INDUCTION AND CONDUCTION THERMOGRAPHY: OPTIMIZING THE ELECTROMAGNETIC EXCITATION TOWARDS APPLICATION J. Vrana1, M. Goldammer2, K. Bailey1, M. Rothenfusser2, and W. Arnold3 1 Siemens Energy, MC Q3-031, 4400 N Alafaya Trail, Orlando, FL 32826, USA 2 Siemens AG, Corporate Technology, Otto-Hahn-Ring 6, D-81730 Munich, Germany 3 Fraunhofer Institute for Nondestructive Testing (IZFP), Campus E 3.1, University, D66123 Saarbrücken, Germany (present address: Department of Materials, Campus D 2.2, D-66123 Saarbrücken, Germany) ABSTRACT. Active thermography, using electromagnetic excitation, allows detecting defects like cracks which distort the flow of current in the component under examination. Like other thermography techniques it is rapid and reliably utilizing infrared imaging. Electric current can be used in two ways for thermography: In induction thermography a current is coupled to the component by passing an AC current through a coil which is in close proximity to the component inspected, while in conduction thermography the current is coupled directly into the component. In this paper, the specific advantages of both coupling methods are discussed, including the efficiency of the coupling and optimization strategies for testing and also the necessary algorithms required to analyze the data. Taking these considerations into account a number of different systems for laboratory and practical application were developed. Keywords: Induction Thermography, Conduction Thermography, Induction Efficiency, Current Density PACS: 87.63.Hg, 81.70.Ex, 81.70.Fy, 44.05.+e, 44.40.+a, 02.60.Cb, 02.70.Dh INTRODUCTION A component can be excited in a thermographic inspection in one of two electromagnetic ways, either inductively or conductively [1]. As illustrated in Fig. 1, inductive excitation is achieved by an AC current running through an inductor placed next to the sample, and conductive excitation is achieved by direct galvanic contacts. In this case an excitation with an AC or DC current is possible. Therefore, with both methods a current is created in the electrically conducting material of the component, and depending on the current density, heating occurs in the sample. If a crack is present, the current is disrupted, resulting in an alternated current density distribution around the cracked area. Therefore, when the heat diffuses to the surface, the crack can be detected with an infrared camera. This process of thermographic crack detection using electromagnetic excitation is discussed in [2] in detail on the example of inductive excitation. However the crack detection mechanism is the same for both excitation methods. CREDIT LINE (BELOW) TO BE INSERTED ON THE FIRST PAGE OF EACH PAPER EXCEPT FOR ARTICLES ON PP. 19 - 32, 217 - 223, 371 - 377, 1091 - 1098, 1099 - 1106, 1190 - 1197, 1719 - 1726, 1792 - 1799, 1870 - 1877, 1878 - 1885, and 1902 – 1909 CP1096, Review of Quantitative Nondestructive Evaluation Vol. 28, edited by D. O. Thompson and D. E. Chimenti © 2009 American Institute of Physics 978-0-7354-0629-2/09/$25.00 518 IR-Camera Direct Contact Inductor Tested part Tested part Current IR-Camera Alternated current density FIGURE 1. Top left: Typical induction thermography measurement set-up. Top right: Typical conduction thermography measurement set-up. Both methods lead to an alternated current density (bottom). In inductive excitation the current density is highest directly under the inductor and the current flows back in the edges of the front and back surface [2]. With DC conductive excitation, the current density is homogeneously distributed if no defects are present. With AC excitation, the current density is highest at the four edges of the conductor and decreases in between. This is called the edge effect (see Fig. 2). Additionally, the skin effect is present as the current density rises from the middle to the sides. When building inspection systems, it is important to identify the appropriate parameters. Factors such as the material under test, the type of induction generator, the type of infrared camera, and the signal processing are critical for a successful system development. Current Dens ity [A/mm²] Thickness AC-Current Width y x ] cm y[ z x[ cm ] FIGURE 2. Current density distribution across the section indicated in grey of a rectangular conductor. 519 OPTIMIZING THE INSPECTION STATION The result of the inspection depends on the inductive or galvanic coupling of the current into the component, the current density distribution created, and the temperature distribution due to the heat diffusion. So, to optimize the inspection station, all these parameters have to be maximized to improve the signal-to-noise ratio, while maintaining fine details and a good quality image. Heating of the Sample The signal-to-noise ratio depends on the heating of the sample Q, which in turn depends on the power P in the part and on the excitation time-span τ: Q = P ⋅τ . (1) The power can be estimated for a galvanic contact by (2) P ≈ I 2R where I is the current flowing through the part at resistance R. For inductive coupling the power can be estimated by [1]: μf (3) P ~ I Inductor 2 σ with the electrical conductivity σ, the permeability μ, the frequency f, and the current flowing through the inductor IInductor. Therefore, to optimize the signal-to-noise ratio, the power should be maximized and as the power is quadratically dependent on the current flowing through the part or on the current flowing through the inductor, respectively, the generator chosen for the inspection should be optimized for high currents. By using standard generators available on the current market, supplying 500 – 2000 A, most electrically conductive materials can be tested. Moreover, even low electrically conductive materials are inspectable. Due to the limited maximum current, materials with higher electric conductivity, like aluminum and copper, require a more sophisticated method of inspection and signal processing. To the contrary materials with a high permeability, due to the skin effect and the increased power, are especially suited for surfacecrack detection using induction thermography. However for every material the generator has to be adapted to the permeability of the material. When utilizing induction thermography, some power is lost in the inductor which requires cooling and can not be used for the excitation. Therefore the efficiency [1] can be stated as: 1 η~ 2h σ P μ I (4) 1+ a μ Pσ I where the distance from the inductor to the surface is h, the radius of the inductor is a, the electrical conductivity is σ, and the permeability of the inductor is μ. Eq. (4) indicates that the permeability of the inductor material should be as close to 1 as possible and that the electrical conductivity of the inductor should be as high as possible. Therefore, most inductors are made out of copper. According to (4), a small distance inductor–sample h is optimal, but with a small distance the distribution of the current density is forced into a small area which must be of reasonable size to detect cracks [1, 2], and hence an optimal distance for a given inspection task has to be found. The cross-section of the inductor, represented in (4) by the radius of the inductor a, should be as large as possible, but an overly inductor will decrease the visibility of the crack. Typical values for these two parameters are h ~ 2 - 10 mm and a ~ 2.5 mm. According to (1), a second factor to optimize the signal-to-noise ratio is the time-span of the excitation. However, the thermal diffusion process leads to the blurring of the image 520 over time, especially with small heat sources as created by cracks. The thermal diffusion length for a pulsed excitation is proportional to: dth ~ α t (5) where α is the thermal diffusivity. As the power of the generator is limited, a compromise is necessary to get enough heat into the component to have a good contrast in the defect image, still maintaining fine details with a sharp image. This compromise is quite easy for most materials and results in excitation times of about 50 – 200 ms. Only materials with a high diffusivity require a fast camera and a more sophisticated method of inspection. This holds especially if a high diffusivity is combined with a high electrical conductivity which is the case for most metals, as thermal and electrical conductivity increase concurrently. Therefore, for a multi-purpose system, the generator should provide the ability to provide a current of about 500 - 2000 A, a pulse duration of about 5 - 2000 ms, and the infrared camera should have a low NETD [2] with a frame rate of at least 100 Hz to capture even fast thermal processes. Signal Processing 2 cm IR Signal [Grayvalues] Besides optimizing the excitation, the resulting images should be optimized to compress all the information available into a single image indicating only the defects. Fig. 3 shows a raw sequence of infrared images of an induction thermography inspection of a turbine blade using a 100 ms long induction pulse. The crack in the turbine blade foot is visible in all the stills. A comparison with a dye-penetrant inspection is provided at the top left. However, this is clearly not the optimal signal desired, especially since several artifacts are visible. Those artifacts are partially caused by the properties of the blade, like the temperature distribution and the emissivity, and therefore even visible before the excitation started. Other artifacts are caused by the excitation, like the heat distribution next to the inductor and the heating of edges and corners. In addition there is blurring of the signal with time. After 260 ms the crack signal is quite blurred and after 597 ms it is hardly visible. 1200 1000 800 600 400 200 0 0 100 200 300 400 500 Time [ms] IR Signal [Grayvalues] 800 0 FIGURE 3. Top left: Turbine blade to be inspected, showing a crack indication obtained with a dyepenetrant inspection; Bottom: Raw infrared image sequence from left to right: Before the induction pulse started, at 20, 99, 260, and 597 ms, respectively. Top right: Typical infrared signal of an active thermography inspection with a 100 ms long electromagnetic excitation. 521 IR Signal [Grayvalues] 800 0 Phase [°] 350 280 FIGURE 4. Top: Background-subtracted infrared image sequence from left to right: Before the induction pulse started, at 20, 99, 260, and 597 ms, respectively. Bottom: Pulse-phase analysis of the infrared sequence shortened to 100 ms excitation time and 100 ms cool-down time. 2200 2750 IR Signal [Grayvalues] IR Signal [Grayvalues] One way to extract the interesting result – the heating of the blade due to the excitation – is to subtract the background. This is accomplished by subtracting a picture before the excitation pulse has started. As it can be seen in Fig 4 (top) the background subtraction reveals the indication. However still some artifacts are visible. Further steps can be taken by post-processing the image sequence containing the heating and cooling phase, using the pulse-phase analysis [3] to reveal the indication. This results in splitting the information contained in the sequence into two images. One of the images contains the amplitude-based information and the other the time-based (phase) information. As the phase image reveals the crack and suppresses the differences in the emissivity of the part and some of the non-uniformities of the heating (Fig.4 bottom), this is the image processing to be used. A similar result can be obtained using the lock-in technique. For the lock-in analysis, a periodical excitation is necessary that yields also a phase and an amplitude image. Therefore, the pulse-phase analysis is equivalent to a lock-in analysis using only one cycle. However, when using only one cycle, the heating will be insufficient in some cases for materials having a high electric conductivity and diffusivity. By using the lock-in approach, a short excitation time can be used and the necessary heating of the part can be accumulated by using a sufficient number of cycles. Even by using pulse-phase or lock-in excitation, some artifacts remain in the resulting image. This can be resolved in automated production testing by using a flawless component as a reference. Fig. 5 shows this in more detail. By subtracting the image of the reference component from the component under test, only the indication remains visible. This can be improved further by combining it with pulse-phase or lock-in excitation [2]. Only some small artifacts due to positioning or size problems remain. 0 0 FIGURE 5. From left to right: Component under test, flawless reference component, and resulting image. 522 SPECIFICATIONS OF THE ELECTROMAGNETIC THERMOGRAPHY SYSTEMS Induction Thermography Fig. 6 shows a stationary multi-purpose induction thermography system built at Siemens in Munich. In this system a 10 kW computer-controlled induction generator is used. The minimal possible pulse length is 5 ms and the system can be adapted to different materials, due to exchangeable condensers. Moreover, the relative position of the inductor and the component to be tested can be adjusted by motors and a lock-in measurement can be carried out by switching the current of the generator on and off. The auxiliary device is connected with the generator by a 2 m long and 5 cm thick stiff cable. In some cases it is necessary to inspect parts in the field. For such cases, a mobile and handheld system capable of producing a high current was developed. This system provides arbitrary positioning and allows immediate identification and marking of defects. This can be achieved by using a different standard induction generator with 10 kW power. The auxiliary device is included at the end of a more flexible and longer (5 m) cable. Therefore, the inductor can be positioned arbitrarily within the radius of the cable. The capacitors of the induction generator are not exchangeable, but the generator is optimized for all non-permeable parts, which results in a maximum usable current of approximately 980 A. Moreover, the control is more versatile and the induction pulse can be modulated by an analog signal up to a frequency of 300 Hz for lock-in measurements. To optimize the design of a portable system, it was integrated into a 19” case (Fig. 7). It includes the induction generator, a computer for system control, a keyboard and a display. The connection between the computer and all periphery equipment is accomplished via a connector box developed for this purpose. The excitation and the post-processing are integrated in the software developed by Siemens CT. This system is used by the service division of Siemens Energy in Pittsburgh, PA. FIGURE 6. Stationary laboratory induction thermography system (top: infrared camera, right: auxiliary device of the induction generator). 523 FIGURE 7. Mobile and handheld induction thermography system. Left: Portable rack including the generator and all necessary equipment. Right: Auxiliary device with a small infrared camera and a touch-screen display. More flexibility of the system is possible by mounting some additional equipment on the auxiliary device. By activating a button to start the measurement, a small infrared camera captures the image displayed on a small screen. This system can be applied at the part location, exciting the component and interpreting the resulting image displayed after the measurement on the small screen. The performance of the small camera is at present limited, but it is sufficient for many inspection cases as the resulting image in Fig. 7 reveals. By using a touchscreen display, the system can be operated remotely, for example by marking and evaluating the indications. Conduction Thermography Besides the induction thermography system, a conduction thermography system was developed. Fastening of the component to be inspected allows the use of pulse-phase and lockin analysis. Therefore the component shown in Fig. 8 is clamped on one side. This fixture also serves as the first electrical contact. The second electrical contact is the clamp shown on the left side. The maximum current produced by the generator is 2000 A which is sufficient for most parts to be inspected with a reasonable excitation time. Phase [°] 220 165 FIGURE 8. Conduction thermography system and resulting image calculated by pulse-phase analysis. The “crack” (several holes drilled into the component next to each other) in the center of the test component is clearly visible. 524 IR Signal [Grayvalues] 200 -10 FIGURE 9. Conduction thermography system and resulting image calculated by pulse-phase analysis. With this system a test-component was inspected containing several holes drilled into the part next to each other. The resulting image (Fig. 8 right) was gained with the pulse-phase analysis suppressing artifacts and showing the indication clearly. The low noise level indicates that nearly the whole part could be inspected for defects. Furthermore, the current unit of a magnetic particle inspection system (Horizontal Wet Unit) was used for the excitation by direct contact and, instead of imaging the distribution of the magnetic particles, an infrared camera is used for detection. Fig. 9 shows a subsurface crack in a non-permeable vane which is clearly discernible. Only a heat-spot, due to the contacting of the part is visible, which can be eliminated by adapting the fixture to the part under test. SUMMARY In this paper we showed that active thermography with electromagnetic excitation can be applied to components by using the inductive and the conductive excitation methods. For both excitation modes the coupling, the current density distribution, and the heating in the part are discussed, including the efficiency. Moreover, it has been demonstrated that a successful application requires post-image processing. REFERENCES 1. J. Vrana, M. Goldammer, J. Baumann, M. Rothenfusser, and W. Arnold, “Mechanisms and Models for Crack Detection with Induction Thermography”, in: Review of Progress in QNDE 27, edited by D.O. Thompson, D.E. Chimenti, AIP Conference Proceedings 975, pp. 475-482 (2008). 2. J. Vrana, “Grundlagen und Anwendungen der aktiven Thermographie mit elektromagnetischer Anregung - Induktions- und Konduktionsthermographie” (“Basics and Applications of Active Thermography Techniques with Electromagnetic Excitation – Induction and Conduction Thermography”), Ph.D. Thesis, Science and Engineering Faculty III, Saarland University (2008), unpublished. 3. X. Maldague and S. Marinetti, “Pulse Phase Infrared Thermography”, J. Appl. Phys. 79, pp. 2694-2698 (1996). 525
© Copyright 2026 Paperzz