MTBF, MTTR and Availability: how to calculate them

In one line

MTBF measures how often a machine fails, MTTR measures how long it takes to repair when it fails, and Availability combines the two into a single number: what percentage of the time the machine was ready to produce.

The formulas

Over a given period of time:

• MTBF (Mean Time Between Failures) = total operating hours ÷ number of failures
• MTTR (Mean Time To Repair) = total repair hours ÷ number of failures
• Availability = MTBF ÷ (MTBF + MTTR)

If a machine operated 500 hours in the month, had 4 failures, and was stopped for repairs for 8 hours in total: MTBF = 125 h, MTTR = 2 h, Availability = 125 / 127 ≈ 98.4%.

Try it with your own numbers

The diagnosis above compares MTBF against MTTR: if the machine fails rarely but each repair weighs a lot relative to that, the real problem is maintainability (repair times), not reliability. If instead repairs are quick but failures are frequent, the problem is reliability (failure frequency) — there it is better to invest in predictive maintenance or RCM before speeding up repairs. The thresholds this calculator uses are this site's own criterion to guide the reading, not a standard — with your real data, a specific analysis of the dominant failure mode is always worth more.

Why a high MTBF can be misleading

Reliability Academy points out two common mistakes when using MTBF:

• Few failures, a lot of statistical noise: with 1 or 2 failures in the period, the calculated MTBF can vary enormously from one month to the next without the machine's real reliability having changed. MTBF is more reliable with a reasonable number of failures in the sample, not with isolated events.
• Infant mortality and wear-out are not the same thing: an average MTBF hides whether failures are early (newly installed or newly repaired equipment, a start-up problem) or late (end of service life, wear). Two machines with the same MTBF may need completely different actions depending on where on the failure curve they are.

What they are used for in practice

• Prioritizing which equipment needs the attention of reliability engineering (low and falling MTBF)
• Detecting whether the real bottleneck is the repair and not the failure (high, stable MTTR with acceptable MTBF) — a sign to review procedures, spare-parts availability or maintenance team training
• Justifying an investment in predictive maintenance or redesign with numbers, instead of asking for it "because this machine gives trouble"

Limitations to keep in mind

• They are averages: they do not distinguish critical failures from minor ones unless calculated separately
• Comparing MTBF between machines of a different nature (a pump against a PLC) says little — it is useful for comparing the same machine against itself over time, or truly comparable machines
• They require a reliable failure record: if the input data (failure hours, operating hours) is weak, so is the indicator

In summary

MTBF, MTTR and Availability are the almost universal starting point for measuring maintenance, but each one answers a different question: how often it fails, how long it takes to repair, and what percentage of the time it was available. Looking at all three together — not just the final Availability — is what tells you whether the real problem lies in the failure frequency or in the repair speed.