What the Spare Parts Consumption Calculator does
This calculator turns a part's issue history into a stocking policy: mean and peak consumption, the variability around it, the demand you should expect during a lead time, safety stock for the service level you choose, a reorder point, an economic order quantity and what all of that costs to hold.
It also tells you when not to trust it. The normal-distribution safety stock everyone uses assumes demand that turns up in most periods. A critical spare that is issued twice in three years is not that: its history is mostly zeros, and the formula will size a buffer that has no relationship to the risk you are actually managing. The page flags lumpy demand rather than quietly producing a number.
Everything is calculated in your browser. Stock figures stay in the tab.
How to use it
- Enter the quantity issued in each period, oldest first. Twelve months is a usable minimum; less than that and the standard deviation means very little.
- Set the lead time in the same periods as the history - two months of lead time against monthly issues, not sixty days.
- Choose a service level. This is the chance of not running out during a replenishment cycle, and 95% is the usual default. One hundred per cent is refused because it needs infinite stock.
- Add the unit cost, the holding rate as a percentage of value per year, and what it costs to raise a purchase order, to get an economic order quantity and a max level.
- Read the lumpy-demand warning if it appears. For a part like that, criticality and lead time matter far more than the arithmetic.
Reading the results
The reorder point is the trigger: raise an order when free stock falls to this number. It covers expected use during the lead time plus a buffer for use being higher than expected.
Safety stock is that buffer. It scales with the standard deviation and with the square root of the lead time, so doubling the lead time increases it by about 41%, not 100%.
The z value is how many standard deviations the service level buys: 1.28 for 90%, 1.64 for 95%, 2.33 for 99%. The jump from 95% to 99% costs 42% more safety stock, which is the honest price of the last four points.
The coefficient of variation - standard deviation over mean - is the tell. Below about 0.5 the part behaves; above 1.0 it is lumpy and the formula is optimistic.
EOQ is flat-bottomed: ordering 20% away from the calculated quantity usually costs only two or three per cent more in total, so round it to a pack size rather than ordering 12.25 bearings.
Worked example: a bearing issued about five a month
Six months of issues are 4, 6, 2, 8, 5 and 5. The mean is 5.0 a month. The deviations are -1, 1, -3, 3, 0 and 0, whose squares sum to 20; divided by five (n - 1) that is 4, so the sample standard deviation is exactly 2.
The lead time is two months, so expected demand during the lead time is 5.0 x 2 = 10 units.
At a 95% service level, z = 1.645. Safety stock is z x sd x sqrt(lead time) = 1.645 x 2 x 1.414 = 4.65, so five units.
The reorder point is 10 + 4.65 = 14.65, so fifteen. Raise an order when free stock reaches fifteen.
The part costs 200, holding is 20% a year - 40 per unit-year - and an order costs 50 to raise. Annual demand is 60, so EOQ is sqrt(2 x 60 x 50 / 40) = sqrt(150) = 12.25, call it twelve. Max level is 15 + 12 = 27, average stock is safety plus half an order, about 11, and that costs roughly 440 a year to hold.
Push the service level to 99% and z rises to 2.326: safety stock becomes 6.6 and the reorder point 17. Four extra points of service cost two more units on the shelf.
Formulas and scoring rules
- Mean and standard deviation
mean = sum / n; sd = sqrt(sum((x - mean)^2) / (n - 1))Sample standard deviation, dividing by n - 1, because the history is a sample rather than the whole population.- Demand during lead time
demandLT = mean x lead time periods- Safety stock
safety = z x sd x sqrt(lead time)z comes from the normal distribution: 1.282 at 90%, 1.645 at 95%, 2.326 at 99%.- Reorder point
ROP = demandLT + safety- Economic order quantity
EOQ = sqrt(2 x annual demand x order cost / annual holding cost per unit)- Average stock and holding cost
average = safety + EOQ / 2; cost = average x unit cost x holding rate- Coefficient of variation
cv = sd / meanAbove 1, or with half the periods at zero, the page calls the demand lumpy and says the normal model does not fit.
Critical spares are a risk decision, not a statistics one
For a gearbox shaft that has been issued twice in five years, there is no meaningful mean and no meaningful standard deviation. Calculating a safety stock of 0.3 units from that history tells you nothing.
The question for a part like that is different: if it fails and there is no spare, how long is the machine down, what does that cost, and how does that compare with the cost of a shaft sitting on a shelf for a decade? That is an expected-cost comparison, and the answer is usually to hold exactly one, insured, regardless of what any consumption formula says.
Where a slow-moving part genuinely needs forecasting, Croston's method separates the size of an issue from the interval between issues and handles intermittent demand far better than the normal approximation used here.
What the service level does and does not promise
A 95% service level in this model is a cycle service level: the probability of not running out during any one replenishment cycle. It is not the percentage of demand you will satisfy, which is a different measure (fill rate) and is usually higher than the cycle service level for the same stock.
It also assumes the lead time itself is fixed. In practice a supplier that occasionally takes twice as long adds far more risk than demand variability does, and no amount of safety stock sized on demand alone covers it. If the lead time is the volatile part, the honest fix is a second supplier or a longer reorder point, not a higher z.
Limitations: what the result does not prove
- Safety stock assumes demand over the lead time is roughly normally distributed and that the lead time is fixed. Neither is usually true of engineering spares.
- It uses the history you give it. A part whose failure rate is rising - a fleet coming to the end of its life - will consume far more next year than the mean suggests, and no amount of arithmetic on past issues will see that coming.
- It does not know criticality. A cheap part that stops the whole plant deserves more stock than the formula gives; an expensive part with a reliable next-day supply deserves less.
- EOQ assumes a constant demand rate, a fixed order cost and no quantity discounts or pack-size constraints. Treat it as a starting point to round, not an answer.
Privacy: where your data goes
Everything you paste, type or drop is processed in this browser tab. It is not uploaded, logged, stored or sent to analytics. Session recording and tag-manager scripts are switched off on this page.
Standards and sources
- ISO 22400-2 - Key performance indicators for manufacturing operations - checked 19 Sep 2026
- IEC 60300-3-11 - Dependability management: reliability centred maintenance
- NIST/SEMATECH e-Handbook of Statistical Methods - normal distribution and percent points
Frequently asked questions
How do I calculate a reorder point?
Expected demand during the lead time, plus safety stock. With a mean of 5 a month and a 2-month lead time that is 10 units of expected demand; adding a 4.65-unit safety buffer for 95% service gives a reorder point of about 15.
How is safety stock calculated?
z times the standard deviation of demand times the square root of the lead time. The square root is the important part: doubling a lead time raises safety stock by about 41%, not by 100%, because the variability partly averages out over the longer window.
What service level should I use for spare parts?
Ninety-five per cent is the common default for consumables. For a part whose absence stops production, the decision is not really about service levels at all - it is about comparing the cost of downtime with the cost of holding one insurance spare.
What is the difference between min/max and reorder point?
The min is the reorder point: the level that triggers an order. The max is the reorder point plus the order quantity, which is roughly what you hold just after a delivery. They are two views of the same policy.
How much history do I need?
Enough to see the pattern, and at least twelve periods for the standard deviation to be worth anything. For seasonal consumption you need two full cycles before the mean means much, and for slow movers no amount of history will make the normal model fit.
What does lumpy demand mean and why does it matter?
Demand that arrives in occasional bursts with long empty gaps - typically half or more of the periods at zero, or a standard deviation larger than the mean. The normal-distribution formula assumes steady demand, so it under-sizes the buffer for exactly the parts where running out hurts most.
Should I hold a spare for a part that has never failed?
That is a downtime question. Estimate the cost of the machine being down for the supplier's lead time, multiply by a realistic probability of failure over the remaining life, and compare it with years of holding cost. For long-lead critical items the answer is very often yes, and no consumption history will tell you so.
Is the economic order quantity worth calculating at all?
As a sanity check, yes. Its real value is the shape rather than the number: the total cost curve is very flat near the optimum, so once you are in the right region the exact quantity barely matters and you can round to a pack size, a pallet or a minimum order without losing anything.
Last reviewed by the A2Z.Tools team against the sources listed above.
