What the MTBF and MTTR Calculator does
This calculator works out mean time between failures, mean time to repair, failure rate and availability from the two things a maintenance record actually contains: how long the asset ran, and how many times it failed. It separates repair time from waiting time, so inherent availability and the availability the plant really sees are both shown.
MTBF is one of the most misused numbers in engineering. A pump with an MTBF of 1,460 hours is not expected to last 1,460 hours - it is a rate expressed backwards, calculated over a population or a period, and it says nothing about when any individual failure will happen. This page gives the figure and states the sample it came from, because with six failures it is an estimate and with two it is barely that.
Everything is calculated in your browser. Nothing is sent anywhere.
How to use it
- Enter the operating hours for the period: hours the asset actually ran, not calendar hours. A standby pump that ran 400 hours in a year has an MTBF based on 400, not 8,760.
- Enter the number of failures - unplanned stoppages that needed a repair. Planned maintenance is not a failure.
- Enter the total repair hours, meaning hands-on-tools time, and separately the total waiting time for parts, permits, people or lifting equipment.
- Read MTBF and MTTR, then compare inherent availability with operational availability. The gap between them is the part of the problem that belongs to planning rather than to the machine.
- Export the figures as JSON for the reliability record, or copy them into a report.
Reading the results
MTBF is operating hours divided by failures. It is a rate, not a lifetime, and it only describes the period you measured.
MTTR is repair hours divided by repairs. Keeping waiting time out of it is deliberate: MTTR should measure the job, so that a rising MTTR means the repairs are getting harder rather than that the stores ran out of gaskets.
Inherent availability, MTBF / (MTBF + MTTR), is what the design allows. Operational availability, which uses mean down time including all the waiting, is what production experiences. A large gap points at spares, permits, shift cover or contractor response, not at the asset.
Below about five failures the estimate is weak enough that one more failure would move it substantially, and the page says so rather than presenting a precise-looking number.
Zero failures gives no MTBF at all. The honest statement is that MTBF is longer than the hours observed - not that it is infinite, and certainly not that the item is perfect.
Worked example: a feed pump over a year
A feed pump ran 8,760 hours and failed six times. Total hands-on repair time was 18 hours; waiting for parts, permits and a fitter added another 30 hours.
MTBF is 8,760 / 6 = 1,460 hours, which is a failure roughly every two months of running. The failure rate is 1 / 1,460 = 6.85e-4 per hour, or 685 failures per million hours.
MTTR is 18 / 6 = 3.0 hours. Inherent availability is 1,460 / (1,460 + 3) = 99.795%.
Mean down time, including the waiting, is (18 + 30) / 6 = 8.0 hours, so operational availability is 1,460 / (1,460 + 8) = 99.455%. The machine is down 2.7 times longer than the repair takes.
That is the finding. Halving MTTR by improving the repair method would recover about 1.5 hours a year; removing the waiting would recover 30. The reliability figure points at the stores and the permit process, not at the pump.
Formulas and scoring rules
- MTBF
MTBF = total operating hours / number of failuresUndefined when there are no failures; the honest statement is then "greater than the hours observed".- MTTR
MTTR = total repair hours / number of repairsRepair time only - hands on tools.- Mean down time
MDT = (repair hours + waiting hours) / number of failures- Failure rate
lambda = 1 / MTBFOften quoted per million hours: 1e6 / MTBF.- Inherent availability
Ai = MTBF / (MTBF + MTTR)- Operational availability
Ao = MTBF / (MTBF + MDT)Always the lower of the two, and always the one production recognises.
MTBF, MTTF and the bathtub curve
MTBF applies to repairable items: the pump fails, is repaired and goes back into service, so there is a between. For something that is discarded rather than repaired - a bearing, a lamp, a sensor - the equivalent figure is MTTF, mean time to failure, and the two are not interchangeable even though the arithmetic looks similar.
Both figures assume a roughly constant failure rate, the flat middle of the bathtub curve. An item in infant mortality or in wear-out has a failure rate that changes with age, and a single MTBF averaged over that period hides the trend completely. When failures are accelerating, the useful analysis is the interval between successive failures over time, not their mean.
Why availability targets are usually about waiting
Most plant reaches an inherent availability well above 99% simply because repairs are short compared with running time. The number that production quotes is almost always lower, and the difference is administrative: no spare on the shelf, no permit until the morning, no contractor until Tuesday.
Separating repair time from waiting time in the record is therefore worth more than another decimal place on MTBF. It converts an argument about whether the equipment is any good into a list of specific delays with owners.
Limitations: what the result does not prove
- It is arithmetic on the period you entered. It is not a prediction, and it carries no confidence interval - a proper reliability estimate from a small number of failures needs a chi-squared interval or a Weibull fit.
- It assumes a constant failure rate. For items in wear-out, where failures come closer together with age, a single MTBF hides exactly the trend you need to see.
- Everything depends on how failures are counted. If short stoppages cleared by an operator are not recorded as failures, MTBF is flattered; if planned maintenance is logged as a failure, it is depressed.
- Availability here is derived from MTBF and repair time. It will not match a downtime-based availability figure that also counts changeovers, breaks and idle time.
Privacy: where your data goes
Everything you paste, type or drop is processed in this browser tab. It is not uploaded, logged, stored or sent to analytics. Session recording and tag-manager scripts are switched off on this page.
Standards and sources
- IEC 60812 - Failure modes and effects analysis (FMEA) - checked 19 Sep 2026
- ISO 22400-2 - Key performance indicators for manufacturing operations - checked 19 Sep 2026
- IEC 60300-3-11 - Dependability management: reliability centred maintenance
Frequently asked questions
How do I calculate MTBF?
Divide the operating hours by the number of failures in the same period. 8,760 running hours with six failures gives an MTBF of 1,460 hours. Use hours the asset actually ran, not calendar hours, or a standby machine will look far more reliable than it is.
What is the difference between MTBF and MTTF?
MTBF is for repairable items, where the equipment fails, is fixed and runs again - so there is a time between failures. MTTF is for items that are replaced rather than repaired, like a bearing or a lamp, where there is only a time to the one failure.
Does MTBF mean the equipment will last that long?
No. It is a rate expressed as a time. An MTBF of 1,460 hours means failures happen at an average rate of one per 1,460 running hours; it does not mean the next failure is due at hour 1,460, and half of the items will have failed well before an MTBF's worth of time in most real distributions.
Why is my availability lower than the MTBF figure suggests?
Almost always because of waiting rather than repairing. Inherent availability uses repair time only; operational availability adds waiting for spares, permits, people and access. On most plant the waiting is several times the repair.
What counts as a failure?
An unplanned loss of the required function that needed a repair. Planned maintenance is not a failure. The definition matters more than the arithmetic: as long as it is written down and used consistently, the trend is meaningful even if the absolute figure is not comparable with anyone else's.
What MTBF can I claim with no failures at all?
None. Zero failures gives no estimate, only a lower bound: MTBF is greater than the hours observed. Reliability work handles this with a one-sided confidence bound rather than by declaring the item perfect, and this page refuses to print a number instead.
How many failures do I need for a meaningful MTBF?
As a rough rule, fewer than five and the figure moves a lot with one more or one fewer event, so treat it as indicative only. Around ten or more it starts to be stable enough to compare periods, provided the failure mechanism has not changed.
How do I reduce MTTR?
Shorten the job itself: standard repair procedures, the right tools staged, isolation points labelled, and components designed to be changed as a module rather than stripped in place. If the number that is large is really mean down time rather than MTTR, the answer is spares and permits instead.
Last reviewed by the A2Z.Tools team against the sources listed above.
