Industry demands are increasing when it comes to the visualization of features present in a pipeline. Higher prediction quality and cost reduction for pipeline dig up activities require a higher reliability and resolution of In-Line Inspection (ILI) data using Magnetic Flux Leakage (MFL) technology. Correct solutions are the result of calculation, not guessing. With modern computing performance, this alternative becomes realistic.

In this article, Johannes Palmer explains how our Deep Field Analysis (DFA) technology enables a true understanding of what is in the ground by providing laser-like 3D metal loss profiles derived from MFL data. ROSEN developed a service package called Virtual–Dig Up (VDU) from this technology, which offers clear benefits for operators: visualization of the true metal loss shape, robust information for maintenance optimization, and integrity planning with higher reliability and lower costs.

Magnetic Flux Leakage (MFL) technology is the most widely established pipeline metal loss inspection method. MFL tools are extremely robust, have compact designs and can easily cope with gaseous products. However, the MFL technique is indirect, i.e. the measurement does not directly deliver the geometry of the pipeline flaw. MFL data anomalies have an enormous repeatability, meaning the measurement is accurate, but the translation of the signals is complex and ambiguous because potential field anomalies allow for more than one correct solution. Moreover, an integrated analysis of all three field components cannot change this.

Today, MFL In-Line Inspection (ILI) data is interpreted based on multifaceted knowledge from pull tests, synthetic data and dig verifications.
The interpretation makes use of machine learning and artificial intelligence; optionally, human expertise can be integrated. The most probable solution is the target of this judgement, based on a particular level of knowledge – a process termed “educated guess.”

Figure 1 – MFL Evaluation Principle – Educated Guess

Figure 1 – MFL Evaluation Principle – Educated Guess

Accepted throughout the industry, this system often relies on thorough interaction with in-situ verification activities, which optionally allow for sizing model improvements and result acceptance for pipeline integrity purposes.

Figure 2 – MFL Sizing Model – Improvement Based on Dig Up

Figure 2 – MFL Sizing Model – Improvement Based on Dig Up

Costly field data verification increases the “education” but does not change the “guessing.” Moreover, the complexity of this environment requires simplification, established with so-called boxes defined by the Pipeline Operators Forum.

Figure 3 – MFL Evaluation Result – Educated Guess and Simplification

CORRECT SOLUTIONS AS A RESULT OF CALCULATION

However, the demands of the industry are increasing. Higher prediction quality and cost reduction for pipeline dig up activities require higher MFL ILI reliability and resolution.

Correct solutions are the result of calculation, not guessing. With modern computing performance, this alternative becomes realistic. Directly deriving the MFL anomaly from a given source geometry is possible. With a loop of repeated approximation, a quasi-calculation becomes possible. This is independent of knowledge and sizing models. The dig up-based refinement is only one loop, strictly speaking valid only for the dig ups – but this refinement can be run several thousand times.

Figure 4 – Groundbreaking MFL Approach – Result Calculation

Figure 4 – Groundbreaking MFL Approach – Result Calculation

In addition, this process model does not require even one verification dig for enhancement. On the contrary, it produces results comparable to a dig up without a dig up. This was the reason for calling this application Virtual–Dig Up (VDU).

Figure 5 – MFL Evaluation Result – High-Resolution 3D Calculation

The resulting solution has the shape of an ultrasonic ILI tool (UT) wall loss map. Principally, it has the resolution of the original MFL ILI measurement data and not of simplified boxes. It does not depend on preset assumptions or the probability of appropriate selection of sizing models.

BLIND TESTING

The VDU application was blind-tested several times, including complex general corrosion, isolated pitting down to pinhole dimension and complex grooving structures with varying directions.

Figure 6 – Blind Test – Complex General Corrosion

Figure 7 – Blind Test – Isolated Pitting

Figure 8 – Blind Test – Complex Grooving

Nonetheless, the principal methodological restriction of ambiguity remains if using only one MFL direction: in that case, VDU would find a correct solution, of course, but not necessarily the true one. This fortunately can be realized by using two field directions: axial MFL (MFL-A) and circumferential MFL (MFL-C). Therefore, shortcomings like the sample calculation in the center of the grooving example do not appear.

For that reason, the VDU service integrates mandatory MFL-A and MFL-C to ensure reliability. VDU is a disruptive new technology that calculates laser-like metal loss profiles from MFL data.

Figure 9 – VDU Output Nature – 3D Result

Figure 9 – VDU Output Nature – 3D Result

BREAKING THE BOUNDARIES

The high-resolution, non-simplified, 3D nature of VDU results facilitates the use of more sophisticated assessment methods directly:

  • Detailed Remaining Strength (RSTRENG)
  • DNV RP-F101
  • Plausible Profiles
  • Finite Element Analysis

VDU, together with these methods, allows for more accurate and reliable safe pressure calculations and remaining life assessments, allowing for less conservatism, i.e. later repair.

Virtual−Dig Up is breaking the boundaries.

A RECENT APPLICATION CASE

Figure 10 – MFL-A and MFL-C results from a recent application case

Figure 10 – MFL-A and MFL-C results from a recent application case

Figure 11 – VDU results from a recent application case

Figure 11 – VDU results from a recent application case

Here are the first results of a recent VDU application:

Challenges

  • Anomaly reported by previous ILI (three different vendors)
  • External corrosion anomaly reported, with a depth exceeding 50% wall thickness in a difficult-to-access location
  • Access to pipe challenging; estimated repair costs high
  • Anomaly was sized independently using both MFL-A and MFL-C
  • Increased length reported by MFL-C technology resulted in an
    ERF > 1
  • Uncertainty over anomaly interaction and results varied depending on technology used
  • RSTRENG assessment based on either MFL-C only or a combination of MFL-C and MFL-A resulted in a critical anomaly

Solution

  • MFL-A and MFL-C data quality and tool set-ups reviewed and confirmed to be suitable for supporting DFA profile calculation
  • Second anomaly on the same pipeline identified for verification purposes; similar morphology but in an accessible location
  • Deep Field Analysis profiles generated for both “target” and “verification anomaly”
  • Detailed comparison of repeat MFL-A data performed to determine whether anomaly is active and estimate a historic growth rate
  • Failure pressure estimates obtained with a range of methods, including the Plausible Profiles approach

Outlook

  • Safe remaining life estimated
  • Active project; awaiting verification anomaly results