From data to decisions: Leveraging destructive testing to unlock the truth about complex cracking in LF-ERW pipelines - Phase II

Pipeline integrity is built on trust, trust in inspection data, trust in verification techniques, and trust in the conclusions operators draw to keep their networks safe. Yet, when complex crack geometries distort that data, trust can falter.

For operators managing vintage low-frequency electric resistance welded (LF-ERW) pipelines, anomalies such as “hook cracks” and other off-axis flaws can challenge the limits of inline inspection (ILI) and non-destructive examination (NDE). A recurring question arises:

When NDE results don’t align with ILI, which do you trust and can either be refined to give the full picture?

This article presents Phase II of a joint investigation by NDT Global and Phillips 66, building on the Phase I study published last year: Novel approach for detecting and identifying complex cracking in LF-ERW pipe, a real case study. Phase I established a systematic method for recognizing ILI signal patterns associated with complex crack geometries particularly hook cracks validated through in-field NDE.

Phase II advances the work by incorporating destructive laboratory testing to provide metallurgical ground truth. This not only tests the accuracy of the Phase I methodology but also reveals where NDE classifications diverge from reality, allowing for more precise defect characterization and better-informed remediation strategies.

How tilt and skew in crack orientation complicate ultrasonic inspection

Phase I Recap: Establishing Pattern Recognition Methodology

The Phase I project began with a high-resolution axial crack ILI of a 6 in, 1951 LF-ERW pipeline. Analysis revealed numerous weld-line anomalies with complex signal patterns. Field NDE using phased array ultrasonic testing (PAUT) confirmed several as hook cracks, alongside other defects.

From these verified digs, NDT Global developed a signal pattern recognition methodology to flag potential hook cracks in the remaining ILI data. Two signature patterns emerged:

1. Amplitude/Pattern Asymmetry: Clockwise (CW) and counterclockwise (CCW) sensors showed markedly different amplitudes or patterns, typical of tilted or skewed flaws.

2. Alternating Reflections: Similar CW/CCW signals, but with alternating internal/external reflections, suggesting multiple reflectors or complex geometry.

By applying these patterns, the team was able to classify unverified anomalies as likely, possible, or unlikely hook cracks. Phillips 66 used this information to prioritize digs, aiming for maximum risk reduction.

B-scan data showing different clockwise and counterclockwise sensor responses, a key signature of complex hook cracks

Phase II: Metallurgical Validation of the Phase I Findings

While Phase I created a robust pattern recognition process, the team knew it was based on NDE field classifications, which, while valuable, are not infallible. The question remained: Were those classifications correct?

To find out, Phillips 66 expanded the verification program in Phase II. This time, they brought in a new NDE vendor using Total Focusing Method (TFM) ultrasonics. Unlike PAUT, TFM synthesizes data from multiple sound paths into a high-resolution image of the inspection zone, potentially revealing more detail about defect geometry.

Nine anomalies from seven pipe sections classified in the field as eight hook cracks and one lack-of-fusion (LOF) defect were selected for destructive metallurgical testing.

Ground Truth: The Lab Tells a Different Story

The destructive testing results were illuminating:

• 6 of 9 anomalies were confirmed as weld bond-line LOF defects, caused by insufficient edge preparation or heating during manufacturing
• 3 of 9 were surface laps sharp unfused folds in the pipe surface near, but not at, the bond line, likely formed during weld upset
• None showed evidence of in-service crack growth

Several anomalies that appeared to be hook cracks in ILI and NDE were, in fact, different defect types. This finding underscored a key limitation of non-destructive methods: even with advanced ultrasonic techniques, geometry and orientation can disguise a defect’s true nature.

Metallurgical cutouts reveal that suspected hook cracks were actually lack-of-fusion and surface lap anomalies.

Refining the Patterns: Distinguishing Hooks from Laps

A detailed review of ILI data showed that surface laps near the bond line can mimic hook cracks in signal behavior different CW/CCW responses, high-amplitude corner echoes on one side, and “cloudy” multiple reflections on the other.

The differentiator was in a reflection path blocking: the lap’s position partially obstructed certain bond line-surface echoes, subtly altering the amplitude distribution. By incorporating this nuance, NDT Global refined the Phase I rules removing one of the original two patterns, which destructive testing showed was unreliable for hook crack identification.

Applying the updated rules to more than 3,000 below-reporting-threshold anomalies reduced the candidate list for complex weld defects to fewer than 20 giving Phillips 66 a leaner, higher-confidence target list for future digs.

Sizing Accuracy: Clearing the ILI’s Name

One of the original concerns in Phase I was that the ILI tool was under-sizing defect depths compared to NDE. Vendor 1’s results suggested a consistent bias.

However, when the metallurgical testing established the true depths, the story changed: the ILI measurements were within their performance specification. The apparent under-call came from NDE sizing inaccuracies not the ILI. Regarding the NDE data against the lab results, it brought both vendors’ measurements into close alignment with ILI sizing curves. Statistical validation per API Standard 1163 confirmed an actual tool tolerance of ±35 mils at 80 % certainty better than the contractual specification.

Lessons from Phase II for the Industry

The progression from Phase I to Phase II delivers several lessons for operators and service providers:

1. Avoid Adjusting ILI Sizing Curves Solely Based on NDE. Without destructive testing, there’s a real risk of introducing bias rather than correcting it
2. The Value of Metallurgical Destructive Testing. Metallurgical destructive testing remains the gold standard for defect characterization and a vital benchmark for validating both ILI and NDE
3. Adherence to a Systematic Methodology. A structured, protocol-driven approach—controlling for NDE limitations and asset-specific conditions—avoids premature conclusions
4. From Hypothesis to Conclusion, Evaluating Accuracy. Probe choice, pipe curvature, and defect geometry can all compromise NDE accuracy
5. Limitations of Field Verification Tools and Techniques. When proven against destructive testing, ILI signal patterns can focus resources where they matter most

For Phillips 66, Phase II delivered tangible benefits:

• Increased confidence in identifying true hook cracks
• Confirmation that ILI met or exceeded specification
• A smaller, better-targeted remediation list for future digs

For NDT Global, the project reinforced the importance of combining high-resolution ILI data with rigorous verification and validation to ensure inspection technology delivers not just data, but actionable insight.

Conclusion

Phase I hypothesized that certain ILI signal patterns could be used to identify hook cracks more accurately, even in complex geometries. Phase II put that hypothesis to the test with the highest standard of proof - destructive metallurgical analysis and found that the ILI was more accurate than initial NDE suggested.

The result is a refined methodology that combines advanced data analytics, targeted field verification, and ground truth validation to produce higher confidence, lower risk, and better-informed integrity decisions.