Critical Spares and Rotable Spares Explained How many to hold, and why.

A critical spare is a part you hold not because you use it often, but because not having it when it fails would cost far more than holding it. It is insurance, and it should be managed like insurance.

A part is usually critical when all three of these are true:

• Failure has a serious consequence: it stops or limits production, or creates a safety or environmental risk.
• There is no redundancy: no standby unit, no bypass, no second line to carry the load.
• Lead time is longer than tolerable downtime: you cannot get one in time from a supplier or another site.

Critical spares are often called insurance spares or capital spares. They frequently show zero usage for years, which is exactly why they get caught up in dead stock clean ups by mistake.

A rotable is a repairable component that goes around a loop: installed on a machine, removed when it fails or hits its service life, sent for repair or rebuild, then returned to the shelf ready to fit again.

Typical rotables are engines, transmissions, final drives, hydraulic pumps, wheel motors and large gearboxes. Because they are expensive and come back, they need different tracking from consumable spares:

A part can be both critical and rotable. A large gearbox with a long rebuild time is usually both.

Score consequence and lead time, then decide by category, not by the loudest voice in the planning meeting. Use a simple matrix like this one and apply it consistently across sites.

Then check the money. If the expected cost of downtime avoided is higher than the cost of holding the part, hold it. The worked example below shows how.

Hold the spare when the expected downtime cost it avoids each year is more than its yearly holding cost. Here is an example with made up numbers.

Example: a drive unit worth $120,000. Holding cost (capital, storage, preservation and insurance) is taken as 20% of value a year. Without a spare, replacement lead time is 21 days. With a spare on site, it can be fitted in 2 days. Downtime costs $40,000 a day. The chance of failure in any given year is 5%.

At a 5% chance of failure, holding it avoids $38,000 a year in expected downtime for $24,000 in holding cost, so you hold it. At a 2% chance, the expected saving is 2% x $760,000 = $15,200, less than the holding cost. Then the better options are sharing one spare across sites, or a supplier holding agreement.

This is a rough model. Real decisions also weigh safety, contract penalties and how confident you are in the failure rate. But it beats holding everything someone once called critical.

Work out how many units are in the repair loop at any time, then add a buffer for variation. The average in repair equals removals per year times repair turnaround in years.

Example with made up numbers: 16 identical pumps are installed across a fleet. Each runs 6,000 hours a year. On average a pump is removed every 4,000 operating hours. Repair turnaround, from removal to back on the shelf, is 45 days.

Holding 3 spares covers the average, but removals come in clumps, so 3 runs out often. A common way to size the buffer is to treat removals during the turnaround as a Poisson count with a mean of 2.96 and pick the service level you want:

For a 95% service level you need 6 spares, so a pool of 22 units in total (16 installed plus 6). In Excel, the chance for each pool size is =POISSON.DIST(spares, 2.96, TRUE).

The biggest lever is usually turnaround, not buying more units. Cut repair turnaround from 45 to 30 days and the mean in repair drops to 24 x 30 / 365 = 1.97, and fewer spares give the same service level.

The most common mistake is flagging too much as critical. When everything is critical, nothing is, and the critical flag becomes a way to protect stock from review.

• Critical spares caught in dead stock write-offs. Zero usage is normal for insurance spares. They need their own review, not the same rule as consumables. See dead stock and obsolete spares.
• Critical spares on automatic min and max. A min of 1 and max of 1 set by hand is fine. Letting an algorithm drop it to zero because of no usage is not.
• Rotables tracked by quantity only. Without serial numbers and condition, you cannot tell a rebuilt unit from one waiting for repair.
• One value for new and rebuilt. It distorts inventory value and the cost charged to each work order.
• Nobody chasing repair turnaround. Units sitting at a repairer for months quietly force you to buy more.
• Cores not returned. If an exchange deal relies on returning the failed unit and it sits in a laydown yard, you pay the core charge.
• The same part held at every site under different numbers. Sites buy their own insurance spare when one could cover several.
• No preservation. A critical spare that has sat for years without inspection or rotation may not work when you need it.

Most ERPs can do this properly. The settings are usually there but not used.

1. Add a criticality field (A, B, C) to every stocked item and make it mandatory for new items. Record who set it and why.
2. Exclude category A items from automatic min and max recalculation and dead stock rules. Give them a yearly review instead.
3. Serialise rotables. Track each unit by serial number, location and status: installed, on shelf, at repairer, awaiting core return.
4. Value by condition. In SAP, split valuation lets one material carry different values for new and repaired stock. Other ERPs have similar options.
5. Track repair turnaround per repairer from removal date to back on shelf, and report it monthly.
6. Link spares to equipment so you can see which machines a critical spare protects, and whether those machines are still on site.

We build stock level tools inside your own environment (your Azure, Databricks or Snowflake) with read only access to your ERP data. The maths is deterministic, every number links back to the line it came from, and a person approves every change. The AI explains; it never decides.

• Stock level engine. Sets min and max from real usage while respecting criticality flags, so insurance spares are not cut by accident. At an international tier 1 mining operator, it matched the site inventory expert's own spreadsheet 99.8% of the time. See the min max inventory engine.
• Cross site parts matching. Finds the same part held under different numbers at different sites, using evidence tiers so it refuses to be confidently wrong. Stock can be shared instead of rebought. See cross site parts matching.
• A data audit first. Before any build, we check the data, which often surfaces problems like missing criticality flags or rotables tracked by quantity. See the AI data audit.

Any savings show up as released working capital, not instant profit.

The first step is a free 15 minute call with Tristan, or send a quick enquiry if you would rather write. No obligation and no pitch. He asks what you run and what is costing you, then tells you straight whether there is something worth fixing. Work is charged at a consulting day rate. The price depends on your sites, systems and data, so you talk first, then Tristan sets out what he would build, how many days it takes and what it costs. An engagement runs like this:

1. A free 15 minute call. What you run, and what is costing you.
2. A look at a sample export of your data, only if it is useful. Read only.
3. A clear scope and price, based on the days the work needs.
4. The build inside your systems.
5. Run, check and support.

See mining spares optimisation for more.