API 579 fitness for service: levels 1 to 3 in practice

A thickness reading below tmin is not a decision. It is the beginning of one. The construction code says what the component needed when it was new; it does not say what a component with a groove, a thin patch or a crack can safely do now. API 579-1/ASME FFS-1 answers that second question, and it does so with a deliberate ladder of effort: a quick conservative screen, a detailed engineering evaluation, and an advanced analysis. Most components stop at the first rung. The ones that do not are the ones that generate turnaround scope, rerates and repair decisions worth millions. This paper describes how the levels work in practice for the common damage types and what an assessment must document to be worth anything after the engineer who ran it has moved on.

What is API 579-1/ASME FFS-1 and when does it apply?

API 579-1/ASME FFS-1 is the joint API and ASME standard for evaluating flaws and damage in pressure vessels, piping and tanks that were designed to a recognized construction code. It is invoked by API 510, API 570 and API 653 whenever inspection finds a condition the construction code did not anticipate, and it answers three questions: can the component continue to operate, at what pressure and temperature, and for how long.

The standard is organized by damage type, with each part carrying its own applicability limits, data requirements and acceptance criteria.

• Part 3: brittle fracture susceptibility
• Part 4: general metal loss
• Part 5: local metal loss (local thin areas and grooves)
• Part 6: pitting corrosion
• Part 7: hydrogen blisters and hydrogen damage (HIC and SOHIC)
• Part 8: weld misalignment and shell distortions
• Part 9: crack-like flaws
• Part 10: components operating in the creep range
• Part 11: fire damage
• Part 12: dents, gouges and dent-gouge combinations
• Part 13: laminations
• Part 14: fatigue

Part 2 sets the general procedure that every part follows: identify the flaw and the damage mechanism, confirm applicability, gather data, apply the assessment technique at the chosen level, evaluate remaining life, decide on remediation, define in-service monitoring, and document. The damage mechanism step is not a formality. A local thin area caused by erosion at an injection point and one caused by under-deposit corrosion have different futures, and the future corrosion allowance in the assessment depends on knowing which it is.

What separates Level 1, Level 2 and Level 3?

The three levels trade conservatism for effort and expertise.

• Level 1 uses screening criteria built to be conservative for the widest range of components. Data requirements are minimal, typically what an inspection already produced, and the assessment can be performed by plant inspection or engineering personnel. Most of the acceptance criteria are simple comparisons or chart lookups.
• Level 2 uses a more detailed evaluation with more refined data. It is performed by engineers experienced in fitness for service. It typically produces a less conservative result because it accounts for component-specific geometry, loads and material properties that Level 1 lumps together.
• Level 3 uses the most detailed analysis: finite element stress analysis, fracture mechanics with measured toughness, ductile tearing analysis, or probabilistic methods. It is reserved for engineers experienced and knowledgeable in fitness for service and is used when Level 2 cannot show acceptability or when the component falls outside Level 1 and 2 applicability.

A failed Level 1 does not mean the component is unfit. It means the screen could not show it fit with minimal data, and the next step is more data and a higher level, or remediation. Programs go wrong in two directions here. Some treat a Level 1 failure as a repair order and spend money the component did not need. Others let a Level 1 pass stand where the applicability limits were not checked, which is how a screening method meant for uniform corrosion gets applied to a groove. The record must show the applicability check for the level used, not just the result.

How does a Part 4 general metal loss assessment work?

Part 4 addresses wall loss that is spread over an area rather than concentrated. The first task is to decide whether Part 4 applies at all.

• Take point thickness readings across the corroded region. Compute the coefficient of variation. If it is ten percent or less, the loss is general and point readings suffice. If it is greater, take critical thickness profiles along the meridional and circumferential directions and use Part 5.
• Establish the future corrosion allowance (FCA), the metal expected to be lost before the next inspection, from the corrosion rate and the planned interval.
• Establish tmin from the construction code for the component, including any supplemental loads at Level 2.

The Level 1 acceptance criteria then read directly. The average measured thickness minus FCA must be at least tmin. The minimum measured thickness minus FCA must be at least the greater of half of tmin and a fixed lower bound of 2.5 millimeters (about 0.10 inch), which protects against very thin spots that averaging would hide. Where profiles are used, the average is taken over a length that depends on the diameter, the thickness and the remaining thickness ratio. If the average criterion fails, the component can be rerated to a reduced MAWP computed from the available thickness, or the interval can be shortened to reduce FCA, or it moves to Level 2.

The record for a Part 4 assessment has to hold the readings themselves, not just their average. An auditor recomputing the coefficient of variation from stored readings is doing exactly what the standard intends; an auditor handed an average and a pass mark has nothing to check.

How do Part 5 local metal loss and Part 6 pitting differ?

Part 5 treats a defined local thin area or groove as a flaw with dimensions. The critical thickness profile gives the length of the flaw in each direction and the minimum thickness within it. Two dimensionless quantities drive the Level 1 screen: the remaining thickness ratio, the minimum measured thickness minus FCA over the required thickness, and a shell parameter that scales the flaw length by the square root of diameter times thickness. The pair is compared against a screening curve; below the curve the flaw is unacceptable at Level 1. Separate limits apply to the minimum thickness and to the distance from the flaw to the nearest structural discontinuity, such as a nozzle or head junction, since the closed-form solutions assume the flaw sits in an undisturbed shell. Where the screen fails, Level 2 divides the profile into subsections and computes a remaining strength factor from the area of metal lost and a Folias-type bulging factor, and the MAWP is reduced in proportion to RSF over the allowable RSF of 0.90.

Part 6 treats pitting, where the metal loss is many small, deep cavities rather than one thin region. The Level 1 method characterizes the pitting with standard pitting charts, graded by pit density and depth, and reads a remaining strength factor from the matching chart. Level 2 computes RSF from measured pit couples, pairs of adjacent pits with their depths, diameters and spacing, and can account for pitting on both surfaces. Pitting confined to a local region is assessed as pitting within a local thin area, combining Parts 5 and 6.

The practical distinction for a program is data type. Part 4 wants a grid of point readings. Part 5 wants profiles with the flaw dimensions and its location relative to discontinuities. Part 6 wants pit geometry or a defensible chart match. Inspection that is planned without knowing which part will be used tends to come back with the wrong data, and the assessment waits for the next outage.

What does a Part 9 crack-like flaw assessment need?

Crack-like flaws are the assessments with the highest consequence of getting wrong, and Part 9 demands correspondingly more.

• Flaw characterization. Length, depth, orientation relative to the principal stress, whether the flaw is surface-breaking or embedded, and its position relative to welds. NDE sizing uncertainty is added to the measured dimensions, and embedded flaws near a surface may be recategorized as surface flaws.
• Stresses. Primary stresses from pressure and supplemental loads, secondary stresses, and residual stresses from welding, which the standard tabulates by weld type and heat treatment condition. Post-weld heat treatment materially changes the residual stress assumption, so the fabrication record matters.
• Material toughness. A fracture toughness at the assessment temperature, from test data where available or from a lower-bound curve indexed to the material's reference temperature, with Charpy correlations where that is all the record offers.

Level 1 uses screening curves that relate allowable flaw length to thickness and reference temperature for common configurations. Level 2 plots the flaw on a failure assessment diagram, with the toughness ratio on one axis and the load ratio on the other, using stress intensity factor and reference stress solutions from the standard's annexes, with partial safety factors on load, toughness and flaw size. A point inside the curve is acceptable. Level 3 extends to ductile tearing, J-integral methods, finite element solutions and probabilistic treatments.

Remaining life for a crack-like flaw is a crack growth calculation, by fatigue, environmental cracking or creep, from the current size to the critical size on the diagram. That calculation depends on a growth law and a cycle count that are themselves recorded assumptions.

What must the FFS record document?

Part 2 makes documentation a step of the assessment procedure, and API 510, API 570 and API 653 require the result to be kept with the equipment record. A complete assessment record carries the following.

• The component identity and the flaw or damage description with the inspection that found it and the NDE data behind it.
• The damage mechanism assigned and its basis.
• The applicability check for the part and level used, including the limitations that were confirmed not to apply.
• All input data with sources: design conditions, material and allowable stress, dimensions, readings or profiles, corrosion rate and FCA, loads, toughness, residual stress assumptions.
• The calculation and its result: RSF or acceptance criterion, any reduced MAWP or temperature, and the remaining life.
• Remediation and monitoring decisions: repair, rerate, shortened interval, or in-service monitoring with its method and frequency.
• The responsible engineer's approval and, where required, the inspector's review.

The reason to hold the inputs is that assessments are revisited. The next inspection produces a new minimum thickness, the process changes, or a Level 3 is commissioned because a rerate is unwelcome. Each of those starts from the prior assessment, and a prior assessment without stored inputs is a prior assessment that has to be redone.

How intelliSPEC handles it

intelliSPEC treats fitness for service as an engineering decision with a data record, not as a button. In intelliINTEGRITY, a Part 4 Level 1 screening compares measured thickness against tMin for the component and reports the ratio, with the tMin itself routed by equipment type to ASME VIII Division 1, ASME B31.3 or API 653 and returned together with the design code and formula. The corrosion summary supplies short-term, long-term and governing rates for the future corrosion allowance, and the remaining life follows as thickness minus tMin over the rate. ASME B31G remaining strength for corroded pipe is available in the original and modified forms, in both the 1991 and 2012 editions.

Two controls are deliberate. Every fitness for service result is provisional until an engineer verifies it, and no result automatically derates a component. And Part 5 Level 1 for local metal loss is not implemented, pending verification by a licensed engineer; the platform does not offer a local thin area screen it cannot yet stand behind. Where a Level 2 or Level 3 assessment is performed outside the platform, its inputs, result and approval attach to the component record with the inspection data that triggered it.

The readings that feed an assessment come from intelliNDT, where seven supported gauge profiles enter thickness without retyping and each CML holds nominal thickness, tMin and its method, material, allowable stress, joint efficiency, grid geometry and three corrosion rates, so a coefficient of variation or a critical thickness profile is computed from stored points rather than from a scanned sheet. Calculators are individually permissioned, a calculator author cannot approve their own work, and every result and approval is written to a hash-chained audit trail that a verify routine can recompute. The refining context for these assessments is described on the oil and gas page, and the terms are defined in the glossary.

Fitness for service is the discipline of deciding, with the arithmetic showing, that a damaged component can keep running. The record is the arithmetic.