More than two thirds of turnarounds exceed their planned cost and schedule by 10%, or trip shortly after startup. 40% overrun by more than 30%.
Maintenance shutdowns overrun for 2 underlying reasons: there is too much work in them, or the work runs inefficiently.
Everything else feeds one of those two.
In this article I'll walk you through the 5 reasons shutdowns underperform, based on what I've seen across oil and gas, mining and chemicals. Then I'll cover a sixth cause that the industry rarely names out loud, even though the benchmarking data points straight at it.
You'll also get the numbers to measure yourself against, and the specific practices that prevent most of this.
Let's start with something more uncomfortable than the causes.
You are probably measuring your shutdown wrong
Ask most teams when their shutdown finished and they'll tell you the date the last work pack was signed off.
That's not when it finished.
A shutdown runs from steady-state production to steady-state production. You ramp down. You make the plant safe. You execute the scope. You pre-commission and commission. You restart, and you ramp back up.
And at that point, and only at that point, is the shutdown event complete.
This matters more than it sounds. In a shutdown that goes badly, shutting the plant down takes longer than anticipated. Making it safe takes longer. Not all the work gets finished before start-up begins. Then start-up is later than promised, and the ramp-up is plagued by trips and interruptions.
Stop the clock at mechanical completion and none of that shows up in your numbers.
So before you ask why your shutdowns overrun, check what you are counting. Plenty of "on-time" shutdowns are only on time because the measurement stopped early.
What the data actually says
You'll see it claimed that 80% of turnarounds go over budget, or that 82% fail to meet performance requirements. Those figures get attributed to a 2019 study by T.A. Cook and Solomon Associates.
I can't find it. Neither can anyone else. There is no published sample, no methodology, no paper.
Here's what is real. Asset Performance Networks maintains a turnaround database of more than 1,350 events across upstream oil and gas, gas processing, refining, chemicals and power generation. Over 60% of them are high or mega complexity events.
From that database:
- More than two thirds of turnarounds exceed their planned cost and schedule by 10%, or have a trip after startup.
- 40% experience a cost overrun or schedule delay of more than 30%.
Those numbers are worse than they first look, and I'll come back to why in a moment.
AP-Networks also names the causes, and their list is short: poor scope control before the shutdown, high rates of discovery during it, poor planning and preparation, and unrealistic cost and schedule targets.
Hold onto that last one.
The 5 reasons shutdowns underperform
This is the framework I teach, and it is built from what actually goes wrong on site rather than from what looks tidy on a slide.
Reason 1: Poor planning and schedule discipline
Poor scope preparation is where this usually starts.
The scope gets agreed but never properly planned, which produces discovery work nobody anticipated and jobs that have to be planned on the fly.
Then there are poor repair or replacement strategies. Sometimes they don't address accessibility or safety properly. Sometimes the chosen strategy simply doesn't deliver a quality outcome.
Job conflicts are the one that catches teams out. During execution you realise 2 jobs scheduled to run concurrently can't actually run at the same time.
Here's a real example. You have a hot work job one level above another worksite. But you have grating instead of solid flooring, so the work site below is suddenly exposed to sparks from above. Now the work can't be done concurrently, and your schedule has a hole in it.
And then there are late scope additions.
This one is almost always a discipline problem rather than a technical one. Very often that scope was known about somewhere in the organisation for months. It just wasn't managed or communicated. People badly underestimate what late scope does to planning and preparation.
Reason 2: Wrong scope
Wrong scope plagues most shutdowns, and it comes in 3 varieties.
Unnecessary scope. A time-based overhaul of a pump when the pump's condition is good and nothing suggests it needs opening.
Scope that could be done online. The principle here is simple: if the work can be done when the equipment is running, then it should be done when the equipment is running. Every job you move out of the shutdown window is a job that can't overrun it.
Missing scope. The one that bites you later, when an integrity or reliability issue you didn't address forces an unplanned or early shutdown.
AP-Networks put it plainly in their benchmarking work: minimising the amount of scope, and the level of scope growth during execution, is the primary driver of competitiveness. Scope is the biggest lever you have.
Reason 3: Flaws in operational support
Operations own more of your shutdown outcome than most maintenance teams admit.
Delays shutting the plant down. Delays draining or decontaminating systems and vessels.
Or worse, not making the plant safe properly, so vessels still contain flammable or toxic material when people open them.
That last one is a safety incident waiting to happen, and it shows up when organisations underestimate the effort involved in making systems safe.
Which is exactly why you want experienced operations people in shutdown preparation. Ideally people who have been through previous shutdowns on that plant.
The rest of this category is familiar: poor isolations, late isolations, delays issuing permits, and delays or incidents during start-up.
Reason 4: Flaws during execution
Insufficient HSE preparation and control, leading to incidents or delays. Insufficient quality control and supervision.
Inefficient logistics for mobilising and demobilising people, equipment and materials.
Poor management of discovery work sits in here too. Sometimes that's slow reporting. More often it's slow decision-making about what to do once something is found.
Late equipment handover from one group to the next is another common one, and it has more to do with poor communication and unclear reporting lines than anything technical.
Then poor sequencing of returning equipment to service, which delays commissioning and start-up.
There are many underlying causes. The most common are insufficient preparation, supervision and control of the work and the workforce.
Reason 5: Organisational aspects and capability
An insufficiently skilled workforce. Not adhering to a proven shutdown management process. Poor collaboration across disciplines, with people working in silos.
Not learning from past shutdowns, either because the lessons were never captured, or they were lost, or worse, they were captured and then ignored.
And a contracting strategy that does not support what the organisation says it wants from the shutdown.
The sixth reason: the target was never achievable
Now the one the industry rarely says out loud.
Some shutdowns don't overrun. They were given a number they were never going to hit.
AP-Networks names this directly as one of their four causes:
"Planned turnaround duration and cost targets are often established by the business far in advance of the turnaround, and are not related to the scope that actually has to be implemented. Turnaround teams have no choice but to live with these targets, realizing that there is little chance of success."
Gordon Lawrence, also at AP-Networks, describes the mechanism. In the capital projects world, you decide how much money is needed to complete a defined scope. In the turnaround world, the budget gets set in stone at the earliest estimate stage, and the question becomes how much scope can be squeezed into a fixed number.
His words: the stage is set for over-optimism about the amount of scope that can be done to clash with the hard barrier of what the finance department wishes to spend.
So the team inherits a target built before anyone knew the scope, and then gets measured against it.
I'm not offering this as an excuse. If your shutdown ran 30% over, that is still 30% of lost production and spent money. But if you're running a review and every finding points at execution, ask a harder question: was the target ever connected to the work?
If it wasn't, no amount of execution discipline was going to save it. And you'll make the same mistake next cycle.
The number to measure yourself against: 19% versus 7%
Here's the benchmark I'd put on the wall.
Across those 1,350-plus turnarounds, industry average scope growth from scope freeze to execution is 19%. Top quartile performers grow by 7%.
That's the whole game in 2 numbers.
Everyone freezes scope. Almost nobody holds it.
The gap between 19% and 7% is the difference between a team that treats scope freeze as a decision and a team that treats it as a suggestion.
And it is measurable. You already have the data. Take your frozen worklist, take what you actually executed, and calculate the growth. If you're at 19% you're average. If you're above it, you've found your biggest single opportunity.
Worth knowing: AP-Networks also found that scope-optimised turnarounds required 30% fewer labour hours than those carrying more scope. And a structured risk-based scope review can strip 30% or more out of a worklist.
Your contingency is almost certainly too small
Remember I said the overrun numbers were worse than they look?
This is why.
At a turnaround industry conference in Amsterdam, 83% of respondents said their turnaround control budget was intended to be a plus or minus 10% estimate. The remaining 17% said it included enough reserve to be a "not to exceed" number.
Not one said their budget was a plus or minus 30% estimate.
Then AP-Networks compared the control budget against actual cost for 133 refinery turnarounds. Those turnarounds overran their estimates by an average of 14%, with an accuracy range closer to plus or minus 25% than the plus or minus 10% everyone assumed.
So the industry builds budgets it believes are accurate to 10%, and delivers something nearer 25%. Meanwhile 40% of turnarounds overrun by more than 30%.
It gets more specific. Only 21% of turnaround teams show allowances as a defined line item in the estimate at all. Of those that do, just 6% break them into separate categories. Teams include an average of 11% allowance, and Lawrence's assessment of that number is blunt: clearly insufficient.
In my course I teach a contingency of 10% to 20%. Based on this data, plan for the upper half of that range, and know why you're doing it.
It helps to split your unknowns into 3 buckets rather than one:
- Emerging work appears after scope freeze but before execution starts. It is either work that should have been captured during scope gathering and was forgotten, or equipment that broke down in the meantime.
- Discovery work is what you find once you open equipment up.
- True contingency covers inefficiency: things that simply take longer or cost more than expected.
Lump them together and you'll under-provide for all 3. Separate them and 2 of the 3 become estimable. If your scope gathering and challenge are done properly, forgotten work should fall towards zero. And breakdown work can be estimated from your own maintenance history. If you normally lose 3 valves a month and you're 9 months out, allow for 27 valves.
Does this apply outside oil and gas?
Fair question, because almost all the published benchmarking data comes from refining and petrochemicals.
In my experience, yes.
Mining, power, chemicals and refining shutdowns differ in scale, complexity and impact. A statutory inspection outage on a gas plant and a mill reline are not the same event. But the principles of effective shutdown management hold across all of them.
I've also used these same principles successfully on short-duration, high-impact outages, where the whole event is measured in days rather than weeks.
What changes is the scale of the consequences. What does not change is that scope discipline, front-end loading and a governed process are what separate the shutdowns that hold their schedule from the ones that don't.
So treat the refining numbers as directional rather than universal. The causes are the same everywhere.
What actually prevents this
You can't eliminate uncertainty from a shutdown. Something unexpected always happens.
What you can do is stop manufacturing your own problems.
Start earlier than feels comfortable. Define your shutdown strategy 24 months out. Start finalising scope 18 months out. Begin detailed planning at least 12 months out. Those dates scale with the size and complexity of your event, but don't dismiss them too quickly. Experience says you can't be early enough.
Front-end load the preparation. Early on, changes are cheap and easy. Find out during preparation that a repair needs a special weld procedure and you can get the procedure in place and your welders qualified. Find out during execution and it costs you money and schedule.
Challenge the scope, properly. Review every item on the worklist and ask whether it is genuinely a shutdown task, whether it's genuinely needed now, and whether it could be done online instead.
Then freeze it, and mean it. Scope freeze aligns every department on the same worklist. After that date, new scope goes through change control or it doesn't go in. Your 19% is decided here.
Put reviews between the phases. At the end of each phase, present readiness to a shutdown steering committee. They review the work, the quality of it, and whether you're ready to move on. Or not. A gate that can't say no isn't a gate.
Decide on discovery work within 24 hours. Discovery work will happen. What kills schedules is slow reporting and slower decisions.
Close out the review within 3 months. Capture lessons learned, review performance against cost, schedule and safety, and run an after action review. For complex shutdowns, get it facilitated by someone impartial. Reviews that drag past 3 months do not get finished, and lessons that are not captured get relearned the expensive way.
What this looks like in practice: Malampaya, 2006
In 2006 I was Head of Maintenance Execution for the Malampaya gas project in the Philippines, responsible for day to day maintenance on the offshore platform and the onshore gas plant.
The shutdown had been in preparation for 18 months, run by a dedicated shutdown manager with a dedicated shutdown team. I wasn't part of that preparation.
But during a shutdown, all routine and corrective work stops and everything gets handed over. So I went offshore as one of the shutdown coordinators, working 12 hour shifts, helping run a crew of 300 people around the clock. Statutory inspections, restoration work and project upgrades, all in one event.
I picked it up about 3 weeks before execution started.
Here's what I walked into. We had 300 people mobilising to a platform, hundreds of work orders, and no mapping between the two. Nobody had assigned people to supervisors, or supervisors to jobs.
So we ran group interviews. Offshore. Three hundred people, none of them known to us, working out who had done what kind of work before so they could be matched to work packs.
That should have been done onshore, months earlier.
It wasn't the only gap. Large diameter flange bolts turned up in poor condition. The replacements we'd specified never arrived, so every bolt going back into a flange had to be tested and checked first. Hired equipment that suppliers had certified as tested failed in service, including a critical item for the flare tip replacement. That job never got completed. And progress reporting hadn't been set up properly, so for a while nobody had a clear picture of where the work actually stood.
Then 2 typhoons came through, in what was supposed to be the good season. The accommodation barge had to leave the platform for several nights and our POB dropped from 300 to around 60.
We still completed 95% of the scope. The organisation called it a success, and given the circumstances it was.
But look at that list again.
The bolts, the hired equipment, the reporting, the crew mapping. Not one of those was caused by a typhoon. Every one was a preparation gap that surfaced during execution, on a shutdown with an 18 month runway and a dedicated team.
You can't plan away the weather. You can plan away the rest.
And that's the point of the 6 causes. Almost everything on that list was decided long before anyone went offshore. Yours will be too.
If you want the full process behind this, SD400: Introduction to Shutdown Management walks through all 6 phases, from strategy through to after action review. It is a short course and you'll get through it in an afternoon.
And if you have a major shutdown coming and you'd rather work through it with your own team in the room, we run an on-site workshop where we do exactly that.
