EOQ MRO spares
MRO spares break EOQ at the assumption layer: annual demand is small (often 5 to 20 units), demand pattern is intermittent (Croston-process[1]), and stockout cost is the cost of a production line stopped. For critical spares the right tool is often a (s, S) policy with a high service level driven by line-down economics, not Wilson Q*[2].
The worked example
| Annual demand D | 14 units |
| Order cost S | $250 / PO (includes expedite) |
| Unit cost C | $480 |
| Carrying cost i | 22% |
| Holding H = i*C | $105.60 / unit / year |
| Annual holding | $430 |
| Annual ordering | $430 |
| Annual total (ex. purchase) | $860 |
Croston’s method for intermittent demand
Croston (1972) splits intermittent demand into demand size and demand interval, then forecasts each separately[1]. The Wilson assumption of constant-D fails when demand is "most weeks zero, occasionally 3 units at once." A Croston-fitted average is the input D you feed into the EOQ; the Wilson formula then applies, but with wider sensitivity bands and a higher safety stock than a continuous-demand SKU.
When (s, S) beats EOQ
For critical spares where a stockout costs $50K per hour of line-down, the optimisation flips. Carrying cost is a rounding error against stockout cost. Set service level to 99.5 percent or higher, order Q = 1 at each trigger, and let safety stock dominate the on-hand calculation. EOQ becomes a one-line check: did you size order cost reasonably? usually yes, so Q = 1.