How MFL Data Fusion Will Enhance Pipeline Integrity Management

Magnetic Flux Leakage (MFL) technology is the most widely used method for identifying metal loss (volumetric) anomalies in pipelines. Our expert, Hazem Rahmah, Service Manager for Data Fusion at ROSEN, dives deeper into how combining the strengths of different MFL technologies will benefit the industry.

ROSEN is the world leader in both MFL-A (axial) and MFL-C (circumferential) technologies for In-Line Inspection (ILI). The different orientations of the magnetic field in these technologies offers distinct advantages in detecting anomalies that are oriented either circumferentially or axially. However, both technologies rely on a unidirectional magnetic signal, which imposes certain limitations in detecting and accurately measuring more complex metal loss morphologies.

As the industry asks for less conservative integrity assessments, there is an increasing need to enhance the accuracy of ILI findings to support more effective pipeline maintenance and repair programs. One approach that offers clear benefits is Data Fusion where data collected from different MFL tools are fused together to improve accuracy.

The Need for MFL Data Fusion

The severity of pipeline incidents, combined with increasing pipeline throughput, commercial demands and the need to extend pipeline lifetimes, has put pressure on the industry to further reduce the risk and cost associated with corrosion. The traditionally accepted levels of uncertainty in ILI findings are no longer sufficient when dealing with heavily corroded pipelines. This has led operators to take conservative approaches, often resulting in high costs for excavations to ensure pipeline safety that turn out to be unnecessary when the actual damage dimensions are known. MFL data fusion can increase certainty, reducing the need for such conservatism.

There is also a growing need to replace the conventional, labor-intensive practice of combining separate MFL-A and MFL-C reports with an innovative solution. While combined reporting of these independently analyzed datasets aims to improve understanding, the process is often subjective and inefficient, creating challenges in interpreting the results. The fusion of MFL data provides an objective solution that offers higher accuracy and additional benefits, such as 3D metal loss depth maps.

Pipeline integrity experts have long desired to merge data from different ILI tools into a single, comprehensive output that maximizes the strengths and minimizes the limitations of each technology. With advancements in technology, including Convolutional Neural Networks (CNNs) and algorithm development, ROSEN has made this possible for MFL-A and MFL-C.

Hazem Rahmah, Service Manager Data Fusion, ROSEN Group

Challenges Addressed by MFL Data Fusion

Avoiding Unnecessary Digs: Uncertainty in ILI results can lead to the unnecessary excavation of defects that do not pose a threat, wasting time, resources, and focus that could be better applied to genuine risks.
Complex Anomaly Identification: Traditional MFL techniques often struggle to detect and size complex anomaly morphologies, such as pitting within general corrosion. Since many corrosion defects are complex, relying on a single unidirectional magnetic field can make it difficult to accurately size the deepest points.
Integrity Management Uncertainty: Uncertainties in defect dimensions lead to conservative integrity decisions, as operators must assume the worst-case scenario, which can inflate costs.
Inefficient Pipeline Inspection Planning: Inaccurate assumptions during inspection planning can result in choosing the wrong ILI technology, leaving some critical defects undetected due to the limitations of the selected method. Not to mention a higher frequency of inspection than is necessary due to uncertainty over both condition and growth rates.

Fusion of MFL-A and MFL-C Data: Joining Forces

Combining the strengths of axial and circumferential magnetic fields improves feature characterization across all POF anomaly dimension classes. ROSEN’s Data Fusion Technology combines signals from both MFL-A and MFL-C tools into a single 3D depth map using a Machine Learning Data Fusion Model. Once data sets are correctly aligned, the pre-trained Convolutional Neural Network (CNN) can be applied to any corrosion anomalies in the pipeline.

Chart showing 3D depth map using data fusion vs. laser scan for advanced data analysis.

Figure 1: Example of 3D depth profile with MFL data fusion vs. Laser scan data for advanced analysis

This approach advances analysis from simple anomaly signal amplitude ‘boxes’ into area grids and 3D profile analysis, allowing more precise defect dimension identification and the generation of a River Bottom Profile (RBP).

As a result, fused data can accurately calculate burst pressures based on actual metal loss depth maps. Advanced failure pressure assessments such as Psqr can help further reduce conservatism. The high-resolution 3D depth map enables the evaluation of every corrosion anomaly on the pipeline, similar to an in-ditch laser scan.

Figure 2: Example of RBP using data fusion vs. Laser scan data for enhanced BP calculation

Benefits of Data Fusion for the Industry

Improved Sizing Accuracy: Data Fusion offers better definition and higher accuracy for each feature, regardless of defect morphology, increasing certainty. For example, depth accuracy of ±5% across POF dimension classes.
Enhanced Safety Through Optimized Dig Programs: By improving prediction accuracy, Data Fusion enables more efficient allocation of budgets for digs and repairs, contributing to enhanced public safety.
Comprehensive Feature Assessment: Fusing data across an entire pipeline enables more proactive asset management, optimizing maintenance schedules and resource allocation.
Detailed 3D Anomaly Mapping: Data Fusion effectively addresses the challenges posed by complex corrosion surfaces in MFL inspections, delivering results regarding geometrical structure and sizing accuracy comparable to ultrasonic testing (UT).
Overcoming Technological Limitations: Data Fusion mitigates the limitations of traditional MFL technology, enabling significant increase in capability and certainty of characterization all metal loss feature types.

Data Fusion Service by ROSEN

Data Fusion service “Performance Specifications” offers an enhanced certainty and accuracy for all metal loss features, regardless of their morphology. The fused ILI data can be delivered in 3D depth map form, very similar to laser scan data but for the entire pipeline, with the ability to derive accurate river bottom profiles (RBP) and different models of failure pressure assessment (Effective area, Psqr) per anomaly or segment.

The two datasets (MFL-A and MFL-C) required for Data Fusion can be collected using separate ILI tools in different runs. Ideally, the time gap between data collections should be short to prevent changes in corrosion morphology from affecting the fusion process. In some instances, for larger diameter pipelines (over 20 inches), both datasets can be collected in a single run, though this is subject to a case-by-case assessment of operational conditions. No particular ILI tool setup or pipeline conditions are required for Data Fusion as such, and previously collected data, such as laser scan data, can be used to fine-tune algorithms and create a model specific to the pipeline of interest.

Next Steps

ROSEN’s Data Fusion Service is currently in the final stages of development. While the core concept, algorithms, and results have already been validated, the fully commercial service will be tested in several pilot projects in early 2025. The initial service model is expected to be available in the second half of 2025.