3. Magnetic NDT technology
Magnetic NDT technologies have been extensively adopted in engineering to ensure the operating safety of ferromagnetic structures and components [30]. In this section, the studies of three typical magnetic NDT technologies (MFL, MBN and MMM) will be summarized. Among them, MFL and MBN techniques may be ascribed to active magnetic test methods in which a strong magnetic field is applied. However, the MMM technique is a weak-field test method in which the Earth’s magnetic field instead of an artificial field is used as the stimulus.
3.1. Magnetic flux leakage
Magnetic flux leakage (MFL) is one of the traditional electro- magnetic NDT techniques, originated from magnetic particle technique. Hoke first discovered the MFL phenomenon in 1918. However, due to the lack of magnetization techniques in the early time, the first application of the MFL technique was performed by Watts in 1933 in assessing the quality of the welded joints. Since 1960s, this technique has been extensively used as an inspection technique in the petrochemical engineering and transportation, energy and metal industries [31–37]. One successful application of the MFL technique is the device, called ‘‘pipeline-pig’’, which is developed to detect the corrosion and metal loss in oil and gas in-service pipelines [38–40]. Fig. 3 shows a scheme of the structure and operating principle of the ‘‘pipeline- pig’’. A strong permanent magnet in the ‘‘pipeline-pig’’ nearly saturates the pipe wall when it is propelled by the oil/gas pressure or driving equipments. No flux is leaked out if the pipe wall is perfect (Fig. 3a). However, the flux ‘‘leaks’’ out of the wall at the location of a metal loss defect (Fig. 3b). The ‘‘leakage flux’’ is detected by an array of circumferentially distributed sensor assembly. The MFL data are sampled and stored using an on- board data acquisition system, and subsequently analyzed offline by trained data analysts.
As a classical NDT method, the principle of the MFL technique is relatively simple. That is, when a strong magnetic field is applied to a ferromagnetic material, any geometrical discontinuity in the test object will cause the field to leak out of the object into the air (see Fig. 4). The flux leakage can be monitored by a magnetic field sensor and used to estimate the dimensions of the defect. Although the MFL phenomenon is easily understood, the design and analysis of MFL systems involve complicated interactions between the excitation field, leakage flux and the defects in the material. There are several important aspects to be considered. Firstly, the level of
the excitation magnetic flux should be large and homogenous to allow the magnetic flux variation to occur at the location of a defect; secondly, the sensors should be located close to the position at which the changes in the magnetic field density, originating from the defect, are distinct from the background noise; addition- ally, developing an effective inversion method to identify the defective characters by the recorded MFL signals is difficult since the defect is irregular.
A great number of efforts have been devoted to develop a simple analytical model to explain the formation of the MFL signal [41–49]. Those developed models may be classified into two types. One was originatedfromFo ̈rstermodel[41]byconsideringthechangeofthe magnetic parameters (e.g. permeability and coercive field) in the local defect region, and the other was developed from Zatsepin– Scherbinin’s model [42] where magnetic dipoles were assumed in the surface of the defect. Clearly, Fo ̈rster model described the MFL phenomenon in the macroscopic view but Zatsepin–Scherbinin’s model in the microscopic view. In order to consider the non-linear magnetic behavior of ferromagnetic materials, more complicated models have been proposed in [43–49].
The reported models can effectively describe the fundamental features and the topography of the MFL field for regular defects (e.g. rectangular slot), but not for irregular defects. In order to analyze the MFL signal generated by irregular defects, some integral equations in describing the defect-induced magnetic charges were given based on the linear approximation of ferro- magnetic materials, which can be solved numerically using the iteration method [50–53].
Compared with the theoretical analysis, the magnetic finite element method (FEM) is a powerful tool for the investigation of the MFL signal due to its flexibility in the simulations of varied irregular geometrical defects. 2D magnetic FEM methods [54,55] provide sufficient information for sharp-shaped defect characteriza- tion, but do not accurately quantify the natural defects, e.g. stress corrosion cracks. Therefore, 3D magnetic FEM has received more attention in recent years. In this aspect, the study in [56] presented comparisons between 2D and 3D models. The studies in [57–60]
Fig. 4. Schematic representation of the flux leakage in the presence of a geometric discontinuity.
Fig. 3. Structure and principle of the ‘pipeline-pig’. (a) No magnetic flux leakage for perfect pipe wall; (b) magnetic flux leakage in defect position.
investigated the characteristics of MFL signals due to corrosion defects. Additionally, the effect of the local dent-induced stresses on the MFL signal was studied using both magnetic FEM modeling and experimental tests in [61]. A 3D simulation, aimed at studying the influence of the defect geometry and lift-off value, was performed in [62]. Further, the study in [63] examined the effects of both dent geometry and localized residual stresses on the MFL signal using a 3D magnetic FEM method.
It may be pointed out that the MFL technique is one of the most popular magnetic NDT techniques and extensively used in various engineering fields. In the MFL test, the key is how to inversely determine the defects of the investigated object using the recorded MFL signals. Various new algorithms (e.g. wavelets, neutral networks and genetic algorithm) were applied in [64,65]. However, there are two primary obstacles in the defect inversion. Firstly, the defect in the reality is commonly complicated and usually characterized by many parameters such as the width, thickness, location and edge condition of the defect, which in turn significantly impact on the measured MFL signals. Clearly, it is generally difficult to characterize every parameter based on the measured MFL signals. Secondly, it is still a challenge to deal with the elastic–plastic zone near the cracks, and idealized material properties are commonly assumed. This is in part due to the fact that the effects of the plastic deformation on the magnetic characteristics of ferromagnetic materials are not thor- oughly understood.