Which Breakdowns Should You Count as Failures in an MTBF Calculator?

MTBF is only as honest as the failure count behind it. Count too many stops and a healthy machine looks unreliable. Count too few and a problem machine looks fine. When you use our MTBF calculator, the "number of failures" field is the one that needs a clear rule. This guide gives you that rule, a list of stops to include and exclude, and a worked example.

The Simple Rule

Count a stop as a failure when:

1. The asset stopped, or could no longer do its job, and
2. The stop was unplanned, and
3. A corrective repair was needed to bring it back.

If a stop fails any of these three tests, it is usually not a failure for MTBF. It may still matter for OEE or downtime, but it does not belong in this number.

Stops to Count

• Breakdowns of the asset itself: a bearing seizes, a motor trips, a gearbox leaks and the machine stops.
• Sensor and control faults that stop the asset: a failed sensor that halts the line is a real equipment failure.
• Failures found during a PM that had already stopped or degraded the asset: the failure happened before the inspection, so count it.
• Damage that needs a corrective repair, even when the cause was an operator error. Count it and tag the cause, so you can see the pattern later.

Stops to Leave Out

• Planned maintenance: lubrication, inspections, calibration and scheduled shutdowns.
• Changeovers and setups: these are production losses, not equipment failures.
• External causes: power cuts, no raw material, no operator and waiting for upstream machines.
• Minor stops cleared in seconds: a jam the operator clears in a minute does not need a repair. Agree on a threshold, for example any stop under five minutes without a work order.
• Trial runs and testing after a repair or modification.

Gray Areas and How to Decide

Write your decisions down once, and apply them to every asset. Consistency matters more than which side of a gray area you choose.

Worked Example: A Packaging Machine

A packaging machine ran for 1,000 operating hours in one quarter. Its log shows ten stops:

Four stops count as failures. The MTBF is 1,000 ÷ 4 = 250 hours.

Now see what happens if you count all ten stops: 1,000 ÷ 10 = 100 hours. The machine looks two and a half times less reliable than it is, and your team would chase problems that are not equipment failures. The right count points you to the bearing, the sensor and the gearbox seal.

Keep Your Rule Consistent

A few habits keep the number trustworthy:

• Use work orders as the source. If a stop led to a corrective work order, it is a candidate for a failure. If not, check why.
• Record a cause code on every stop. Mechanical, electrical, operator, external and planned are enough to start.
• Apply the same rule across assets so you can compare them fairly.
• Review the rule once a year, and note the date if you change it, since it will shift your trend.

For the full definitions of MTBF, MTTR and availability, see our MTTR, MTBF and availability formulas guide.

Frequently Asked Questions

Should planned shutdowns count as failures in MTBF?

No. Planned shutdowns are scheduled work, not unplanned stops, so leave them out of the failure count.

What if I'm not sure whether a stop was a failure?

Ask whether a corrective repair was needed. If a technician had to fix something to restart the asset, count it. If the operator simply restarted it, treat it as a minor stop.

Does a failure caused by an operator count?

Yes, if it damaged the asset and needed a repair. Record the cause as operator-related, so you can see whether training or guarding would reduce it.

How many failures do I need for a reliable MTBF?

The more the better. With only one or two failures, MTBF is a rough guide, so use a longer period before drawing conclusions.

Count Carefully, Then Act on the Result

Decide your counting rule once, write it down and apply it the same way every quarter. Then your MTBF shows real reliability, and the causes behind each failure tell you what to fix first.

If you want MTBF and MTTR worked out automatically from every work order, with each failure coded by cause, PlantOps360 can show you how it works. Book a free 30-minute demo and we will use a scenario from your plant.