When ILI Geometry Becomes an Engineering Input: What Does a Benchmark Dataset Actually Strengthen?
Current industry developments examined through lifecycle engineering evidence.
VERIFIED — PUBLIC-SOURCE ANALYSIS
Core distinction
Consistency is not validity. And validity is not sufficiency.
The distinction matters when an ILI-derived geometry measurement becomes an engineering input.
PRCI releases a new benchmark for dent-geometry parameter measurement
On 5 November 2026, Pipeline Research Council International released PR214-253820-R01 — ILI Benchmark Data to Improve Consistency in Mechanical Damage Parameter Measurement.
The report is authored by Sanjay Tiku and Arnav Rana of BMT Canada Ltd., is listed by PRCI as a 46-page report, and carries DOI 10.55274/R0000213.
The work addresses a technically important transformation:
PRCI describes work on noisy caliper-data smoothing, ill-defined dent axial boundaries, a benchmark dataset derived from PRCI NDE-4-18 data, intermediate calculations associated with the PRCI Level 2 Shape Parameter approach, and examples of anomalous caliper data.
The stated purpose is to improve consistency in dent-geometry parameter extraction between operators and ILI service providers.
What the public record establishes
Benchmark objective
PRCI frames the work around improving consistency in mechanical-damage geometry parameter measurement and supporting more standardized analysis and reporting.
Scale of related analysis
BMT's public IPC 2026 description reports approximately 1,500 dent ILI caliper readings in the smoothing analysis and 200+ field dent ILI datasets in the axial-limit work, supplemented by an extensive finite-element dataset.
Standards boundary
PRCI says the findings and recommendations can be incorporated into API RP 1183 Second Edition. That is not the same as saying they have already been incorporated.
API currently describes the second edition of RP 1183 as under development. This distinction should remain visible when the research is discussed.
What exactly becomes stronger when geometry extraction becomes more consistent?
A benchmark can make results more reproducible between analysts and ILI service providers. It can expose differences associated with smoothing, dent-boundary selection and calculation implementation. It can also make the transformation from inspection data into an assessment parameter more transparent.
Those are meaningful gains. But they do not automatically establish that the entire engineering conclusion has been validated.
Is the result reproducible?
Can different analysts or service providers process equivalent source data and obtain sufficiently consistent geometry parameters?
Is the measurement valid for the claim?
Was the underlying inspection technology appropriate for the feature, conditions and measurement task, and does the parameter represent what the assessment assumes it represents?
Is the evidence sufficient for the decision?
Even if the parameter is reproducible and locally valid, is the total evidence adequate for the particular integrity decision or action?
Research boundary
Reproducibility → Validity → Sufficiency
Related, but not interchangeable.
What should travel with the parameter?
Origin
The source of the geometry measurement should remain identifiable: inspection run, ILI system, relevant raw or processed data, tool specification and acquisition conditions.
Binding
The derived parameter must remain bound to the correct physical feature. Numerical precision cannot compensate for uncertain feature identity or correspondence.
Context
Tool applicability, smoothing rules, boundary-selection method, geometry assumptions and assessment applicability should remain visible.
Continuity
The transformation should remain reconstructable from source measurement through preprocessing, geometry reconstruction, parameter extraction, assessment input and engineering conclusion.
Currency
Method, software implementation, guidance and standard version matter. The result should remain linked to the version actually used.
Sufficiency
The final question is decision-specific: does the complete evidence justify the engineering consequence being considered?
What the benchmark can strengthen
Reproducibility. A stronger common reference for parameter extraction.
Comparability. Differences in processing and implementation can be examined against shared data.
Analytical transparency. Intermediate calculations can help locate where analytical divergence occurs.
Process standardization. Operators and service providers gain a stronger basis for aligned analysis and reporting.
What it does not establish by itself
It does not by itself establish universal inspection applicability, detection capability, correct feature identification or classification, sizing performance outside the demonstrated scope, correct feature binding, field confirmation, complete interacting-damage treatment, downstream uncertainty sufficiency, or the adequacy of the total evidence for a specific integrity decision.
Nor should the work be represented as already incorporated into API RP 1183 Second Edition.
Method, version and applicability context cannot disappear
API's 2024 Addendum 1 to RP 1183 removed Table 6, Dent Fatigue Life Spectrum Severity Criteria, after practical application indicated that the screening approach was not the most conservative method in every case.
That is a different technical issue from the new PRCI benchmark and should not be conflated with it.
White Paper No. 24 — Preserving the Evidence Boundary from ILI Measurement to Engineering Assessment
The earlier DrKu.net research describes an ILI Evidence Reconciliation Chain:
It also separates detection, identification, classification, sizing and location rather than compressing them into the single word “accuracy”.
PR214-253820-R01 is especially relevant to the analysis → engineering-parameter portion of this chain.
Its broader significance is therefore not simply that analysts may reproduce a number more consistently. It creates an opportunity to make the transformation from inspection evidence to engineering input more explicit and reviewable.
A benchmark dataset should reduce inter-analyst variability in derived ILI geometry parameters, but improved reproducibility alone will not necessarily reduce uncertainty in the final engineering decision unless the parameter remains connected to its provenance, transformation history, feature identity, applicability conditions and validation evidence.
This suggests two different measures of improvement: parameter consistency and decision-evidence sufficiency.
When an ILI-derived parameter enters an integrity assessment, what minimum evidence must travel with it so that the eventual engineering conclusion remains reconstructable, reviewable and defensible?
A candidate minimum package may need to preserve: origin + feature binding + processing lineage + assumptions + uncertainty + verification status + method/version context .
Engineering Signal
A better benchmark should not only help us reproduce the number. It should help preserve the evidence explaining how that number became an engineering input.
Pipeline Research Council International
PR214-253820-R01 — ILI Benchmark Data to Improve Consistency in Mechanical Damage Parameter Measurement
Released 5 November 2026 · DOI 10.55274/R0000213
BMT
Public IPC 2026 description of the related work on consistent dent-geometry parameter measurement.
American Petroleum Institute
API RP 1183 Addendum 1 and current second-edition development context.
Related DrKu.net Research
Preserving the Evidence Boundary from ILI Measurement to Engineering Assessment, Version 1.3.0.
This Signal extends the research question first raised in the LinkedIn discussion published on 7 November 2026: “Consistency is not validity. And validity is not sufficiency.”
Status: VERIFIED — PUBLIC-SOURCE ANALYSIS
Published: 7 November 2026
Update trigger: Revisit when API RP 1183 Second Edition is published, when the PRCI recommendations are formally incorporated into a standard or guidance document, or when material new validation evidence becomes available.