It’s day 9 of a 21-day turnaround. Inspection has just called in a crack-like indication at a nozzle-to-shell weld on a vessel that wasn’t scheduled for anything beyond routine UT thickness surveys. The unit is on the critical path and startup is booked. Now someone only has a few hours to decide whether this vessel runs without formal repair, gets a welded repair, or comes out of service for replacement.

This is not a hypothetical. It is one of the most common—and most poorly managed—moments in a turnaround. When done properly under API 579-1/ASME FFS-1, a fitness-for-service (FFS) assessment is a methodical process that does not care whether a vessel is on a 21-day critical path or if deferred startup is costing the plant money every hour. The result is a predictable and avoidable failure mode: rigorous engineering gets compressed into hours, decisions get made on incomplete data, and full replacement is rarely a real option in the moment as fabrication and delivery lead times for major components almost never fit inside a turnaround window. Thus, the default under pressure tends to run the other way: a rushed repair or run-as-is call gets pushed through without the full assessment it warrants, or an overly conservative derate gets applied simply because it’s the fastest thing to justify on paper. Neither outcome is really engineering judgment, but rather whatever can be defended fastest.

The fix is not to make engineers work faster under stress; it is to do the engineering before the stress arrives — to pre-build the decision trees, calculation templates, and approval pathways so that when an unplanned finding surfaces mid-turnaround, the team has a plan to execute instead of being forced to improvise.

Why This Keeps Happening

Turnaround scope is built from risk-based inspection (RBI) history and planned NDE, but opening equipment always surfaces the unknown. RBI predicts likely damage mechanisms based on service history and prior findings, not the local thin area (LTA) that shows up in a location no one expected or the indication that does not match any documented corrosion pattern for that unit. A few root causes make this structurally difficult rather than just occasionally unlucky:

• Inspection scope is sampled, not exhaustive. RBI-driven NDE targets the highest-probability locations; it cannot cover every square foot of every vessel.
• Turnaround schedules allocate for known, scoped repairs, not engineering work that hasn’t been identified yet. When an unplanned finding shows up, engineering becomes the unbudgeted item on the critical path.
• Level 2 and Level 3 assessments for complex LTA geometries, crack-tip evaluations under Part 9, and remaining-life calculations for unusual damage mechanisms often require calculation effort that was never scoped or resourced before the turnaround started.
• Decision authority is frequently unclear in the moment. Who has the standing to approve “do nothing,” run “as-is” with a de-rate, perform a temporary repair, or leave the equipment out of service until it can be replaced, and how fast can that approval actually move at 2 AM?

The Core Argument: Separate What Can Be Anticipated From What Can’t

The productive distinction is between pre-turnaround engineering (the work that can be anticipated and done in advance) and in-turnaround engineering (the irreducible core of genuinely novel findings that must be resolved in real time). The goal of a decision-tree program is not to eliminate the second category; it’s to shrink it as far as possible so that the majority of findings, even ones not explicitly predicted, fall into a pathway that was already thought through calmly, either weeks or months earlier.

For each component and damage mechanism combination that can reasonably be anticipated from RBI data and operating history, the pre-turnaround engineering package should include:

• The applicable FFS Part and assessment level, with acceptance criteria pre-calculated against nominal and expected operating conditions, not derived from scratch after the finding is discovered
• Pre-approved repair details (weld overlay, patch plate, sleeve, or clad restoration) with WPS and PQR already qualified, so a repair decision doesn’t stall on procedure qualification during the outage
• Maximum allowable working pressure (MAWP) and re-rate calculation templates with plant- and equipment-specific inputs already loaded, so the engineer on shift is dropping in field data, not building a calculation model under time pressure
• Pre-identified decision authority: who can approve which category of outcome, at what schedule or dollar threshold, with what documentation trail

One useful way to test whether this exercise is complete: pick any credible finding for the unit’s known damage mechanisms and ask whether the path from discovery to decision is already documented. If the answer requires a meeting to figure out, the tree isn’t finished.

Worked Example: An LTA Near a Nozzle

Consider a shell course showing an LTA near a nozzle reinforcement zone discovered during routine UT inspection. With a pre-built decision tree, the path looks like this: the finding is sized and UT-mapped against a pre-defined grid; the data is run through a Level 1 screening curve for that geometry and service, already validated during turnaround planning. If it passes, the vessel returns to service under existing documentation with sign-off from the inspector and superintendent. This is a Tier 1 outcome, resolved in the field within the shift.

If the LTA fails Level 1 screening because of an unfavorable length-to-depth ratio or proximity to the nozzle reinforcement pad, the pre-built Level 2 calculation template is pulled, plant-specific geometry and material properties already populated. This requires sign-off from the FFS engineer and engineering manager: a Tier 2 outcome. The finding is escalated to Tier 3 only if it falls outside every anticipated pattern or is a crack-like indication requiring fracture mechanics evaluation or damage mechanism of which the unit has no documented history. Tier 3 necessitates a Level 3 assessment, potentially involving finite element analysis (FEA), with plant leadership needing to weigh the finding against real cost-of-delay numbers rather than an abstract sense of schedule pressure.

The value of the tree is not in eliminating judgment but in front-loading any judgment calls that can be. This means judgment having to be exercised in the moment is applied to the genuinely novel 10% rather than re-litigated for the routine 90%.

A Tiered Escalation Matrix

Tier	Trigger	Assessment Path	Approval Authority
Tier 1	Finding fits pre-approved envelope (known damage mechanism, within pre-calculated trigger points)	Level 1 screening against pre-loaded acceptance criteria	Turnaround Superintendent + Inspector
Tier 2	Exceeds Tier 1 envelope; requires calculation beyond screening (e.g., LTA with irregular geometry, MAWP re-rate)	Level 2 assessment using pre-built calculation template	Engineering Manager + FFS Engineer
Tier 3	Novel damage mechanism, crack-like indication requiring fracture mechanics, or Level 2 fails acceptance	Level 3 assessment, may require FEA	Plant Manager / VP, with cost-of-delay data in hand

What It Takes to Build This

Bring FFS engineering into turnaround planning early

This works only if FFS engineering is part of turnaround planning 6 to 12 months in advance, not a reactive phone call during execution. The planning team needs to sit down with inspection history, RBI data, and operating history to review the credible damage mechanisms for this unit and whether paths have been pre-built for each.

Pre-negotiate rapid-turnaround vendor support

Some findings will require material verification, hardness testing, or metallurgical replication that cannot be pre-calculated. Having lab and vendor agreements in place before the turnaround with committed response times prevents a Tier 3 finding from also becoming a logistics problem.

Build a war-room FEA capability

For units with known complex geometries, such as nozzle intersections, non-standard heads, historically problematic welds, etc., pre-building parametric finite element models means a Level 3 assessment becomes a data input exercise during the turnaround rather than requiring a model needing to be built from scratch. The difference is often measured not in hours but days.

The ROI Argument

The case for building this program is a straightforward comparison: the fully loaded cost of one additional day of extended turnaround, deferred production, contractor standby, and schedule penalties against the cost of assembling a decision-tree program during the planning phase, when time is not scarce and calculations can be double-checked.

But note that the schedule argument—while getting the budget approved—understates the real risk. A rushed FFS decision made under time pressure with incomplete documentation is also the decision most likely to fail scrutiny later in an internal audit, a process hazard analysis (PHA), or worse, a subsequent inspection that finds the “run-as-is” call didn’t hold up. Pre-built decision trees are not just a schedule tool—they are a mechanism for ensuring engineering judgment exercised under pressure is still engineering judgment: documented, defensible, and consistent with what the same team would have concluded with unlimited time.