MTBF and MTTR: prepare a consistent breakdown log
Define operating time, failure count and repair time consistently before calculating equipment reliability and availability.
Why consistent definitions matter more than the formula
MTBF (mean time between failures) and MTTR (mean time to repair) are simple ratios, so the result is only as good as the definitions behind the inputs. Two sites can report different MTBF for identical machines because one counts a micro-stop as a failure and the other does not, or because one includes waiting for spares in repair time and the other does not. Before calculating, decide in writing what a failure is, which hours count as operating time and what the repair clock covers. MTBF applies to repairable equipment; for parts that are replaced and not repaired, a life or Weibull analysis fits better.
It also helps to separate the two ideas that availability combines. MTBF describes how often a machine stops; MTTR describes how long each stop lasts. A line with a long MTBF and a long MTTR and a line with a short MTBF and a short MTTR can show similar availability, yet they need different actions: better spares and repair planning in the first case, root-cause work on the recurring cause in the second.
The formulas the calculator uses
The MTBF, MTTR and availability calculator takes three inputs for one reporting window: total operating time, number of failures and total repair time.
- MTBF = operating time ÷ failures
- MTTR = repair time ÷ failures
- Availability = MTBF ÷ (MTBF + MTTR) × 100 %, which equals operating time ÷ (operating time + repair time)
Operating time excludes downtime. Because the denominator is operating plus repair time, this is a simplified inherent availability: planned shutdowns, waiting time and idle periods must be left out of both inputs. If you want availability against calendar or scheduled time, the method differs and you should state which basis you used. With zero failures the calculator refuses to give a result, because an observed MTBF cannot be formed from no events.
Worked example 1: one pump over 30 days
A transfer pump is scheduled to run continuously for 30 days, 720 h. It fails three times and each repair is timed from stoppage to hand-back to operations, totalling 14 h. Operating time is 720 − 14 = 706 h.
| Result | Calculation | Value |
|---|---|---|
| MTBF | 706 ÷ 3 | 235.33 h |
| MTTR | 14 ÷ 3 | 4.67 h |
| Availability | 706 ÷ (706 + 14) | 98.06 % |
Three failures in 30 days is a small sample. The MTBF is a description of this window, not a prediction of the next 235 hours.
Worked example 2: how the repair-time definition shifts the result
Take the same pump, but a maintenance planner also counts 9 h spent waiting for a spare part across the three events. Repair time becomes 14 + 9 = 23 h, and operating time becomes 720 − 23 = 697 h.
MTBF = 697 ÷ 3 = 232.33 h. MTTR = 23 ÷ 3 = 7.67 h. Availability = 697 ÷ 720 = 96.81 %.
Neither answer is wrong, but they answer different questions. The first isolates hands-on repair; the second includes logistics delay. A second case with a conveyor over a quarter: operating time 2 010 h, five failures, 31.5 h of repair gives MTBF 402 h, MTTR 6.3 h and availability 98.46 %. Comparing these two assets is valid only if both used the same definitions.
The calculator will return the same figures for these inputs, with MTBF and MTTR displayed to four decimal places at most.
Decide what repair time includes
Pick one boundary and apply it to every record. Common options are: from failure until the equipment is safe to work on; from the start of work until the repair is complete; or from failure until production resumes. Waiting for the technician, permits, isolation, parts, restart and quality verification either belong inside the clock or outside it, but not sometimes one and sometimes the other. Also decide whether a failure is a loss of function, a stop above a minimum duration, or any unplanned work order. A short stop of two minutes and a gearbox failure of two days count equally in MTBF, so averages can hide a few long events.
Common mistakes
- Including planned maintenance, changeovers or idle time in operating time.
- Counting calendar time and then also subtracting repair time inconsistently.
- Mixing start and end timestamps from different systems, such as operator and technician logs.
- Combining different machines with different duty into one MTBF.
- Reading MTBF as guaranteed life or time to first failure.
- Reporting from a very small number of failures without saying so.
- Dropping minor failures that nobody raised a work order for.
Checks before you trust the result
- Operating hours plus repair hours should reconcile with the scheduled time for the window.
- The failure count should match the number of breakdown rows in the log.
- Look for outliers: one 40 h event can double MTTR. Report the median repair time next to the mean.
- Verify that failures with no end time were closed out before the period ended.
- Zero failures does not mean infinite MTBF. Report operating hours with no failures and extend the window.
- Compare two windows, for example this month and the trailing quarter, to see whether one event dominates the average.
- State the sample size next to every MTBF you publish. Two failures and twenty failures support very different confidence.
FAQ
Can I use MTBF to schedule preventive maintenance?
Only with caution. A mean says nothing about whether failures are random or wear-related; the Weibull calculator is more suitable for that question.
Is a higher availability always better?
It is better against the same definition. Compare only like with like.
Why does the calculator need at least one failure?
MTBF divides operating time by failures, so zero failures has no observed value. Report the failure-free operating hours instead and keep collecting data.
Should planned shutdowns count as downtime?
Not in this method. Leave them out of both operating and repair time, and report them separately.
Recording the data
A breakdown log needs one row per event with equipment tag, failure start, repair start, repair end, return to service, failure description, cause category, and a column marking whether the event counts under the agreed definition. Totals by month then feed the calculator directly. The free breakdown log template on the site provides this layout, and the downtime cost guide shows how to add a financial column to the same record.