Mean Time to Repair

Mean time to repair is the responsiveness half of equipment reliability. Where mean time between failures measures how often things break, MTTR measures how quickly the team can get the machine back. Both halves matter, and the right one to attack depends on the failure pattern in the shop. A machine that breaks rarely but takes a day to recover has very different problems than a machine that breaks weekly but recovers in 20 minutes. MTTR is what separates those cases.

"Reliability is half the picture. Recovery is the other half."

How mean time to repair works

The calculation is total repair time divided by the number of failures over the same window. The window has to be long enough to give a meaningful sample, usually a month for a busy machine or a quarter for one that fails less often. The repair time per event starts when the machine stops because of the failure and ends when the machine is back in production making good parts. This is wider than just the technician's wrench time. It includes:

• Diagnostic time, the period spent figuring out what failed.
• Wait time for the technician to arrive at the machine.
• Wait time for parts to arrive.
• Actual repair work.
• Verification and restart, including the first parts that get checked.

The biggest MTTR reductions in a small shop usually come from the non wrench time. Parts that should be on the shelf but are not. Procedures that should be written down but are in someone's head. Wait time that could be eliminated by training a second person to do the diagnostic. A shop that tracks MTTR by category often finds that the actual repair is the shortest segment; the rest is the system.

MTTR combines with mean time between failures to produce availability. The formula: availability equals MTBF divided by the sum of MTBF and MTTR. Improving MTTR is often the faster lever because it does not require redesigning the equipment. It requires improving the maintenance system around the equipment.

Where mean time to repair fits on the shop floor

Picture a 30 person fab shop with a laser cutter that is the bottleneck for a third of the work. The laser has decent reliability, with an MTBF around 200 hours, but each failure takes an average of six hours to recover. The two most common failures account for 80 percent of the events: a lens contamination issue and an optical alignment fault. Both have known fixes; both currently take far longer than they should.

A focused MTTR effort would do three things. One: pre stage the parts most likely to be needed (lenses, alignment tools, a spare bellows kit) on a labeled cart next to the machine. Two: write a one page procedure for each of the two common failures so a competent operator can start the recovery while waiting for the maintenance lead. Three: set up a vendor agreement for the next tier of parts that beats the current two day wait. After the project, MTTR on those two failures drops from six hours to under two. Availability climbs from 97 percent to 99 percent, and the bottleneck capacity that was being lost now belongs to the shop.

Common mistakes with mean time to repair

• Starting the clock at technician arrival. Reaction and travel are part of the recovery. Start at the failure event.
• Excluding parts wait time. Waiting two days for a bearing is two days of MTTR. Pretending otherwise hides the problem.
• Averaging across all failure types. A spindle rebuild and a sensor swap give no actionable average together. Bucket by failure type when volume allows.
• Tracking MTTR without MTBF. You need both to know which side to invest in.
• Treating MTTR as a maintenance department metric only. Operators, schedulers, and parts staff all affect MTTR. Make ownership shared.

Mean time to repair and related Lean tools

MTTR pairs with mean time between failures to compute availability, the A in OEE. Every minute of MTTR is a minute of unplanned downtime. MTTR is most effectively reduced by combining strong preventive maintenance procedures with pre staged parts and standard repair work that anyone trained can execute.