Mean time metrics are the average intervals that describe how often equipment fails and how long it takes to recover. The core three for production are MTBF (mean time between failures), MTTR (mean time to repair), and mean time between changeovers. Together they explain the availability side of OEE: MTBF sets how often you stop, MTTR sets how long each stop lasts.
These metrics matter because availability, the first factor of OEE, is not one number but the product of two behaviors: reliability and speed of recovery. A line can lose the same availability from rare long breakdowns or from frequent quick ones, and the fix is completely different. Mean time metrics separate those two stories so you attack the right one. This post defines each metric, shows the arithmetic that links them to availability, explains where they overlap and where they mislead, and connects them to the OEE on your board.
What are the main mean time metrics in production?
The main mean time metrics are MTBF, MTTR, MTTF, and mean time between changeovers, each an average that turns a pile of downtime events into a single interval you can trend. They fall into two families: how long things run before stopping, and how long stops last.
- MTBF (mean time between failures) average operating time between one failure and the next, for repairable equipment. It measures reliability. MTBF = total operating time ÷ number of failures.
- MTTR (mean time to repair) average time to restore a failed asset to running, from stop to running again. It measures maintainability. MTTR = total repair downtime ÷ number of repairs.
- MTTF (mean time to failure) average life of a non-repairable item before it fails and is replaced, such as a bearing or a belt. Used for spares, not repairs.
- Mean time between changeovers average run time between product or format changes; it frames how often setup losses hit the schedule.
The two you watch daily are MTBF and MTTR, because they are the two levers behind availability. Raise MTBF and you stop less often; cut MTTR and each stop hurts less. The site has deep dives on each, MTBF and MTTR but their real power shows up when you read them together.
How do MTBF and MTTR feed the availability side of OEE?
They combine into availability through one formula: availability = MTBF ÷ (MTBF + MTTR). That ratio is exactly the availability factor that multiplies into OEE, so mean time metrics are not a separate maintenance world, they are the engine underneath the first term of your OEE calculation.
Work an example. A machine runs an average of 40 hours between failures and takes 2 hours to fix each time. Availability = 40 ÷ (40 + 2) = 95 percent. Now suppose reliability slips and MTBF falls to 20 hours while repairs still take 2 hours: availability = 20 ÷ (20 + 2) = 91 percent. Same repair speed, but failing twice as often cost four points of availability, which flows straight into OEE. Alternatively, hold MTBF at 40 but let MTTR balloon to 8 hours: availability = 40 ÷ (40 + 8) = 83 percent. The lesson is that availability has two independent leaks, and the mean time metrics tell you which one is open.
By the numbers. MTBF is defined as total operating time divided by the number of failures, and MTTR as total repair time divided by the number of repairs; combined, steady-state availability equals MTBF ÷ (MTBF + MTTR) (Mean time between failures, Wikipedia). That availability figure is the first of the three factors in Overall Equipment Effectiveness, alongside performance and quality, so reliability and repair speed convert directly into OEE points (OEE.com, OEE factors).
How do you calculate mean time metrics from a downtime log?
Every mean time metric is a total divided by a count, so a clean downtime log is all you need. The steps below turn a shift's stop records into trendable numbers.
- Define a failure versus a stop. Decide what counts as a failure: an unplanned breakdown, but usually not a planned changeover or a break. Consistency here decides whether MTBF means anything.
- Sum operating time and count failures. Over your window, total the hours the machine actually ran and count the failures. MTBF = operating hours ÷ failures.
- Sum repair time and count repairs. Total the hours spent restoring the machine and count the repair events. MTTR = repair hours ÷ repairs. Be clear whether repair time starts at the stop or at the moment a technician arrives.
- Compute availability. Combine them: availability = MTBF ÷ (MTBF + MTTR), and check it against the availability in your OEE.
- Trend, do not snapshot. A single shift is noisy. Plot MTBF and MTTR over weeks so a real reliability decline is not lost in one bad day.
The subtle trap is the clock definition. If MTTR starts when the technician arrives rather than when the machine stopped, you hide response and wait time and flatter the number. Pick one definition, ideally stop-to-running, and hold it, the same discipline that keeps the line between loading time and operating time honest.
Where do the mean time metrics overlap and get confused?
The most common confusion is MTBF versus MTTF. MTBF is for repairable equipment and counts time between failures; MTTF is for things you replace rather than fix, like a bearing or a lamp. Using MTBF for a consumable, or MTTF for a repairable machine, quietly corrupts your spares and reliability planning. A second overlap is MTTR itself: "repair" can mean the wrench time only, or the whole stop including detection, waiting, and testing, and those can differ by hours.
The table pins down each metric so the right one gets used for the right decision.
| Metric | Formula | Applies to | Tells you |
|---|---|---|---|
| MTBF | Operating time ÷ failures | Repairable equipment | How often it fails (reliability) |
| MTTR | Repair time ÷ repairs | Any repair | How long recovery takes (maintainability) |
| MTTF | Total life ÷ units | Non-repairable parts | Expected life before replacement |
| MTBC | Run time ÷ changeovers | Scheduled changeovers | How often setup losses hit |
MTBF and MTTR also do not describe variability. Two machines with identical 40-hour MTBF can behave very differently if one fails on a predictable schedule and the other at random, which is why reliability work pairs these averages with distribution thinking. Treat them as direction-finders, not guarantees for any single week.
Why can two lines have the same availability but need opposite fixes?
Because availability is a single number produced by two independent behaviors, so the same percentage can hide very different problems. One line loses availability to rare, long breakdowns: high MTBF, high MTTR. Another loses the same availability to frequent, short stops: low MTBF, low MTTR. The OEE availability figure looks identical, but the mean time metrics underneath tell opposite stories and demand opposite responses.
Consider two lines both at roughly 90 percent availability. Line A fails once a week and takes most of a shift to recover, a maintainability problem screaming for spares and faster diagnosis. Line B never has a big breakdown but stalls a dozen times a shift for a few minutes each, a reliability and minor-stop problem calling for root-cause work on chronic nuisances. Spend Line A's budget on Line B's fix and you waste it. This is why availability alone is not actionable; the split into MTBF and MTTR is what makes it a diagnosis.
How do mean time metrics guide maintenance strategy?
They point maintenance at the right problem. A low MTBF says the equipment fails too often, which is a reliability problem, better preventive and predictive work, root-cause analysis, and design fixes. A high MTTR says recovery is too slow, which is a maintainability and response problem, spares availability, faster diagnosis, better procedures, and closer parts stocking. Reading the two together tells you whether to invest in preventing failures or in recovering from them faster.
This is the bridge from OEE to equipment reliability and total productive maintenance. When availability drags OEE down, mean time metrics diagnose why, and the maintenance program responds to the specific leak rather than to a vague "we have too much downtime." See machine downtime for capturing the events these metrics summarize, and the availability rate formula for the OEE-side view.
How does live capture make mean time metrics trustworthy?
Every mean time metric is only as good as the downtime timestamps beneath it, and hand-logged stops miss the short ones and blur the start and end of the long ones. That understates failure counts and repair time in inconsistent ways, so MTBF and MTTR wander for reasons that have nothing to do with the equipment. Automatic capture through machine monitoring records the exact moment each stop begins and ends, so the counts and durations behind every mean time metric are measured, not remembered.
With trustworthy timestamps, MTBF and MTTR become early-warning signals: a falling MTBF flags a machine drifting toward trouble before it blows a shift, and a rising MTTR flags a spares or skills gap. That is the connected-floor view behind the CLS case study; run the availability and OEE math for your own line with the OEE calculator.