Fitting a Drone Inspection Into a Shutdown or Turnaround Window

A good deal of drone inspection work exists precisely to avoid an outage. You fly the flare stack while the unit runs, you keep the scaffold crew at home, and the comparison that justifies it is set out in the article on drone versus scaffolding for elevated inspections.

This is about the other situation. The turnaround is already on the calendar. The budget is approved, the contractors are booked, and the question is not whether to avoid the outage but how to get useful drone work done inside one. That is a scheduling problem more than a technical one, and it is lost or won well before the unit comes down.

Three windows, and they answer different questions

Almost every avoidable disappointment on turnaround drone work comes from flying the right scope in the wrong window. There are three, and each is suited to different things.

• Before the outage, asset running. The asset is hot, pressurised and energised. Thermal is at its most useful, external visual condition can be captured without competing with anyone, and the output is a work list. This window is the most under-used and usually the most valuable.
• Inside the outage, asset down. The asset is cold, isolated and opened up. Internals become reachable, confined spaces can be entered or flown, and surfaces normally hidden by insulation or operating conditions are exposed. Thermal is largely useless here.
• After restart. The asset is hot again. Verification that the repair worked, and a fresh baseline for the next cycle. This is where a thermal scan earns its place a second time.

Name the window for each scope in writing. A line item that just says drone inspection will get scheduled into whichever window has a gap, which is how a thermal survey ends up flown on a cold unit.

The pre-outage flight is the one that changes the work list

If a turnaround is going to include drone work at all, the flight that pays for itself most reliably is the one before the outage starts, because it is the only one whose findings can still change what the outage does.

What it produces:

• Evidence-based scope. A work list built from current condition rather than from the last cycle's assumptions and the loudest opinion in the planning meeting.
• Scope you can remove. Just as useful as scope you add. An item carried for three cycles because nobody could see the asset is an item you can drop once somebody can.
• Access planning. Knowing what the repair actually involves before booking the scaffold, the crane and the trades.
• A dated pre-outage record. Condition at the point of handover to the turnaround organisation, which settles later questions about what was damaged during the work.

What makes it work is that external capture on a running asset competes with nobody. There is no permit queue, no crane conflict, no simultaneous operations problem. It is the cheapest window to fly in and the only one whose output can still move money.

Your deadline is the scope freeze, not the outage start

This is the single most important scheduling point and it is routinely missed. A turnaround work list locks well before the unit comes down, because materials have to be ordered, contracts let and crews scheduled. After that date, adding work is a change with a cost attached.

So the deliverable has to be in a planner's hands before the freeze, which means working backwards through four steps rather than two:

• Your own review and decision time, which is usually the step people forget to allow for.
• Report production.
• Processing, which on a large asset with thermal is not a same-day activity. What drives it is set out in how long from flight to deliverable.
• The flight itself, plus a contingency date, because wind and visibility do not negotiate. The thresholds are discussed in the article on wind limits on a drone survey.

Add those up, add a weather margin, and the resulting date is when the inspection has to be commissioned. It is almost always earlier than people expect, and it is the reason pre-outage drone work gets squeezed out: not cost, but a deadline nobody back-calculated.

A cold unit has nothing for a thermal camera to find

Worth stating plainly because it is counterintuitive to anyone who has not worked with radiometric imagery. Thermal inspection finds anomalies by detecting a temperature difference: a hot spot, a cold spot, heat escaping where it should not, a pattern that differs from its neighbours. The defect becomes visible because of the heat flowing through it.

A unit that has been down for days has equalised. There is no flow, no load, no energisation and therefore no differential. The camera works perfectly and sees nothing of interest. The conditions that make thermal work at all are set out in the article on the conditions that make or break a drone thermal inspection, and an outage removes most of them.

The consequence for planning is firm rather than preferential:

• Thermal on electrical gear, insulation, refractory, steam systems and solar arrays belongs in the pre-outage or post-restart window, under load.
• Inside the outage, thermal is useful only where a genuine differential still exists, which is a narrower set of cases than people assume.
• If a thermal scan is specified for a turnaround and the only slot offered is during the outage, that is worth raising rather than accepting, because the deliverable will be technically compliant and practically empty.

Internal capture is locked to the outage

The mirror image of the thermal problem. Some scopes cannot be flown at any other time, and those are the ones that should be competing for space inside the window.

Vessel and tank interiors, columns, stacks, flare tips, ducting, silos, ore passes and similar enclosed assets are only reachable when the asset is down and opened. A collision-tolerant aircraft can cover them without a person going in, which is the whole argument set out in the article on confined space drone inspections.

Two things are worth being precise about:

• The drone replaces the entry, not the isolation. Lockout and isolation, blinding, purge, and atmospheric testing all still apply before anything goes into a vessel, and the confined space is still a confined space. What is removed is a person inside it, along with the rescue standby and much of the access structure.
• It is still an activity on a schedule. An internal flight needs the asset open, cleaned enough to see, lit or self-illuminated, and free of other work for its duration. That is a slot, and slots inside a turnaround are contested.

Inside the outage, the drone competes with everyone

A drone inspection is low impact. It is not no impact, and a turnaround is the one environment where that distinction bites.

• Permit to work. The flight needs one, raised in advance, like any other activity. A crew arriving to discover the permit has not been raised loses the day.
• Overhead space. Crane lifts, man baskets and scaffold erection all occupy the same volume the aircraft wants. These are not negotiable in the moment and the drone will lose.
• Simultaneous operations. Hot work, grit blasting, chemical cleaning and pressure testing each create exclusion zones that may rule out a flight regardless of who booked first.
• People density. A turnaround puts far more people on site than normal operations, and bystander standoff requirements are driven by how many of them are in the operating area and whether they can be cleared.
• Escorts and orientation. The crew needs site-specific induction and, on most plants, an escort. That is a person taken off something else, which is exactly what a turnaround organisation does not have spare. The general version of this is covered in how to prepare your site for a drone survey.

The fix is organisational rather than technical: get the flight onto the turnaround schedule as a named activity with a time slot, a permit and an escort assigned, in the same system as every other job. Work that sits beside the schedule as a visiting contractor gets displaced on the first bad morning.

The data has to come back faster than the decision window

A standard inspection reporting turnaround is often longer than the window in which the finding could have been acted on. Inside an outage that mismatch makes an otherwise good inspection worthless, because the repair crew demobilised on Thursday and the report arrived on Monday.

The answer is to agree a staged deliverable before the work starts:

• Immediate flagged findings. Within hours of the flight, a short list of anything that needs a decision inside the outage, with locations and imagery and no formatting. Enough to act on, not a report.
• The full report afterwards. Complete, annotated, severity-ranked, formatted for the records and the next cycle, delivered on the normal timeline.

Say this in the scope, because it changes how the contractor resources the job. It is also worth agreeing what a flagged finding means, so the immediate list is short enough to be useful. What a defensible full report should contain either way is covered in how to tell a good drone inspection report from a bad one.

After restart: verification and the next baseline

The third window is the cheapest to schedule and the easiest to drop, which is a shame because it does two jobs.

It verifies. A repair carried out inside the outage can be confirmed under load, and thermal works again because the differential is back. A refractory repair, an insulation replacement or an electrical termination can be checked doing its actual job rather than sitting cold.

It also resets the baseline. The post-restart capture is the reference the next cycle's pre-outage flight gets compared against, which is what turns one-off inspections into a condition trend. Setting that interval sensibly is the subject of how often you should re-fly a site.

What to settle before the outage

• Which window each scope belongs in, written down, with thermal explicitly placed under load.
• The scope freeze date, and the commissioning date back-calculated from it with a weather contingency.
• The flight as a named activity on the turnaround schedule, with a slot rather than a general availability.
• Permit to work raised in advance, and the named escort.
• Known simultaneous operations conflicts identified against the proposed slot.
• For internal work: who confirms isolation and purge, and who certifies the space before the aircraft goes in.
• A two-stage deliverable, with the definition of a flagged finding agreed.
• Who receives the immediate list and who has authority to act on it inside the outage.
• Whether a post-restart verification flight is in the budget, because it is far easier to include now than to approve afterwards.

None of this is about the aircraft. A turnaround is a scheduling machine, and drone work succeeds inside one when it is treated as an activity in that machine rather than as a technology that shows up. Put it on the schedule, put the thermal under load, and make sure the findings land before the work list locks.