MTTF

What is MTTF?

MTTF (Mean Time To Failure) is the expected average operating time until a non-repairable component fails for the first time. It applies to parts that are replaced rather than repaired when they fail, such as a sensor, a fuse, a lamp or a battery. MTTF is a statistical average over a large population, not a lifetime guarantee for a single unit. Together with MTBF and MTTR it is one of the basic figures of reliability engineering, asset management and machine safety.

🧠 What does MTTF tell you, and what does it not?

MTTF describes how long a type of component lasts on average under the stated conditions. Those conditions matter: manufacturers often assume a fixed ambient temperature and nominal load. A sensor in a hot cabinet next to a variable frequency drive rarely reaches the datasheet value.

• An average, not a minimum - with a constant failure rate, about 63% of the population has already failed at t = MTTF and only 36.8% (e⁻¹) is still working
• Useful life only - a datasheet MTTF usually describes the period of random failures, not wear-out at the end of life
• No repair - after the first failure the component is at end of life; for systems that return to service after a fault you use MTBF
• A population figure - an MTTF of 1,000,000 hours (over 114 years) does not mean a sensor lasts a century, but that a large fleet will see few failures per year

🧮 How is MTTF calculated?

With a constant failure rate λ the arithmetic is simple:

• MTTF = 1 / λ - a failure rate of 2 × 10⁻⁶ per hour gives an MTTF of 500,000 hours
• Reliability R(t) = e^(−t / MTTF) - the probability that a component is still working after t hours
• FIT - electronics manufacturers often state the failure rate in FIT (failures in time): the number of failures per 10⁹ operating hours. Then MTTF = 10⁹ / FIT, so 5,000 FIT equals 200,000 hours

From test results or field data you estimate MTTF as: total operating time of all units / number of failed units. If 10 sensors each ran for 10,000 hours before failing, the MTTF is (10 × 10,000) / 10 = 10,000 hours. Note that you must also count the running hours of units that have not failed yet. This estimate is only valid if the failure rate is constant.

If the failure rate changes over time, you use a Weibull distribution with shape parameter β and scale parameter η. Then MTTF = η × Γ(1 + 1/β). With β < 1 early failures dominate, β = 1 reduces to the exponential distribution, and β > 1 means wear-out prevails.

🔄 What is the difference between MTTF, MTBF, MTTR and MTTFd?

Some sources define MTBF as MTTF + MTTR. Because the repair time is usually small compared with the operating time, the difference is often negligible. MTTF itself says nothing about repair, only about the time to the first failure.

📉 Where does MTTF fit in the bathtub curve?

The failure rate of many components follows the well-known bathtub curve with three phases:

1. Early failures - manufacturing and assembly defects; the failure rate decreases. Burn-in and careful commissioning absorb this phase.
2. Useful life - random failures at an almost constant failure rate. This is where MTTF = 1/λ applies.
3. Wear-out - the failure rate rises, for example in electrolytic capacitors, fans, batteries and relay contacts.

A high datasheet MTTF therefore does not protect you against wear-out. For a UPS battery or a fan in an embedded controller, the stated service life or replacement interval often matters more than the MTTF. Condition monitoring and predictive maintenance focus precisely on recognising that wear-out phase.

🔐 What role does MTTF play in functional safety?

Within functional safety the concern is not every failure, but failures that disable a safety function. In machine safety, ISO 13849-1 works with the MTTFd: the mean time to a dangerous failure per channel. The standard classifies it as low (3 to 10 years), medium (10 to 30 years) or high (30 to 100 years). Higher values are capped at 100 years per channel; only in Category 4 have values up to 2,500 years been allowed, since the 2015 edition (and still in the current 2023 edition). Together with the category, diagnostic coverage and CCF, the MTTFd determines the PFHd and therefore the Performance Level.

For wearing parts such as a door switch or safety relay, the MTTFd follows from the manufacturer’s B10d value: MTTFd = B10d / (0.1 × n_op), where n_op is the number of operations per year. In addition, T10d = B10d / n_op is the point at which you replace the component preventively.

In the process industry, IEC 61508 and IEC 61511 do not use MTTF itself for a SIL, but failure rates per failure mode, in particular dangerous undetected failures (λDU). From these follows the PFDavg or PFH of a SIS. IEC 62061 also works with failure rates. There, MTTF is a derived figure rather than an input.

🏭 How do you use MTTF in practice?

Typical components for which MTTF is the right measure: sensors, relays, lamps, fuses, batteries, embedded electronics that cannot be serviced, and fail-safe actuators that are replaced as a whole. A practical approach:

1. Collect the values - take MTTF, FIT or B10d from datasheets. Where these are missing, a supplier often uses a prediction handbook such as MIL-HDBK-217F (last updated in 1995 and relatively conservative), Telcordia SR-332, Siemens SN 29500 or IEC 61709.
2. Correct for conditions - temperature, vibration, load and switching frequency change the actual failure rate.
3. Scale to your installed base - expected failures per year = number of units × operating hours per year / MTTF.
4. Translate into spare parts - link the expected failures to inventory management and lead times, and keep backup units for critical positions.
5. Compare with field data - record failures in your CMMS and use root cause analysis when reality clearly deviates.

Worked example - a water utility runs 500 pressure sensors with an MTTF of 1,000,000 hours in continuous operation. Expected failures: 500 × 8,760 / 1,000,000 ≈ 4.4 sensors per year. Keeping four to five spare sensors in stock covers normal random failures. After roughly 105,000 hours (12 years), the exponential model predicts that 10% of the original sensors have failed, but wear-out can bring that point forward.

Worked example for a door switch - a switch with B10d = 2,000,000 cycles on a door that opens every two minutes, 16 hours a day and 220 days a year, performs 105,600 operations per year. Its MTTFd is then about 189 years (capped at 100 years outside Category 4, so high), but T10d is 18.9 years. That is shorter than the usual 20-year mission time, so plan replacement after 18.9 years at the latest.

MTTF also feeds into FMEA (the likelihood of a failure mode), RAMSHEEP analyses and the justification of redundancy or high availability. For LOPA the link is indirect: there the focus is on failure frequencies and the probability of failure on demand per protection layer.

❓ Frequently asked questions

Does an MTTF of 100 years mean a component lasts 100 years?

No, MTTF is an average over a large population and not a lifetime guarantee for a single unit. With a constant failure rate, about 63% of components have already failed at t = MTTF. Moreover, a datasheet MTTF usually only applies to the useful-life phase, before wear-out sets in.

What is the difference between MTTF and MTBF?

MTTF applies to non-repairable components that are replaced after a failure, such as sensors and fuses. MTBF applies to repairable systems that return to service after a failure, such as pumps and machines. The calculation is similar; the difference lies in what happens after the failure.

How do I convert FIT to MTTF?

FIT is the number of failures per billion (10⁹) operating hours. The MTTF in hours is 10⁹ divided by the FIT value, assuming a constant failure rate. A component with 1,000 FIT therefore has an MTTF of 1,000,000 hours, over 114 years.

What is MTTFd?

MTTFd (Mean Time To dangerous Failure) is the mean time to a failure that compromises a safety function. ISO 13849-1 uses the MTTFd per channel to determine the Performance Level of a safety function, with the classes low (3 to 10 years), medium (10 to 30 years) and high (30 to 100 years).

How do I determine the MTTF of a relay or switch?

For electromechanical wearing parts, the manufacturer usually gives a B10 or B10d value in switching cycles instead of an MTTF. The MTTF then depends on usage: MTTF = B10 / (0.1 × operations per year), and for dangerous failures MTTFd = B10d / (0.1 × operations per year). The more often a relay or switch operates, the lower its MTTF.

Why does real-world MTTF differ from the datasheet?

Manufacturers determine MTTF under reference conditions, such as a fixed temperature and nominal load. Higher temperatures, vibration, humidity and a high switching frequency shorten the actual service life. It is therefore wise to check the datasheet MTTF against failure data from your own maintenance system.

📌 In summary

MTTF is the expected average operating time until a non-repairable component fails: a population average, not a lifetime guarantee.

With a constant failure rate, MTTF equals 1/λ and by that point 63% of components have already failed. Use MTTF to plan spare parts and inventory management, combine it with wear data such as B10d, and use the MTTFd according to ISO 13849-1 in machine safety. For repairable systems, MTBF and MTTR are the right measures.