How to Plan an Industrial Shutdown That Finishes On Time

A turnaround is the largest single project most plants execute, and it is the one they are worst at.

The numbers are unflattering and consistent. Industry benchmarking of maintenance turnarounds finds that more than two-thirds exceed their planned cost or schedule by at least 10%, or suffer a trip after startup, and that roughly 40% experience an overrun of more than 30%. Not 10% over. Thirty.

What makes this striking is that turnarounds are not novel work. The same plant does one every few years, largely the same scope, often with the same contractors. The engineering is well understood. And yet the outcome distribution looks like a research programme rather than a repeated industrial process.

The reason is that overruns are almost never caused by the technical work. They are caused by decisions made, or not made, in the year before anybody picks up a spanner. This article sets out what turnaround planning actually requires, why scope discipline is the single dominant variable, and how to build a shutdown that lands when it was supposed to.

Why Turnarounds Overrun

Strip the post-mortems down and the same root causes appear in nearly every one.

Scope added after the freeze date. This is the dominant cause, and it is worth being precise about the mechanism. A late addition does not merely add its own duration. It disrupts a schedule that had already been optimised, competes for the same craft resources as planned work, requires materials nobody ordered, and forces re-sequencing that idles crews elsewhere. A job that would take two days in isolation can cost five days of critical path when inserted into a running turnaround.

Discovery work. Some of this is genuinely unavoidable. You cannot see inside a closed vessel or a sealed turbine casing, and inspection reveals things. But the proportion that is true discovery should be modest if pre-shutdown inspection, condition monitoring and history review were done properly. Where discovery work is large, it usually indicates that the plant did not know the condition of its own equipment going in.

Poor front-end loading. Turnarounds that are planned late are planned badly. Materials are ordered against assumptions rather than scope, contractors are engaged without adequate briefing, and the schedule is built at a level of detail too coarse to manage.

Resource shortfalls. Skilled turnaround labour is finite and regionally concentrated, and outage seasons cluster because everybody wants the same low-demand window. A plan that assumes a crew size the market cannot supply is a plan that fails at execution.

Weak decision authority. When a finding emerges at two in the morning, someone must decide whether it is repaired now, deferred, or run to the next outage. If that decision requires three approvals through people who are asleep, the crew waits. Across a three-week turnaround these delays accumulate into days.

Startup failures. A turnaround that finishes on schedule and then trips the unit within a week has not succeeded. Commissioning is part of the turnaround, not an afterthought, and rushing it to recover schedule is how a well-executed outage becomes a forced outage.

The Phases, and How Long Each Really Takes

Turnaround management is organised into phases, and the most common planning error is compressing the early ones because they produce no visible progress.

Long-range planning begins twelve to twenty-four months before execution for a major event. This is where the window is fixed against production and market conditions, the outline scope is established, the budget is framed, and long-lead procurement is identified. Siemens Energy advises approximately eighteen months of lead time for major turbine overhaul planning, and that figure is representative of large rotating equipment generally.

Scope development runs from roughly twelve months out. Every candidate job is identified, justified, challenged and either accepted or rejected. This is where the turnaround is won or lost, and it is covered in detail below.

Scope freeze happens typically four to six months before execution. After this date, additions require formal change control with named authority, and the bar should be genuinely high. This date is the single most important date in the entire process.

Detailed planning follows the freeze. Every accepted job is broken into tasks with durations, craft requirements, materials, tooling, permits, isolation requirements and safety documentation. The schedule is built and the critical path identified. Contractors are contracted against a defined scope rather than a hope.

Pre-shutdown work captures everything that can be done while the plant is running: scaffolding erected, materials staged and kitted per job, isolations prepared, permits drafted, contractors mobilised and inducted. Every hour moved out of the shutdown window is an hour of production recovered, and this is one of the highest-leverage activities available.

Execution is the shutdown itself. If the preceding phases were done properly, execution is largely a matter of managing progress against plan, handling discovery work within contingency, and keeping the critical path moving.

Startup and commissioning returns the plant to service in a controlled sequence with function testing of protection systems.

Post-turnaround review captures what was found, what it cost, what overran and why, and what the next turnaround should do differently. This phase is skipped more than any other and it is the only mechanism by which a plant gets better at turnarounds.

Planning a major outage with no scope defined yet? Our maintenance and asset management team builds inspection-driven turnaround scopes and work packages.

Benchmarked turnaround performance across industry. Overruns are driven by planning decisions rather than by the technical work. Source: Benchmarking and Optimizing Maintenance Work Scope for Turnarounds, Asset Performance Networks.

Scope Discipline Is the Whole Game

If a plant improves one thing about its turnaround practice, it should be scope challenge, because scope drives duration, duration drives cost, and lost production usually dwarfs the maintenance spend.

The failure mode is familiar. A worklist is compiled from every request submitted since the last outage. Nobody wants to be the person who rejected the job that later failed, so items accumulate. The list arrives at planning bloated with work that is not condition-driven, is not risk-driven, and in some cases is not necessary at all.

A disciplined process challenges every item against a small set of questions. What is the evidence this work is needed, and is that evidence condition data or is it habit? What is the consequence of not doing it this outage? Can it be done with the plant running, either now or later? Does it genuinely require the shutdown, or has it simply always been done during one? Is it on the critical path, and if so, is its value proportionate to the production it costs?

That last question is the one plants ask least and should ask most. A job that adds two days to the critical path costs two days of output. If the plant's daily contribution is large, the value of that job needs to be very large indeed.

Condition evidence is what makes this challenge possible. A plant that has vibration trends, oil analysis, thermographic surveys and performance data can say with confidence that a particular pump does not need opening this cycle. A plant without that evidence can only guess, and guessing conservatively means opening everything. This is the strongest practical argument for condition monitoring: not that it catches failures, though it does, but that it lets you take work out of a turnaround with your eyes open. The relationship between condition evidence and intervention decisions is the same one set out in our guide to preventive versus predictive maintenance.

The related discipline is contingency for discovery. Because some findings are genuinely unforeseeable, the schedule and budget must carry explicit allowance for them, and the authority to approve discovery work must be delegated to someone on site with the competence to decide. Pre-agreeing decision rules for common discovery scenarios, before the outage starts, removes hours of delay from the critical path.

Long-Lead Items Decide the Date

The most avoidable overruns come from materials. A turnaround does not finish when the work is complete. It finishes when the last part arrives, and parts do not compress their lead times because a plant is losing money.

Diaphragms, turbine blades, large bearings, specialised valves, heat exchanger bundles, transformers, custom fabrications and certain instrumentation carry lead times measured in months. These must be ordered against anticipated scope during scope development, before findings confirm them, which means accepting that some ordered material may not be used.

That feels wasteful and it is not. The alternative is a crew standing idle for six weeks. The correct comparison is the carrying cost of a spare against the cost of lost production, and for most plants that comparison is not close.

The same logic applies to the specialists. Turbine overhaul crews, balancing engineers, non-destructive testing technicians, certified welders for specific procedures, and heavy lift contractors are finite and heavily booked in outage season. Securing them early is part of planning, not part of procurement.

Staged procurement, in which material is ordered in sequence against the schedule and delivered to a kitted, job-numbered laydown area, is what separates a turnaround where crews start on time from one where a supervisor spends the first morning of every shift looking for gaskets.

The turnaround planning sequence, with the scope freeze at four to six months out as the single most important date. Source: MIMAH engineering practice; lead-time guidance consistent with Siemens Energy overhaul planning recommendations.

Building a Schedule You Can Actually Manage

A turnaround schedule has one job: to tell you, at any moment, whether you are ahead or behind and on which path.

Identify the critical path explicitly and manage it separately. Most of the work in a turnaround has float. A small proportion does not, and every hour lost on that path is an hour added to the outage. Critical path activities deserve the best crews, the closest supervision, and the first call on any resource.

Plan to the shift, not to the day. Turnarounds run continuously. A schedule at daily granularity cannot detect a half-shift slip, and half-shift slips are how a three-week outage becomes a four-week one without anybody noticing until week three.

Sequence around isolations and permits, not just around craft. In practice the constraint is frequently not whether a fitter is available but whether the system is safely isolated, the permit is issued and the confined space is certified. These processes have their own duration and their own bottlenecks, typically a small number of authorised people, and a schedule that ignores them will not survive contact with the plant.

Hold a short daily review with authority in the room. Progress against critical path, discovery items requiring a decision, resource conflicts, and safety observations. Decisions made in that meeting rather than deferred are what keep the schedule intact.

Track completion by work package, not by percentage. Percentage complete on a turnaround is an opinion. Work packages signed off as complete, with quality documentation attached, are a fact.

Startup Is Part of the Turnaround

A depressing share of turnaround failure is measured after the plant is nominally back. A unit that trips within days of startup has converted a planned outage into an unplanned one, and it usually happens for one of a few reasons: commissioning was compressed to recover schedule, protection systems were not function tested, an isolation was left in place, or a control setting changed during the outage was never restored.

Guard against this deliberately. Function test every protection and shutdown system rather than assuming it works. Verify alignment hot as well as cold on rotating equipment. Re-baseline vibration at the new condition so that any future change is measured against reality rather than against pre-outage data. Confirm every temporary modification, blind, jumper and forced signal has been removed, using a register maintained through the outage rather than memory.

And measure performance after startup. A turnaround that included efficiency work should show it: heat rate improvement from restored turbine seal clearances or a cleaned condenser is measurable, and given that a single percentage point of heat rate is worth roughly $700,000 a year on a 500 MW coal unit, it is worth verifying that the money was actually delivered. This is where turnaround work and the power plant energy audit discipline meet, and where the turbine work described in our steam turbine overhaul guide proves its value.

One more item that is routinely forgotten: the standby power that carries the site through the outage. A shutdown frequently transfers critical loads onto generators for extended periods, which is precisely the duty those machines are least often prepared for. The considerations are covered in our diesel generator maintenance guide, and the time to discover a battery problem is not the morning the main supply is isolated.

Need turnaround scoping, work packaging or execution support? Our industrial engineering team covers outage scheduling, staged procurement and turbine and controls work packages. Our project record includes a plant audit at KRPC in Nigeria that generated four subsequent turbine and controls contracts, and a major steam turbine generator overhaul for NNPC.

The five questions that should be asked of every item on a turnaround worklist. Condition evidence is what makes it possible to take work out with your eyes open. Source: MIMAH engineering practice.

Frequently Asked Questions

How far in advance should turnaround planning start? For a major event, twelve to twenty-four months, with the scope freeze four to six months before execution. Smaller outages compress proportionally, but the sequence does not change. The most common and most costly error is starting late and then compressing the planning phases, which produces the appearance of being on track right up until execution.

What is a scope freeze and why does it matter so much? It is the date after which no work is added without formal change control and named approval authority. It matters because a late addition disrupts a schedule that had been optimised, competes for resources already committed, and often requires materials nobody ordered. Late scope is the single largest driver of overrun in benchmarked turnarounds.

How much contingency should a turnaround carry? Enough that discovery work does not consume the critical path, with the amount informed by how much is genuinely known about equipment condition going in. A plant with strong condition monitoring and recent inspection data can carry less; a plant opening equipment it has not seen in a decade should carry considerably more. Contingency should be explicit and visible, not hidden inside optimistic durations.

Should turnaround work be done by contractors or in-house staff? Almost always a mixture. In-house staff carry plant knowledge and stay after the contractors leave. Contractors provide the surge capacity that no plant can justify carrying year-round, plus specialist capability. What matters is that contractors are engaged against a defined scope well in advance rather than mobilised against an evolving worklist.

How do we know whether our turnaround was successful? Four measures, and all four are needed. Did it finish within schedule and budget? Did the plant start up cleanly and stay up? Was the work actually completed to specification, with documentation? And did performance improve where the scope said it would? A turnaround that finished early because work was quietly dropped is not a success.

What should a post-turnaround review capture? Actual against planned duration and cost by work package, all discovery work with its root cause, every schedule slip with its reason, materials that arrived late or were never used, and the specific decisions that would be made differently. This document is the input to the next turnaround's scope development, and skipping it guarantees repeating the same overruns.

The Outage Is Won Before It Starts

By the time the plant comes down, the outcome is largely determined. The scope is what it is, the materials are ordered or they are not, the contractors are briefed or they are guessing, and the schedule is either built to a level of detail that can be managed or it is a bar chart on a wall.

Execution matters, and good supervision recovers real time. But no amount of execution excellence rescues a turnaround whose scope was frozen too late, whose long-lead items were ordered against confirmed findings rather than anticipated ones, and whose plant did not know the condition of its own equipment before the first flange was broken.

The plants that consistently land on schedule are not the ones with the best contractors. They are the ones that start planning a year out, challenge every job on the worklist against evidence, freeze the scope and mean it, order material against anticipated scope, and hold a review afterwards that is honest enough to be useful.

Have a turnaround approaching and a worklist nobody has challenged yet? Talk to our engineering team. We will review the scope against equipment condition, identify what can come off the critical path, and build work packages that can actually be executed in the window you have.