Executive summary
- The pillar has eight stages, counting step zero: preparation.
- Steps 1 and 2 cut MTTR. Only from step 3 onward does MTBF move.
- Without a failure database, every later step is guesswork.
- Machine waiting time has seven components; most of them are not repair time.
- It ends with an audit system, not with a software purchase.
The planned maintenance pillar — Professional Maintenance in many plants — is the one the maintenance department owns outright, and the one most often run out of order. The correct sequence does not begin with rewriting the PM schedule. It begins with knowing what actually fails, how often, and where the time goes once it has failed.
Why MTTR and MTBF must be separated
MTTR is the average time it takes to get the machine running again. MTBF is the average time it runs before it fails again. They respond to completely different levers. MTTR falls through management — spare parts, information, work procedure. MTBF only stretches through engineering, that is, by actually fixing the weakness in the machine.
A plant that announces a PM overhaul and starts by rewriting every check sheet is trying to stretch MTBF without knowing why the machine fails. The result is a thicker check sheet, more technician hours consumed, and numbers that do not move.
Step 0 — Preparation
Step 0 gets skipped because it does not look like work, yet it decides whether the other seven steps can be measured at all. Before starting you need one equipment register with one numbering scheme across the plant, a criticality ranking, a definition of breakdown that production and maintenance both accept, and baseline MTTR and MTBF figures recorded before anything changes.
An undefined breakdown cannot be counted
If production counts an eight-minute stop as a breakdown but maintenance does not because no work order was raised, the two sets of figures will never agree, and every meeting will end in an argument about the data rather than a decision about the machine.
Step 1 — The machine failure database
This step builds an equipment failure history that can actually be used, not a pile of old work orders. The minimum record for one event is: which machine, which component, what the symptom was, when it stopped, when it ran again, who attended, what they did, and what was replaced.
- Record every event, including the two-minute stop the technician cleared himself.
- Classify by component, not by machine name — otherwise all you learn is that line 3 breaks down.
- Keep symptom separate from cause. The symptom is what was seen; the cause comes later.
- Collect at least three months before drawing any trend conclusion.
Once this database has been running for about three months, nearly every plant finds the same thing: most of the downtime concentrates on a handful of components, and several of those items have been repaired repeatedly.
Step 2 — Improving machine waiting time (the seven MWTs)
When a machine stops, very little of the lost time is repair. Most of it is waiting. Breaking that waiting into seven categories shows immediately where to attack first — and almost all of it can be fixed without touching the machine.
| Waiting time | Usually caused by | Addressed by |
|---|---|---|
| Waiting to be told | The operator tries to fix it first, or does not know whom to call | A clear call-out route and a rule on when to escalate immediately |
| Waiting for a technician | The technician is in another building or on another job | Zoned responsibility and a job priority rule |
| Waiting for diagnosis | No history for this machine, so the fault is traced from scratch every time | The step 1 database and a machine-specific troubleshooting guide |
| Waiting for a spare | Not held, held in the wrong place, or held but unknown | A criticality-based spares list and a store you can search |
| Waiting for a tool | The special tool has been borrowed, or has no home position | Visual tool boards located in the area |
| Waiting for approval | The job exceeds a limit, or a signature is needed | A delegated spend limit for emergency work |
| Waiting for hand-back | The repair is done but nobody has inspected or trial-run the machine | A hand-back procedure with a named owner |
The outcome of steps 1 and 2 is a visible fall in MTTR within a few months, while MTBF may not move at all. That is normal and correct: nothing done so far has addressed a single cause of failure.
Step 3 — Corrective maintenance and maintenance prevention
This is the first step aimed squarely at MTBF. Corrective maintenance means modifying the machine or the component so the same failure cannot recur. Maintenance prevention means feeding what you learned from the existing machine back into the specification for the next one, before it is ordered.
- Pick from the step 1 top five, not from the supervisor’s impression.
- Every corrective job states how the result will be measured, and when.
- Record the modification in the equipment register, or in three years nobody will know why this machine differs from the one beside it.
- The lesson has to reach the purchasing specification for new equipment.
Step 4 — The maintenance management system
Four parts that have to move together: the maintenance information system, spare parts management, budget management, and scheduling. The point is that this is step four, not step one. A CMMS installed before the data is any good becomes a repository nobody trusts.
Software does not improve the data. It only makes bad data available faster.
Step 5 — Improving the PM programme
Only now is there enough evidence to rebuild the PM schedule rationally. The work is not adding checks; it is deleting the checks that have never found anything and moving the frequency onto the components the failure history says matter.
| Question to ask of every PM task | If you cannot answer |
|---|---|
| Which failure mode does this task prevent? | Delete it, or find the real cause first. |
| What has it actually found in the last twelve months? | Reduce the frequency, or convert it to a condition check. |
| Where did this interval come from? | The manufacturer’s manual is a starting point, not the answer. |
| Does the task require a shutdown? | If not, move it to a running check. |
Step 6 — Predictive and condition-based maintenance
This step measures machine condition while it runs, to forecast when it will fail: vibration, thermography, oil analysis, ultrasound. It sits at six rather than one because these tools only tell you something is deteriorating; they do not tell you what to do about it. A plant that cannot get a spare part within two weeks gains very little from two weeks of warning.
Step 7 — The maintenance audit system
The last step is what stops the previous six from decaying: an audit against written criteria, carried out by someone who does not own the work, on a fixed cycle. What is audited is not whether the paperwork exists, but whether a technician picked at random can explain what his machine fails on most often, and what was done about it last time.
Sequence beats speed
A plant that does steps 1 and 2 properly for six months will always end up further ahead than one that announces all seven steps at once in a single quarter.