A warning that arrives before a failure is only useful if you can act on it in time. For many parts, the real limit is not the repair; it is the wait for the part. Linking the forecast to purchasing turns a warning into a plan.
The core rule
Compare two numbers:
- Time to failure: how long the machine is expected to run before the part has to be replaced.
- Lead time: how long it takes to get the part to the plant, including approval, order, production or stock at the supplier, transport and receiving.
Then the latest sensible order date is:
order date = expected failure date − lead time − safety margin
The safety margin covers delays, an inspection on arrival, and the time to schedule a stop at a convenient moment.
An illustration (made-up numbers). A bearing is expected to reach its limit in 14 days. The supplier delivers in 5 days. You need 3 days of margin to plan the stop. The order must go out within 14 − 5 − 3 = 6 days.
If lead time plus margin is longer than the warning time, ordering at the warning is already too late. For such parts you need stock on the shelf.
Plan with the earliest likely date, not the average
A forecast is a range, not a point. "Between 10 and 20 days" means you plan for 10. Showing the uncertainty of the estimate next to the date is not decoration; it decides the order date.
Not all parts are equal: three classes
| Class | Typical part | Stock policy |
|---|---|---|
| A: critical | Stops the line, long or uncertain delivery | Keep a spare on site, reorder when it is used |
| B: important | Stops one machine, delivery is days | Order on warning, hold a minimum stock for the most common ones |
| C: standard | Cheap, available quickly | Order when needed or hold in bulk |
The class follows from two questions: what does an hour of stop cost, and how long does the part take to arrive? The first is the cost of an unplanned stop, which is worth working out for your own plant.
The trade-off
Spare stock costs money: capital tied up, storage, parts that age. A stop costs money too. The sensible amount of stock is where these meet, and a forecast moves the balance: with reliable early warning you can hold less stock for parts that deliver quickly, and keep the stock for the parts that cannot be bought in time.
What you need to make this work
- A parts list per machine (a bill of materials), so a warning on a machine points to the exact parts.
- Real lead times. Use what the supplier actually delivered over the last year, not what the catalogue says. Record every delivery.
- Consumption history, so you know how often each part is really used.
- A purchasing process the warning can feed: an order proposal that a person approves, with the date by which it must be ordered.
When supply is uncertain
If delivery times are long or vary a lot, the margin has to grow, and the number of parts that need stock grows too. Measure the variation, not just the average: a supplier who delivers in 5 days on average but sometimes in 20 needs a margin of 20 for critical parts.
Where possible, qualify a second source for the critical parts and note compatible alternatives in the parts list in advance.
Putting it into practice
- List the critical machines and their A-class parts.
- Record the real lead time for each part.
- For each warning, calculate the latest order date and put it in front of the person who decides.
- After the repair, compare the forecast with what happened and adjust the margin.
Our platform recommends when to order a part from the forecast and the supplier's lead time, and can pass the order proposal on to your ERP system, so the decision is made on a date and not in a panic.