MTBF — the reliability metric that tells you when something will fail (if you use it right)

The calculation that requires more data than most plants have

Mean Time Between Failures is the average operating time between unplanned failure events. A machine running 2,000 hours with four failures has an MTBF of 500 hours. The International Electrotechnical Commission's reliability engineering guidelines specify that meaningful MTBF estimates require a minimum of 20 to 30 failure events. For equipment with high MTBF — machines that fail once or twice per year — this means several years of data are needed before the estimate stabilises. A plant using MTBF figures based on two or three events is making decisions based on statistically unreliable numbers.

This is why the six KPIs that SMRP research links to actual performance improvement include MTBF tracked by equipment category — not just overall. The category-level view is where the actionable signal lives.

What world-class MTBF looks like in practice

SMRP benchmark data shows world-class manufacturers achieving MTBF above 1,000 hours for critical production equipment, compared to an industry average of 400 to 600 hours. IEEE Std 493 (the Gold Book) publishes failure rate data for electrical equipment in industrial power systems. The OREDA handbook provides MTBF distributions for rotating and static equipment based on field data across oil and gas facilities — widely used as a reference in continuous process industries. The gap between average and world-class represents approximately two to three times as many unplanned failure events per year.

Which maintenance interventions the research shows actually increase MTBF

A 2022 study in the International Journal of Industrial Engineering found the strongest MTBF improvements from: lubrication management calibrated to condition monitoring rather than calendar intervals (34 percent improvement), precision alignment and balancing at installation and after maintenance interventions (28 percent), and contamination control through filtration, sealing, and environmental controls (41 percent combined with lubrication management). Calendar-based PM without condition-based adjustment produced approximately 12 percent improvement over reactive-only maintenance.

The lubrication failure finding connects directly to bearing failure diagnosis — SKF data shows lubrication failure as the cause of 36 percent of premature bearing failures. Improving lubrication management addresses the single largest driver of bearing MTBF degradation.

When time-based PM does not improve MTBF — and why

The failure rate curve of a component determines whether time-based PM is rational. For wear-based failures — bearings, seals, liners — failure probability increases with time, making a PM interval before the probability rises steeply a rational intervention. For random failures — electronic components, pneumatic valves — failure probability is constant with respect to time. EPRI and OREDA data show that many industrial electronic component failures are random. Changing them on a calendar schedule provides no reliability improvement. This finding is the basis for reliability-centred maintenance recommendations to apply time-based PM selectively rather than universally.

YAFEX helps maintenance teams track MTBF by equipment type and identify highest-frequency failure assets in their plant's own data. Talk to us.