Lean tools

Autonomous maintenance (jishu hozen)

In French: Maintenance autonome

Autonomous maintenance entrusts routine upkeep — cleaning, inspection, lubrication, retightening — to those who run the equipment, so that degradation is detected before the breakdown.

The principle: cleaning is inspecting

The founding idea comes down to one sentence: you cannot clean a machine without touching it everywhere, and you cannot touch it everywhere without noticing what has changed. An incipient leak, abnormal play, a loose bolt, overheating are thus discovered well before they stop production.

Cleaning therefore has no cosmetic purpose here. It is the means of a regular inspection, done by the person who knows the normal behaviour of the equipment best because they spend their days with it.

That is also what makes the first deep clean so instructive. It reveals in one go all the accumulated degradation, and the list that comes out of it forms the programme for the following months.

Sharing with maintenance

The approach does not transfer maintenance to the operators: it redistributes. To the operators, the daily gestures and the detection of anomalies; to maintenance, the technical interventions, failure analysis and equipment improvement.

That sharing frees up maintenance time, and that is where success is decided. If the time freed is absorbed by other emergencies, nothing changes lastingly; if it is reinvested in prevention and improvement, the number of breakdowns falls and the approach sustains itself.

It assumes an explicit agreement on the boundary — which gestures, how far, and from what sign one calls. Left implicit, that boundary produces either hazardous interventions, or operators who dare not touch anything.

How it is put in place

It starts with one piece of equipment, not with the workshop. A pilot machine makes it possible to build the standards, measure the effect and train the first operators, who then become the best relays for the next ones.

It requires short, visual standards: what is checked, how often, what is normal, and what to do when it is not. A long, exhaustive document is not consulted; a sheet displayed on the machine, with visible marks on the points to check, is.

It finally requires making the check points accessible. Many inspections are not done because a guard has to be removed or one has to contort oneself: opening an inspection window, fitting a level gauge or moving a grease point costs little and determines whether the gesture will really be done.

The common mistakes

Launching the approach to cut the maintenance headcount. The intention is perceived immediately and cooperation disappears; maintenance then has every interest in the approach failing, and it fails.

Writing standards without making the checks accessible. A check point behind a screwed-on guard will not be checked, and the sheet will nonetheless be signed — which is worse than no sheet at all.

Stopping after the initial deep clean. It is the most visible and the most rewarding step, and it produces no lasting effect if the daily standards do not take over.

On the shop floor — Plastics processing

The first complete clean of a press brought out eleven minor hydraulic leaks, none of which was reported. The operators knew them all and regarded them as normal, the equipment having run for years in that state. Two of them were the origin of the repeated stoppages that maintenance had been looking for elsewhere for six months.

The other tools illustrated by this trade: plastics and chemicals.

Frequently asked questions

Which gestures are entrusted to the operators?

Cleaning, visual inspection, lubrication at the designated points, retightening and the small adjustments defined in advance, as well as reporting any departure from normal behaviour. What requires dismantling, a diagnosis or an intervention on safety components stays with maintenance — and that boundary must be written, not assumed.

How do you get the maintenance teams on board?

By setting out from the start what the freed time will make possible, and sticking to it. The legitimate fear is that of a headcount reduction; the credible answer is the move from endured troubleshooting to prevention and improvement work, which is also more interesting. That promise is worth something only if it survives the first busy quarter.

How much time should be devoted to it each day?

A few minutes at the start or end of the shift are enough for the routine checks, provided that time is recognised in the workload and not taken from production. Unplanned time is the first sacrificed as soon as the day gets tight, and the approach then dies out without anyone having decided to stop it.

Does autonomous maintenance replace preventive maintenance?

No, they complement each other: one provides daily monitoring and detects incipient degradation, the other intervenes according to a plan on components that require specific skills and means. What the first changes is the quality of the information the second has to adjust its plan.