Before adding another job to the shutdown, ask what it displaces

A planned shutdown creates an attractive opportunity: the equipment will already be stopped, so why not complete the extra work at the same time?

Imagine a site preparing a major maintenance window. The agreed scope includes inspections, component replacements and work needed before restart. Then a department adds a useful modification. A contractor proposes another repair while access is available. Someone remembers a recurring defect that never reached the formal list.

Each suggestion may be reasonable. Together, they can compete for the same scaffolding, isolation boundary, lifting equipment or specialist crew.

AI can help examine the scope. It needs to make those dependencies visible before it recommends more work.

Protect the work that cannot be negotiated away

Start by identifying statutory inspections, safety requirements and other obligations confirmed by the responsible specialists. Keep their evidence and applicable deadlines attached to the relevant work.

These jobs must not enter a generic ranking exercise where a low estimated financial return can push them out. The approved engineering and safety process determines whether any change is permissible.

For the remaining work, ask what justifies doing it during this particular shutdown. Some tasks require the equipment to be unavailable. Others can be completed during routine operations or a later planned stop. Some requests still lack enough information to judge.

Preserving that distinction gives the review a useful starting point. It also avoids turning every submitted idea into an apparent commitment.

Examine readiness as well as desirability

A desirable intervention may be poorly prepared. Materials have not arrived. The contractor has priced labour but excluded access. The inspection could reveal additional work for which no contingency exists.

An assistant can compare requests with purchase orders, drawings, work history and contractor scope. It can highlight missing prerequisites and conflicting assumptions. It should leave unanswered questions visible rather than filling them with plausible estimates.

The scope owner then has a choice: resolve the missing preparation, adjust the intervention or defer it where permitted. “Ready pending confirmation” needs a named confirmation and a deadline. Otherwise, it becomes a polite way of hiding unfinished preparation.

Ask what an addition changes

For a proposed extra job, examine four practical effects:

• Does it extend an isolation or prevent another crew from entering?
• Does it require a resource already committed to a time-critical activity?
• Could it reveal a condition that changes the restart plan?
• What would be removed, moved or resourced differently to accommodate it?

A scheduling model can make these conflicts easier to see. The quality of its recommendation depends on the dependencies it has been given. If the lifting operation is absent from the plan, the model cannot reliably protect it.

Keep the original scope and the revised proposal side by side. A change record should explain the operational consequence, the responsible approval and the resulting schedule. Counting accepted additions alone rewards growth without showing its cost.

Handle discoveries without pretending the plan was complete

Once equipment is opened, new information will appear. The point of scope control is to make those decisions manageable, not to deny the condition found on site.

Give discoveries a defined route. The competent specialist assesses the finding, the shutdown team examines execution implications, and the authorised owner decides the action. Safety and restart requirements remain binding.

An assistant can collect photographs, inspection notes, related history and the affected schedule items for that decision. It can also show which earlier assumptions are now invalid. It should not quietly merge the discovery into the plan as though it had always been agreed.

Use the next shutdown to improve the estimates

At close-out, compare the planned scope with the work actually performed. Separate added work caused by newly discovered conditions from additions caused by incomplete preparation. Record why planned work was removed and what obligation or follow-up remains.

Review duration estimates against comparable activities. A longer intervention may reflect an additional repair rather than a poor estimate for the original task. Without that distinction, the next model learns from a misleading history.

The useful output is a better preparation process and clearer future choices. A shorter shutdown is valuable only if the equipment is safely returned with the required work completed and outstanding items explicitly handled.

Bring a shutdown scope and its late changes to a Strategy Session. We can examine where AI would help expose dependencies before the window starts.