MTBF (Mean Time Between Failures) equals total operating time divided by the number of failures. MTTR (Mean Time To Repair) equals total downtime time divided by the number of repairs. Both are averages over a period you define, and both need a failure log behind them.
- MTBF = Total operating hours ÷ Number of unplanned failures
- MTTR = Total downtime hours ÷ Number of repairs
Here is what that looks like on one injection molding press over a quarter. It ran 1,000 operating hours and broke down three times, so MTBF = 1,000 ÷ 3 = 333.3 hours between failures. Those three breakdowns cost 14.0 hours of downtime, so MTTR = 14.0 ÷ 3 = 4.67 hours to restore.
The arithmetic takes about a minute. The definitions are where plants get into trouble, because whether you include a changeover, a Saturday planned service, or an hour spent waiting on a parts courier changes the number enough to make the comparison meaningless. Experienced maintenance teams argue the hard part is agreeing on what counts, not dividing.
This guide walks through how to calculate MTBF and MTTR end to end: what data you need, the five calculation steps, the spreadsheet formulas, which time buckets belong inside MTTR, how the two numbers combine into an availability figure, and the maintenance decisions worth making with the result. If you also need a consistent basis for the operating hours side of the calculation, our guide to how to calculate machine hour rate is worth reading first.
Table of Contents
- What You Need
- Step-by-Step: How to Calculate MTBF and MTTR
- Step 1: Define the Equipment and Measurement Period
- Step 2: Collect Operating Hours and Failure Data
- Step 3: Calculate MTBF
- Step 4: Calculate MTTR
- Step 5: Interpret the Results
- How to Calculate MTBF and MTTR in Excel or Google Sheets
- What Time Counts Inside MTTR
- Turn MTBF and MTTR Into Availability
- What Counts as a Good MTBF and MTTR Target
- MTBF, MTTF and the Rest of the Metric Family
- Common Mistakes
- Frequently Asked Questions
- Conclusion
What You Need
You need five things before the formula is worth anything: a defined asset, a defined period, operating hours for that asset, a count of unplanned failures, and a downtime total with timestamps. Anything less produces a number you cannot repeat next quarter.
The asset list
MTBF only means something when every asset in the calculation is the same kind of asset. The rule is blunt: never mix an injection molding press with a 40-ton press and a servo machine in one average. Same model, same configuration, ideally same age band.
Write down the asset ID, model, serial range and commissioning date for every unit in the group. If you are tracking a fleet of eight conveyor lines, all eight go in together or not at all.
Operating hours
Operating hours are the hours the machine was actually available to produce, which is why they exclude planned downtime. Get this figure from the machine controller, the production reporting system or a shift log, not from the calendar. A machine running two shifts is not a three-shift machine because someone was on site.
If your plant records run time in months or shifts rather than hours, convert before you calculate. A press that ran 22 days at 16 hours gives 352 operating hours, not 720 calendar hours for the month. Units have to line up or the MTBF is fiction.
A failure event
A failure is any unplanned stop where the equipment could not perform its intended function. Define it in writing and put that definition on the maintenance work order. Without a written definition, one shift calls a short alarm stop a failure and the next shift logs nothing.
Downtime timestamps
For each failure you need the time the machine stopped and the time it was released back to production. Those two stamps give you total downtime. If you want MTTR that reflects your crew, you also want repair start and repair complete, because the gap between them is wrench time and the gap outside them is where most plants lose hours.
A failure code
A short code per event, such as hydraulic, heater, sensor, controls or mechanical. You will want this later to see whether one failure mode is eating your MTBF on its own. Three failures all coded heater changes the maintenance conversation entirely.
A way to mark planned work
Every work order needs a planned flag. Scheduled preventive maintenance, a planned changeover and a planned utility outage all count as planned. Anything else is unplanned downtime and belongs in your MTTR numerator. One flag column removes most of the argument later.
What stays out of both numbers
Keep planned preventive maintenance, planned changeovers, scheduled shutdowns, quality hold time and production starvation out of the operating-hours numerator. Keep them out of the MTTR numerator too, unless your definition of MTTR deliberately covers all downtime. State which convention you are using and stick to it across every period you report.
Step-by-Step: How to Calculate MTBF and MTTR
Step 1: Define the Equipment and Measurement Period
Fix the asset boundary and the period before you touch any data, because the answer changes with both. Choose a period long enough to contain a meaningful number of failures. One quarter is a reasonable default for production equipment; a week with two breakdowns tells you almost nothing.
Write four definitions on the front of the worksheet: what counts as a failure, what counts as a repair, what counts as operating time, and what is excluded. Then note the observation window, for example April 1 through June 30.
Decide also whether you will report per asset or per fleet. Per asset is more useful for maintenance work; per fleet is more useful for a plant-level availability number. Do both, but keep the group identical every time.
Step 2: Collect Operating Hours and Failure Data
Pull operating hours from the controller or production system, then pull every unplanned work order in the window from the CMMS. Here is a complete raw dataset you can reproduce the calculations from:
| Event | Date | Failure code | Downtime start | Downtime stop | Downtime hours |
|---|---|---|---|---|---|
| F1 | Apr 9 | Heater | 06:12 | 10:42 | 4.5 |
| F2 | May 21 | Hydraulic | 14:05 | 21:05 | 7.0 |
| F3 | Jun 18 | Controls | 22:30 | 01:00 | 2.5 |
Note that F3 crosses midnight. That is exactly the kind of record that gets mis-entered by hand and quietly loses 22.5 hours from your total. Timestamp arithmetic in a spreadsheet handles it; a person reading a paper log usually does not.
Total unplanned downtime across the quarter is 4.5 + 7.0 + 2.5 = 14.0 hours. Total operating hours for the press is 1,000. Failure count is 3.
Step 3: Calculate MTBF
MTBF = Total operating hours ÷ Number of unplanned failures. For this press, 1,000 ÷ 3 = 333.3 hours between failures, which is about 13.9 days of run time on a two-shift schedule.
Three things belong in the numerator: operating time for the asset, across the whole period, with planned downtime removed. Three things belong in the denominator: the number of unplanned failures in that same period. Nothing else goes in either.
Two boundary cases catch people. If a machine is down for planned maintenance at the end of the period, the hours after that do not count as operating time. And if you replace a failed sensor, that is a repair, not a new asset starting a fresh clock. MTBF is per asset identity, so a repaired press keeps accumulating hours across its life.
Step 4: Calculate MTTR
MTTR = Total downtime hours ÷ Number of repairs. Using the same three events, 14.0 ÷ 3 = 4.67 hours to restore, which is 4 hours and 40 minutes per failure.
Now break the 14 hours down by bucket so you know where the time actually goes:
| Time bucket | F1 heater | F2 hydraulic | F3 controls | Total |
|---|---|---|---|---|
| Detection | 0.25 | 0.50 | 0.25 | 1.00 |
| Travel to the machine | 0.25 | 0.25 | 0.25 | 0.75 |
| Diagnosis | 0.75 | 1.00 | 0.50 | 2.25 |
| Waiting for parts | 1.50 | 3.00 | 0.25 | 4.75 |
| Repair and replacement | 1.25 | 1.75 | 1.00 | 4.00 |
| Testing and return to service | 0.25 | 0.25 | 0.00 | 0.50 |
| Administrative | 0.25 | 0.25 | 0.25 | 0.75 |
| Total downtime | 4.5 | 7.0 | 2.5 | 14.0 |
Wrench time is only the repair and replacement row: 4.00 hours across three failures, or 1.33 hours per event. That is the number your technicians actually influence with tools. The remaining 10 hours were detection, travel, diagnosis, waiting and paperwork, which is where most of the improvement opportunity sits.
The F2 hydraulic failure alone cost 3 hours waiting for a seal kit. One part, three hours of an eight-hour shift, and a store room that should have had it on the shelf.
Step 5: Interpret the Results
A 333-hour MTBF on a press that ran 1,000 hours in a quarter means roughly three unplanned stops per quarter, and that is what your PM interval, spares coverage and shift staffing should be built around. Compare it against your own history for the same asset first, then against sibling assets in the group, then against the OEM’s design life figure.
Be careful with manufacturer numbers. Published MTBF figures are often modelled or lab-derived, and sometimes optimistic by an order of magnitude. Treat a spec sheet as a lower bound, not as your operating reality. Where you only have a spec, say so in the report rather than presenting it as measured performance.
Two honest limits. First, small samples: with three failures the confidence interval is enormous, and you can honestly only say the MTBF is at least 333 hours. Second, the constant-failure-rate assumption behind a simple average, which will not reflect the bathtub curve, where young equipment fails from installation defects and old equipment fails from wear.
Turn the numbers into money before you act on them. Costing those 14 hours against contribution margin, or folding the maintenance burden into factory energy cost per part and total cost of ownership for a part, is what gets a reliability project funded rather than filed away.
How to Calculate MTBF and MTTR in Excel or Google Sheets
Spreadsheet steps are the fastest route to a repeatable number. Set up one work order log with these columns: A Asset ID, B Failure date, C Downtime start, D Downtime stop, E Downtime hours, F Failure code, G Planned, H Repair start, I Repair complete, J Wrench hours.
Put this in E2 and fill down: =(D2-C2)*24. That converts the timestamp difference into hours and handles the midnight crossing automatically. In J2 use =(I2-H2)*24 for wrench time.
Keep an asset list on a second sheet with Asset ID in column A and operating hours for the period in column C. On a summary sheet, put the asset ID in A2 and then:
=IFERROR(VLOOKUP(A2,'Asset List'!A:C,3,FALSE)/COUNTIFS(WorkOrders!$A$2:$A$500,A2,WorkOrders!$G$2:$G$500,"No"),0) gives MTBF in hours. The COUNTIFS does two jobs: it counts only unplanned events, and it points at exactly which rows are in the numerator.
=IFERROR(SUMIFS(WorkOrders!$E$2:$E$500,WorkOrders!$A$2:$A$500,A2,WorkOrders!$G$2:$G$500,"No")/COUNTIFS(WorkOrders!$A$2:$A$500,A2,WorkOrders!$G$2:$G$500,"No"),0) gives MTTR in hours. Swapping the E range for the J range gives you average wrench time instead, which is a more honest target for the crew.
For availability, add a cell that divides MTBF by the sum of MTBF and MTTR. The IFERROR wrapper matters, because a brand-new asset with zero unplanned failures returns a division error instead of a number.
One formatting habit pays for itself: store Asset ID and Failure code as plain values rather than merged cells or free text, or SUMIFS silently returns zero and your MTTR collapses to a believable-looking blank.
What Time Counts Inside MTTR
This is the single most disputed definition in plant maintenance, so settle it explicitly for your site and write it on the worksheet. The full downtime number includes all seven buckets below. Some plants report wrench time separately as a second metric, which is why the two numbers you see in different reports may differ.
- Detection. From the moment the machine stops to the moment anyone knows. Include it. A silent alarm nobody sees for an hour is a real loss of production.
- Travel. From the desk to the machine. Include it, especially on night shift when the technician is on another line.
- Diagnosis. Fault finding, checking the code, swapping a suspect sensor. Include it.
- Waiting for parts. Include it in MTTR. Excluding it makes parts availability look free when it is often the largest single cost of an unplanned stop.
- Repair and replacement. The hands-on work itself. Include it; this is wrench time.
- Testing and return to service. First-article checks, warm-up, proving the fault is cleared. Include it. Cutting cycles run the risk of a repeat failure.
- Administrative. Raising the work order, shift handover notes, permits. Include it if your process needs it, then track it so it shrinks rather than hides.
What never belongs in MTTR: planned preventive maintenance, planned changeovers, quality hold time, and time lost upstream because a feeder ran dry. Those sit outside the failure definition by design.
Turn MTBF and MTTR Into Availability
The two metrics are useful together because they combine into a single availability figure: A = MTBF ÷ (MTBF + MTTR). With our press at 333.3 hours MTBF and 4.67 hours MTTR, availability is 333.3 ÷ 338.0 = 0.986, or 98.6%.
This is inherent availability in its simplest form, driven only by failures and their repair times. Operational availability adds scheduled downtime into the denominator, and because operating hours already exclude planned time, you get availability = operating hours ÷ (operating hours + downtime hours). Plugging the same numbers in gives 1,000 ÷ 1,014 = 98.6% as well, because the ratio holds when the two inputs come from the same period.
The practical read is that this press is already highly available, and the 4.67-hour MTTR has little room to improve its availability further. Halving MTTR to 2.33 hours would lift availability to 99.3%, a 0.7 point gain. Extending MTBF instead, say cutting failures from three to two, moves it to 99.3% too. When both numbers are already strong, the cheaper intervention wins, and that is usually spares coverage rather than new equipment.
What Counts as a Good MTBF and MTTR Target
There is no universal number, and anyone quoting one is selling something. A good target depends on asset class, criticality and how the failure affects output, so benchmark in this order: your own history for the same asset, the same asset across a fleet, the OEM design life, then a peer plant in your industry if you can get access.
Process-critical assets such as injection molding presses, compressors and main conveyors justify far higher MTBF and tighter MTTR targets than non-critical auxiliaries, because an hour down on a bottleneck line costs far more than an hour down on a lab compressor. A sensible first target for a production-critical machine is to halve the MTTR you measured over the last two quarters, then hold it.
Track the number as a trend rather than a pass or fail. Three consecutive quarters of 5.0, 4.4 and 4.67 hours says something about your crew and your parts supply; a single quarter of 4.67 hours says little on its own.
MTBF, MTTF and the Rest of the Metric Family
MTBF applies only to repairable assets, since it assumes the machine goes back into service and keeps accumulating hours. For a non-repairable item like a bearing set or a lamp, you use MTTF, mean time to failure, calculated as total operating hours divided by the number of units that failed, with no repairs in the picture.
Be careful with MTTR itself. The acronym is overloaded across repair, recovery, resolve, respond and restore, and two departments can report different numbers while both calling it MTTR. Write the full words next to it whenever the number travels outside maintenance.
The neighbouring metrics split that restore timeline into pieces:
- MTTD, mean time to detect: failure occurs to the failure being noticed.
- MTTA, mean time to acknowledge: detection to a human accepting the ticket.
- MTTI, mean time to identify: acknowledgement to the root cause being known.
- MTRS, mean time to restore: root cause known to production resumed.
- MTBSI, mean time between service incidents: MTBF plus MTTR, which is a cycle time. Do not average it or compare it to MTBF.
Common Mistakes
Mixed planned and unplanned downtime. Including preventive maintenance in MTTR inflates it and hides the real breakdown cost. Flag planned work at entry and filter on it.
Calendar time instead of operating time. Dividing total elapsed hours by failures understates MTBF whenever a machine runs fewer than 24 hours a day. Use operating hours.
Counting one failure twice. When a technician resets a machine, pulls a part and the fault returns, that is still one failure event. Log each event once, with all the time attached to it.
Confusing MTTR with mean time to recovery. Recovery includes waiting for the next job, material or shift. Recovery is a production number, not a maintenance number.
Averaging different assets. A press and a fan in one MTBF tell you nothing. Filter to identical equipment.
Undercounting downtime to make a target. Stopping the clock when the technician walks away leaves the waiting-for-parts hours out, and the following quarter looks like a win that never happened. The bucket breakdown in Step 4 exists precisely to make that visible.
Reporting a vendor spec as your result. A published figure is a claim, not a measurement. Keep the two labelled separately on the dashboard.
Frequently Asked Questions
What is the MTBF and MTTR formula?
MTBF equals total operating hours divided by the number of unplanned failures. MTTR equals total downtime hours divided by the number of repairs. Both are averages across a period you define in advance, and both require a failure log to exist before you can compute them.
How do I calculate MTBF step by step?
Set the asset and the period first, then total the operating hours with planned downtime excluded. Count the unplanned failures in that same window. Divide hours by failures. On a press with 1,000 operating hours and 3 breakdowns that gives an MTBF of 333.3 hours between failures.
How do I calculate MTTR and MTBF in Excel?
Put asset ID, timestamps and a planned flag in one sheet, add a helper column with =(D2-C2)*24 for downtime hours, then use SUMIFS over the downtime column and COUNTIFS over the asset and planned columns. Divide one by the other for MTTR, and divide operating hours from your asset list by the same COUNTIFS for MTBF.
Do you include waiting for parts and travel time in MTTR?
Include them. Detection, travel, diagnosis, waiting for parts, hands-on repair, testing and administration together make up full downtime MTTR. Excluding waiting for parts flatters the metric and hides a real production cost. Report wrench time, the hands-on portion only, as a separate number if you want to see crew performance.
What is the availability formula using MTBF and MTTR?
Availability equals MTBF divided by the sum of MTBF and MTTR. With an MTBF of 333.3 hours and an MTTR of 4.67 hours, availability is 333.3 divided by 338.0, or 98.6%. Adding planned downtime into the denominator gives operational availability, which is the figure most plants report to leadership.
What is a good MTBF target?
There is no single good MTBF. Process-critical assets such as injection molding presses, compressors and conveyors justify far higher targets than non-critical auxiliaries. Benchmark against your own history for the same asset and against siblings in the same fleet first, and treat any manufacturer specification as a lower bound rather than an expected result.
Conclusion
To calculate MTBF and MTTR properly, start with one asset group and one quarter. Pull the operating hours and every unplanned work order from the same period, split the downtime into its buckets, and divide. Then take the two numbers to the people who can act on them.
Maintenance reads the numbers as spare parts coverage and work order flow. Production reads them as PM intervals and shift staffing. Finance reads them as lost contribution. Hand the same 333 hours and 4.67 hours to all three and the next decision writes itself.
Recalculate every quarter, keep the definitions written down and unchanged, and within a year you will have the trend line that no vendor brochure can give you.