The Eight Pillars

Concept one

The eight, and what each is actually for

1. Autonomous Maintenance. Lesson six. The operator owns the daily condition of the equipment.

2. Focused Improvement. A cross-functional team takes one named loss and removes it, with data, to a conclusion. This is where the OEE you learned to measure finally gets used, and it is the pillar most often skipped — plants measure diligently and then improve nothing in particular.

3. Planned Maintenance. Lesson seven. Strategy, intervals, spares, MTBF and MTTR.

4. Quality Maintenance. Not inspection. The aim is to identify the equipment conditions under which defects cannot occur, then hold those conditions — temperature, alignment, pressure, tool condition. Inspection finds bad parts after you have paid to make them; quality maintenance prevents the conditions that produce them.

5. Education and Training. A skill matrix that says who can do what, and a plan to close the gaps. Without it, autonomous maintenance is operators told to inspect equipment nobody ever taught them to read.

6. Early Equipment Management. Feeding everything you learned about maintaining this machine into the specification for the next one. Maintainability, access, standard parts, the ability to clean it without dismantling it. This is the only pillar that prevents a loss from being purchased in the first place.

7. Office TPM. The same loss thinking applied to the administration around the plant — scheduling, procurement, planning. A machine idle for want of a purchase order is a maintenance loss with an office cause.

8. Safety, Health and Environment. Not a pillar in the sense of being optional. An improvement that raises output and lowers safety is not an improvement, and TPM treats zero accidents as a condition of the programme rather than a benefit of it.

Concept two

They hold each other up, and 5S holds all of them

Underneath the eight sits 5S — sort, set in order, shine, standardise, sustain. It is not a ninth pillar and it is not a cleaning campaign. It is the floor the pillars stand on: you cannot notice an abnormality in a place where everything is abnormal, and you cannot halve MTTR when the tool is somewhere in the shop.

The pillars are worth naming separately, but they do not work separately.

Autonomous Maintenance without Education and Training gives you operators instructed to inspect equipment nobody taught them to read. They tag nothing, and the programme is judged a failure of enthusiasm when it was a failure of teaching.

Planned Maintenance without Early Equipment Management means you get very good at maintaining a machine that should never have been bought in that configuration — and then you buy another one exactly like it.

Focused Improvement without measurement is a suggestion scheme. Measurement without Focused Improvement is a dashboard. Neither one moves a number.

Quality Maintenance without Planned Maintenance is holding process conditions on equipment that is drifting out from under you.

This is why TPM is described as a system rather than a toolkit. Any single pillar produces a small, real improvement and then stops, and the plant concludes TPM does not work — when what it actually tried was one eighth of it.

Concept three

Attack the largest loss, not every loss

Eight pillars is a lot of programme, and the commonest way to fail is to launch all of it at once. The alternative is not to pick a favourite pillar. It is to pick a loss, and let the loss tell you which pillar you need.

That is what the five previous lessons were for. You can now split a machine's downtime into six named losses, and each of those losses has an owner: breakdowns to Planned Maintenance, small stops to Autonomous Maintenance, setup to Focused Improvement, defects and startup to Quality Maintenance and Early Equipment Management.

Then the arithmetic decides. Halving your largest loss is almost always worth more than halving anything else, and frequently more than halving several smaller ones together. This is the same logic as a bottleneck: improving anything other than the constraint makes the effort look busy and the number stay still.

There is one qualification worth carrying, from lesson five. Where a defect is caught matters as well as how large it is, because a part rejected at final inspection carried the whole chain with it. So the rule is: attack the largest loss, and break the tie in favour of the loss that occurs latest.

Do that once, finish it, and prove it with the number. A single closed loss with evidence behind it will buy you more support for the next one than any launch event ever will.