Every pressure boundary carries flaws. The engineering question after a rupture is never whether a flaw existed, but whether the flaw that failed should have been found, sized and acted on before it reached critical dimensions. That question is governed by two overlapping frameworks, fitness-for-service assessment and integrity management, and the records those frameworks generate are usually the most contested material in the matter. What they actually require, and what they do not, is what separates a supported attribution from an assumption about inspection.
Two frameworks doing different work
Fitness-for-service assessment answers a question about one flaw at one moment: given its size, the material and the stress, is continued operation acceptable and on what basis. Integrity management answers a programme question: which threats apply to this asset, what inspection addresses each, how often, and what response is triggered by a finding.
A rupture usually implicates both. The flaw was either never found, found and assessed as acceptable, or found and not acted on, and each of those routes leads to a different set of documents.
How the fitness-for-service standard is structured
API 579-1/ASME FFS-1 organises assessment into tiers of increasing rigour, from conservative screening rules through more detailed analysis to numerical modelling, with separate parts for general and local metal loss, pitting, crack-like flaws, creep, fire damage and other damage classes.
That structure matters forensically because it records the level of analysis actually performed. A screening-level acceptance and a detailed crack-like flaw assessment carry very different assumptions, and the inputs used at the chosen level are recoverable and checkable.
Running the assessment backward
The same methodology used prospectively to justify continued service is used after the fact to explain why the flaw did not survive. Measured flaw dimensions from the fracture surface, toughness from Charpy or CTOD testing of the failed material, and stress from the reconstructed operating pressure produce a critical flaw size, which can then be compared against what the flaw measured at each earlier inspection.
This is where an inspection history becomes analytically useful rather than merely suggestive. It converts the record from a list of dates into a growth history.
Detection threshold, tolerance and probability of detection
An inspection tool is not a measuring device with unlimited resolution. Every method has a detection threshold, a sizing tolerance, a stated confidence level and a coverage limit, and in-line inspection performance is qualified against those declared specifications under API 1163.
The forensic comparison is therefore not simply whether the flaw appeared in the report. It is whether a flaw of that type, orientation and size at that location was within the capability of the tool that was run, at the confidence the vendor specified, on the coverage that was actually achieved.
Tool selection is itself a finding
Metal-loss tools and crack-detection tools address different threats, and a programme that ran only one of them has answered only one question. A line inspected repeatedly and competently for wall loss may have no meaningful record at all regarding axial cracking.
Where the failure mechanism is one the selected tools were not designed to find, the issue moves upstream to the threat assessment that chose them.
Interval versus growth rate
A reassessment interval is defensible only if it is short relative to how fast the flaw grows. For gas transmission, ASME B31.8S sets out the integrity management approach and the basis for reassessment intervals, and the federal integrity management requirements in 49 CFR Part 192 and Part 195 impose their own timing obligations for covered segments.
Comparing the interval that was used against the growth rate derived from the failed flaw is often the single most probative calculation in the case, and it is one the operator's own data usually supports or undermines without outside assumption.
Pressure limits and the operating record
Maximum allowable operating pressure and maximum operating pressure are established from design, testing and regulatory history, and the record supporting them can be as contested as the metallurgy. Where establishment relied on records that no longer exist or on a pressure test whose documentation is incomplete, that is an issue independent of the failure mechanism.
Cyclic severity matters as much as peak pressure. A line operating well within its limit but cycling aggressively can drive fatigue growth that a static assessment never contemplated.
Fixed equipment runs on its own codes
Vessels, process piping and storage tanks are governed by their own inspection regimes: API 510 for pressure vessels, API 570 for in-service piping and API 653 for aboveground storage tanks, each prescribing inspection types, interval-setting rules and the treatment of findings, often informed by risk-based inspection methodology.
Those codes generate dated, signed documents with named findings and recommendations. Whether a recommendation was closed, deferred or left open is frequently more decisive than the condition it described.
What a clean prior inspection does and does not prove
A prior inspection reporting nothing reportable is not proof the flaw was absent. It establishes that nothing exceeding the reporting threshold was detected by that tool, on that coverage, at that time. Whether the flaw was below threshold, outside coverage, or of a type the method could not see are separate questions with separate answers.
Opinions in this area are challenged on exactly that gap. Work that states the tool specification, the coverage achieved, the reporting threshold, the growth rate assumed and its source will hold where a bare assertion that the flaw was or was not detectable will not.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.