MTTR (Mean Time To Repair) is the average time a machine stands still when it breaks down — from the moment it stopped to the moment it runs again. It answers the question: “How long does it last when it breaks down?”
Formula
MTTR = total downtime due to failures ÷ number of failures
Example
Press PR-04 had 5 failures in one week, lasting 12, 35, 18, 9 and 26 minutes. A total of 100 minutes. MTTR = 100 ÷ 5 = 20 minutes.
MTTR is not just repair
When you break it down, MTTR almost always shows that the repair itself is the smaller part. A typical 26-minute failure:
- Waiting to be noticed — 4 min (the operator first tries on their own).
- Waiting for a technician — 9 min (they did not hear the call, they were at another machine).
- Diagnosis — 5 min.
- Waiting for a part — 6 min (the part is in the warehouse at the other end).
- Repair and a test cycle — 2 min.
Only 2 of the 26 minutes are “repair”. The rest is response time and organisation — and that can be solved without a new tool or a new technician.
How to reduce it
- A fast, clear call — Andon that goes straight to the technician’s phone, with the machine and the reason, instead of searching around the hall.
- Escalation — if nobody accepts within 10 minutes, the call goes to the next level.
- Parts at hand — a stock of the parts that are used most often, based on history.
- Recording the cause — the next technician knows what happened last time.
Response and repair separately
It is useful to look at MTTR in two parts: response (from the call to the technician’s arrival) and repair (from arrival to resolved). If the response is long, the problem is organisation. If the repair is long, the problem is knowledge, parts or the machine itself.
In MachineTakt
The technician accepts the call on their phone and marks “Arrived” and “Resolved” with the real cause. This way response and repair are measured separately, without any writing. MTTR and MTBF are visible by machine in the Maintenance module.