- Email: [email protected]
Velocity effect analysis of dynamic magnetization in high speed magnetic flux leakage inspection
NDT&E International 64 (2014) 7–12
Contents lists available at ScienceDirect
NDT&E International journal homepage: www.elsevier.com/locate/ndteint
Velocity effect analysis of dynamic magnetization in high speed magnetic flux leakage inspection Ping Wang a, Yunlai Gao a,n, GuiYun Tian a,b, Haitao Wang a a b
College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu, 210016, China School of Automation Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan, 611731, China
art ic l e i nf o
a b s t r a c t
Article history: Received 22 September 2013 Received in revised form 29 January 2014 Accepted 13 February 2014 Available online 20 February 2014
The investigation described in this paper focuses on the velocity effect of dynamic magnetization and magnetic hysteresis due to rapid relative motion between magnetizer and measured specimens in highspeed magnetic flux leakage (MFL) inspection. Magnetization intensity and permeability of ferromagnetic materials along with the duration of dynamic magnetization process were analyzed. Alteration of the intensity and distribution of magnetic field leakage caused by permeability of specimen were investigated via theoretical analysis and finite-element method (FEM) combined with the actual highspeed MFL test. Following this, a specially designed experimental platform, in which motion velocity is within the range of 5 m/s–55 m/s, was employed to verify the velocity effect and probability of a highspeed MFL test. Preliminary results indicate that the MFL technique can achieve effective defect inspection at high speeds with the maximum inspection speed of about 200 km/h being verified under laboratory conditions. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Velocity effect Dynamic magnetization Magnetic hysteresis MFL High-speed inspection
1. Introduction Magnetic flux leakage testing is an efficient electromagnetic non-destructive testing (NDT) [1] technique that has been extensively used for defect inspection and characterization of ferromagnetic materials such as pipelines, pressure vessels, rail tracks and wheels, ropes or cables, etc. Magnetic flux lines generated by a magnetizer are coupled into test specimens with air coupling. Any geometrical discontinuity or local anomalies are manifested as an abrupt change of magnetic permeability [2–4] and force magnetic flux to leak out of the specimen in the poles of yoke in the air. Leakage magnetic field which contains information of defect is collected by magnetic field sensors and used to evaluate the defect dimensions and structural performance. For the advantages of its simplicity, low cost, air coupling and non-contact application, MFL testing is extremely suitable for the automated in-line and real time defect inspection. Some advancing works such as 3-D sensing of magnetic field [5–7], pulsed electromagnetic method (PMFL/ PMR) [8–10] and orthogonal magnetization [11] are currently under different stages of development and application for description of the shape and dimensions of defect. Although this method has a high probability of defect inspection, it still fraught with problems associated with the sensitivity n
Corresponding author. E-mail address: [email protected] (Y.L. Gao).
http://dx.doi.org/10.1016/j.ndteint.2014.02.001 0963-8695 & 2014 Elsevier Ltd. All rights reserved.
and interpretation of MFL signals to many factors, such as the condition of magnetization, inspection velocity, the B–H curves [4] of a specimen, lift-off, etc. Additionally, a conventional magnetostatic model is unsuitable for the high-speed MFL test. Eddy current distributed in conductors induced by relative movement between the MFL probe and a specimen will alter the profile and intensity of magnetic field leakage and distort the profile of MFL signals [12–17]. It also brings about difficulty in the signal interpretation and description of the defect. The target of improving the probability and accuracy of defect inspection has been attempted in previous work. Many numerical simulations based on 2-D or 3-D transient FEM models were carried out to simulate the distribution of motion-induced eddy currents and analyze their effect on MFL signals [13–20]. In addition, some methods on compensation and velocity invariance of MFL signal to minimize the velocity-induced eddy current effect have been introduced in previous papers [3,17–19]. Motion-induced eddy currents have also been utilized in the description of stress corrosion cracks in terms of measuring perturbation fields [21]. The magnetic flux lines flowing into a test specimen and the magnetization intensity are very important for the sensitivity of MFL signals and the defect inspection ability [2–4]. However, numerical simulation in previous work of the velocity effect only concentrated on the motion-induced eddy current and their influence on MFL signals; the factors of dynamic magnetization and the hysteresis effect during the high speed MFL test were
8
P. Wang et al. / NDT&E International 64 (2014) 7–12
neglected. The hysteresis effect exists in the dynamic magnetization process accompanying the magnetic domain rotation and domain wall movements which shows Barkhausen noise (BN) and magnetic hysteresis loop. The duration of the magnetization process decreases with the increase of inspection velocity because of the constant distance between sensor and the starting point of magnetization. If the higher inspection velocity and shorter time of the magnetization process bring about impact on the dynamic magnetization process, magnetization intensity and permeability of specimen as well as leakage magnetic field will be altered. This paper addressed the problem on the velocity effect of dynamic magnetization and magnetic hysteresis during high speed MFL inspection. In dynamic magnetization process, the distribution of magnetic resistance around defect and the velocity effect on specimen permeability and detection signals during the high speed MFL inspection have been analyzed, and verified in FEM simulation and actual high speed test. The rest of this paper is organized as follows: Section 2 describes the theoretical analysis of the dynamic magnetization and its effect on magnetic resistance and field distribution. Section 3 presents FEM simulation on magnetic field leakage along with the permeability of specimen on the basis of magnetization due to velocity effect. Section 4 elaborates on the experimental study with a specially designed high-speed MFL inspection platform. And then, derived conclusions will be given in Section 5.
2. Theoretical analysis of the dynamic magnetization effect 2.1. Ferromagnetic magnetization Ferromagnetic material is composed of many small spontaneous magnetization areas called magnetic domains, and the transition area between adjacent magnetic domains is called the magnetic domain wall. Magnetic domains distributed in the spontaneous magnetization direction and the ferromagnet do not show magnetism outward when the object is not magnetized. When an extra external magnetic field is applied to the specimen, magnetic domains flip and rotate in the direction of the applied magnetic field accompanied with the domain wall movements. Fig. 1 illustrates the initial magnetization curve [22] of the ferromagnetic material with four stages showing the dynamic change during magnetization and the variation of micro-level magnetic domain and domain wall structures [23]. When all the magnetic moments of domains tends to be consistent with applied magnetic field direction and the magnetization intensity no longer
Fig. 1. Initial magnetization curve of the ferromagnetic material, (I) reversible magnetic domain wall motion, (II) irreversible domain wall motion, (III) reversible domain wall motion and domain magnetization rotation, and (VI) only domain magnetization rotation [22].
increases, it means that the specimen is to be saturation magnetized. The magnetization process is governed by the following equations M¼
∑m0 ¼ χmH V
B ¼ μ0 H þ μ0 M ¼ μ0 ð1 þ χ m ÞH ¼ μ0 μr H ¼ μH
ð1Þ ð2Þ
where M, H, B and V, respectively, represent magnetization intensity, magnetic field intensity, magnetic flux density and volume of the magnetized specimen; ∑m0 and χ m , respectively, are the sum of magnetic moments in a certain volume and the magnetic susceptibility of specimen; μ0 and μr denote the permeability of air and relative magnetic permeability of materials with respect to air, and μ represents the absolute permeability of the medium. During the magnetization process, the variation of magnetic flux density B in a specimen lags behind the applied field H by a phase-shift because of the internal magnetic damping and some energy losses in specimen according to Jiles–Atherton model [24–26] and Landau–Lifshitz–Gilbert (LLG) equation [27,28] which is famous for the description of dynamic magnetization process. After the applied field is revoked, ferromagnetic specimen can still keep part of original magnetism, which is called the magnetic hysteresis phenomenon [29–31]. This phenomenon is the result of irreversible migration of the magnetic domain wall, which is affected by the internal friction of magnetic materials [23], the damping effect of micro eddy currents around a moving domain wall [23], the stress of specimen, material hardness and impurities, lattice defects, and so on. It can cause the resistance of the migration of magnetic domain wall and rotation of magnetic domains. According to the magnetic hysteresis phenomenon, the ∑m0 will be decreased due to the velocity effect of dynamic magnetization. It will lead to the decrease of χ m and the permeability of magnetized specimen [22–31]. 2.2. Dynamic magnetization effect of the high-speed MFL inspection A conventional MFL testing model [6] using a yoke-electromagnet is illustrated below in Fig. 2. The MFL probe incorporates a magnetizer and the sensor travels on the specimen along the scanning direction at a certain velocity (V). MFL signals are the magnitudes of leakage magnetic field measured by sensors positioned in the middle of the two magnet poles and at a constant distance over specimen (lift-off). The distance between two yokeelectromagnet poles is a constant, represented as L in Fig. 2, which is the valid magnetization distance in MFL inspection. The dynamic magnetization processes during the high speed MFL inspection is like a magnetization with a sinusoidal current of a certain angular frequency ð2π Þ=T due to the applied magnetic field induced by a moving yoke-electromagnet. The magnetic field acting on the magnetized specimen first is zero, then increased to the strength of the south-pole, it changes to the strength of the horizontal field between the two poles, and decreases further to the strength of the north-pole before reducing absolutely to zero
Fig. 2. A conventional magnetic flux leakage testing model [6].
P. Wang et al. / NDT&E International 64 (2014) 7–12
again. Based on the valid magnetization distance, the duration from the beginning to the end of dynamic magnetization process, represented as T, can be described by Eq. (3) below, and the T decreases with the increase of inspection velocity on the basis of Eq. (3). According to the magnetic hysteresis phenomenon, the χ m and permeability of a specimen will decrease due to the dynamic magnetization in the high speed MFL inspection. T ¼ L=V
ð3Þ
As is shown in Fig. 2, magnetic resistance around the defect is divided into two parts consisting of (a) and (b) in the boundaries of the specimen surface. According to Ohm’s law for magnetic circuit, magnetic resistance and flux can be described as shown in the following equations: NI ¼ ϕ U Rm ¼ ϕðaÞ U RðaÞ ¼ ϕðbÞ U RðbÞ ¼ ϕðbÞ U
lðbÞ
μðbÞ SðbÞ
ϕ ¼ ϕðaÞ þ ϕðbÞ
ð4Þ ð5Þ
where NI, ϕ and Rm , respectively represent magnetomotive force, magnetic flux and magnetic resistance in the whole magnetic
9
circuit, and ϕðaÞ , ϕðbÞ , RðaÞ and RðbÞ are the same concepts as described above in position (a) and (b); lðbÞ and SðbÞ , respectively, represent the length and cross-sectional area of magnetic circuit in position (b); μðbÞ denotes permeability of specimen in position (b). The permeability of a yoke-magnetizer and specimen is greater than air, so resistance of the whole magnetic circuit mainly depends upon the air gap. Magnetic flux of the whole circuit basically remains unchanged regardless of the constant air gap and velocity-induced eddy current effect in high-speed MFL inspection when the same defect is detected. With the decrease of μðbÞ and the increase of RðbÞ due to the raise of inspection velocity, ϕðbÞ is decreased and the opposite ϕðaÞ is increased according to Eqs. (4) and (5) and Fig. 2. This phenomenon is just like the situation of the difference between existence and absence of defect in MFL inspection. From the above, magnetic flux leakage increases as the increase in inspection velocity takes the dynamic magnetization and magnetic hysteresis effect into account. Additionally, the effect of motion-induced eddy current due to the relative movement between applied magnetic field and specimen should be taken into account in the high speed MFL inspection.
3. Simulation on the dynamic magnetization effect in high speed MFL inspection
Fig. 3. Simulation model for the analysis of velocity effect of dynamic magnetization and magnetic hysteresis in high speed MFL inspection.
The electromagnetic FEM simulation models have been widely employed in solving Maxwell equations and providing guidance to solve the engineering problems. The domain wall dynamics in conjunction with microstructure is ongoing research to bridge the gaps of micro and macro observation and modeling [32]. However, the simulation of a static and transient FEM model does not take the dynamic magnetization and magnetic hysteresis effect that is present with the magnetic domain rotation and domain wall movements into account. Hence, simulation on the velocity effect of dynamic magnetization in high speed MFL inspection should combine theoretical analysis and the actual test.
Table 1 The properties of simulation model. Components Excitation coil
Magnetizer
Specimen
Defect
Materials Properties
Ferrite-yoke Permeability μr ¼ 5000; Conductivity¼0.01 S/m
Steel_1008 Permeability μr : having BH curve, and variation with the scope of constant 1–5000; Conductivity ¼ 2e6 S/m
Rectangular air slot Permeability μr ¼1; Conductivity¼ 0 S/m
Copper Permeability μr ¼1; Conductivity ¼ 5.8e7 S/ m; Current source¼ DC 20000A
Fig. 4. Simulation results, (a) the leakage magnetic field of Bx component, (b) the leakage magnetic field of By component.
10
P. Wang et al. / NDT&E International 64 (2014) 7–12
3.1. Simulation set-up Ansoft Maxwell EM V12 [33], one of the leading finite-element software packages for numerical simulation, was employed to analyze velocity effects of dynamic magnetization and magnetic hysteresis in high speed MFL inspection. As illustrated in Fig. 3, the model is set up in 2-D with X–Y coordinates representing the crosssection of MFL probe and specimen. The simulation involves investigation of the characterization of MFL signals with variations of permeability in specimen due to the magnetic hysteresis phenomenon and inspection velocity. The dimensions and properties of probe, specimen and defect are illustrated in Fig. 3 and Table 1.
inspection velocity in high speed MFL inspection whilst the velocityinduced eddy current effect is not taken into account.
4. Experimental investigation of the high-speed MFL inspection For verifying experiments, a series of high-speed test on specimens using MFL were carried out with the intention of measuring 3-D components of leakage magnetic field in different inspection velocities with the scope of 5 m/s–55 m/s. A specially designed test platform and experimental models were employed to give us some initial results that illustrate the capabilities of the high speed MFL inspection.
3.2. Simulation results and analysis 4.1. Experimental set-up According to theoretical analysis of the dynamic magnetization and magnetic hysteresis effect, permeability of the specimen changes with the magnetizing duration, whilst leakage magnetic field alters with motional velocity due to the variation of magnetic resistance around the defect in high-speed MFL inspection. Magnetic intensity and permeability of the specimen is an approximate constant under the condition of a fixed inspection velocity or a constant state of magnetization. Hence, a constant equivalent permeability of magnetized material is proposed and permeability of the specimen changing with the scope of 1–5000 compared with the B–H curve of steel_1008 which is implemented in the simulation. The magnitudes of leakage magnetic field are represented as the components of Bx and By along the line over specimen and are observed and illustrated in Figs. 4 and 5. As is shown in Fig. 4(a), the magnitude of Bx where there is no defect increases with the decrease of equivalent permeability of specimen. From the profile of leakage magnetic flux intensity illustrated in Fig. 4(a) and (b) and Fig. 5, it is noticeable that magnitudes of Bx and By components increase first and then decrease with the decrease in permeability. The maximum magnitude shown in Fig. 5 is located at the turning point where the equivalent permeability is 50. Moreover, the magnitude of MFL signals with the specimen of steel_1008 having the B–H curve is close to the situation that equivalent permeabilities are 20 or 150, respectively. According to the magnetism and magnetization of ferromagnetic materials, the equivalent permeability of steel_1008 is approximate close to 150, certainly, the actual permeability of magnetized steel_1008 is very low due to the B–H curve. Hence, on the basis of simulation results that equivalent permeability of specimen is greater than 50, the magnitude of MFL signals increases with the decrease in permeability of specimen due to the increase in
Fig. 5. Simulation results: the magnitudes of Bx and By against equivalent permeability of specimen.
The sample is a circular turntable specimen that contains surface-breaking defects on the edge of circle. The test specimen was designed completely according to the shape and size of rail head and material of specimen is same as one of U71Mn rail. A series of different types of artificial cracks are distributed on the specimen surface. Rotation linear velocity of the edge of circle is within the range of 5 m/s–55 m/s under the drag of the motor in the high speed testing platform, with the block diagram and actual
Fig. 6. The block diagram of high speed MFL test system.
Fig. 7. The photograph of test platform.
P. Wang et al. / NDT&E International 64 (2014) 7–12
platform photograph as shown in Figs. 6–8. The MFL test probe consists of electromagnet of yoke-1 with DC excitation and 3-D hall sensors are positioned at a constant distance of 1 mm over specimen to detect leakage magnetic field and the defects indication is recorded. The surface of the magnet pole has a good coupling with sample based on the radian and shape of specimen. The relative movement is realized by the rotating turntable specimen and by keeping the MFL probe still to simulate the actual high speed MFL test. Two opposite yoke-electromagnets were employed to magnetize the specimen to avoid the repeated magnetization during multicycle rotation and increase the difficulty of magnetization before the test. The MFL inspection devices include six parts: magnetizer of yoke-1, reverse magnetizer of yoke-2, 3-D hall sensors, signal conditioning circuit, data acquisition card and the PC for data processing and display as shown in Figs. 6–8. Two yokeelectromagnets made of silicon steel sheet material cover the whole surface of specimen. Excitation of the copper coils around
Fig. 8. The simplified experimental model (top view) of high-speed MFL test in the platform.
11
the yoke-1 and yoke-2 arm is from a DC stabilized voltage supply and the generated magnetic field of yoke-2 is greater than that of yoke-1. The 3-D hall sensors are composed of three UGN3503 elements which were positioned closely and perpendicular to each other in order to measure three components of magnetic field. The Bz component is perpendicular to the surface of specimen, Bx and By components are both parallel to the material surface, and parallel and orthogonal to the applied magnetization field, respectively. The amplifier AD620 is employed to magnify the MFL signals by 100 times. A high-speed data acquisition card of DAQ2204 is used to obtain data, which is then processed in MATLAB, and the sampling frequency is set to 150 kHz according to the space and time according to Shannon’s sampling theorem. 4.2. Experimental results & analysis As is shown in Fig. 8, a surface-breaking artificial crack, with the depth 8 mm and width 0.4 mm, was selected to analyze. The MFL signals from 3-D hall sensors against inspection velocity were illustrated in Fig. 9. Inspection velocity is in the range from 5 m/s to 55 m/s, and the step is 5 m/s. As is shown in Fig. 9(a)–(d), the magnitudes of three components of magnetic signals where there is no defect increase with the increase in inspection velocity whilst the increasing gradient is monotone decreasing. The profiles of MFL signals are analogous with the variation of inspection velocity because of the tiny width of defect, but it can be seen that the concave–convex profile of Bz is not symmetrical. Due to the fact that the defect is approximately perpendicular to the applied magnetic field, the magnitude of By component is less than Bx and Bz. With the increase in inspection velocity, the magnitudes of the three components of MFL signals are nonlinear monotone increasing and the increasing gradient of
Fig. 9. Experimental results of the MFL test in different inspection velocities, (a) the Bx component of MFL signals, (b) the By component of MFL signals, (c) the Bz component of MFL signals, (d) the magnitudes of MFL signals against inspection velocity.
12
P. Wang et al. / NDT&E International 64 (2014) 7–12
Bx and Bz is greater than By. The magnitudes of MFL signals M MFL signals which contain Bx, By and Bz components are affected by many factors such as lift-off Llif t of f , excitation current I excitation of yoke-electromagnet, parameters of defect P def ect , thickness of specimen H specimen , inspection velocity V inspection , the sensitivity of magnetic sensors Ssensor , the magnification of amplifying circuit M amplif y , and so on. The relationship between these impact factors and magnitudes of MFL signals is described in Eq. (6) with a function f . The simple relationship between inspection velocity and magnitude of detection signals is difficult to specify due to the complex changes of other impact factors in different inspection conditions. M MFL signals ¼ f ðLlif t of f ; I excitation ; P def ect ; H specimen ; V inspection ; Ssensor ; Mamplif y ; :::Þ
ð6Þ
On the basis of the theoretical analysis on dynamic magnetization and its effect on the magnetic resistance and field distribution in high speed MFL inspection, the experimental results under the range of inspection velocity from 5 m/s to 55 m/s are in good agreement with simulation results. 5. Conclusion The relative movement between the applied magnetic field and measured specimen can shorten the duration of the magnetization process and induce eddy current in conductive ferromagnetic materials in high speed MFL inspection. It will alter the distribution and intensity of magnetic field and influence MFL signals. This paper has presented the theoretical analysis of dynamic magnetization and magnetic hysteresis and their effects on magnetic resistance and field distribution around the defect over the specimen. Then, an FEM simulation has been carried out to analyze the velocity effect of dynamic magnetization and permeability of specimen. Finally, a specially designed experiment has been implemented to verify the velocity effect and probability of the high-speed MFL inspection. Initial investigation results indicate the applicability of high speed MFL inspection and the maximum inspection velocity (approximately 200 km/h) has been verified under laboratory conditions. With the increase in velocity, the magnitudes of MFL signals are nonlinear monotone increasing and the sensitivity of signals is more significant. Under the experimental conditions in this paper, the magnetic hysteresis and dynamic magnetization effect are the main factors affecting leakage magnetic field relative to the blocking effect of velocityinduced eddy current in velocity effect. Further domain wall dynamics observation and modeling, dynamics of magnetization and quantitative velocity effects in microstructure and cracks will be studied [32,34]. Acknowledgment The work in this paper was supported by the FP7 “NDE and SHM for Health Monitoring of Offshore Wind Farm (HEMOW)” (FP7-PEOPLE-2010-IRSES) project, National Science Foundation of China (50907032/E070104 and 51377015), “Simulation and realization of integration of electromagnetic NDT methods for online high speed railway inspection”, and Shanghai railway bureau project funding “High-speed railway seamless rail temperature stress and the crack detection system”. References [1] Sophian A, Tian GY, Taylor D, Rudlin J. Electromagnetic and eddy current NDT: a review. Insight 2001;43(5):302–6.
[2] Katoh M, Masumoto N, Nishio K, Yamaguchi T. Modeling of the yokemagnetization in MFL-testing by finite elements. NDT E Int 2003;36 (7):479–86. [3] Mandayam S, Udpa L, Udpa SS, Lord W. Invariance transformations for magnetic flux leakage signals. IEEE Trans Magn 1996;32(3):1577–80. [4] Katoh M, Nishio K, Yamaguchi T. The influence of modeled B–H curve on the density of the magnetic leakage flux due to a flaw using yoke-magnetization. NDT E Int 2004;37(8):603–9. [5] Dutta SA, Ghorbel FH, Stanley RK. Simulation and analysis of 3-D magnetic flux leakage. IEEE Trans Magn 2009;45(4):1966–72. [6] Li Y, Wilson J, Tian GY. Experiment and simulation study of 3D magnetic field sensing for magnetic flux leakage defect characterisation. NDT E Int 2007;40 (2):179–84. [7] Wilson JW, Tian GY. 3D magnetic field sensing for magnetic flux leakage defect characterisation. Insight 2006;48(6):357–9. [8] Sophian A, Tian GY, Zairi S. Pulsed magnetic flux leakage techniques for crack detection and characterisation. Sens Actuators A-Phys 2006;125(2):186–91. [9] Wilson JW, Kaba M, Tian GY, Licciardi S. Feature extraction and integration for the quantification of PMFL data. Nondestruct Test Eval 2010;2 5(2):101–9. [10] Wilson JW, Tian GY. Pulsed electromagnetic methods for defect detection and characterisation. NDT E Intl 2007;40(4):275–83. [11] Sun YH, Kang YH. High-speed magnetic flux leakage technique and apparatus based on orthogonal magnetization for steel pipe. Mater Eval 2010;68 (4):452–8. [12] Niikura S, Kameari A. Analysis Of eddy-current and force in conductors with motion. IEEE Trans Magn 1992;28(2):1450–3. [13] Du ZY, Ruan JJ, Peng Y, Yu SF, Zhang Y, Gan Y, et al. 3-D FEM simulation of velocity effects on magnetic flux leakage testing signals. IEEE Trans Magn 2008;44(6):1642–5. [14] Gan Z, Chai X Numerical simulation on magnetic flux leakage testing of the steel cable at different speed title. In: Proceedings of the International Conference on Electronics and Optoelectronics, 2011. 3316-3319. [15] Li Y, Tian GY, Ward S. Numerical simulation on magnetic flux leakage evaluation at high speed. NDT E Int 2006;39(5):367–73. [16] Li Y, Tian GY, Ward S. Numerical simulations on electromagnetic NDT at high speed. Insight 2006;48(2):103–8. [17] Katragadda G, Sun YS, Lord W, Udpa SS, Uidpa L. Velocity Effects and Their Minimization in Mfl Inspection of Pipelines: A Numerical Study. New York: Plenum Press Div Plenum Publishing Corp; 1995. [18] Park GS, Park SH. Analysis of the velocity-induced eddy current in MFL type NDT. IEEE Trans Magn 2004;40(2):663–6. [19] Yang LJ, Ma FM, Wang DW, Gao SW. Velocity effect and signal process in highspeed magnetic flux leakage inspection. Hong Kong: International Academic Publishers Ltd; 2005. [20] Chen ZJ, Xuan JQ, Wang P, Wang HT, Tian GY Simulation on High Speed Rail Magnetic Flux Leakage Inspection.In: Proceedings of the IEEE International Instrumentation and Measurement Technology Conference.2011. 760-764. [21] Yang S, Sun Y, Udpa L, Udpa SS, Lord W. 3D simulation of velocity induced fields for nondestructive evaluation application. IEEE Trans Magn 1999;35 (3):1754–6. [22] Stefanita Garmen-Gabriela. Magnetism Basics and Applications. New York: Springer-Verlag Berlin Heidelberg; 2011. [23] Cullity BD, Graham CD. Introduction to Magnetic Materials. 2nd Edition. New Jersey: John Wiley & Sons, Inc; 2009. [24] Jiles DC, Atherton DL. Ferromagnetic hysteresis. IEEE Trans Magn 1983;19 (5):2183–5. [25] Laudani A, Fulginei FR, Salvini A. Comparative analysis of Bouc–Wen and Jiles– Atherton models under symmetric excitations. Physica B 2014;435:134–7. [26] Miltat J, Albuquerque G, Thiaville A. An introduction to micromagnetics in the dynamic regime. In: Hillebrands B, Ounadjela K, editors. Spin Dynamics in Confined Magnetic Structures I, Topics Applied Physics, 83; 2002. p. 1–34. [27] Gilbert TL. A phenomenological theory of damping in ferromagnetic materials. IEEE Trans Magn 2004;40(6):3443–9. [28] Cimrák I, Slodička M. An iterative approximation scheme for the Landau– Lifshitz–Gilbert equation. J Comput Appl Math 2004;169:17–32. [29] Jiles DC, Atherton DL. Theory of ferromagnetic hysteresis. J Appl Phys 1984;55:2115–20. [30] Jiles DC. Hysteresis models: non-linear magnetism on length scales from the atomistic to the macroscopic. J Magn Magn Mater 2002;242:116–24. [31] Hauser Hans. Energetic model of ferromagnetic hysteresis. J Appl Phys 1994;75:2584–97. [32] Batista L, Rabe U, Altpeter I, Hirsekorn S, Dobmann G. On the mechanism of nondestructive evaluation of cementite content in steels using a combination of magnetic Barkhausen noise and magnetic force microscopy techniques. J Magn Magn Mater 2014;354:248–56. [33] ANSOFT Corporation. Maxwell EM V12 Manual; 2008. 〈http://www.ansoft. com/〉. [34] Ferré Jacques, Metaxas Peter J, Mougin Alexandra, Jamet Jean-Pierre, Gorchon Jon, Jeudy Vincent. Universal magnetic domain wall dynamics in the presence of weak disorder. Comptes Rendus Phys 2013;14(8):651–66.
Contents lists available at ScienceDirect
NDT&E International journal homepage: www.elsevier.com/locate/ndteint
Velocity effect analysis of dynamic magnetization in high speed magnetic flux leakage inspection Ping Wang a, Yunlai Gao a,n, GuiYun Tian a,b, Haitao Wang a a b
College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu, 210016, China School of Automation Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan, 611731, China
art ic l e i nf o
a b s t r a c t
Article history: Received 22 September 2013 Received in revised form 29 January 2014 Accepted 13 February 2014 Available online 20 February 2014
The investigation described in this paper focuses on the velocity effect of dynamic magnetization and magnetic hysteresis due to rapid relative motion between magnetizer and measured specimens in highspeed magnetic flux leakage (MFL) inspection. Magnetization intensity and permeability of ferromagnetic materials along with the duration of dynamic magnetization process were analyzed. Alteration of the intensity and distribution of magnetic field leakage caused by permeability of specimen were investigated via theoretical analysis and finite-element method (FEM) combined with the actual highspeed MFL test. Following this, a specially designed experimental platform, in which motion velocity is within the range of 5 m/s–55 m/s, was employed to verify the velocity effect and probability of a highspeed MFL test. Preliminary results indicate that the MFL technique can achieve effective defect inspection at high speeds with the maximum inspection speed of about 200 km/h being verified under laboratory conditions. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Velocity effect Dynamic magnetization Magnetic hysteresis MFL High-speed inspection
1. Introduction Magnetic flux leakage testing is an efficient electromagnetic non-destructive testing (NDT) [1] technique that has been extensively used for defect inspection and characterization of ferromagnetic materials such as pipelines, pressure vessels, rail tracks and wheels, ropes or cables, etc. Magnetic flux lines generated by a magnetizer are coupled into test specimens with air coupling. Any geometrical discontinuity or local anomalies are manifested as an abrupt change of magnetic permeability [2–4] and force magnetic flux to leak out of the specimen in the poles of yoke in the air. Leakage magnetic field which contains information of defect is collected by magnetic field sensors and used to evaluate the defect dimensions and structural performance. For the advantages of its simplicity, low cost, air coupling and non-contact application, MFL testing is extremely suitable for the automated in-line and real time defect inspection. Some advancing works such as 3-D sensing of magnetic field [5–7], pulsed electromagnetic method (PMFL/ PMR) [8–10] and orthogonal magnetization [11] are currently under different stages of development and application for description of the shape and dimensions of defect. Although this method has a high probability of defect inspection, it still fraught with problems associated with the sensitivity n
Corresponding author. E-mail address: [email protected] (Y.L. Gao).
http://dx.doi.org/10.1016/j.ndteint.2014.02.001 0963-8695 & 2014 Elsevier Ltd. All rights reserved.
and interpretation of MFL signals to many factors, such as the condition of magnetization, inspection velocity, the B–H curves [4] of a specimen, lift-off, etc. Additionally, a conventional magnetostatic model is unsuitable for the high-speed MFL test. Eddy current distributed in conductors induced by relative movement between the MFL probe and a specimen will alter the profile and intensity of magnetic field leakage and distort the profile of MFL signals [12–17]. It also brings about difficulty in the signal interpretation and description of the defect. The target of improving the probability and accuracy of defect inspection has been attempted in previous work. Many numerical simulations based on 2-D or 3-D transient FEM models were carried out to simulate the distribution of motion-induced eddy currents and analyze their effect on MFL signals [13–20]. In addition, some methods on compensation and velocity invariance of MFL signal to minimize the velocity-induced eddy current effect have been introduced in previous papers [3,17–19]. Motion-induced eddy currents have also been utilized in the description of stress corrosion cracks in terms of measuring perturbation fields [21]. The magnetic flux lines flowing into a test specimen and the magnetization intensity are very important for the sensitivity of MFL signals and the defect inspection ability [2–4]. However, numerical simulation in previous work of the velocity effect only concentrated on the motion-induced eddy current and their influence on MFL signals; the factors of dynamic magnetization and the hysteresis effect during the high speed MFL test were
8
P. Wang et al. / NDT&E International 64 (2014) 7–12
neglected. The hysteresis effect exists in the dynamic magnetization process accompanying the magnetic domain rotation and domain wall movements which shows Barkhausen noise (BN) and magnetic hysteresis loop. The duration of the magnetization process decreases with the increase of inspection velocity because of the constant distance between sensor and the starting point of magnetization. If the higher inspection velocity and shorter time of the magnetization process bring about impact on the dynamic magnetization process, magnetization intensity and permeability of specimen as well as leakage magnetic field will be altered. This paper addressed the problem on the velocity effect of dynamic magnetization and magnetic hysteresis during high speed MFL inspection. In dynamic magnetization process, the distribution of magnetic resistance around defect and the velocity effect on specimen permeability and detection signals during the high speed MFL inspection have been analyzed, and verified in FEM simulation and actual high speed test. The rest of this paper is organized as follows: Section 2 describes the theoretical analysis of the dynamic magnetization and its effect on magnetic resistance and field distribution. Section 3 presents FEM simulation on magnetic field leakage along with the permeability of specimen on the basis of magnetization due to velocity effect. Section 4 elaborates on the experimental study with a specially designed high-speed MFL inspection platform. And then, derived conclusions will be given in Section 5.
2. Theoretical analysis of the dynamic magnetization effect 2.1. Ferromagnetic magnetization Ferromagnetic material is composed of many small spontaneous magnetization areas called magnetic domains, and the transition area between adjacent magnetic domains is called the magnetic domain wall. Magnetic domains distributed in the spontaneous magnetization direction and the ferromagnet do not show magnetism outward when the object is not magnetized. When an extra external magnetic field is applied to the specimen, magnetic domains flip and rotate in the direction of the applied magnetic field accompanied with the domain wall movements. Fig. 1 illustrates the initial magnetization curve [22] of the ferromagnetic material with four stages showing the dynamic change during magnetization and the variation of micro-level magnetic domain and domain wall structures [23]. When all the magnetic moments of domains tends to be consistent with applied magnetic field direction and the magnetization intensity no longer
Fig. 1. Initial magnetization curve of the ferromagnetic material, (I) reversible magnetic domain wall motion, (II) irreversible domain wall motion, (III) reversible domain wall motion and domain magnetization rotation, and (VI) only domain magnetization rotation [22].
increases, it means that the specimen is to be saturation magnetized. The magnetization process is governed by the following equations M¼
∑m0 ¼ χmH V
B ¼ μ0 H þ μ0 M ¼ μ0 ð1 þ χ m ÞH ¼ μ0 μr H ¼ μH
ð1Þ ð2Þ
where M, H, B and V, respectively, represent magnetization intensity, magnetic field intensity, magnetic flux density and volume of the magnetized specimen; ∑m0 and χ m , respectively, are the sum of magnetic moments in a certain volume and the magnetic susceptibility of specimen; μ0 and μr denote the permeability of air and relative magnetic permeability of materials with respect to air, and μ represents the absolute permeability of the medium. During the magnetization process, the variation of magnetic flux density B in a specimen lags behind the applied field H by a phase-shift because of the internal magnetic damping and some energy losses in specimen according to Jiles–Atherton model [24–26] and Landau–Lifshitz–Gilbert (LLG) equation [27,28] which is famous for the description of dynamic magnetization process. After the applied field is revoked, ferromagnetic specimen can still keep part of original magnetism, which is called the magnetic hysteresis phenomenon [29–31]. This phenomenon is the result of irreversible migration of the magnetic domain wall, which is affected by the internal friction of magnetic materials [23], the damping effect of micro eddy currents around a moving domain wall [23], the stress of specimen, material hardness and impurities, lattice defects, and so on. It can cause the resistance of the migration of magnetic domain wall and rotation of magnetic domains. According to the magnetic hysteresis phenomenon, the ∑m0 will be decreased due to the velocity effect of dynamic magnetization. It will lead to the decrease of χ m and the permeability of magnetized specimen [22–31]. 2.2. Dynamic magnetization effect of the high-speed MFL inspection A conventional MFL testing model [6] using a yoke-electromagnet is illustrated below in Fig. 2. The MFL probe incorporates a magnetizer and the sensor travels on the specimen along the scanning direction at a certain velocity (V). MFL signals are the magnitudes of leakage magnetic field measured by sensors positioned in the middle of the two magnet poles and at a constant distance over specimen (lift-off). The distance between two yokeelectromagnet poles is a constant, represented as L in Fig. 2, which is the valid magnetization distance in MFL inspection. The dynamic magnetization processes during the high speed MFL inspection is like a magnetization with a sinusoidal current of a certain angular frequency ð2π Þ=T due to the applied magnetic field induced by a moving yoke-electromagnet. The magnetic field acting on the magnetized specimen first is zero, then increased to the strength of the south-pole, it changes to the strength of the horizontal field between the two poles, and decreases further to the strength of the north-pole before reducing absolutely to zero
Fig. 2. A conventional magnetic flux leakage testing model [6].
P. Wang et al. / NDT&E International 64 (2014) 7–12
again. Based on the valid magnetization distance, the duration from the beginning to the end of dynamic magnetization process, represented as T, can be described by Eq. (3) below, and the T decreases with the increase of inspection velocity on the basis of Eq. (3). According to the magnetic hysteresis phenomenon, the χ m and permeability of a specimen will decrease due to the dynamic magnetization in the high speed MFL inspection. T ¼ L=V
ð3Þ
As is shown in Fig. 2, magnetic resistance around the defect is divided into two parts consisting of (a) and (b) in the boundaries of the specimen surface. According to Ohm’s law for magnetic circuit, magnetic resistance and flux can be described as shown in the following equations: NI ¼ ϕ U Rm ¼ ϕðaÞ U RðaÞ ¼ ϕðbÞ U RðbÞ ¼ ϕðbÞ U
lðbÞ
μðbÞ SðbÞ
ϕ ¼ ϕðaÞ þ ϕðbÞ
ð4Þ ð5Þ
where NI, ϕ and Rm , respectively represent magnetomotive force, magnetic flux and magnetic resistance in the whole magnetic
9
circuit, and ϕðaÞ , ϕðbÞ , RðaÞ and RðbÞ are the same concepts as described above in position (a) and (b); lðbÞ and SðbÞ , respectively, represent the length and cross-sectional area of magnetic circuit in position (b); μðbÞ denotes permeability of specimen in position (b). The permeability of a yoke-magnetizer and specimen is greater than air, so resistance of the whole magnetic circuit mainly depends upon the air gap. Magnetic flux of the whole circuit basically remains unchanged regardless of the constant air gap and velocity-induced eddy current effect in high-speed MFL inspection when the same defect is detected. With the decrease of μðbÞ and the increase of RðbÞ due to the raise of inspection velocity, ϕðbÞ is decreased and the opposite ϕðaÞ is increased according to Eqs. (4) and (5) and Fig. 2. This phenomenon is just like the situation of the difference between existence and absence of defect in MFL inspection. From the above, magnetic flux leakage increases as the increase in inspection velocity takes the dynamic magnetization and magnetic hysteresis effect into account. Additionally, the effect of motion-induced eddy current due to the relative movement between applied magnetic field and specimen should be taken into account in the high speed MFL inspection.
3. Simulation on the dynamic magnetization effect in high speed MFL inspection
Fig. 3. Simulation model for the analysis of velocity effect of dynamic magnetization and magnetic hysteresis in high speed MFL inspection.
The electromagnetic FEM simulation models have been widely employed in solving Maxwell equations and providing guidance to solve the engineering problems. The domain wall dynamics in conjunction with microstructure is ongoing research to bridge the gaps of micro and macro observation and modeling [32]. However, the simulation of a static and transient FEM model does not take the dynamic magnetization and magnetic hysteresis effect that is present with the magnetic domain rotation and domain wall movements into account. Hence, simulation on the velocity effect of dynamic magnetization in high speed MFL inspection should combine theoretical analysis and the actual test.
Table 1 The properties of simulation model. Components Excitation coil
Magnetizer
Specimen
Defect
Materials Properties
Ferrite-yoke Permeability μr ¼ 5000; Conductivity¼0.01 S/m
Steel_1008 Permeability μr : having BH curve, and variation with the scope of constant 1–5000; Conductivity ¼ 2e6 S/m
Rectangular air slot Permeability μr ¼1; Conductivity¼ 0 S/m
Copper Permeability μr ¼1; Conductivity ¼ 5.8e7 S/ m; Current source¼ DC 20000A
Fig. 4. Simulation results, (a) the leakage magnetic field of Bx component, (b) the leakage magnetic field of By component.
10
P. Wang et al. / NDT&E International 64 (2014) 7–12
3.1. Simulation set-up Ansoft Maxwell EM V12 [33], one of the leading finite-element software packages for numerical simulation, was employed to analyze velocity effects of dynamic magnetization and magnetic hysteresis in high speed MFL inspection. As illustrated in Fig. 3, the model is set up in 2-D with X–Y coordinates representing the crosssection of MFL probe and specimen. The simulation involves investigation of the characterization of MFL signals with variations of permeability in specimen due to the magnetic hysteresis phenomenon and inspection velocity. The dimensions and properties of probe, specimen and defect are illustrated in Fig. 3 and Table 1.
inspection velocity in high speed MFL inspection whilst the velocityinduced eddy current effect is not taken into account.
4. Experimental investigation of the high-speed MFL inspection For verifying experiments, a series of high-speed test on specimens using MFL were carried out with the intention of measuring 3-D components of leakage magnetic field in different inspection velocities with the scope of 5 m/s–55 m/s. A specially designed test platform and experimental models were employed to give us some initial results that illustrate the capabilities of the high speed MFL inspection.
3.2. Simulation results and analysis 4.1. Experimental set-up According to theoretical analysis of the dynamic magnetization and magnetic hysteresis effect, permeability of the specimen changes with the magnetizing duration, whilst leakage magnetic field alters with motional velocity due to the variation of magnetic resistance around the defect in high-speed MFL inspection. Magnetic intensity and permeability of the specimen is an approximate constant under the condition of a fixed inspection velocity or a constant state of magnetization. Hence, a constant equivalent permeability of magnetized material is proposed and permeability of the specimen changing with the scope of 1–5000 compared with the B–H curve of steel_1008 which is implemented in the simulation. The magnitudes of leakage magnetic field are represented as the components of Bx and By along the line over specimen and are observed and illustrated in Figs. 4 and 5. As is shown in Fig. 4(a), the magnitude of Bx where there is no defect increases with the decrease of equivalent permeability of specimen. From the profile of leakage magnetic flux intensity illustrated in Fig. 4(a) and (b) and Fig. 5, it is noticeable that magnitudes of Bx and By components increase first and then decrease with the decrease in permeability. The maximum magnitude shown in Fig. 5 is located at the turning point where the equivalent permeability is 50. Moreover, the magnitude of MFL signals with the specimen of steel_1008 having the B–H curve is close to the situation that equivalent permeabilities are 20 or 150, respectively. According to the magnetism and magnetization of ferromagnetic materials, the equivalent permeability of steel_1008 is approximate close to 150, certainly, the actual permeability of magnetized steel_1008 is very low due to the B–H curve. Hence, on the basis of simulation results that equivalent permeability of specimen is greater than 50, the magnitude of MFL signals increases with the decrease in permeability of specimen due to the increase in
Fig. 5. Simulation results: the magnitudes of Bx and By against equivalent permeability of specimen.
The sample is a circular turntable specimen that contains surface-breaking defects on the edge of circle. The test specimen was designed completely according to the shape and size of rail head and material of specimen is same as one of U71Mn rail. A series of different types of artificial cracks are distributed on the specimen surface. Rotation linear velocity of the edge of circle is within the range of 5 m/s–55 m/s under the drag of the motor in the high speed testing platform, with the block diagram and actual
Fig. 6. The block diagram of high speed MFL test system.
Fig. 7. The photograph of test platform.
P. Wang et al. / NDT&E International 64 (2014) 7–12
platform photograph as shown in Figs. 6–8. The MFL test probe consists of electromagnet of yoke-1 with DC excitation and 3-D hall sensors are positioned at a constant distance of 1 mm over specimen to detect leakage magnetic field and the defects indication is recorded. The surface of the magnet pole has a good coupling with sample based on the radian and shape of specimen. The relative movement is realized by the rotating turntable specimen and by keeping the MFL probe still to simulate the actual high speed MFL test. Two opposite yoke-electromagnets were employed to magnetize the specimen to avoid the repeated magnetization during multicycle rotation and increase the difficulty of magnetization before the test. The MFL inspection devices include six parts: magnetizer of yoke-1, reverse magnetizer of yoke-2, 3-D hall sensors, signal conditioning circuit, data acquisition card and the PC for data processing and display as shown in Figs. 6–8. Two yokeelectromagnets made of silicon steel sheet material cover the whole surface of specimen. Excitation of the copper coils around
Fig. 8. The simplified experimental model (top view) of high-speed MFL test in the platform.
11
the yoke-1 and yoke-2 arm is from a DC stabilized voltage supply and the generated magnetic field of yoke-2 is greater than that of yoke-1. The 3-D hall sensors are composed of three UGN3503 elements which were positioned closely and perpendicular to each other in order to measure three components of magnetic field. The Bz component is perpendicular to the surface of specimen, Bx and By components are both parallel to the material surface, and parallel and orthogonal to the applied magnetization field, respectively. The amplifier AD620 is employed to magnify the MFL signals by 100 times. A high-speed data acquisition card of DAQ2204 is used to obtain data, which is then processed in MATLAB, and the sampling frequency is set to 150 kHz according to the space and time according to Shannon’s sampling theorem. 4.2. Experimental results & analysis As is shown in Fig. 8, a surface-breaking artificial crack, with the depth 8 mm and width 0.4 mm, was selected to analyze. The MFL signals from 3-D hall sensors against inspection velocity were illustrated in Fig. 9. Inspection velocity is in the range from 5 m/s to 55 m/s, and the step is 5 m/s. As is shown in Fig. 9(a)–(d), the magnitudes of three components of magnetic signals where there is no defect increase with the increase in inspection velocity whilst the increasing gradient is monotone decreasing. The profiles of MFL signals are analogous with the variation of inspection velocity because of the tiny width of defect, but it can be seen that the concave–convex profile of Bz is not symmetrical. Due to the fact that the defect is approximately perpendicular to the applied magnetic field, the magnitude of By component is less than Bx and Bz. With the increase in inspection velocity, the magnitudes of the three components of MFL signals are nonlinear monotone increasing and the increasing gradient of
Fig. 9. Experimental results of the MFL test in different inspection velocities, (a) the Bx component of MFL signals, (b) the By component of MFL signals, (c) the Bz component of MFL signals, (d) the magnitudes of MFL signals against inspection velocity.
12
P. Wang et al. / NDT&E International 64 (2014) 7–12
Bx and Bz is greater than By. The magnitudes of MFL signals M MFL signals which contain Bx, By and Bz components are affected by many factors such as lift-off Llif t of f , excitation current I excitation of yoke-electromagnet, parameters of defect P def ect , thickness of specimen H specimen , inspection velocity V inspection , the sensitivity of magnetic sensors Ssensor , the magnification of amplifying circuit M amplif y , and so on. The relationship between these impact factors and magnitudes of MFL signals is described in Eq. (6) with a function f . The simple relationship between inspection velocity and magnitude of detection signals is difficult to specify due to the complex changes of other impact factors in different inspection conditions. M MFL signals ¼ f ðLlif t of f ; I excitation ; P def ect ; H specimen ; V inspection ; Ssensor ; Mamplif y ; :::Þ
ð6Þ
On the basis of the theoretical analysis on dynamic magnetization and its effect on the magnetic resistance and field distribution in high speed MFL inspection, the experimental results under the range of inspection velocity from 5 m/s to 55 m/s are in good agreement with simulation results. 5. Conclusion The relative movement between the applied magnetic field and measured specimen can shorten the duration of the magnetization process and induce eddy current in conductive ferromagnetic materials in high speed MFL inspection. It will alter the distribution and intensity of magnetic field and influence MFL signals. This paper has presented the theoretical analysis of dynamic magnetization and magnetic hysteresis and their effects on magnetic resistance and field distribution around the defect over the specimen. Then, an FEM simulation has been carried out to analyze the velocity effect of dynamic magnetization and permeability of specimen. Finally, a specially designed experiment has been implemented to verify the velocity effect and probability of the high-speed MFL inspection. Initial investigation results indicate the applicability of high speed MFL inspection and the maximum inspection velocity (approximately 200 km/h) has been verified under laboratory conditions. With the increase in velocity, the magnitudes of MFL signals are nonlinear monotone increasing and the sensitivity of signals is more significant. Under the experimental conditions in this paper, the magnetic hysteresis and dynamic magnetization effect are the main factors affecting leakage magnetic field relative to the blocking effect of velocityinduced eddy current in velocity effect. Further domain wall dynamics observation and modeling, dynamics of magnetization and quantitative velocity effects in microstructure and cracks will be studied [32,34]. Acknowledgment The work in this paper was supported by the FP7 “NDE and SHM for Health Monitoring of Offshore Wind Farm (HEMOW)” (FP7-PEOPLE-2010-IRSES) project, National Science Foundation of China (50907032/E070104 and 51377015), “Simulation and realization of integration of electromagnetic NDT methods for online high speed railway inspection”, and Shanghai railway bureau project funding “High-speed railway seamless rail temperature stress and the crack detection system”. References [1] Sophian A, Tian GY, Taylor D, Rudlin J. Electromagnetic and eddy current NDT: a review. Insight 2001;43(5):302–6.
[2] Katoh M, Masumoto N, Nishio K, Yamaguchi T. Modeling of the yokemagnetization in MFL-testing by finite elements. NDT E Int 2003;36 (7):479–86. [3] Mandayam S, Udpa L, Udpa SS, Lord W. Invariance transformations for magnetic flux leakage signals. IEEE Trans Magn 1996;32(3):1577–80. [4] Katoh M, Nishio K, Yamaguchi T. The influence of modeled B–H curve on the density of the magnetic leakage flux due to a flaw using yoke-magnetization. NDT E Int 2004;37(8):603–9. [5] Dutta SA, Ghorbel FH, Stanley RK. Simulation and analysis of 3-D magnetic flux leakage. IEEE Trans Magn 2009;45(4):1966–72. [6] Li Y, Wilson J, Tian GY. Experiment and simulation study of 3D magnetic field sensing for magnetic flux leakage defect characterisation. NDT E Int 2007;40 (2):179–84. [7] Wilson JW, Tian GY. 3D magnetic field sensing for magnetic flux leakage defect characterisation. Insight 2006;48(6):357–9. [8] Sophian A, Tian GY, Zairi S. Pulsed magnetic flux leakage techniques for crack detection and characterisation. Sens Actuators A-Phys 2006;125(2):186–91. [9] Wilson JW, Kaba M, Tian GY, Licciardi S. Feature extraction and integration for the quantification of PMFL data. Nondestruct Test Eval 2010;2 5(2):101–9. [10] Wilson JW, Tian GY. Pulsed electromagnetic methods for defect detection and characterisation. NDT E Intl 2007;40(4):275–83. [11] Sun YH, Kang YH. High-speed magnetic flux leakage technique and apparatus based on orthogonal magnetization for steel pipe. Mater Eval 2010;68 (4):452–8. [12] Niikura S, Kameari A. Analysis Of eddy-current and force in conductors with motion. IEEE Trans Magn 1992;28(2):1450–3. [13] Du ZY, Ruan JJ, Peng Y, Yu SF, Zhang Y, Gan Y, et al. 3-D FEM simulation of velocity effects on magnetic flux leakage testing signals. IEEE Trans Magn 2008;44(6):1642–5. [14] Gan Z, Chai X Numerical simulation on magnetic flux leakage testing of the steel cable at different speed title. In: Proceedings of the International Conference on Electronics and Optoelectronics, 2011. 3316-3319. [15] Li Y, Tian GY, Ward S. Numerical simulation on magnetic flux leakage evaluation at high speed. NDT E Int 2006;39(5):367–73. [16] Li Y, Tian GY, Ward S. Numerical simulations on electromagnetic NDT at high speed. Insight 2006;48(2):103–8. [17] Katragadda G, Sun YS, Lord W, Udpa SS, Uidpa L. Velocity Effects and Their Minimization in Mfl Inspection of Pipelines: A Numerical Study. New York: Plenum Press Div Plenum Publishing Corp; 1995. [18] Park GS, Park SH. Analysis of the velocity-induced eddy current in MFL type NDT. IEEE Trans Magn 2004;40(2):663–6. [19] Yang LJ, Ma FM, Wang DW, Gao SW. Velocity effect and signal process in highspeed magnetic flux leakage inspection. Hong Kong: International Academic Publishers Ltd; 2005. [20] Chen ZJ, Xuan JQ, Wang P, Wang HT, Tian GY Simulation on High Speed Rail Magnetic Flux Leakage Inspection.In: Proceedings of the IEEE International Instrumentation and Measurement Technology Conference.2011. 760-764. [21] Yang S, Sun Y, Udpa L, Udpa SS, Lord W. 3D simulation of velocity induced fields for nondestructive evaluation application. IEEE Trans Magn 1999;35 (3):1754–6. [22] Stefanita Garmen-Gabriela. Magnetism Basics and Applications. New York: Springer-Verlag Berlin Heidelberg; 2011. [23] Cullity BD, Graham CD. Introduction to Magnetic Materials. 2nd Edition. New Jersey: John Wiley & Sons, Inc; 2009. [24] Jiles DC, Atherton DL. Ferromagnetic hysteresis. IEEE Trans Magn 1983;19 (5):2183–5. [25] Laudani A, Fulginei FR, Salvini A. Comparative analysis of Bouc–Wen and Jiles– Atherton models under symmetric excitations. Physica B 2014;435:134–7. [26] Miltat J, Albuquerque G, Thiaville A. An introduction to micromagnetics in the dynamic regime. In: Hillebrands B, Ounadjela K, editors. Spin Dynamics in Confined Magnetic Structures I, Topics Applied Physics, 83; 2002. p. 1–34. [27] Gilbert TL. A phenomenological theory of damping in ferromagnetic materials. IEEE Trans Magn 2004;40(6):3443–9. [28] Cimrák I, Slodička M. An iterative approximation scheme for the Landau– Lifshitz–Gilbert equation. J Comput Appl Math 2004;169:17–32. [29] Jiles DC, Atherton DL. Theory of ferromagnetic hysteresis. J Appl Phys 1984;55:2115–20. [30] Jiles DC. Hysteresis models: non-linear magnetism on length scales from the atomistic to the macroscopic. J Magn Magn Mater 2002;242:116–24. [31] Hauser Hans. Energetic model of ferromagnetic hysteresis. J Appl Phys 1994;75:2584–97. [32] Batista L, Rabe U, Altpeter I, Hirsekorn S, Dobmann G. On the mechanism of nondestructive evaluation of cementite content in steels using a combination of magnetic Barkhausen noise and magnetic force microscopy techniques. J Magn Magn Mater 2014;354:248–56. [33] ANSOFT Corporation. Maxwell EM V12 Manual; 2008. 〈http://www.ansoft. com/〉. [34] Ferré Jacques, Metaxas Peter J, Mougin Alexandra, Jamet Jean-Pierre, Gorchon Jon, Jeudy Vincent. Universal magnetic domain wall dynamics in the presence of weak disorder. Comptes Rendus Phys 2013;14(8):651–66.