Magnetic flux leakage (MFL) is a non-destructive testing (NDT) technique used to assess materials for corrosion, pitting, and deformations. MFL testing can be applied to pipelines, storage tanks, railways, cables, and other ferromagnetic materials. Inline Inspection (ILI) refers to the use of inspection devices commonly called intelligent pigs. When MFL techniques are applied to storage tank floors or other surfaces, floor scanners or other portable MFL technology can be employed to find defects.
ILI devices can be customized with several configurations for cleaning, inspecting, or other purposes. Intelligent, or smart pigs are outfitted with sensors and other tools to collect a wide variety of data. Configurations can include acoustic resonance technology, calipers/geometry sensors, electromagnetic acoustic transducers (ultrasonic), MFL sensor arrays, or a combination of these technologies. Improvements in computing and data analytics continue contribute to remarkable advances in detecting defects. Neural networks, machine learning, and developing technologies promise a future rich in advancements for MFL examinations.
2. Smart Pigs
Driven by product flow through a pipeline, pigs can come in a variety of materials, shapes, and sizes, each design best suited to a specific task or pipeline. Batching or separating differing products, cleaning, and inspecting are the most common uses for pigs. Before ILI begins, pipelines are often cleaned with such utility pigs.
The inline inspection devices known as smart pigs are ideally suited to performing examinations without halting flow of product through a pipeline. Technologies used in intelligent pigs vary depending on the pipeline and pipeline product. For instance, the use of ultrasonic technology such as an electromagnetic acoustic transducer in a smart pig is best suited to a pipeline carrying a liquid product. MFL equipped intelligent pigs are best utilized in a gas pipeline, while geometry pigs are ideally suited to either pipeline product.
The first commercial smart pig hit the market in 1964, though it lacked circumferential sensors, odometer wheels, and recorded data on digital magnetic audio tape. Gaining low resolution circumferential MFL in 1966 and hi-res MFL in 1976, the boundaries of early smart pig technology grew rapidly.
Today’s intelligent pigs are an efficient combination of technologies:
Large Sensor Arrays
80% and above Probability of Detection (POD)
Differentiate between internal and external detections
Employ rare-earth magnets for continuous MFL
Fully grade through casings and other structures
Employ gyros to determine clock positions for detections
Better navigation through radius bends
Often combine hi-res geometry for deformations and damage
Speed control for maintaining desired product flow
Distance tracked with Odometer wheels, GPS and above ground markers
3. MFL Principle
MFL testing relies on the uniform magnetization of a ferromagnetic material. Once magnetized near saturation, the magnetic field’s flux lines are observed. When material is missing, or damaged magnetic resistance increases at that site. The increased resistance causes a leak, distorting the magnetic flux lines generated by the MFL equipped ILI device. Sensors aligned perpendicular or parallel to the defect and perpendicular to the measurement surface detect the size and shape of the anomaly. Electrical signals which correlate to these sites are recorded and analyzed to determine the precise location and properties of a defect. Assuming proper magnetization and sensors for the pipeline, the final factor is data analysis and interpretation.
Mathematically modeling the magnetic leakage in using MFL technique grants this type of nondestructive testing a significant appeal. Since the technologies involved in MFL were first developed, analytical models have been pioneered to allow visual representation, higher POD percentages, and further advancement of MFL techniques. Analytical and numerical models are used for these purposes. The specific methods, software, and analytical approaches are often proprietary intellectual property of the company performing MFL testing.
4.1. Analytical Modeling
The magnetic leakage field, anomaly shape correlating to magnetic field strength, and magnetic permeability of material was first related by the magnetic dipole model. This model could be used to simulate the effects holes, pits, and other defects in a two-dimensional model. Relegated to detecting regular, simple defects despite the development of modern electromagnetic theory, analytical modeling is a helpful, convenient, and continually developing method of interpreting MFL data. More commonly used is the method of numerical modeling.
4.2. Numerical Modeling
Several methods of numerical modeling such as boundary element, finite difference, finite element, hybrid, and others are routinely used to numerically model defects within pipelines and storage tanks. The finite element method ranks among the most commonly used for numerical modeling. First introduced in 1975 by Lord and Hwang, the finite element method spurred tremendous research progress for calculating magnetic leakage from the magnetic field. Complex electromagnetic field vectors calculated through differential equations were transformed into algebraic equations, allowing for the modeling of these complicated magnetic field leakage shapes. Several key researchers would follow, each adding their critical input to calculating magnetic field leakage using the finite element method. These advancements, at first theoretical were simulated and experimentation began. The feedback iteration method is another numerical modeling method which allowed further refinement and simulation for natural gas pipeline defects. Advances in measuring techniques, computer technology, and sensors further improved POD rates and MFL testing overall.
5. Methods of Measurement
Measuring a magnetic field accurately and precisely is significantly aided by advancements in computing, electronics, and the sensors used in generating these measurements. The primary methods of measurement include:
Electromagnetic Induction Method
The Hall Effect Method
Magnetic Resistance Effect Method
Magnetic Resonance Imaging
Ground Marking Method
Quantitative Analysis of MFL detection
Statistical Identification of Defects
3D Finite Element Neural Network Method
The electromagnetic induction method is a straightforward measurement method, measuring AC/DC and pulsed magnetic fields. Electronic integrators, induction coils, impact galvanometers, flux meters, and vibrating coil magnetometers are used in measuring with this method.
The Hall Effect method relies on observing the change of magnetic field intensity. Changes occur as the electromotive force (generated by electric current in the magnetic field) is recorded by Hall sensors on a smart pig. Hall sensors are the most commonly used sensor with in-line non-destructive testing.
The magnetic resistance effect method examines the change of material resistance under the effect of magnetic fields. Sensors include ferromagnetic thin film and semiconductor reluctance elements.
The magnetic resonance imagining method utilizes frequency specific electromagnetic waves within the magnetic field to induce resonance, the intensity of which is discerned by observing the degree of resonance. Several types of devices are used to absorb or radiate particles for this purpose, including electron spin resonance magnetometers, nuclear magnetic resonance magnetometers, and others.
The magneto optical method takes advantage of magneto-optical and magneto-stricture effects to measure magnetic forces, though this method is restricted to use in non-harsh environments.
The ground marking method is crucial to successful MFL testing. As the smart pig navigates a pipeline—potentially one tens of thousands of kilometers long—odometer wheels accumulate errors due to rotation, sliding, and other factors. Mitigating these errors with precise tracking systems such as above ground markers (AGM) is essential to accurate tracking and detection of defects. Non-magnetic sensors play a significant role in the precise and accurate record of these locations, though static magnetic field technology and ultra-low frequency electromagnetic field technology are widely employed to determine the location of the MFL device’s detections.
Quantitative analysis of MFL detections is the process of analyzing and identifying magnetic leakage signals after compensations are made. This process contains several subdivisions of analysis and is highly important to development of a defect library. Developing an extensive library is essential to signal analysis or pattern classification, and dovetails into algorithm and neural network development. Quantitative analysis is an iterative process of refining detection data for greater degrees of accuracy in detections.
The statistical identification of defects explores the relation between height, length, and width of defects. This complex process evaluates length the easiest, while the width and depth remain complicated. Examining statistical data in this manner produces excellent results in practice, though it requires extensive testing and simulation. Computer neural networks are routinely applied to process and analyze vast amounts of statistical data toward this end.
The 3D finite element neural network method uses the finite element calculation model in conjunction with a neural network to render a three-dimensional model or diagram. A neural network is a form of parallel computing that functions as a simplified facsimile of the human brain. A network of computers is designed to analyze, process, and sometimes learn or adapt enormous quantities of data within given parameters.
A wide variety of magnetic sensors with various measurement ranges, strengths, and weaknesses are available for use in MFL testing. Their basic function is to translate magnetic signals into electrical signals.
Additional non-magnetic sensors typically take the form of differential pressure sensors, hydraulic sensors, odometers, pneumatic sensors, and temperature sensors. The data from these sensors is compressed and stored while noise reduction techniques are applied to increase the accuracy of data analysis.
6.1. Selection of Sensors
Selecting the right sensors for non-destructive MFL testing is critical to the success of the examination. One sensor is likely to be preferred in a given environment over another, for detection of buried defects or surface defects for example.
The signal pattern of MFL testing is dependent upon the sensor used; for example, when examining change in a magnetic field or the magnetic field itself. Certain sensors are more durable while displaying less sensitivity, some have greater dynamic range than others. Sensors are always selected based on the expected circumstances, defects, and the magnetic leakage fields they create. Correctly using a combination of sensors may provide the best results.
6.2. AMR Sensors
Anisotropic magnetoresistance (AMR) sensors measure a phenomenon generated by applying an external magnetic field which rotates magnetization, changing resistance. This occurrence depends on the angle between the electric current and the direction of magnetization. Commercially available and less prone to noise when compared to giant magnetoresistance (GMR) sensors, AMR sensors can be integrated into an MFL configuration as needed.
6.3. Magnetic Fluxgate Sensors
Magnetic fluxgate sensors are comprised of a ferromagnetic core, a primary excitation coil, and a secondary pickup coil (drive, output, and control). When the core is not fully saturated it presents a low magnetic resistance path to the magnetic flux lines of the external magnetic field. When saturated, the resistance of the core increases, producing excess magnetic flux lines.
6.4. GMR Sensors
Giant magnetoresistance sensors consist of alternating ferromagnetic and non-ferromagnetic thin film multilayers and operates on the GMR effect. It detects large changes in electrical resistance to a magnetic field. This change is the result of spin-dependent scattering of electrons. Boasting high spatial resolution and sensitivity at low magnetic fields, they excel at the rapid scanning of surfaces when integrated into a sensor array.
6.5. Hall Sensors
Hall sensors are widely used in MFL technique, operating on the Hall Effect by measuring the normal component of the magnetic field. Inexpensive, relatively small and exhibiting high linearity, these sensors are a standby in ILI devices.
Magnetodiodes are thin rectangular semiconductor plates that measure magnetoresistance. More sensitive than Hall sensors, these sensors have difficulty operating in high magnetic fields.
6.7. Pick-Up Coil
Pick-up coils are simple copper wire, wound around a core. They are capable of measuring change in magnetic fields in the presence of anomalous defects. Inexpensive and highly adaptable for a variety of sensor configurations, these sensors are less sensitive than others.
7. Capabilities and Limitations
Non-destructive MFL presents several well-established capabilities, including:
Automatic MFL of components in production lines
Detection of corrosion, cracks, dents, and gouges in pipelines and tanks
Detection of flaws and cross-sectional area in wires and wire ropes
High sensitivity detection of surface cracks w/o paint or rust removal
Non-destructive testing of hot objects and materials
MFL testing presents a few limitations due to its nature, including:
Magnetization must be perpendicular to expected anomalies to attain maximum leakage field, thus detection
MFL signals are constrained by stress and velocity, requiring exceptional experience and skill to properly interpret data
MFL testing is restricted to ferromagnetic materials
Measuring normal and tangential parameters of a defect is a complex requirement of MFL technique
8. Recent Trends and Future Directions
Artificial Intelligence and Expert Systems
Machine learning or artificial intelligence (AI) relies on a core of information supplied by highly knowledgeable and experienced experts in a given field. This core of information is called the expert system. The expert system is comprised of a comprehensive database, inference engine, a knowledge base, knowledge acquisition modules, and a human-computer interaction interface.
The comprehensive database stores a wide variety of information involved in the reasoning process of an AI system, including assumptions, intermediate results, objectives, and other functions.
An inference engine coordinates and controls the system, using data input, existing knowledge, and reasoning strategies. Typical reasoning strategies include forward, backward, and hybrid reasoning.
Knowledge acquisition modules allow growth within the system and improve both accuracy and problem solving for the AI. These modules provide additional knowledge to the system, expanding and modifying the AI.
The human-computer interface allows users to interact with the AI system to adjust, adapt, specify, and otherwise submit requests to the system.
Advances in technology are at the core of non-destructive examination methods like MFL. The decades following its development have seen relentless innovation by practitioners, scholars, and engineers alike. Highly experienced engineers and technical personnel have contributed their accumulated knowledge to further development of sophisticated analytical methods. In addition to improvements in virtually every aspect of MFL testing, technologies that were theoretical a decade ago are presently in use, driving innovation faster than ever before.
The further use of neural networks, expert systems within AI, dual field MFL technology, robotic inspection tools, and UAV technology ensures a host of future developments in MFL and non-destructive testing.
The use of MFL technology for examination is critical for maintaining pipeline integrity and remains the most prevalent method for doing so. As a non-destructive examination method, utilizes inline inspection devices known as smart pigs to travel the length of pipelines, detecting defects with magnetic sensors. This technique can—with 80% or higher probability of detection—observe defects on the internal and external surfaces of a pipeline. Hall sensors are favored for measuring magnetic leakage fields and ground marking systems for accurately determining location of defects. Once data is gathered, compressed, and stored, complex statistical identifications methods are employed to MFL signals to defect size and shape. Advances in software, computing, and artificial intelligence have paved the way for new advances to follow, as innovations continue to drive improvements in non-destructive testing methods.
We are launching courses in Pipeline Design and Business & Project Development soon. Sign up and will notify you.