification on its own pallet.

A condition record can open a spares review. It cannot decide the shelf. The reason is not caution — it is coverage: the warning covers the modes it covers, while the shelf covers everything that can happen.

Spares is a room with eight people in it

Spares policy sits where reliability, supply, maintenance support, finance, engineering, preservation, contract and operational resilience all have standing. Condition evidence is one input among those, and it is usually the newest one in the room.

What it cannot do is convert an uncertain warning proposition into permission to reduce, release, move, pool, purchase or defer stock.

The defence against that conversion is a written record. Eight fields, and the third column is the one that does the work.

Review field	Evidence to retain	What this field cannot establish alone
Demand and exposure	Asset population, service, event definitions, repeated events, censoring, installed-base changes, usage history	A portable demand rate or a stock quantity
Consequence and coverage	Required function, affected production, safety and integrity controls, covered and uncovered modes	That high consequence proves detectability, or that monitoring covers the exposure
Warning evidence	The exact proposition, method, operating state, acquisition, comparator, uncertainty, misses, false alerts, confirmation and response-time evidence	A guaranteed P–F interval, an order date, or a delivery buffer
Supply and repair	Dated supplier evidence, lead- and repair-time distributions, capacity, logistics, customs, quality, alternates, disruption scenarios	A stable lead time or assured emergency supply
Interchangeability	Controlled part identity, revision, configuration, software or firmware, material, certification, preservation, substitution approvals	Fit, function, conformity, or permission to substitute
Response capacity	Diagnosis, engineering, procurement, permits, labour, tooling, access, outage and return-to-service constraints	That an alert can be converted into work before need
Portfolio options	Local holding, repairable pool, sharing, consignment — with their contracts and failure dependencies	Availability, suitability, value, or risk transfer
Authority and review	Named owners, assumptions, source dates, decision, dissent, expiry, change triggers	Automatic procurement, stock, maintenance or operating authority

Read the second column as the reason the review exists and the third as the reason it stays a review.

”Warning beats lead time” is not one inequality

The failure mode of most spares arguments is arithmetic that collapses the whole question into a comparison of two numbers. Warning time beats lead time, therefore no stock.

Keep the distributions instead.

Warning time, confirmation, planning, procurement, transport, preparation and the work window all vary — and they are frequently dependent. The week that delays the diagnosis is often the same week that has no crew. The supplier who is slow is often slow because the whole sector is buying at once, which is also when your own contractors are unavailable. Independence is the assumption that makes the one-line arithmetic work, and it is the assumption least likely to hold in the conditions that matter.

So a nominal lead time shorter than an estimated warning period does not establish that a part can be obtained, accepted and made ready before it is needed. The reference puts the asymmetry in one sentence: ninety days of warning is generous only if the replacement does not take a hundred to arrive.

And the reverse holds just as firmly. An uncertain or absent warning does not by itself select a stock level either. Consequence, resilience options, policy and authorised judgement are still doing that work.

The recent past supplies the caution rather than the number. Through 2021 and 2022 the Institute for Supply Management’s monthly Report on Business surveys recorded average lead times for production materials and capital equipment at the longest levels in that series’ history, well above pre-2020 norms. Those are self-reported averages from a survey of U.S. purchasing managers. They are not decomposable to your bill of materials, and they set no lead time for any specific part, supplier or region. What they establish is only this: a lead-time assumption with no date on it is not evidence.

The shelf has its own governing basis

A stored item carries requirements of its own — environmental, preservation, inspection, handling, shelf-life, configuration, traceability, readiness. They are specific to the item, its packaging, its environment, the contract, the OEM and the site.

Which is why this article gives no handling, movement, rotation, energisation, lubrication, environmental-control, inspection or test instruction for stored equipment, and neither does the reference behind it. Approved manufacturer and site procedures, the site’s own risk controls, competent people and named owners govern that.

A spare that was never preserved to those requirements is not a spare. It is a delay with a part number.

The same boundary applies to shared, pooled and supplier-held items. A commercial arrangement is a promise about liability and price. It does not by itself establish physical compatibility, present condition, logistics, legal availability, software or firmware compatibility, certification, quality acceptance or timely access. Each of those is a separate, dated proposition — and each has ended an outage badly at least once.

What condition evidence is actually good for here

It is good at raising questions and dating assumptions.

It can support asking whether an exposure class deserves review, which stocking assumptions have gone stale, or whether a time-bounded supply option is worth analysing at all. Those are real contributions, and most storerooms would benefit from any of them.

What it cannot show is that every relevant mode is detectable, that a warning observed once will recur, that a method survives a rebuild or a re-rate, or that the response capacity exists to use the warning at all. A score, an alert, a diagnosis, a prognosis, an FMEA field or an illustrative P–F window is not a stock instruction. A spares decision taken as though it were is a decision taken on one mode’s evidence about all modes’ consequences.

In the Meridian composite — the reference’s illustrative site, not an operating record — no stocking outcome is claimed. No seal, airend or motor moves from shelf to order-on-alert. No pool shrinks. No capital is released. What the composite demonstrates is the record, and its most useful rows are the ones still empty.

For orientation on the underlying data structures, ISO 14224 covers reliability and maintenance data collection and exchange, and IEC 60300-3-14 covers maintenance and maintenance support. Consult the current editions. Naming them is not a conformity claim, and no method or result from either is adopted here.

An exercise that costs an afternoon

Take one long-lead item somebody has recently been asked to justify holding.

Fill the eight rows above against your own records — not from memory, and not from the vendor’s brochure. Put a date beside every piece of supply evidence. Write the exact warning proposition in the warning row, and beside it, the failure modes that proposition does not cover. Take the part identity down to revision and firmware. Then name a person against the decision and a date when it gets revisited.

Count the empty rows.

Two of them stay empty longest in most plants. The first is supply evidence with a date on it, because the lead time everyone quotes was learned from a conversation in a year nobody can name. The second is coverage — which modes the warning does not see — because it is the one column that makes the storeroom slide stop working.

Which failure modes does your warning not cover, and does your spares argument know that?

The spares review record, its evidence fields and the three review states above are from Predictive Maintenance: Practitioner Reference Frameworks and Planning Guide (Part 11: Maintenance Workflow and CMMS Integration).