Spare parts inventory: how many of each part to keep

A maintenance storeroom usually has two problems at once: shelves full of parts nobody has touched in years, and no stock of the one part that just stopped the line. Both have the same cause: inventory decided by habit, not by usage and risk.

Two types of parts, two kinds of logic

Before any formula, split the parts into two groups. Each one calls for a different calculation:

Group	Examples	How to decide
Regular usage	Common bearings, belts, filters, lubricants, fuses	By average usage and lead time
Insurance spares (critical)	Gearbox on a single-path line, special motor, discontinued circuit board	By risk: cost of running out × cost of holding

The common mistake is treating both the same way: applying a reorder point to a part that fails once every five years, or leaving the most-used belt in the plant with no minimum stock.

Regular-usage parts: the reorder point

The reorder point tells you when to buy: the moment the balance drops to a level that covers usage until the new part arrives, plus a margin.

Reorder point = average usage × lead time + safety stock

A simple way to calculate safety stock, without statistics, uses the worst case you've already seen:

Safety stock = maximum usage × maximum lead time − average usage × average lead time

Example with a bearing used in several motors across the plant:

• Average usage: 3 per week; highest weekly usage in the last year: 5.
• Average supplier lead time: 2 weeks; longest delay seen: 3 weeks.
• Safety stock = 5 × 3 − 3 × 2 = 15 − 6 = 9 units.
• Reorder point = 3 × 2 + 9 = 15 units.

When the balance reaches 15, you order. The order quantity depends on price and space, but a practical reference is to buy one to two months of usage.

Critical spares: the risk calculation

For a part that is almost never used, the reorder point makes no sense: average usage is practically zero. The question becomes whether it's worth keeping a spare on the shelf.

Compare two numbers per year:

1. Cost of holding: the value of the part multiplied by the annual inventory carrying cost. That cost includes tied-up capital, space, insurance and obsolescence, and usually falls between 20% and 30% per year.
2. Expected cost of not holding: the probability of the failure happening in the year multiplied by the cost of downtime during the lead time.

Example: a gearbox worth 45,000 on a line with no backup equipment. Values are in local currency.

• Cost of holding: 45,000 × 25% = 11,250 per year.
• The gearbox fails, on average, once every three years: a probability of about 1/3 per year.
• Without a spare, the part takes 10 days to arrive. With the line running partially, the loss is 30,000 per day: 300,000 per failure.
• Expected cost of not holding: 300,000 × 1/3 = 100,000 per year.

Keeping the gearbox in stock is clearly the better option. If the line had a backup gearbox installed, or if the supplier delivered in one day, the math would change and the part could come off the shelf. To work out the cost of downtime more carefully, see what an hour of downtime costs.

Practices that cut inventory without more stockouts

• Standardize. Five motors with the same bearing need one stock, not five.
• Reserve parts for planned work orders. Next week's preventive task already knows what it will use: the part is set aside and doesn't count as available balance.
• Build kits. For repeat jobs, have the work order kit ready (bearings, seals, grease). The technician spends less time at the storeroom counter, one of the hidden components of MTTR.
• Review the critical spares list once a year. Replaced equipment leaves orphan parts on the shelf.
• Use equipment criticality. A critical part for class C equipment rarely justifies stock.

In taskfield, parts live in the technical catalog linked to equipment classes, and every closed work order records the parts used. That way, usage per asset and per period stops being an estimate.