Building a Refinery Turnaround Inspection Scope, From Premise to Mobilisation

The scope starts as a due list, not a wish list

A turnaround scope is not assembled by asking operations what it wants opened. It starts from the inspection due list generated out of the integrity management system: every item whose examination interval expires before the next scheduled window, with its current corrosion rate, its remaining life calculation and its last recorded findings. That list is arithmetic, and arithmetic is what survives a scope challenge.

The premise document sits alongside it and states what this particular turnaround exists to achieve, which is what allows an item to be defended or dropped. Items retaining an operating licence are non-negotiable. Items maintaining production capability compete on economics. Items upgrading the asset compete against capital. An inspection item that cannot be sorted into one of those categories will be cut, and cutting it is the right decision on the information presented.

This is the first place data structure converts into money. An item whose interval is supported by a documented corrosion rate, a stable CML register and a traceable history is defended in ninety seconds. An item supported by a technician's recollection and a folder of PDFs is cut, comes back as emergent work mid-execution, and costs several times what it would have cost in the plan. The evidence discipline is set out at /blog/proving-rbi-inspection-intervals-to-a-regulator.

The scope challenge is where inspection earns its budget

Every major US refinery runs a formal scope challenge in which each proposed item is defended in front of a group whose job is to remove it. The challenge is not hostile; it exists because an unchallenged scope grows without limit and every added item consumes hours from a window measured in weeks. AP-Networks characterises that window as three to five weeks against a four to six year cycle.

Inspection items are defended on three grounds and no others: the interval expires within the cycle, the corrosion rate or damage mechanism makes deferral untenable, or prior findings created a specific commitment. Each of those is a data claim, and each can be evidenced or not. Items defended on the fourth ground, that it has always been done, are the ones that disappear.

The output is an examination register: which items, which methods, what extent, what acceptance criteria, and what access each requires. That register is the document the whole turnaround plans against, and it is the reason an inspection contractor should read it before pricing. A register that specifies extent but not access is a register that will be repriced in the field, and repricing in the field is the most expensive conversation on site.

Scope freeze is a commercial event, not a technical one

Freeze is the point at which the scope stops moving so that materials can be ordered, contracts awarded, crews sized and the schedule built to a resource-loaded network. Everything after freeze is either an approved change with a visible cost or an unapproved change that appears as a schedule surprise. It is the single most consequential date on the calendar and it is defended for commercial reasons, not because the technical picture is complete.

The data says how well the industry holds it. AP-Networks reports average scope growth of 19 percent between freeze and execution, with top-quartile performers holding to around 7 percent. Nearly half of turnarounds change their start date at least once. On budget accuracy, Gordon Lawrence's analysis of 93 turnarounds published in PTQ found budgets underestimating actual cost by about 16 percent on average, with estimates intended as plus or minus 10 percent behaving in practice like plus or minus 30 percent.

One widely circulated figure deserves correction. The claim that 80 percent of turnarounds exceed budget by more than 10 percent, generally attributed to a 2019 T.A. Cook study, cannot be traced to a published source. The traceable AP-Networks position is that more than two-thirds of turnarounds exceed planned cost and schedule by 10 percent or trip after startup, and 40 percent overrun by more than 30 percent. Use the figures that can be cited.

Why findings drive repair scope and therefore the critical path

Before a vessel is opened, its repair scope is an estimate carried in a contingency line. After it is opened, cleaned and examined, the repair scope is a fact with a bill of materials attached. This is the structural reason inspection sits on the critical path: no other function converts unknown work into known work, and everything downstream waits for that conversion.

The chain from a single finding is long and every link is serial. A thickness reading below minimum triggers an engineering assessment, which triggers a repair decision, which triggers material procurement, welder qualification, fit-up, welding, post-weld heat treatment where the code requires it, and re-examination of the completed repair. Each step needs the previous one finished. A finding on day three of a twenty-eight day window is absorbable; the same finding on day nineteen sets the start-up date.

The planning consequence is that examination sequence matters more than examination speed. Items with the highest probability of generating repair scope should be opened, cleaned and read first, even when access is harder and the crew would rather start elsewhere. Front-loading discovery buys the repair chain the days it needs, and it is the cheapest schedule intervention available to a turnaround team.

The discovery loop, and where it actually stalls

The execution loop is short and it repeats hundreds of times: examine, report, disposition, repair, re-examine, accept. Teams under schedule pressure instinctively try to accelerate the first step by adding technicians, and the first step is the one least often responsible for the delay. Examination capacity is straightforward to add, because certified technicians can be mobilised from other regions at short notice. Disposition capacity cannot be added the same way, because disposition requires a qualified engineer with the authority to accept a condition or mandate a repair.

The stall shows up as a queue of completed examinations waiting for a decision while a repair crew stands idle at full rate. It is measured in hours per finding, and it compounds because the engineer making dispositions is the same person the scope challenge and the daily plan meeting need. Resourcing the disposition step at the same intensity as the examination step is the intervention that shortens outages.

Reporting latency is the other half. A finding that reaches the disposition engineer at the end of shift as a scanned form has already lost half a day. Structured field capture that puts a reading in front of the decision-maker within minutes converts directly into schedule, which is why field data capture practice at /blog/offline-field-data-capture-for-inspection-crews and the tooling comparison at /ndt-inspection-software are turnaround topics rather than back-office ones.

What the contractor must have ready before mobilisation

Mobilisation readiness is document readiness. Procedures and method statements approved by the client, not merely submitted. A personnel qualification matrix showing every technician certified for the methods assigned, with certifications and eye examinations valid through the last day of the window rather than the first. Instrument calibrations current with traceability documented. Radiation safety documentation, source logistics and exclusion planning where radiography is in scope.

The certification expiry trap is the one that recurs. A technician whose certification lapses on day eleven of a twenty-eight day turnaround invalidates every examination performed after that date, and the reports have to be reissued by someone qualified, if that is even possible after the unit is closed. Tracking expiry against the window rather than against the mobilisation date is the whole discipline, and it is what /erp/ndt-personnel-certification-tracking and /blog/tracking-asnt-certification-expiry-at-scale exist to enforce.

Client-side pre-mobilisation checks in US refineries are consistent and demanding: written practice, procedures matched to the applicable codes, qualification records, calibration certificates, insurance, safety statistics, and a demonstration that reports can be issued in the client's required format from day one. The full checklist is at /blog/operator-pre-mobilisation-checks-what-they-ask-for, and being ready for it is worth more to a bid than a lower rate.

PSM obligations run straight through the outage

The process safety management standard does not pause for a turnaround. Every examination on covered equipment produces documentation that must identify the date, the name of the person who performed it, the serial number or other identifier of the equipment, a description of the examination and the result, under 29 CFR 1910.119(j)(4)(iv). That is five fields on every report, and a report missing the equipment identifier is a regulatory defect as well as a filing problem.

Frequency is governed by 1910.119(j)(3), which requires inspection and test frequency consistent with manufacturers' recommendations and good engineering practices, and more frequently where prior operating experience demands it. Prior operating experience is the history file. Repairs, re-rates and material substitutions decided during the outage run through management of change under 1910.119(l), which means the turnaround generates PSM documentation at a rate no other period does.

OSHA inspects against exactly this. Instruction CPL 03-00-021, effective 17 January 2017, applies the PSM Covered Chemical Facilities National Emphasis Program across every covered facility including petroleum refineries, and it replaced the separate refinery emphasis programme that ran under federal jurisdiction until 2011. Mechanical integrity is where the standard meets the inspection function directly, and it is the paragraph a turnaround exercises hardest. Programme structure and the gaps that get cited are covered at /consulting/osha-psm-mechanical-integrity-ndt and /consulting/ndt-program-audit-gap-assessment.

Where spatial models earn their place in a turnaround

Turnarounds are where a maintained model stops being a visualisation and starts saving days. Scaffold reach, access sequencing, isolation boundaries and examination point locations are all spatial questions, and drawings answer them badly on a unit that has been modified for thirty years. A model carrying inspection points bound to geometry lets the access register be built against real elevations rather than against an assumption.

The second use is sequencing discovery. If the model shows which items historically produce findings and where they sit, the front-loading argument becomes a plan rather than an opinion. The third is the repair loop itself: showing a fabricator the actual as-built geometry of a tie-in before a spool is cut removes the most common source of rework at the exact moment rework is unaffordable.

This is the use case behind the refinery views at /digital-twins/refinery and /digital-twins/refinery-houston, and it is deliberately narrow. The model does not certify anything, and the inspection record stays where it belongs. It shortens the distance between a finding and the people who have to act on it, which during a three to five week window is the only measure that matters.

Handover, and the record file that outlives the outage

The last seventy-two hours are where inspection records get damaged. Crews demobilise, laptops go home, and the final reports get written from memory in the following fortnight. Every gap an auditor finds two years later was created in that window, and the cost of closing it afterwards is disproportionate because the people involved have moved to the next job or the next employer.

Handover means the record file is complete on the day: every examination reported, every revision closed with a reason, every disposition recorded with its authorising engineer, every repair linked to its re-examination, every PSM field populated. API Standard 653 requires a complete record file per tank covering construction, inspection history and repair and alteration history, and equivalent expectations apply across pressure equipment and piping. The API interpretation is explicit that the code sets no retention period, so the operating regime sets it, and on PSM-covered equipment that means the lifetime of the process.

Contractors who hand over a complete, equipment-keyed record file on the last day get invited back, because the client's own audit exposure drops. That is the commercial argument for structuring inspection data properly, made concrete in the record structure at /blog/ndt-data-management-best-practices and the back-office spine at /erp. Atlantis supports US turnaround programmes through ASNT Level III consulting at /consulting/asnt-level-iii-consulting-services and NDT training at /training-usa, with scoping and quotes available on request via /contact.

Request a consultation