Skip to content
STOIntelligence
All insights

What expires when you defer a turnaround

Tom MankowskiFounder & PrincipalAugust 26, 20266 min read

A turnaround plan looks like one thing. It is not. It is a stack of separate commitments, each validated at a different moment, each good for a different length of time, and every one of them quietly assuming the event happens roughly when it was said it would. Move the date and you find out which of them were perishable.

Most teams find out in the field. That is the part of a deferral nobody prices.

Why the question is live right now

In August 2026, US diesel crack spreads hit a record above $100 a barrel, far above historical levels. Refiners have been running near 95% utilization, about as hard as the fleet physically goes, and reported maintenance downtime early in the year came in well below the prior two years. When a day of runtime is worth several times its normal margin, the case for pushing planned work to the right writes itself, and some operators are making that call.

The economics can be real. None of what follows is an argument against deferring. It is an argument about what you owe the plan afterward.

Two boundaries decide whether deferral is even on the table

Integrity first. Economic deferral only exists inside inspection intervals, mechanical-integrity limits, safety cases, and regulatory commitments. Margin does not move a statutory inspection or an end-of-run reliability limit. Where those bind, the date is not yours to move, and no business case changes that.

And it is a site-specific call, not an industry position. Equipment condition, unit configuration, contractor commitments already signed, and permit obligations all vary. A deferral that is sound at one plant is reckless at the next. “Everyone is pushing to the right” is a market story, not a plan.

A deferral is a new event, not the old one on a later date

Deferring does not remove the work. It moves it, and it grows it. More runtime means more condition findings, more deferred inspections, and more accumulated scope. When the unit finally comes down, the event is larger than the one you deferred, and the plan you are executing was built for the smaller, earlier one.

The shape is familiar. Refiners who ran hard through the high margins of 2022 contributed to a pile-up of planned and unplanned outages the following year, as deferred work and normal cycles landed in the same window. COVID-era deferrals did the same thing on a longer lag. The risk does not disappear when you defer. It concentrates.

What actually expires

Here is the more useful question. Stop asking whether the plan is still good, and start asking which parts of it have expired, because they do not expire together and they do not expire on a clock.

Scope

The premise
Expires when the reason for the event changes. Re-validated by restating the premise for the new date and re-testing every discretionary item against it, not by inheriting the old one.
Condition data and inspection findings
Expire with runtime. The findings that justified the original list were taken at one point on the corrosion and fatigue curve, and you have moved along it. Re-validated by fresh data, not by the previous report.
The discretionary list
Expires the moment items are added to it. Re-validated by re-scoring the whole list on one scale, not by bolting additions onto a ranking made under different conditions.

Schedule logic

Walkdowns and work packages
Expire on change, not on time. Isaac Limon, a lead turnaround planner, put it better than I have heard it put: a deferred event needs every package re-walked, not just re-dated. Drawings drift, material gets reallocated, and durations and resources that were verified become assumptions again while the schedule still shows them as verified. Re-validated by walking the job again.
Material and long-lead items
Expire when someone else needs the steel. A reservation held for a deferred event competes with a live one. Re-validated by re-confirming the delivery, not the purchase order.
Contractor commitments and the resource curve
Expire on the contractor’s calendar, not yours. Crews booked for the original window get sold, and the market you priced is not the market you will execute in.
The seasonal window
Does not expire so much as invert. A spring event pushed into autumn or winter is a different execution risk on weather, daylight and craft availability, while the schedule still carries the old productivity assumptions.

Readiness gate

The risk register and contingency
Expire when the scope changes. Contingency sized for the smaller event is not contingency for the larger one, it is a shortfall with a name on it.
The funding envelope
Expires with the estimate underneath it.
The gate decision itself
Expires immediately. It was a decision about a specific event, of a specific size, on a specific date, and that event no longer exists. Re-validated only by taking the decision again on the event you now have.

What is deliberately not on that list is the statutory and integrity boundary. That is not a line item to re-validate. It is the fence the whole decision sits inside.

Deferral moves more than your event

The second-order effect gets missed almost every time, because an event does not sit alone on a site.

Added or moved scope does not sit quietly beside the existing plan, it changes how the neighboring jobs execute. Access someone was counting on gets taken by work now running in the same area. A tower that was never going internal is now getting blinded, which changes the isolation and testing plan on a piping job tied off the same nozzle. And the version of this that gets missed most often is the one Isaac Limon raised: a unit that no longer comes down, because it was deferred, changes the isolation plan for the event next door, one that nobody deferred and nobody re-planned.

So the cost of a deferral does not land only on the deferred event. Part of it lands on the events around it, which is exactly where nobody is looking for it.

The half of the business case nobody writes

A deferral case is usually built one way: margin captured by continuing to run, against margin foregone by taking the unit down. That is a clean comparison and it is the wrong one, because it has no cost column.

What belongs in that column:

  • Re-validation effort. Re-walking packages, re-inspecting, re-estimating. Real hours, from the same people who are already loaded.
  • Scope growth. The additional work the extended run generates, and the additional window it will need.
  • Remobilization and market exposure. Rebooking crews and re-sourcing material at a price you do not control.
  • Interface cost on neighboring events. Rarely counted, and it lands on teams who had no say in the decision.
  • The probability of an unplanned outage during the extended run, priced against the very margin you extended the run to capture. That is the sharp end: the conditions that make deferral attractive are the conditions that make an unplanned outage expensive.

Put those in and the comparison changes shape. You are not comparing running against not running. You are comparing a smaller event now against a larger event later, plus the cost of re-underwriting the plan in between.

So how long is a plan good for?

I put that question back to him: how long does a walkdown stay good before you re-walk instead of re-date? His answer was that it is not a length of time. Any scope change large enough to matter sends him back out to walk the job, even late in planning, on an event whose date never moved.

That instinct is right and it generalizes. Expiry here is triggered by change, not by elapsed time. A six-month deferral that adds nothing expires less of the plan than a two-month deferral that adds one major work item. Which means the question is never how stale the plan is. It is what changed, and what that change quietly invalidated.

Where this lands

None of this is an argument against deferring. It is an argument that a deferral you did not pair with a reopened scope, a re-tested schedule, and a fresh readiness gate is a deferral you are only half managing.

Which is why moving an event raises the bar on independent assurance rather than lowering it. Scope development and challenge re-tests the list on one scale instead of accepting additions onto a ranking made under different conditions. Schedule assurance stress-tests the logic, the critical path and the resource basis against the event you now have. And a readiness review gives leadership an independent read on the larger, later event, while the plan can still change.

The plan you inherited was built for a different event. Re-underwriting it is the half of the deferral that actually protects the restart.

Want a defensible read on your next event?

Start a conversation