MTBF — Mean Time Between Failures
Definition
Average operating time between consecutive failures of a repairable asset. Reliability metric — higher MTBF = more reliable. For repairable systems only. For non-repairable items (one-shot failure), use MTTF instead.
Formula
MTBF = Total Operating Time / Number of Failures
Required data
- Total operating time over the observation window (running hours, not calendar hours)
- Count of failures in the same window
- A consistent definition of “failure” (the hard part)
Possible data sources
- PLC fault tags + state machine
- CMMS work-order log (failure-type work orders only)
- Operator-logged stops with failure reason codes
Factory example
Auto Tier-2 stamping press, monthly review:
- Operating time: 600 hours
- Failures (unplanned stops requiring maintenance): 4
MTBF = 600 / 4 = 150 hours
Trend: if MTBF drops from 150h → 90h month-over-month, the asset is degrading even if no single failure looks catastrophic.
Economic impact
- Predicts unplanned downtime risk → maintenance budget + capex planning
- Drives spare-parts holding policy
- Inputs to RUL (Remaining Useful Life) models in Stage 7
Common mistakes
- Mixing planned and unplanned stops in the failure count → MTBF collapses meaninglessly
- Resetting the clock incorrectly after preventive maintenance
- Computing across a too-short window → high variance, no signal
- Reporting MTBF without paired MTTR → can’t talk about availability honestly
- Using calendar hours instead of operating hours → MTBF looks worse than it is
See also
- MTTR · OEE · First Pass Yield
- NOTQIN Industrial Intelligence Operating Thesis 2026 §5 Stage 1, Stage 7
- ISO 22400-2