MTBF – What Does the Number Really Mean, and What Does It Not?

One Number, Many Misunderstandings

Few metrics in the engineering world are cited more often than MTBF – and few are less understood. It appears in data sheets, procurement specifications, contractual requirements, and marketing presentations. However, when you ask a team to explain exactly what the number means, the common answer is: "How long the component will last."

That is precisely the mistake.

What is MTBF?

MTBF – Mean Time Between Failures – is the average time between failures of a repairable item, within a population of items and under defined operating conditions.

For non-repairable items, the term MTTF – Mean Time To Failure – is typically used. In practice, the two terms are often used interchangeably, so it is important to check exactly what is defined in the document before you.

The most important word in the definition is Mean. MTBF describes the behavior of a population, not a guarantee regarding a single component.

How many components survive until the MTBF?

When assuming a constant failure rate – what is known as an exponential distribution – the reliability at time t is calculated as follows:

R(t) = e^(−t / MTBF)

If we set t = MTBF, we get R = e^(−1) ≈ 0.37.

In other words, only about 37% of components are expected to survive until reaching a time equal to the MTBF. Approximately 63% will fail before then.

A supplier that states an MTBF of 50,000 hours does not guarantee the component will work for 50,000 hours. Under constant failure rate conditions, it means that roughly two-thirds of the components will fail before that point.

MTBF is Not Lifespan

The second mistake is confusing MTBF with useful life.

A component can exhibit a very high MTBF during its useful life, yet reach wear-out long before that number "runs out." A fan, an electrolytic capacitor, a bearing, or a battery may show an MTBF of hundreds of thousands of hours during normal operation – and enter a wear-out phase after only a few years.

Therefore, when making decisions regarding preventive maintenance, lifespan, and spare parts planning, one must also examine the useful life and wear-out mechanisms – not just the MTBF.

When is the Constant Failure Rate Assumption Valid?

The Bathtub Curve describes three stages in a product's life:

• Infant Mortality – A decreasing failure rate, usually due to manufacturing and assembly defects
• Useful Life – A relatively constant failure rate, with random failures
• Wear-out – An increasing failure rate

The assumption of a constant failure rate – and the simple MTBF calculations that rely on it – is mainly suitable for the second stage. For mechanical components, components with clear wear-out patterns, or at the beginning of a new product's life, this assumption can be misleading. In such cases, a more appropriate analysis is required, such as a Weibull analysis on actual failure data.

Where does the number come from?

An MTBF value can come from very different sources:

• Prediction – Calculation according to models and databases such as MIL-HDBK-217F, 217Plus, or Telcordia SR-332
• Testing – Reliability testing, including accelerated testing
• Field Data – Actual failures collected from systems in operation, for example through FRACAS

These three numbers are not necessarily equal. Predictions based on a different model, different environmental conditions, or a different operating profile can lead to significant differences. When comparing suppliers, it is important to ask: How was the number calculated, under what conditions, and according to which model?

How to use MTBF correctly?

• Ask about the assumptions – The source of the number, the prediction model, temperature, operating environment, and duty cycle
• Calculate reliability for the mission duration – Not "how long will it last," but what is the probability that the system will survive the duration of the mission or the warranty period
• Examine wear-out separately – Consumables and mechanical components require a dedicated lifespan analysis
• Close the loop with field data – FRACAS allows you to check if the promised number is being realized in reality
• Do not look at MTBF alone – Availability also depends on the time to repair (MTTR), logistics, and maintenance

Summary

MTBF is an important tool – when one understands what it represents. It describes the mean of a population, usually under the assumption of a constant failure rate, and does not guarantee a lifespan for an individual component.

A good reliability engineer does not just ask "What is the MTBF?" but also "How was it calculated, are the assumptions appropriate, and what do the field data say?"

Further Reading

AI and Failure Analysis – Can we identify the next failure before it occurs?

FMEA – Identifying the failure before it happens

CRE Course – Certified Reliability Engineer

Want to dive deeper into MTBF, reliability prediction, Weibull analysis, and FRACAS?

These topics are a key part of the CRE track at ALD Academy, including practical exercises on case studies from the aviation, rail, and defense industries.

For details on the CRE track