After an annual shutdown, failures cluster in the first days of production. The machine has been opened, adjusted and reassembled while the team was reduced, and the work done at that moment is rarely written down. The restart reveals all at once everything that passed without a trace.
The pattern repeats every year, on sites that otherwise master their maintenance plan. Industrial intelligence platforms such as Mimorian model equipment and support fault diagnosis for field teams, so that adjustments made during a shutdown remain available at restart.
What actually happens during an annual shutdown?
The annual shutdown concentrates into a few weeks what the rest of the year spreads out: removing components, replacing wear parts, reworking wiring, changing parameter settings, returning equipment to service. It is the period when the machine changes the most.
It is also the period when the team is at its smallest. Some technicians are on leave, contractors work on specific lots, and the systems integrator who knows the machine in detail spends two days on site then leaves. Each person does the right thing, and nobody holds the full thread of what has moved on the equipment.
The result shows at restart. The first production batch comes out, faults appear, and the technician on duty looks for the origin of behaviour that did not exist before the shutdown.
Why the restart concentrates faults
Three mechanisms stack up at restart.
The first is the untracked change. A parameter adjusted on a panel, a sensor moved back by a few millimetres, a contact bridged for the duration of a test: this work happens fast, under the pressure of the return to service, and is rarely recorded. At restart, the technician works from a drawing that no longer matches the machine.
The second is the hand that did the work. According to a Vanson Bourne study for ServiceMax, 23% of unplanned downtime in manufacturing comes from human error, against 9% in other sectors [Vanson Bourne/ServiceMax, 2017]. A period when unfamiliar people work on equipment they barely know brings exactly those conditions together.
The third is accumulation. The faults that appear at restart come from the sum of the shutdown interventions rather than from a single action. Without a record of each one, diagnosis starts from a blank page while the answer sits in what was done three weeks earlier.
The knowledge missing at restart
The case is clear on imported machines, where the PLC program is commented in the manufacturer's language and the original parameter settings were never explained to the team. A site running that kind of equipment depends on a handful of people: the maintenance manager who learned the machine on the job, and the outside integrator who commissioned it.
When those two people are on leave at the same time, which is exactly what happens during a shutdown period, the remaining team faces a machine that has moved and nobody can say what changed. The knowledge exists, but it lives in two heads that are unavailable.
That cost is measurable at company scale. Panopto's workplace knowledge report puts at $47 million a year the productivity lost in a large organisation through inefficient knowledge sharing [Panopto, 2018]. On a restart, that missing sharing is paid in extra downtime hours, at the very moment production has to ramp back up.
Prepare the restart during the shutdown, not after
The useful window sits during the shutdown itself, while the work is still fresh. Three practices change the restart.
Record each change as it happens, with the drawing page concerned, the parameter touched and its value before and after. Voice dictation makes that action workable in the field, where a form to fill in at the end of the day stays empty.
Attach each record to the equipment rather than to the intervention. A technician arriving at the machine at restart looks for the history of the machine, not the report of a job they took no part in.
Make the whole thing readable by anyone, including on the night and the weekend of the restart. Shutdown knowledge is useful precisely when the people who produced it are away.
When those conditions are met, restart diagnosis starts from what was done during the shutdown. The technician finds the list of changes on the equipment concerned and compares the current state with the expected state. The first reflex becomes reading, rather than reconstruction.
Frequently asked questions
Does every action during the shutdown need documenting, even the smallest?
The most useful actions to record are those that change the state of the machine: a parameter, a sensor position, wiring, a part replaced with a different reference. An intervention that follows the preventive plan is recorded in one line.
Our contractors do not fill in our tools, what can we do?
The record can come from the person on site who signs off the intervention, by voice, at the end of the visit. What matters is that the information reaches the equipment, not that the contractor enters it.
Our restarts go well, is this still useful?
A restart that goes well often rests on the presence of one person who knows. The record becomes valuable the day that person is away, on leave or has left the site.
Conclusion
Three points to remember.
An annual shutdown changes the machine more than any other period of the year, with the smallest team on site.
Restart faults come mostly from changes made during the shutdown that left no record.
The window for action sits during the shutdown, at the moment of the work, before diagnosis has already begun at restart.
A concrete next step for a maintenance manager: list the interventions from the last shutdown on the most critical piece of equipment, and check how many of them left a record that can still be found today.
To place this practice in a wider approach, see our guide to capitalising maintenance know-how. On what an hour of downtime really costs, see the real cost of unplanned downtime in industry.