What Is MTTR?

MTTR is the mean time to repair: the average time a repairable system is out of service after a failure, measured from the moment it stops doing its job to the moment it is doing it again. It is a measure of maintainability, meaning how quickly a failure can be put right, not of how often failures happen.

The letters are used in more than one way. The IEC dependability vocabulary, IEC 60050-192, uses mean time to restoration, and IT teams often say mean time to recovery, respond or resolve. The idea is the same in each case, but the start and stop points differ, so two MTTR figures can only be compared if they are measured the same way.

For a computer on a production line, a vehicle or a trackside cabinet, the time that counts is usually the whole outage. That includes noticing the fault, getting someone to site, finding a replacement, restoring the software and configuration, and proving the system works, not just the minutes spent with a screwdriver.

Add up the downtime caused by failures over a period and divide it by the number of repairs. If a line PC failed three times in a year and was out for 2 hours, 6 hours and 1 hour, its MTTR was 9 ÷ 3 = 3 hours.

The figure is only as honest as the clock behind it. A record that starts when the engineer arrives will look far better than one that starts when the line stopped. A complete MTTR usually includes:

The two metrics look at opposite halves of a system’s life. MTBF, mean time between failures, is the average running time from one failure to the next, and a higher figure is better. MTTR is the average time spent down after each failure, and a lower figure is better. MTTF, mean time to failure, is the MTBF equivalent for items that are replaced rather than repaired, such as a fan or a fuse.

They are also usually produced differently. Supplier MTBF figures are typically predictions calculated from component data using methods such as MIL-HDBK-217F or Telcordia SR-332, as our guide to what MTBF means for industrial computers explains. MTTR is mostly a property of the site: the same computer can have an MTTR of 30 minutes in a plant with spares on the shelf and two weeks at a remote site that waits for a return.

Where no site records exist yet, maintainability can be estimated at the design stage. MIL-HDBK-472 sets out procedures that add up the task times for fault isolation, removal, replacement and checkout. For a computer that is swapped as a whole, that estimate is usually dominated by logistics and software restore, not by hands-on time.

Availability combines the two. For a repairable system, inherent availability is A = MTBF ÷ (MTBF + MTTR), one of the expressions set out in IEC 61703. The table holds MTBF at 50,000 hours and changes only the repair time.

The figures assume a constant failure rate, so they are averages across a fleet, not a promise for one machine. The point they make holds anyway: with a sound MTBF, repair time decides how much downtime you see. Cutting MTTR from two weeks to a few hours is usually cheaper than buying a system with a higher MTBF.

Most of the gain comes from decisions made before anything fails.

As standard, we build each system to a documented configuration and image, so an identical replacement can be prepared and swapped in with as little downtime as possible.