sensors Article

Eddy Current Pulsed Thermography with Different Excitation Configurations for Metallic Material and Defect Characterization Gui Yun Tian 1,2 , Yunlai Gao 2,3, *, Kongjing Li 2 , Yizhe Wang 1 , Bin Gao 1 and Yunze He 2,4 1

2 3 4

*

School of Automation Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China; [email protected] (G.Y.T.); [email protected] (Y.W.); [email protected] (B.G.) School of Electrical and Electronic Engineering, Newcastle University, Newcastle upon Tyne, NE1 7RU, UK; [email protected] (K.L.); [email protected] (Y.H.) College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China College of Electrical and Information Engineering, Hunan University, Changsha 410082, China Correspondence: [email protected] or [email protected]; Tel.: +44-191-208-5151

Academic Editor: Vittorio M. N. Passaro Received: 20 April 2016; Accepted: 31 May 2016; Published: 8 June 2016

Abstract: This paper reviews recent developments of eddy current pulsed thermography (ECPT) for material characterization and nondestructive evaluation (NDE). Due to the fact that line-coil-based ECPT, with the limitation of non-uniform heating and a restricted view, is not suitable for complex geometry structures evaluation, Helmholtz coils and ferrite-yoke-based excitation configurations of ECPT are proposed and compared. Simulations and experiments of new ECPT configurations considering the multi-physical-phenomenon of hysteresis losses, stray losses, and eddy current heating in conjunction with uniform induction magnetic field have been conducted and implemented for ferromagnetic and non-ferromagnetic materials. These configurations of ECPT for metallic material and defect characterization are discussed and compared with conventional line-coil configuration. The results indicate that the proposed ECPT excitation configurations can be applied for different shapes of samples such as turbine blade edges and rail tracks. Keywords: eddy current pulsed thermography (ECPT); material characterization; nondestructive evaluation (NDE); multi-physical phenomenon; excitation configuration

1. Introduction Eddy current pulsed thermography (ECPT) [1], as an emerging nondestructive evaluation (NDE) method, has been investigated for detection and characterization of structural degradation. Material degradations and failures such as defect, fatigue, corrosion, and residual stress have been detected and evaluated using ECPT [2–7]. The configuration of ECPT is one of the major issues for induction thermography, which greatly respond to the heating effect and evaluation of structural states from thermal image sequences. Wilson et al. [2] proposed a line-coil-based excitation to inductive heating on sample surfaces for rolling contact fatigue crack detection. Tian et al. [5] employed a rectangular single turn coil to evaluate the contact fatigue of gear. Cheng et al. [3] investigated rectangular multi-turns coil of ECPT to detect surface cracks of carbon-fiber-reinforced plastic (CFRP) materials. There are some limitations associated with these excitation configurations. The first limitation is the influence of non-uniform heating, which means the abnormal defect is covered by a severe temperature gradient. One solution is scanning thermography [8], which makes uniform heating by a constant moving of coil. However, the scanning configuration makes the system complicated. Another solution is Sensors 2016, 16, 843; doi:10.3390/s16060843

www.mdpi.com/journal/sensors

Sensors 2016, 16, 843

2 of 11

the application of phase information through eddy current pulsed phase thermography (ECPPT) based on an FFT algorithm [9,10] and thermal pattern separation and analysis [6,11]. However, these algorithm innovations cannot eliminate the non-uniform heating from the sources. The second limitation is the field of view, which is blocked by the coil. A possible solution is the configuration of the transmission mode where the infrared (IR) camera and coil are placed on opposite sides [12]. However, the transmission mode is not suitable for most metallic components due to the great thickness. An alternative solution is the use of lateral heat conduction excited by line coil [13]. This approach, however, has a directional sensitivity and is only sensitive to the cracks which are parallel to the coil. In order to solve these problems, Li et al. [4] proposed the Helmholtz coil excitation of ECPT for the state detection of bond wires in an insulated gate bipolar transistor (IGBT) sample. Peng et al. [14] investigated the Helmholtz-coil-type excitation of ECPT to produce approximate uniform heating on railhead surfaces and provide a larger detection area without the coil shielding problem. The Helmholtz coil excitation configuration does achieve the uniform heating and open-view imaging. After analyzing the principle of electromagnetic excitation, it is speculated that the ferrite-yoke-based excitation configuration is an effective means for uniform heating and open-view imaging. This idea derives from magnetic flux leakage (MFL) [15], where Gao et al. proposed a ferrite-yoke-based model to combine MFL and ECPT after a comparison of multiple crack evaluations. With the development of ECPT [15–17] for challenging nondestructive evaluation (NDE) applications such as free-form samples, multiple cracks such as Rolling Contact Fatigue Cracks (RCFs) and Stressed Corrosion Cracks (SCCs), gears and bearings, the limitations of non-uniform heating, and restricted views have been the some of the greatest obstacles in the application of ECPT. To overcome a non-uniform induction field due to sample geometry, e.g., defects at edges or corners and the block view of excitation coils, this paper introduces two recently developed ECPT configurations: Helmholtz-coil-based ECPT for the uniform heating of free-form samples and ferrite-yoke-based ECPT with magnetic flux and eddy current thermography for ferromagnetic material uniform heating and open-view infrared imaging. After a comparison of these configured ECPTs through simulation and experimental results, the multi-physics and characteristics of them are discussed for inverse problems and the quantitative NDE of metallic materials. The rest of this paper is organized as follows: Section 2 introduces theory background and methodology for two different excitation configurations. Section 3 describes the numerical simulations and experimental studies results of different ECPT configurations, the multi-physical effect of hysteresis losses, stray losses, and eddy current heating are conducted and implemented for ferromagnetic and non-ferromagnetic materials. After a comparison and discussion of different ECPT configurations, a conclusion and future works are derived in Section 4. 2. Theory and Methods 2.1. Theory Background of ECPT ECPT involves multi-physical interactions [17] with electromagnetic-thermal phenomena including induced eddy currents, Joule heating, and heat conduction. Figure 1 shows a schematic of the operation principle of ECPT for inspecting metallic components. After being triggered by a pulse signal from a pulse generator, the induction heater generates an excitation signal, which is a period of high-frequency alternating current with high amplitude. The current is then driven into an inductive coil positioned on a sample with a small distance. When the current passes through the coil, it induces eddy currents in the components. These eddy currents are governed by a subsurface penetration depth, δ, based on an exponentially damped skin effect. The latter can be calculated from δ “ pπµσ f q´1{2

(1)

where f is the frequency of the excitation signal, σ is the sample electrical conductivity (S/m, siemens per meter), and µ is the magnetic permeability (H/m, henries per meter) of the sample. The temperature

Sensors 2016, 16, 843 Sensors 2016, 16, 843

3 of 11 3 of 11

temperature of conductive material increases due to resistive heating from the induced eddy current, which is known as Jouleincreases heating. It cantoberesistive expressed by thefrom equation: of conductive material due heating the induced eddy current, which is known as Joule heating. It can be expressed by 1 the2 equation: 1 2

Q

 1

Js  2

1

E

(2)

2

Q “ |Js | “ |σE| (2) σ where the sum of generated heat Q is proportional toσ the square of the eddy current density, J s . It where the sum ofskin generated Q is proportional the square of the dozens eddy current density, Js . It is is concluded that depthsheat for metallic materials istovery small (about of μm under 100-kHz concluded that skin metallic materials is veryissmall (about dozens of µm under heat 100-kHz excitation). Thus, thedepths heatingfor style for metallic materials surface heating [8]. The resistive will excitation). the heating style for metallic materials surface heating heat will diffuse as a Thus, time transient until a heat balance is restoredisbetween the bulk[8]. andThe its resistive surface, at which diffuse a time transient a heatstate. balance restored between theisbulk and itsby surface, at which time theassignal will come tountil a steady The is heat diffusion process governed time the signal will come to a steady state. The heat diffusion process is governed by T    2T  2T  2T  1   q  x, y, z, t  (3) ˆ  2 2  2 2  22 ˙  t  C  C  x  y  z λ p B T B T B T BT p1 ` ` “ ` q px, y, z, tq (3) Bt ρC p Bx2 ρC p By2 Bz2 where T  T ( x, y, z, t ) is the temperature distribution,  is the thermal conductivity of the material, where T“ Tpx, y,and z, tq is temperature distribution, conductivity of the material, q( x,λyis , z,the t ) thermal C the density, is the internal heat generation functionρ  is the p is the specific heat, and is the density, and C is the specific heat, and qpx, y, z, tq is the internal heat generation function per per unit volume andp unit time. In this process, the temperature field on the surface of components unit be volume and by unit In this as process, the temperature on on theasurface oftypical components will be will captured antime. IR camera a sequence of an imagefield saved PC. One temperature captured by an IR camera as a sequence of an image saved on a PC. One typical temperature response response at a pixel is shown in Figure 1. It can be divided into two phases: the heating phase and the at a pixel is shown in Figure 1. It can be physical divided into two phases: heating phase and the cooling cooling phase. If there is a defect, these processes will bethe interrupted. The aforementioned phase. If there is a defect, thesewhich physical processes will be is the basic principle of ECPT, is also illustrated ininterrupted. Figure 1. The aforementioned is the basic principle of ECPT, which is also illustrated in Figure 1.

Figure 1. Schematic diagram of eddy current pulsed thermography (ECPT). Figure 1. Schematic diagram of eddy current pulsed thermography (ECPT).

eddy When an induced eddy current encounters geometric discontinuity or local anomalies, the eddy current will be forced to divert, resulting in distorted regions of eddy current densities. The The eddy eddy diversion leads to heat distribution with different thermal contrasts between defective and current diversion defect-free area, which makes the defect visible using an IR camera. The thermal transient responses during the the heating heating and andcooling coolingstages stagesinclude includerich richpattern patterninformation, information, which employed during which cancan be be employed to to evaluate crack detection andclassification, classification,and andstructural structuraldegradation degradation through through thermal thermal image evaluate crack detection and sequences. However, However, current current ECPT ECPT with with single single line-coil line-coil excitation excitation [1–3] [1–3] is is not not good good enough enough for sequences. complex geometry geometry structures, structures, e.g., gears, turbine blades, rail tracks, etc. complex etc. Different configurations of ECPT are introduced and compared under different different test test samples. samples. 2.2. The The Methodology Methodology of of the the Proposed Proposed Configurations Configurations 2.2. This paper paperproposes proposes different ECPT configurations with Helmholtz-coiland This two two different ECPT configurations with Helmholtz-coiland ferrite-yokeferrite-yoke-based structures to improve the capability of ECPT through uniform field excitation for based structures to improve the capability of ECPT through uniform field excitation for inductive inductive heating and open-view imaging. The schematic diagram of the Helmholtz-coil configuration heating and open-view imaging. The schematic diagram of the Helmholtz-coil configuration of ECPT of ECPT with magnetic flux distribution is illustrated inThe Figure 2. The Helmholtz coiltoisgenerate known to with magnetic flux distribution is illustrated in Figure 2. Helmholtz coil is known a generate a uniform magnetic field in a wide region around a center point of the coil pair axis, as uniform magnetic field in a wide region around a center point of the coil pair axis, as shown in shown 2b. in Figure 2b. Compared with line-coil-based ECPT, here, the Helmholtz-coil is employed to Figure Compared with line-coil-based ECPT, here, the Helmholtz-coil is employed to generate generate superpatterns heating patterns in the sample or sample edges between twofor coils for defect evaluation. super heating in the sample or sample edges between two coils defect evaluation. The

Sensors Sensors 2016, 2016, 16, 16, 843 843

of 11 11 44 of

Sensors 2016, 16, 843

4 of 11

defects in the sample or sample edges are detected through a thermal contrast from the original The defects in the sample or sample edges are detected through a thermal contrast from the original defects the or sample or sample edges detected through a thermal contrast from original uniforminfield heat distribution, whichare can be recorded by an IR camera towards thethe open-view uniform field or heat distribution, which can be recorded by an IR camera towards the open-view uniform field or heat distribution, which can be recorded by an IR camera towards the open-view space between two coils. space between two coils. space between two coils.

Figure Proposed Helmholtz-coil configuration of Schematic ECPT. (a)diagram Schematic diagram of Figure 2.2.Proposed Helmholtz-coil configuration of ECPT. (a) of Helmholtz-coilHelmholtz-coil-based system. (b) Magnetic flux distribution with uniform field between based ECPT system.Helmholtz-coil (b)ECPT Magnetic flux distribution uniform field between twoofcoils. Figure 2. Proposed configuration ofwith ECPT. (a) Schematic diagram Helmholtz-coiltwo coils. based ECPT system. (b) Magnetic flux distribution with uniform field between two coils.

The schematic diagram of the ferrite-yoke-based configuration of ECPT and its magnetic flux configuration of ECPT ECPT The schematic diagram the ferrite-yoke-based ferrite-yoke-based configuration of and its magnetic flux distribution are illustrated in of Figure 3. This configuration combines both the advantages of magnetic Figure 3.flux distribution Thisgenerated configuration combines both the coil advantages of magnetic flux leakage are andillustrated ECPT. Theinmagnetic by the circle excitation is transmitted to the leakage andand ECPT. The magnetic fluxgenerated generated the circle excitation is transmitted to flux leakage and ECPT. The magnetic byby the circle coilcoil isa transmitted to the sample surface near the surface flux between the two poles of excitation yoke through magnetic circuit, the sample surface near the surface between two poles yokeflux through magnetic circuit, sample surface andand near surface between thethe two poles ofofyoke through aa magnetic circuit, including ferrite-yoke, thetheair gap, and the sample. The magnetic is relatively uniformly ferrite-yoke, the air air gap,gap, and the The magnetic flux is relatively distributed including ferrite-yoke, the andsample. the sample. The flux is uniformly relatively uniformly distributed in the ferromagnetic material, which leads to a magnetic relatively uniform eddy current field in the ferromagnetic material,flux, which leads which to to a relatively current field to distributed in material, leads touniform a relatively eddyorthogonal current field orthogonal to the the ferromagnetic magnetic according the Faraday's laweddy for uniform electromagnetic induction. the magnetic flux, according to the Faraday's law for electromagnetic induction. Meanwhile, both orthogonal to the magnetic flux, according to the Faraday's law for electromagnetic induction. Meanwhile, both the magnetic flux and eddy current generate heating through hysteresis loss and the magnetic flux the and eddy current generate heating through loss andhysteresis eddy The current loss Meanwhile, magnetic fluxsamples and eddy current generatehysteresis heating through loss and eddy currentboth loss in ferromagnetic due to alternating dynamic magnetization. leakage in ferromagnetic samples dueinteractions to alternating dynamic magnetization. magnetic field eddy current in ferromagnetic samplesofdue to alternating dynamic magnetization. Thelosses leakage magnetic fieldloss induced by the the applied field and cracksThe alsoleakage lead to stray for induced the induced interactions of the applied field andapplied cracks alsobetween lead stray losses fortodefect-free different heating magnetic field by the interactions of the field and to cracks also lead stray losses for differentby heating responses around defects. Thermal contrast defective and areas responses around defects. Thermal contrast between defective and defect-free areas can beECPT. imaged different heating responses aroundindefects. Thermal contrast defective areas can be imaged with an IR camera open-view, which avoidsbetween line-coil blocks inand the defect-free current with animaged IR camera open-view, avoids line-coil blocks line-coil in the current can be withinan IR camerawhich in open-view, which avoids blocksECPT. in the current ECPT.

Figure 3. Proposed ferrite-yoke-based configuration of ECPT. (a) Schematic diagram of ferrite-yokebased ECPT system. (b) Magnetic flux configuration distribution inofferromagnetic material betweenof yoke poles. of Figure 3.3.Proposed ferrite-yoke-based ECPT.of(a)ECPT. Schematic ferrite-yokeFigure Proposed ferrite-yoke-based configuration (a) diagram Schematic diagram based ECPT system. (b) Magnetic flux distribution in ferromagnetic material between yoke poles. ferrite-yoke-based ECPT system. (b) Magnetic flux distribution in ferromagnetic material between

3. Simulation yoke poles.and Experimental Studies 3. Simulation and Experimental Studies As described in Section 2, three ECPT excitation configurations are illustrated. To understand 3. Simulation and Studies As described inExperimental Section 2,behind three ECPT excitation configurations areand illustrated. To understand their operational mechanism, the multi-physical coupling effect characteristics for NDE, their operational mechanism, behind the multi-physical coupling effect characteristics forwith NDE,a simulations and experiments were conducted with different samples. Aand typical ECPTTo system As described in Section 2, three ECPT excitation configurations are illustrated. understand simulations and experiments were conducted with different samples. A typical ECPT system with a line-coiland ferrite-yoke-based ECPT were applied for ferromagnetic material. Helmholtz-coiltheir operational mechanism, behind the multi-physical coupling effect and characteristics for NDE, line-coiland ferrite-yoke-based ECPT were applied for ferromagnetic material. Helmholtz-coilbased ECPT was applied to a non-ferromagnetic sample. The different ECPT structures and their simulations and experiments were conducted with different samples. A typical ECPT system with a based ECPT was applied to a non-ferromagnetic sample. The different ECPTHelmholtz-coil-based structures and their advantages disadvantages were and line-coilandand ferrite-yoke-based ECPTinvestigated were applied foridentified. ferromagnetic material. advantages and disadvantages were investigated and identified.

Sensors 2016, 16, 843

5 of 11

ECPT was applied to a non-ferromagnetic sample. The different ECPT structures and their advantages Sensors 2016, 16, 843 5 of 11 and disadvantages were investigated and identified.

3.1. Simulation and Results 3.1. Simulation and Results Numerical modeling for ECPT simulations with different configurations was established in Numerical modeling for ECPT simulations with different configurations was established in COMSOL Multiphysics 4.3b. Three different configurations of ECPT are illustrated in Figure 4. The COMSOL Multiphysics 4.3b. Three different configurations of ECPT are illustrated in Figure 4. geometrical and material physical parameters, e.g., permeability, electrical and thermal conductivity The geometrical and material physical parameters, e.g., permeability, electrical and thermal of the sample, and excitation are referenced in [4,14,15]. For comparison, three types of coil conductivity of the sample, and excitation are referenced in [4,14,15]. For comparison, three types of configurations and defects in different samples as illustrated in Figure 4 were simulated, e.g., a coil configurations and defects in different samples as illustrated in Figure 4 were simulated, e.g., a surface crack in steel_1008 under a line-coil with 0.5 mm of lift-off, as shown in Figure 4a, a crack surface crack in steel_1008 under a line-coil with 0.5 mm of lift-off, as shown in Figure 4a, a crack edge in Ti-alloy between two coils, as shown in Figure 4b, and an angular crack at 45° in steel_1008 edge in Ti-alloy between two coils, as shown in Figure 4b, and an angular crack at 45˝ in steel_1008 between two poles of ferrite-yoke, as shown in Figure 4c. The size of the sample is 150 mm in length, between two poles of ferrite-yoke, as shown in Figure 4c. The size of the sample is 150 mm in length, 90 mm in width, and 10 mm in depth. The size of the crack is 10 mm in length, 5 mm in depth, and 90 mm in width, and 10 mm in depth. The size of the crack is 10 mm in length, 5 mm in depth, and 0.5 mm in width. The inner diameter of the Helmholtz-coil and circular excitation coil wound on the 0.5 mm in width. The inner diameter of the Helmholtz-coil and circular excitation coil wound on the yoke is 50 mm with a wire outer diameter of 6.35 mm. The refined mesh is setup for the calculation yoke is 50 mm with a wire outer diameter of 6.35 mm. The refined mesh is setup for the calculation of of a 3D model. The current 380 A/256k Hz flows into the excitation coil to generate magnetic flux for a 3D model. The current 380 A/256k Hz flows into the excitation coil to generate magnetic flux for induction heating. The duration of the heating is 200 ms, which is enough to generate an effective induction heating. The duration of the heating is 200 ms, which is enough to generate an effective thermal response for defect evaluation. thermal response for defect evaluation.

Figure for ECPT ECPT simulation simulation with with different different configurations. configurations. (a) (a) Line-coil Line-coil ECPT. ECPT. Figure 4. 4. Numerical Numerical modeling modeling for (b) Helmholtz-coil ECPT. (c) Ferrite-yoke-based ECPT. (b) Helmholtz-coil ECPT. (c) Ferrite-yoke-based ECPT.

After a short short period period of of 100 100 ms ms for for induction induction heating, heating, simulation simulation results, results, which can typically demonstrate the theoperational operational characteristics of three the three proposed ECPT configurations, were demonstrate characteristics of the proposed ECPT configurations, were obtained obtained and are illustrated in The Figure 5. The multi-physical interactions and effect coupling were and are illustrated in Figure 5. multi-physical interactions and coupling wereeffect analyzed analyzed and arewith discussed with the flux, thecurrent, induced current, and the and are discussed the magnetic flux,magnetic the induced eddy andeddy the heat distribution forheat the distribution for theIndefect evaluation. In a typical systemofinFigure the first column of Figure 5, linedefect evaluation. a typical ECPT system in the ECPT first column 5, line-coil ECPT generates ECPT generates magnetic flux (red vectors) around the coilthe and interacts with the sample with acoil magnetic flux (red avectors) around the coil and interacts with sample with a limited area close a limited area close to the coil, as shown in the side view of the top-left image. The induced eddy to the coil, as shown in the side view of the top-left image. The induced eddy current is parallel to current paralleland to the coil direction and distributed inthe theline-coil; sample close to it the line-coil;ahence, it the coil is direction distributed in the sample close to hence, generates limited generates a limited heating area along with the line-coil. In the middle column of Figure 5, the heating area along with the line-coil. In the middle column of Figure 5, the Helmholtz-coil generates a Helmholtz-coil generates a uniform magnetictwo field (redas vectors) two coils,image. as shown the uniform magnetic field (red vectors) between coils, shown between in the top-middle Thein eddy top-middle image. The current flux is orthogonal to the magnetic parallel to theuniform sample current is orthogonal to eddy the magnetic and parallel to the sample flux edgeand with a relatively edge with a relatively uniform distribution, which is positioned in the center the coil pair. This distribution, which is positioned in the center of the coil pair. This leads to aofrelatively uniform leads toarea a relatively heating around edge, asimage. shownIn inthe theright bottom-middle heating around uniform the sample edge,area as shown in the the sample bottom-middle column of image.5,Ina ferrite-yoke-based the right column configuration of Figure 5, agenerates ferrite-yoke-based generates a relatively Figure a relativelyconfiguration uniform magnetic field between two uniform magnetic field between two poles of yoke with the orthogonally distributed induction eddy poles of yoke with the orthogonally distributed induction eddy current. The heating response of this current. The heating response of this configuration distributed on a large scale between yoke poles configuration distributed on a large scale between yoke poles accompany four heater areas around the accompany four heater areas around the yoke corner, as shown in the bottom-right image in yoke corner, as shown in the bottom-right image in Figure 5. Figure 5.

Sensors 2016, 16, 843 Sensors 2016, 16, 843

6 of 11 6 of 11

with different configurations such as line-coil-, Helmholtz-coil-, Figure 5. 5. Simulation Simulationresults resultsofofECPT ECPT with different configurations such as line-coil-, Helmholtzand Ferrite-yoke-based excitation on a multi-physical-phenomenon and coupling effect.effect. coil-, and Ferrite-yoke-based excitation on a multi-physical-phenomenon and coupling

To verify the the capabilities capabilities of To verify of the the three three different different ECPT ECPT configurations configurations for for the the defect defect evaluation, evaluation, different defects were employed, and the simulation results are illustrated in Figure 6. As As shown shown in in different defects were employed, and the simulation results are illustrated in Figure 6. Figure is demonstrated demonstrated through Figure 6a, 6a, the the crack crack profile profile is through thermal thermal contrast, contrast, and and heat heat density density around around the the crack is decreasing away from the excitation coil. The strong thermal responses under the line-coils crack is decreasing away from the excitation coil. The strong thermal responses under the line-coils are are partially partially blocked blocked in in practical practical applications, applications, so so it it is is difficult difficult to to have have aafull fullpicture pictureof ofcrack crackresponses. responses. As shown in Figure 6b, the crack edge is clearly identified by the heat concentration in the crack As shown in Figure 6b, the crack edge is clearly identified by the heat concentration in the crack open open position and and the the heat heat distribution distribution around around the the crack crack edge. edge. Even away from position Even heat heat density density is is decreasing decreasing away from the sample edge. As shown in Figure 6c, the angular crack is presented through the heat distribution the sample edge. As shown in Figure 6c, the angular crack is presented through the heat distribution around with the the heater heater area area and and the the thermal thermal contrast contrast around around the the crack crack edges. edges. around the the crack crack tips tips with

Figure 6. 6. Simulation Simulation results of ECPT ECPT with with different different configurations. configurations. (a) Figure results of (a) Line-coil Line-coil ECPT ECPT for for the the crack, crack, (b) Helmholtz-coil-based ECPT for the crack edge. (c) Ferrite-yoke-based ECPT for the angular crack.

3.2. Experiments Experiments and Results To verify the the simulation simulation above To verify above and and characteristics characteristics of of the the three three configurations configurations of of ECPT ECPT for for defect defect identification, experiments were conducted with the ECPT system including the Easyheat224 identification, experiments were conducted with the ECPT system including the Easyheat224 induction induction the high hollow conductive hollow copper tube coil, a device, water-cooling device, Optris heater, the heater, high conductive copper tube coil, a water-cooling the Optris PI400the IR camera PI400 IR camera (Optris GmbH, Berlin, Germany), and a PC, the detailed parameters of which are (Optris GmbH, Berlin, Germany), and a PC, the detailed parameters of which are referenced in [15]. referenced in [15]. Three types of configurations with the line-coil-, Helmholtz-coiland ferrite-yokeThree types of configurations with the line-coil-, Helmholtz-coil- and ferrite-yoke-based structures based structuresthe were coil parameters of which same as Two thosesamples of the were employed, coil employed, parameters the of which were the same as thosewere of thethe simulation. simulation. Two samples with rail specimens of ferromagnetic material, including multiple cracks with rail specimens of ferromagnetic material, including multiple cracks and blade specimens of and blade specimens of Ti-alloy including defects, were in used for 7, thethe test. As Ti-alloy material including naturalmaterial edge defects, were natural used foredge the test. As shown Figure ECPT shown Figure 7, parameters, the ECPT system structures, parameters, testasprocedures were the same as system in structures, and test procedures were theand same [15]. The excitation current [15]. The excitation current 380 A/256k Hz was used for the induction heating. The temperature 380 A/256k Hz was used for the induction heating. The temperature resolution of the Optris PI400 resolution of the Optris PI400 IR camera was 0.08 K with a sensing spectral range of 7.5–13 μm. The size for the focus spot about 150 mm × 120 mm area was used for the thermal imaging of the defect.

Sensors 2016, 16, 843

7 of 11

Sensors 2016, 16, 843 Sensors 2016, 16, 843

7 of 11 7 of 11

IR camera was 0.08 K with a sensing spectral range of 7.5–13 µm. The size for the focus spot about A frame rate ofmm 80 Hz forwas the used IR camera with a 382imaging × 288 array usedAtoframe recordrate thermal 150 mm ˆ 120 area for the thermal of thewas defect. of 80images Hz for A frame rate of 80 Hz for the IR camera with a 382 × 288 array was used to record thermal images within a 200-ms duration of heating. In Figure 8 three different ECPT excitation configurations the IR camera with a 382 ˆ 288 array was used to record thermal images within a 200-ms durationare of within a 200-ms durationmetallic of heating. In Figure three different ECPT excitation configurations are illustrated different samples and 8defects, e.g., oneare edge defect in one blade from the heating. In with Figure 8 three different ECPT excitation configurations illustrated with different metallic illustrated with different metallic samples and defects, e.g., one edge defect in one from the Nanjing ofe.g., Aeronautics Astronautics (NUAA) and the RCF cracks in blade theAeronautics rail, which samples University and defects, one edgeand defect in one blade from the Nanjing University of Nanjing University of Aeronautics and Astronautics (NUAA) and the RCF cracks in the rail, which is theAstronautics same sample(NUAA) used in [2,14,15]. Thecracks RCF multiple cracks areisdistributed in a gauge corner of the and and the RCF in the rail, which the same sample used in [2,14,15]. is the same sample usedsurface in [2,14,15]. Thewhere RCF multiple cracks are are distributed inmm, a gauge corner of are the rail head with complex shapes, the crack lengths up to 20 the distances The RCF multiple cracks are distributed in a gauge corner of the rail head with complex surface shapes, rail head with surface shapes, where thethe crack lengthsangles are upare to about 20 mm, the distances are about mmcomplex among are multiple RCF cracks, and orientation 45°–60° against the where0.5–2 the crack lengths up to 20 mm, the distances are about 0.5–2 mm among multiple RCF cracks, about 0.5–2 mm among multiple RCF cracks, and the orientation angles are about 45°–60° against the rail direction.angles The edge the˝blade include two distinct natural notches with lengths andtravel the orientation are defects about 45in˝ –60 against the rail travel direction. The edge defects in the railabout travel5–8 direction. Thethe edge defects in thedepths blade of include two distinct natural notches with lengths of mm along blade edge and about 1–5 mm inside the blade. blade include two distinct natural notches with lengths of about 5–8 mm along the blade edge and of about 5–8 mm along the blade edge and depths of about 1–5 mm inside the blade. depths of about 1–5 mm inside the blade.

Figure 7. Experimental system of ECPT. Figure 7. 7. Experimental Experimental system system of of ECPT. ECPT. Figure

(a) (a)

(b) (b)

(c) (c)

Figure Figure 8. 8. ECPT ECPT with with different different configurations. configurations. (a) (a) Line-coil Line-coil ECPT. ECPT. (b) (b) Helmholtz-coil Helmholtz-coil ECPT ECPT and and Figure 8. ECPT with different configurations. (a) Line-coil ECPT. (b) Helmholtz-coil ECPT and (c) Ferrite-yoke-based structure for ECPT. (c) Ferrite-yoke-based structure for ECPT. (c) Ferrite-yoke-based structure for ECPT.

During experiments, thermal images were obtained by using the IR camera for crack During experiments, thermal images werewere obtained by using IR camera for crack identification. During experiments, thermal images obtained bythe using theanalysis IR camera for crack identification. The results after a heating duration of 100 ms were selected for and discussion, The results after a heating duration of 100 ms were selected for analysis and discussion, as shown in identification. The results after a heating duration of 100 ms were selected for analysis and discussion, as shown in Figure 9, which is identical to all simulation results. As shown in Figure 9a, rail multiple Figure 9, which is identical to all simulation results. As shown in Figure 9a, rail multiple cracks are as shown Figuredemonstrated 9, which is identical to thermal all simulation results. As shown in Figure 9a, rail multiple cracks areinclearly through contrast between defective and defect-free areas, clearly are demonstrated through thermal contrast between defective and defect-free areas, but defects cracks demonstrated thermal contrast defective areas, but defectsclearly under line-coils couldthrough not be imaged due to thebetween coil blocking effect.and Thedefect-free heating density under line-coils could not be imaged due to the coil blocking effect. The heating density was also but defects under line-coils could not be imaged due to the coil blocking effect. The heating density was also decreasing away from the line-coil. As shown in Figure 9b, the natural edge defect profile decreasing away from the line-coil. As shown in Figure 9b, the natural edge defect profile in the blade was decreasing from the line-coil. As shown in Figure 9b, thethrough naturalopen-view edge defectimaging profile in thealso blade sample isaway well presented by IR imaging towards the sample sample is well presented bypresented IR imaging towards the towards sample through open-view imaging between the in the blade sample is well by IR imaging the sample through open-view imaging between the two coils. The defect-free heating area is relatively uniformly distributed. As shown in two coils. The defect-free heating area is relatively uniformly distributed. As shown in Figure 9c, rail between9c,therail two coils. The defect-free heating area is relatively uniformly distributed. shown in Figure multiple cracks are fully demonstrated in larger areas between yokeAspoles with multiple cracks are fully demonstrated in larger areas between yoke poles with open-view imaging Figure 9c, imaging rail multiple are fully demonstrated in largerwith areas yoke in poles with open-view undercracks relatively uniform heating. Compared thebetween three images Figure 9, under relatively uniform heating. Compared with the three images with in Figure 9, theimages heatinginuniformity open-view imaging under relatively uniform heating. Compared the three Figure 9, the heating uniformity and open-view imaging area of Figure 9c are better than Figure 9a,b. Two new the heating uniformity and open-view imaging area of Figure 9c are better than Figure 9a,b. Two new configurations of ECPT with Helmholtz-coil- and ferrite-yoke-based excitation demonstrate better configurations for of ECPT with Helmholtz-coiland ferrite-yoke-based demonstrate better characteristics complex structural edge defect detection and RCFexcitation multiple crack visualization characteristics for complex structural edge defect detection and RCF multiple crack visualization

Sensors 2016, 16, 843

8 of 11

and open-view Sensors 2016, 16, 16, 843 843 imaging Sensors 2016,

area of Figure 9c are better than Figure 9a,b. Two new configurations of of 11 88 of 11 ECPT with Helmholtz-coil- and ferrite-yoke-based excitation demonstrate better characteristics for without aastructural blocking effect effect than conventional conventional line-coil ECPT. Comparing Comparing Figure 9c 9cwithout with the the previous complex edge defect detection and RCF multiple crack visualization a previous blocking without blocking than line-coil ECPT. Figure with resultsthan withconventional ECPT based based on on lateralECPT. heat conduction conduction [13], the advantages advantages ofprevious the open openresults view are are apparent. effect line-coil Comparing Figure 9c with theof with ECPT results with ECPT lateral heat [13], the the view apparent. Three tests can view multiple cracks and defects at edges. As illustrated in Figure 9c, the heating based on lateral heat conduction [13], the advantages of the open view are apparent. Three tests can Three tests can view multiple cracks and defects at edges. As illustrated in Figure 9c, the heating distribution pattern near the ferrite-yoke poles has stronger heating than areas, e.g., in the central view multiple cracks near and defects at edges. poles As illustrated in Figure 9c, the heating distribution pattern the ferrite-yoke has stronger heating than areas,distribution e.g., in the pattern central area, as simulated in the bottom of the right column of Figure 5, which will be further evaluated for near the ferrite-yoke poles has stronger heating than areas, e.g., in the central area, as simulated in the area, as simulated in the bottom of the right column of Figure 5, which will be further evaluated for quantitative NDE. bottom of theNDE. right column of Figure 5, which will be further evaluated for quantitative NDE. quantitative

Experimental results results of of ECPT ECPT with with different configurations. (a) Line-coil for multiple cracks. Figure 9. 9. Experimental with different different configurations. configurations. (a) (a) Line-coil Line-coil for for multiple multiplecracks. cracks. Figure edges and and (c) (c) Ferrite-yoke Ferrite-yoke excitation excitation for for multiple multiple cracks. cracks. (b) (b) Helmholtz-coil Helmholtz-coil for for crack crack edges edges and (c) Ferrite-yoke excitation for multiple cracks.

For quantitative quantitative comparison comparison of of three three different different ECPT ECPT excitation excitation configurations, configurations, experimental experimental For ECPT excitation configurations, results on the same sample with RCF multiple cracks are demonstrated in Figure 10. The signal-toresults on onthe thesame samesample samplewith with RCF multiple cracks demonstrated in Figure 10. signal-to-noise The signal-toRCF multiple cracks areare demonstrated in Figure 10. The noise(SNR) ratio (SNR) (SNR) [18,19] or thermal thermal pattern contrast are calculated calculated in different different defective regions to noise ratio or are in regions to ratio [18,19][18,19] or thermal patternpattern contrastcontrast are calculated in different defectivedefective regions to evaluate evaluate performance of defect detection and characterization of three ECPT configurations with each evaluate performance defect detection and characterization threeconfigurations ECPT configurations withother. each performance of defectof detection and characterization of threeofECPT with each other. SNR describes the thermal contrast between defective region considered as “signal” and other. SNR describes the thermal contrast between defective region considered as “signal” and SNR describes the thermal contrast between defective region considered as “signal” and non-defective non-defective region defined as “noise”. The calculation of SNR in dB is defined below: non-defective defined “noise”. The calculation SNR inbelow: dB is defined below: region definedregion as “noise”. Theascalculation of SNR in dB isofdefined 𝑆 𝑇𝑚𝐷 −𝑇 𝑇𝑚𝑁 𝑚𝐷 − 𝑚𝑁 ˙ ˆ(𝑇 𝑆𝑁𝑅 = =S𝑆 = = 20𝑙𝑜𝑔 20𝑙𝑜𝑔10 [𝑑𝐵] (4) 10 ( T ´ T)mN)) [𝑑𝐵] 𝑆𝑁𝑅 (4) mD 𝑁 “ 20log10 𝜎(𝑇𝑚𝑁 𝑚𝑁) SNR “ 𝑁 rdBs (4) 𝜎(𝑇 N σ pTmN q where 𝑇 𝑇𝑚𝐷 and 𝑇 𝑇𝑚𝑁 are average average temperature temperature in in defective defective and and non-defective non-defective regions, regions, respectively. respectively. 𝑚𝐷 and 𝑚𝑁 are where where TmD T are average temperature in defective andrelevant non-defective regions,regions. respectively. The σ(𝑇 σ(𝑇 )and is the the temperature standard deviation deviation in the the relevant non-defective regions. The mN temperature 𝑚𝑁) The is standard in non-defective The 𝑚𝑁 The σ pT q is the temperature standard deviation in the relevant non-defective regions. The defective defective regions are marked in Figure 10 including the defective region with dotted line enclosed defectivemN regions are marked in Figure 10 including the defective region with dotted line enclosed regions marked in Figure 10 including the defective region withregions dotted selected line enclosed and two and two twoare local defective regions A and and B. B. Relevant Relevant non-defective regions selected are same same areaslocal for and local defective regions A non-defective are areas for defective regions A and B. Relevant non-defective regions selected are same areas for the three the three configurations. The SNR values of these selected regions are illustrated in Table 1 for the three configurations. The SNR values of these selected regions are illustrated in Table 1 for configurations. The SNRof of these selected regions are illustrated in Table 1 for quantitative quantitative comparison comparison ofvalues three ECPT ECPT configurations. quantitative three configurations. comparison of three ECPT configurations.

Figure 10. Experimental Experimental results of ECPT for same the same sample withmultiple RCF multiple cracks(a)using Figure 10. 10. results of ECPT ECPT for the the sample with RCF RCF cracks using using lineFigure Experimental results of for same sample with multiple cracks (a) line(a) line-coil. (b) Helmholtz-coil and (c) ferrite-yoke excitation configurations. coil. (b) Helmholtz-coil and (c) ferrite-yoke excitation configurations. coil. (b) Helmholtz-coil and (c) ferrite-yoke excitation configurations.

Table 1. 1. Defect Defect detection detection performance performance comparison comparison of of different different ECPT ECPT excitations excitations with with SNR SNR in in dB. dB. Table Region Region Excitation Excitation Line-coil Line-coil Helmholtz-coil Helmholtz-coil Ferrite-yoke Ferrite-yoke

SNR(dB) of of Whole Whole SNR(dB) Defective Area Defective Area

SNR(dB) of of SNR(dB) Region A Region A

SNR(dB) of of SNR(dB) Region B Region B

4.88 4.88 3.86 3.86

8.42 8.42 8.33 8.33

2.42 2.42 4.03 4.03

10.53 10.53

7.45 7.45

13.31 13.31

Sensors 2016, 16, 843

9 of 11

Table 1. Defect detection performance comparison of different ECPT excitations with SNR in dB. Region Excitation Line-coil Helmholtz-coil Ferrite-yoke

SNR (dB) of Whole Defective Area 4.88 3.86 10.53

SNR (dB) of Region A 8.42 8.33 7.45

SNR (dB) of Region B 2.42 4.03 13.31

As shown in Figure 10a, rail multiple cracks can be identified by thermal contrast in the regions close to line-coil. However, thermal patterns are influenced by the line-coil non-uniform heating, which are shown that higher thermal intensity close to coil and lower intensity with the distance away from coil. It also can be proved by the results that SNR values of the region A are better than the region B, which is away from coil, as illustrated in Table 1. As shown in Figure 10b, a relatively uniform heating area is produced between Helmholtz-coil pairs for defect detection. It proves better performance for blade edge defects inspection as shown in Figures 6b and 9b. However, the defective region close to coil is also influenced by the non-uniform heating that higher thermal intensity close to coil as shown in Figure 10b. The similar SNR values to line-coil also prove its non-uniform influence on defect detection and characterization. Because of the uniform heating and open-view imaging of ferrite-yoke excitation of ECPT, the whole defective regions are clearly demonstrated with better thermal contrast as shown in Figure 10c. The higher SNR values of ferrite-yoke excitation also prove the better defect characterization than other two because of its superior characteristics. Overall, it is identified that different ECPT configurations own individual strengths for different shape samples and defects characterization. Recent improvement of ECPT with multi-physical fields interactions for uniform heating and open-view imaging provide potential capabilities for quantitative evaluation of complex metallic structures. Based on the above simulation and experimental results, all the crack orientations can be detected by these three ECPT configurations [13–16]. The perpendicular and oblique cracks produce large electrical resistance to change the distribution of induced eddy currents, which are flowing parallel to line-coil and blade edge as shown in Figure 5. The parallel cracks can be identified through lateral heat conduction as described in [13]. Thermal patterns for these defects have been illustrated in Figures 6, 9 and 10 to prove defect orientations characterization. The limitations of line-coil excitation are the blocking effect and non-uniform heating that influence the defect detection and characterization. The Helmholtz-coil and ferrite-yoke have been proved suitable for complex shaped edge and multiple cracks inspection as shown in Figures 9 and 10, respectively, with higher SNR values. However, the defect detection and characterization of Helmholtz-coil is still influenced by its non-uniform heating in the region close to the Helmholtz-coil. The ferrite-yoke excitation is the best one in these three ECPT configurations because of the superior characteristics and high SNR values for the two regions. Similarly, its non-uniform heating regions also exist in the area close to the ferrite-yoke poles. The different configurations of ECPT can be applied for metallic material and defect characterization e.g., material stress [20,21], different defects e.g., fatigue [5], RCF [14,15], corrosion [22] and stress corrosion crack (SCC) via multiple physics characterization and fusion [17,23]. Relative uniform excitation through Helmholtz-coil or ferrite-yoke-based excitation provides better thermal images for the material characterization including quantitative NDE. 4. Conclusions and Future Works This paper reports three different excitation configurations of ECPT systems for complementary NDE of complex metallic structures and defects characterization. Simulation and experimental results demonstrate different ECPT configurations own individual strengths for different shape samples and defects detection applications. One proposed configurations of ECPT with Helmholtz coil with relatively uniform inductive heating can be applied for crack edge [4,24] detection, and other proposed

Sensors 2016, 16, 843

10 of 11

ferrite-yoke-based excitation can be applied for the characterization of RCF cracks in the rail track heads [14,15] with uniform heating and open-view thermal imaging. The open-view field imaging can be applied for large area inspection in contrast to the ECPT with the line-coil. A quantitative comparison of three ECPT configurations through SNR values is also provided to evaluate the defect detection and characterization performance. In conclusion, the relatively uniform area of ECPT configuration with the Helmholtz coil can easily be affected by the inspected sample inside the coil. Reasonably small samples can be inspected. For the ECPT configuration using ferrite-yoke for excitation can robustly control the uniform excitation field and direction. The inspected samples affecting the excitation field is relative small. The linear coil excitation can reach a local maximum with a strong influence on the surface curvature of the tested samples. Although rail multiple cracks and blade natural edge defects have well detected and verified the capabilities of differently configured ECPTs, quantitative NDE needs further consideration and processing due to the geometric superimposition on the heating pattern [23,25,26]. Their inspection limitations of electro-conductive materials only, constant lift-off, and the requirement of sample scanning need to be further addressed. Acknowledgments: This work is funded by the National Natural Science Foundation of China (Grant Nos. 61527803 and 51377015), the Fundamental Research Funds for the Central Universities and the Funding of Jiangsu Innovation Program for Graduate Education (Grant No. KYLX_0253), the EU FP7 “Health Monitoring of Offshore Wind Farms (HEMOW)” (FP7-PEOPLE-2010-IRSES, No. 269202) project, and the Fundamental Research Funds for Sichuan province science and technology innovation talent project (Grant No. 2015056). We thank the China Scholarship Council for sponsoring Yunlai Gao's visit to Newcastle University, UK. Author Contributions: All authors contributed to this work. Gui Yun Tian invented the original ECPT systems and acted as Academic Supervisor to the Yunlai Gao, and both of them have specifically contributed to the different configurations and multi-physics phenomena of the ECPT. Yunlai Gao conceived and designed the simulation and experiments; Yunlai Gao performed the simulations and experiments and analyzed the data; Kongjing Li, Yizhe Wang, Bin Gao and Yunze He contributed to the simulation design and the discussion of the results; Yunze He contributed to the literature review and the principles of ECPT; all the authors contribute to the writing and discussion of the paper. Conflicts of Interest: The authors declare no conflict of interest.

References 1.

2. 3. 4. 5.

6. 7. 8. 9.

Tian, G.Y.; Wilson, J.; Cheng, L.; Almond, D.P.; Kostson, E.; Weekes, B. Pulsed Eddy Current Thermography and Applications. In New Developments in Sensing Technology for Structural Health Monitoring; Mukhopadhyay, S.C., Ed.; Springer Berlin Heidelberg: Berlin, Germany, 2011; pp. 205–231. Wilson, J.; Tian, G.; Mukriz, I.; Almond, D. PEC thermography for imaging multiple cracks from rolling contact fatigue. NDT&E Int. 2011, 44, 505–512. Cheng, L.; Tian, G.Y. Surface crack detection for carbon fibre reinforced plastic (CFRP) materials using pulsed eddy current thermography. IEEE Sens. J. 2011, 11, 3261–3268. [CrossRef] Li, K.; Tian, G.Y.; Cheng, L.; Yin, A.; Cao, W.; Crichton, S. State detection of bond wires in IGBT modules using eddy current pulsed thermography. IEEE T. Power Electr. 2014, 29, 5000–5009. [CrossRef] Liu, J.; Ren, W.; Tian, G.Y.; Gao, B.; Wang, Y.; Zhang, J.; Shaw, B.; Yin, A.; King-Alale, N.O. Nondestructive evaluation of early contact fatigue using eddy current pulsed thermography. IEEE Sens. J. 2015, 15, 4409–4419. [CrossRef] Gao, B.; Bai, L.; Woo, W.L.; Tian, G. Thermography pattern analysis and separation. Appl. Phys. Lett. 2014, 104, 251902. [CrossRef] He, Y.; Tian, G.Y.; Pan, M.; Chen, D. Impact evaluation in carbon fiber reinforced plastic (CFRP) laminates using eddy current pulsed thermography. Compos. Struct. 2014, 109, 1–7. [CrossRef] Yang, R.; He, Y. Optically and non-optically excited thermography for composites: A review. Infrared. Phys. Technol. 2016, 75, 26–50. [CrossRef] He, Y.; Tian, G.Y.; Pan, M.; Chen, D. Eddy current pulsed phase thermography and feature extraction. Appl. Phys. Lett. 2013, 103, 084104. [CrossRef]

Sensors 2016, 16, 843

10. 11.

12. 13. 14. 15. 16.

17. 18. 19. 20. 21.

22. 23. 24. 25.

26.

11 of 11

Yang, R.; He, Y. Eddy current pulsed phase thermography considering volumetric induction heating for delamination evaluation in carbon fiber reinforced polymers. Appl. Phys. Lett. 2015, 106, 234103. [CrossRef] Cheng, L.; Gao, B.; Tian, G.Y.; Woo, W.; Berthiau, G. Impact damage detection and identification using eddy current pulsed thermography through integration of PCA and ICA. IEEE Sens. J. 2014, 14, 1655–1663. [CrossRef] He, Y.; Yang, R. Eddy current volume heating thermography and phase analysis for imaging characterization of interface delamination in CFRP. IEEE Trans. Ind. Inform. 2015, 11, 1287–1297. [CrossRef] Yang, R.; He, Y.; Gao, B.; Tian, G.Y.; Peng, J. Lateral heat conduction based eddy current thermography for detection of parallel cracks and rail tread oblique cracks. Measurement 2015, 66, 54–61. [CrossRef] Peng, J.; Tian, G.Y.; Wang, L.; Zhang, Y.; Li, K.; Gao, X. Investigation into eddy current pulsed thermography for rolling contact fatigue detection and characterization. NDT&E Int. 2015, 74, 72–80. Gao, Y.; Tian, G.Y.; Li, K.; Ji, J.; Wang, P.; Wang, H. Multiple cracks detection and visualization using magnetic flux leakage and eddy current pulsed thermography. Sens. Actuators A Phys. 2015, 234, 269–281. [CrossRef] Almond, D.P.; Weekes, B.; Li, T.; Pickering, S.G.; Kostson, E.; Wilson, J.; Tian, G.; Dixon, S.; Burrows, S. Thermographic techniques for the detection of cracks in metallic components. Insight 2011, 53, 614–620. [CrossRef] Yin, A.; Gao, B.; Tian, G.Y.; Woo, W.L.; Li, K. Physical interpretation and separation of eddy current pulsed thermography. J. Appl. Phys. 2013, 113, 064101. [CrossRef] Chatterjee, K.; Tuli, S.; Pickering, S.G.; Almond, D.P. A comparison of the pulsed, lock-in and frequency modulated thermography nondestructive evaluation techniques. NDT&E Int. 2011, 44, 655–667. Liu, J.; Wang, Y.; Dai, J. Research on thermal wave processing of lock-in thermography based on analyzing image sequences for NDT. Infrared. Phys. Techn. 2010, 53, 348–357. [CrossRef] Morozov, M.; Tian, G.Y.; Withers, P.J. Elastic and plastic strain effects on eddy current response of aluminium alloys. Nondestruct. Test. Eval. 2013, 28, 300–312. [CrossRef] Bai, L.; Tian, G.Y. Stress measurement using pulsed eddy current thermography. In Proceedings of the 51st Annual Conference of the British Institute of Non-Destructive Testing (BINDT 2012), Northamptonshire, UK, 11–13 September 2012. He, Y.; Tian, G.Y.; Pan, M.; Chen, D.; Zhang, H. An investigation into eddy current pulsed thermography for detection of corrosion blister. Corros. Sci. 2014, 78, 1–6. [CrossRef] Gao, B.; Woo, W.L.; Tian, G.Y. Electromagnetic thermography nondestructive evaluation: Physics-based modeling and pattern mining. Sci. Rep. 2016, 6, 25480. [CrossRef] [PubMed] Zhou, X.; Zhou, J.; Tian, G.; Wang, Y. Research on defects inspection of solder balls based on eddy current pulsed thermography. Sensors 2015, 15, 25882–25897. [CrossRef] [PubMed] Lahiri, B.B.; Bagavathiappan, S.; Soumya, C.; Mahendran, V.; Pillai, V.P.M.; Philip, J.; Jayakumar, T. Infrared thermography based defect detection in ferromagnetic specimens using a low frequency alternating magnetic field. Infrared. Phys. Techn. 2014, 64, 125–133. [CrossRef] Jäckel, P.; Netzelmann, U. The influence of external magnetic fields on crack contrast in magnetic steel detected by induction thermography. Quant. InfraRed Thermogr. J. 2013, 10, 237–247. [CrossRef] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).