Analyzing MFL signals

Figure 1. POF metal-loss feature definitions.¹

Magnetic flux leakage is one of the most common and robust methods of ILI (In Line Inspection) used by pipeline operators worldwide to detect and characterize a broad range of pipeline features and anomalies. Magnetic flux leakage data signals are influenced by the shape and configuration of metal loss zones; with the feature shapes used as a basis for algorithm development and interpretation techniques developed to provide estimates of metal loss lengths, widths and depths. For pipelines, primary lengths and widths of features may be oriented in any random direction with respect to the pipeline axis and applied magnetic fields.

In addition to field levels, material properties, velocity effects, length, width, and depth, feature orientation will also influence the magnetic field data acquired for typical pipeline metal loss anomalies.  Difficulties encountered in characterizing specific classes of metal loss features have led to the development, implementation, and use of tools employing multiple axis sensors in an effort to improve performance for these categories of metal loss features. To demonstrate the effects of feature orientation, several magnetic flux leakage (MFL) data sets were analyzed. The results provided insight into the signal response for metal loss zones as the angular orientation of the features are varied within the applied magnetic field.

Figure 2. Magnetic vector components.

Maintaining safe operations
Worldwide, pipeline systems are utilized for the transportation of a broad range of products, ranging from crude oil and refined products to chemical stock; pipelines are an integral part of the energy and products distribution systems serving all market segments. Production and gathering facilities both inland and offshore are served by pipeline systems that collect and transport products, liquids and gases to processing plants or distribution facilities. Pipelines are among the safest and most economical means of transporting the vast quantities of energy products that are delivered and consumed daily by both domestic and global users.

While economic considerations do impact decision making with respect to day-to-day operation of pipeline systems, safety and continued safe operation of these systems are a priority of pipeline operators. With the implementation of integrity management programs by operators worldwide, greater emphasis has been placed upon monitoring these systems to insure continued safe operation. Several pipeline incidents have focused attention upon pipeline safety, resulting in legislative initiatives and rulemaking intended to improve upon an exemplary safety record.

Figure 3. Metal loss feature rotation – 0, 22.5, 45, 67.5, and 90 degrees.²

Integrity management programs targeted to reduce incidents quite often will focus on corrosion and mechanical damage type anomalies, using processes and methods to identify and monitor any areas that may be subject to either. Among the various methods available, Inline Inspection (ILI) tools using Magnetic Flux Leakage (MFL) are among the most widely accepted and used technique for detecting and monitoring pipeline corrosion. Focusing on volumetric or non-crack like features, MFL tools are capable of providing accurate assessments of corrosion features. Pipelines quite often are located in harsh environments that over time can result in coating degradation or disbondment, leading to corrosion; which, if left unabated, may impact the structural integrity of the pipeline. In addition to external features, internal surfaces are subject to corrosive products or mixtures capable of creating environments conducive to corrosion resulting from microbiological or other sources.

The growth regimen of corrosion zones and patches will result in metal loss zones forming, commonly referred to as pitting, typically being spherical in shape. Individual corrosion features can enlarge or coalesce into larger patches, creating elongated or irregular shaped zones, the orientation of which may be randomly distributed, often resulting in grooving and channeling type features.  In addition, operational and environmental conditions present within individual segments may also influence the shape of metal loss zones.

Figure 4. MFL amplitude change with degrees of rotation.

Metal loss features

    Typically, metal loss features are grouped and categorized with respect to surface dimensions, usually resulting in references to axial or circumferential extents. Figure 1 groups and defines metal-loss categories commonly referenced when describing corrosion zones. As indicated by the Figure, seven distinct groups are defined based upon extents, orientation and aspect ratio. Algorithms and analysis techniques developed for the interpretation of MFL metal loss signals may be optimized for features within each of the zones presented.  

The methods employed in describing metal loss zones through the interpretation of MFL signals will correlate the features present in MFL data signals to actual lengths and widths of the individual zones, using these to produce depth estimates for metal loss zones. During the development of algorithms and analysis techniques, varying the lengths of features with similar depths and widths will provide an indication of the effect of length on the MFL data signals and field profiles acquired at anomaly locations. Similarly, the depths and widths may be varied to establish the effects each of these will have on metal loss signals. The results are employed to establish conversion factors and coefficients for use in interpretative algorithms, ranging from statistical fits, inverse solution methods to neural networks.

As a supplemental feature, for metal loss zones with a defined principal axis, the angular orientation may be estimated and used as a variable for determination of corrosion zone parameters.  Modeling of MFL data displaying each individual axis along with data results previously reported2 can be compared to determine the relationship for several features. Within the data sets, circumferential grooving will be shown relative to axial grooving, rotating the features in approximately 22-degree increments. 

MFL data contains three components and may be described using each of the individual components, x, y, and z, commonly referred to as axial, radial and circumferential. Figure 2 provides a relative description of the individual components, with the axial or tangential direction typically aligned with the pipe axis, assuming an axially oriented applied magnetic field direction. The radial or normal component refers to the field direction perpendicular to the pipe surface. The third component is the circumferential or transverse direction referencing the leakage component arising at magnetic anomaly features, directed around the circumference of the pipe; the Magnitude (Bm) or total magnetic field being the summation of the thee vectors.

Data description
    The metal loss feature in Figure 3 is used here to illustrate the response to angular displacement. The metal loss zone was placed into a flat plate sample using a chemical etching process, the resulting area being non-uniform in depth accompanied by varying widths as visible in the photos of Figure 3. The flux leakage data amplitudes are given as the flat plate is rotated in 22.5-degree increments, with the 0 degree position oriented as axial grooving, ending in a circumferential grooving position. For this feature, the depth profile varies along the length of the metal loss zone, with the individual depth variations visible in the MFL signals present at the surface.
    The peak-to-peak amplitudes for each of the components and angular position are referenced in Figure 4. For the axial and radial components, the flux leakage amplitudes can be seen to increase monotonically as the metal loss feature is rotated. For this metal loss feature, the circumferential component displays a non-monotonic response with orientation, each angle position representing a percentage of either the axial or radial peak-to-peak amounts.
    Within the ranges through which the feature is rotated, the leakage amplitudes of each magnetic component can be evaluated; for the y axis, a change of nearly 6:1 is observed, for the x axis a 7:1 change is observed, with an 11:1 range for the z axis, these ratios representing the peak to peak maximum to minimum amounts. The metal loss zone with respect to volume removed has remained the same, the effective length and width changing as the defect is rotated through the angle range. Among several features modeled, the results for a rectangular groove are presented and described in the following graphs. The model uses a circular plate with a thickness of .250-in. incorporating a rectangular metal loss feature that measures .500-in. x 2.00-in. with a depth of 50% (0.125-in.). A permanent magnet assembly is used in the model to supply the magnetization field present within the flat circular plate. Each of the solutions is performed for the static case, ignoring velocity effects, for a far side feature. For the magnetic circuit, a magnetizing force of  >100 Oe (Oersted) is present within the defect area.  

Figure 5. Axial field data model surface plots.  

    In Figure 5 are the results of the modeled metal loss feature displaying the flux leakage surface profiles as the groove is rotated from an axial position to a circumferential orientation using 22.5-degree increments. Each individual panel is identified with the rotation amount noted below each plot. The arrow below the graphs indicates the magnetization direction for the sequence. Color coding for the group is consistent across all angle positions, providing a relative relationship for the amplitudes and magnetic flux leakage profile as the groove is rotated. Within each of these, in the absence of known orientation as derived from within the models, a principal axis can be inferred and an orientation angle may be estimated.
    Displayed in Figure 6 is the metal loss signal amplitudes derived from the model. The amplitudes for each axis and angular position are graphed, providing an overview of the effects of rotation on the individual data sets.

Discussion of findings
Driven in part by the aspect ratio of metal loss features, it can be seen that greater ratios would be present for slotting type features. For each of the data sets in Figure 4 and Figure 6, similar trends are present with respect to angular orientation. The experimental and modeled data provide similar results as the angular rotation for each feature is varied. Within the range of rotation, the severity of the features increases as the metal loss region changes from circumferential to axial orientation.3

Figure 6. MFL model amplitude change.

Figure 7. Signal amplitude corrections.

Using the principal axis of the metal loss signals as shown in Figure 5, an angular orientation may be estimated. The angular orientation may be used as a parameter in normalizing the amplitudes of metal loss signals when analyzing data associated with metal loss zones. Figure 7 provides a series of results derived from the known rotation angles for each of the metal loss signals. 

Using the peak-to-peak amplitudes, the values are normalized based upon the angle of the principal axis of the metal loss feature. Within this data set, the known angle is used in the general outline form in Equation [1] with coefficients for each axis. For this feature it is possible to produce an estimated amplitude correction across a very broad signal amplitude range. Additional refinements could include supplemental factors and coefficients based upon estimated aspect ratios, magnetization levels, surface extents, zone interactions with respect to length and width, etc.  

(Pk_Pk*((Cos (Ø -Ky1) + Ky2) + Cy1) * Ky3) - (Pk_Pk * ((Cos (Ø – Ky4)/2)*Ky5)) [1]

Conclusions
Magnetic flux leakage signals require interpretation in order to provide estimates of metal loss features.  From the data presented, it can be seen that angular orientation within a metal loss feature can have a significant influence on the flux leakage signal present. Angular displacements for each of the individual magnetic components, or combinations, may be used to generate additional parameters for use in the interpretation of metal loss signals. Additional efforts in both modeling and experimental data acquisition can be used in continuing improvement of analysis techniques and interpretation methods as applied to magnetic flux leakage data signals.     

Acknowledgment
Based on a paper presented at the NACE Corrosion 2009 Conference & Expo, held in Atlanta, Georgia, March 22-26, 2009.

References
1. “Specification and Requirements for Intelligent Pig Inspection of Pipelines,” Ver. 3.2,  January 2004, European Pipeline Operator Forum.
2. Simek, James C., “Effects of Angular Displacement on Magnetic Flux Leakage Multi-Axis Data”  NACE Central Area Conference,  (Sept 22,2007)
3. Coulson, K.E.W., and Worthington, R.G., “New Guidelines Promise More Accurate Damage Assessment”, Oil and Gas Journal, (April 1990).