Key points
- MTBF tells you how often an asset fails. MTTR tells you how quickly you recover. Availability combines the two.
- Agree on what counts as a failure before you measure. Planned servicing is not a failure.
- State which hours you used: operating hours from the meter, or calendar hours. Both are valid, but they give different numbers.
- A median repair time is often more honest than the mean, because one long repair can dominate the average.
The three formulas
MTBF = operating time ÷ number of failures
MTTR = total repair time ÷ number of repairs
Availability = uptime ÷ (uptime + downtime) ≈ MTBF ÷ (MTBF + MTTR)
The approximation in the third line holds when downtime is mostly repair time. If machines also stand idle waiting for an operator or for work, measure availability directly from recorded downtime rather than deriving it.
A worked example
The figures below are invented to show the arithmetic. One excavator over a 30-day month, working a single 10-hour shift, 26 working days:
| Item | Value |
|---|---|
| Scheduled operating time | 260 h |
| Failures (unplanned stops needing repair) | 3 |
| Repair times | 2 h, 4 h and 14 h |
| Total downtime | 20 h |
- Uptime = 260 − 20 = 240 h
- MTBF = 240 ÷ 3 = 80 operating hours between failures
- MTTR = 20 ÷ 3 ≈ 6.7 h (the median repair is 4 h)
- Availability = 240 ÷ 260 ≈ 92.3%
Note how the single 14-hour repair pulls the mean repair time well above the typical one. Reporting both the mean and the median tells the reader whether the problem is slow repairs in general or one bad week, usually waiting for a part.
What counts as a failure
MTBF is only comparable across assets and months if everyone counts the same events. A workable definition:
- Count: any unplanned stop where the asset could not do its job and needed a repair.
- Do not count: planned services, inspections, waiting for work or an operator, and reports raised in error.
- Decide once: whether a repeat of the same fault within a few days is a new failure or the same one reopened, and apply it consistently.
Operating hours or calendar hours?
Operating MTBF uses the hours the machine actually ran, from its hour meter. Calendar MTBF uses elapsed time. Operating MTBF is the better measure of reliability, because a machine that worked twice as hard should be allowed to fail twice as often. Calendar MTBF is easier to get, because it needs no meter readings. Either is fine; mixing them in one report is not. Label the basis next to the number.
Common mistakes
- Averaging small numbers. One failure in a month gives an MTBF equal to the whole month. Use a longer window or group similar assets before drawing conclusions.
- Repair time that is really waiting time. Split downtime into response, waiting for parts and actual repair. MTTR alone hides which one is slow.
- Counting from when a repair was logged, not when the machine stopped. If faults are reported late, downtime is understated.
- Including off-site assets. Machines that have been transferred or are off hire are not part of your availability figure.