Study on fatigue crack detection for hollow axle in high-speed railway by phased array UT and quantification

Transcription

1 Study on fatigue crack detection for hollow axle in high-speed railway y phased array UT and quantification Jianping Peng 1, Menglin Zhao 1, Xiaorong Gao 1, Yu Zhang 1, Kai Yang 1, Bo Zhao 1 1 NDT research center, Southwest Jiaotong University, Chengdu, China, More info aout this article: Astract adams.peng@home.swjtu.edu.cn Hollow axle is one of issue component in high speed railway, which is inspected using UT to keep safety in service. In this paper, the shortest path geometry model of ultrasonic propagation under doule layer media is studied. Comined with Snell's law and Ferrari method, the position of the incident point of the ultrasonic eam with the interface plane is calculated, so as to realize the total focus method (TFM) imaging in the doule-layer medium. The experimental results show that this method can e provided the exact solution of the coordinates of the incident point of the linear interface and accurately reconstruction the fatigue crack image. To improve the model through increasing the secondary reflection of the path calculation and waveform conversion status, it can e achieved with the hollow axle s real crack imaging. The experimental results can clearly present the shape and tendency of cracks, due to the phenomenon of diffuse reflection of the crack tip, causing some impact on the quantitative. Keywords: Doule layer medium; Ferrari method; TFM; Phased array ultrasonic testing; Hollow axle 1 introduction The hollow axle in the whole world is one of important component and widely used in high-speed train. The hollow axle fatigue is one of critical danger source for railway [1]. To keep the safety of hollow axle, the Non-destructive Testing(NDT) methods, such as Ultrasonic testing, has een employed widely in the world railway. But the conventional ultrasonic detection technology has meet some challenges in the impact of system disturance, detecting depth [2]. The conventional UT comined with SAFT thechnology can improve the signal-to-noise ratio and high resolution ultrasonic image y DB [3]. AREVA has added phased array proes to carry out several angle eams to scan the hollow axle transverse cracks, so as to improve the signal-to-noise ratio of the transverse crack detection [4]. But the accurately location and quantification of crack defects are still a ig chanllenge for UT testing. In 2005, Caroline, Holmes and so on use the full matrix acquisition (FMC) to collect the complete data set of the ultrasonic scanning line, and uses the total focus method (TFM) to imaging [5, 6]. FMC-ased TFM algorithm uses a complete data set containing all the information of defects to image, can e sensitive to small defects and accurate quantitative and positioning [7, 8], suitale for hollow axle angle crack detection quantitative analysis [9, 10]. Miles Weston et al. Calculated the position coordinates of ultrasonic at the junction of the doule-layer media y Snell's law, and studied the TFM imaging of the wedge in the angle and depth direction [ID108] 1

2 caliration and compensation methods to achieve image quality consistency in different regions [11, 12]. The wedge and the hollow axle make up the doule-layer medium, and the TFM needs to otain the propagation delay of the ultrasonic and the hollow axle respectively. Therefore, the ultrasonic propagation model ased on the doule-layer medium can e otained according to Snell's law and Fermat's theorem. In this paper, the ultrasonic propagation model mentioned aove will e studied and e employed to detect the hollow fatigue crack with simulation and expereimental results. The results show that the algorithm has excellent performance in quantitative analysis of tilt crack detection. 2 Principles and methods The ultrasonic waves are emitted from the inner and outer surfaces of the hole and the hollow axle is sujected to nondestructive testing in a spiral motion manner. It is possile to improve the image signalto-noise ratio y adding the phased array proe to the spiral data or using SAFT to process the spiral data ased on the conventional ultrasonic to improve the image signal-to-noise ratio. Because the collected crack imaging information is incomplete, the length of the defect and the growth trend are accurately descried. The Total Focus Method (TFM) is that a single array element is transmitted in turn, and all phased array elements receive this process at the same time as full matrix acquisition (FMC). Through the full matrix acquisition technology, it is possile to otain all the transmit-receive A-scan data cominations in the proe, which provides the most complete data information for the post-pointfocused imaging of post-data processing. 2.1 Doule-layer media ultrasonic propagation geometry model The shortest propagation path ased on Fermat's theorem also satisfies the ultrasonic propagation model in the doule-layer medium, and the refraction of ultrasonic at the interface of the doule layer needs to conform to Snell's law. Ultrasonic wave from the first layer of media (plexiglass wedge) incident to the second layer of media (steel) eam propagation as shown in Figure 1 show. Beam emission position X,Z 0 0 coordinates, eam focus point position X, s Zs coordinates, whereq i and qr respectively represent the ultrasonic incident angle and the refraction angle. X,Z The ultrasonic coordinates at the oundary of the doule layer media is. Let according to the trigonometric function Snell law we can see: = v / v, [ID108] 2

3 ( X0, Z0 ) θi V1 ( X, Z ) V2 θr ( Xs, Zs ) Figure 1: ultrasonic refraction in doule-layer media. X -X X -X s = s - + s - Z Z X X Z Z X X (1) In equation (1), other parameters are known, in order to simplify the calculation of the definition of the media at the oundary of the coordinates ordinate X,0 X of the incident point can e otained.. Using the Ferrari method, the value of the horizontal 2.2 Comparisons of total focus method and total focus method of wedge (WTFM) A single array element is transmitted in turn, and all phased array elements receive this process at the same time as full matrix acquisition (FMC). Through the full matrix acquisition technology, it is possile to otain all the transmit-receive A-scan data cominations in the proe, which provides the most complete data information for the post-point-focused imaging of post-data processing. Figure 2 FMC acquisition process The time at which the ultrasonic wave is emitted from the element i through the focus P to the element j am: Known x i, x j and t (, ) p x z ( ( x- x ) ) 2 ( - ) 2 ( - ) 2 ( - ) 2 i + y yi + x xj + y yj = (2) c t, we can get the amplitude information of each A-scan line at point P, and p( x, z) ring in the formula (2) to otain the amplitude information of each pixel in the detection area and superimpose it to otain the image of the whole detection area. [ID108] 3

4 ( ) ( xp zp) = åå e xi xj tpxz (, ) (3) N N I,,, i= 1 j= 1 The full focus imaging algorithm uses a complete data set ased on the FMC method to otain etter imaging quality than other time-delay superposition focusing algorithms, as shown in figure 3. In figure 4, it is shown that total focus imaging the same defects through the wedge. Figure 3: total focus imaging Figure 4: total focus imaging with wedge. 3 Hollow axles of ultrasonic testing 3.1 Conventional detection of transverse crack on the online surface of the axle The hollow axle flaw detection equipment uses 40 ~70 degrees transverse wave proe, and otains the information of the fatigue crack in the axle outline through the reflection of the corner angle. The coupling oil is used as a variale wedge to flexily adjust the angle of incidence and detect the different regions. When the refraction angle is 40 degrees, the Shear wave energy is stronger than the longitudinal wave, and the echo signal has a etter signal-to-noise ratio. It is suitale for detecting the transverse crack (groove) of the axle outer diameter, as shown in figure 5. The effect of the conventional proe on detecting the crack outline the axle is shown in Figure 6. The true depth of the crack in the hollow axle is 2.5mm, while the B scanning interface shows a 12mm length. Cannot reflect the crack shape and length, which makes it difficult to determine the level of post injury. [ID108] 4

5 3.2 WTFM transverse crack detection experiment The transverse crack detection of the hollow axle outline, WTFM involves the ultrasonic signal corner angle reflection prolem, and the ultrasonic propagation path will have the secondary reflection phenomenon. If the ultrasonic propagation path is directly used y the time of flight principle, the delay data will e superimposed and the imaging of the crack will e dislocated. Therefore, it is necessary to improve the WTFM and increase the path calculation method of the secondary reflection to reconstruct the ultrasonic propagation path under the condition of transverse cracks. Figure 7: WTFM transverse crack detection propagation path. N N és1+ s2 s3+ s4 + AB' ù I ( xp, zp) = åå eê +, i, jú i= 1 j= 1 ë c1 c (4) 2 û To improve the geometric model of ultrasonic propagation path of the wedge, as shown in Figure 7, in (4), " and # are the propagation velocity of ultrasonic wave in the wedge and test lock, and %,' is the signal received y the ith element to transmit the jth element. ", # is the two-part of ultrasonic propagation in the wedge. ), * is the two-part propagation path of the ultrasonic wave in the test lock. The mirror reflection of ' B is B y AB ' = AB. Using the SA12-N55S wedge with curved surface, the defect of the transverse crack at the outer surface of the hollow axle is 2.5mm, and the hollow axis is 50mm thick. (a) () Figure 8: FMC data acquisition of wedge lock; (a) wedge lock phased array proe and hollow axle semicircle test lock; () transverse crack on the outer surface of axle. [ID108] 5

6 (a) original result () filtering image Figure 9: WTFM transverse crack In Figure 9(a), the defects coincide with the true crack shape, ut due to the presence of a diffuse reflection phenomenon at the crack tip, a small numer of urrs is generated, and the crack in the imaging view is longer than the true crack. For the transverse crack resulted y WTFM is the second reflected wave sound path (Figure 7) to ensure that the crack at the amplitude of the superposition of the correct, the real show the shape of the crack. But only the ultrasonic signal at the crack needs to calculate the secondary reflected wave sound path. Without knowing where the transverse crack exists, the improved WTFM algorithm uses the secondary reflected wave sound path for all imaging positions, resulting in the size of the crack near the range of superposition of the index to get defect information and high energy, low signal to noise ratio. Figure 9() is an image result y application with circular filter; the results show the effective removal of the diffuse phenomenon and the secondary reflection wave error and the emergence of the tip of the thorn and pseudo-crack information. Tale 1: comparison of WTFM and conventional ultrasonic on transverse crack Imaging. Parameters Origin crack B mode WTFM & Filter Crack depth(mm) PSNR(dB) It can e seen from Tale 1 that the WTFM can reflect the crack shape and extract the depth information of the crack effectively, and locate and quantify the hollow axle defects compared with the conventional hollow axle crack detection technology. After the circular filtering treatment, the WTFM can further Improve the imaging signal to noise ratio while reducing the crack depth error. [ID108] 6

7 4 Conclusions In this paper, the geometrical model of ultrasonic propagation path under doule layer and TFM is analyzed. According to the Fermat's theorem, the API coordinates show that the effect of WTFM on defect imaging and TFM are consistent with the imaging effect of defect. Experiments show that the improved WTFM can e used to detect the growth trend of defects in the transverse crack surface of the hollow axle. The true reflection of the defect location and size greatly improves the signal-to-noise ratio of the test results. Later studies can e ased on the rule of two-layer media imaging technology, through the doule total focus method (the first surface curve fitting, the second adaptive delay eam) to achieve complex surface detection oject imaging algorithm. References [1] C. Klinger, D. Bettge, Axle fracture of an ICE3 high speed train, Vol. 35, No. 15, pp , Dec [2] C. Richter et al, Development of Components for an Industry Approved Hollow-Axle Testing System, proceedings of the 18th World Conference on Nondestructive Testing, pp , [3] U. Völz, P. Heilmann and M. Spieker, New Approach for a More Detailed Visualisation of Ultrasonic Testing Data of Railway Hollow Axles proceedings of the 19th World Conference on Nondestructive Testing, [4] P. Marty et al, Latest development in the UT inspection of train wheels and axles, proceedings of the Proceedings of the 18th World Conference on Nondestructive Testing, [5] C. Holmes, B. W. Drinkwater and D. W. Ilcox, Post-processing of the full matrix of ultrasonic transmit receive array data for non-destructive evaluation, NDT & E International, vol. 8, no. 38, pp , [6] C. Holmes, B. W. Drinkwater and D. P. Wilcox, Advanced post-processing for scanned ultrasonic arrays: application to defect detection and classification in non-destructive evaluation, Ultrasonics, vol. 6-7, no. 48, pp , [7] L. Moreau, A. J. Hunter and B. W. Drinkwater, Efficient imaging techniques using an ultrasonic array, Proceedings of SPIE - The International Society for Optical Engineering, vol. 7650, pp , [8] M. Sutcliffe et al, Real-time full matrix capture for ultrasonic non-destructive testing with acceleration of post-processing through graphic hardware, NDT & E International, vol. 10, no. 51, pp , [9] J. Zhang et al, Defect detection using ultrasonic arrays: The multi-mode total focusing method, NDT & E International, vol. 2, no. 43, pp , [10] R. Long, J. Russell and P. C. awley, Ultrasonic phased array inspection using full matrix capture, Insight-Non-Destructive Testing and Condition Monitoring, vol. 7, no. 54, pp , [11] M. Weston et al, Time efficient auto-focussing algorithms for ultrasonic inspection of duallayered media using full matrix capture, NDT & E International, vol. 7, no. 47, pp , [ID108] 7

8 [12] M. Weston et al, Caliration of ultrasonic techniques using full matrix capture data for industrial inspection, vol. 11, no. 54, pp , [13] M. Sutcliffe et al, Full matrix capture with time-efficient auto-focusing of unknown geometry through dual-layered media, Insight-Non-Destructive Testing and Condition Monitoring, vol. 6, no. 55, pp , [14] J. Zhang, B. W. Drinkwater and P. D. Wilcox, The use of ultrasonic arrays to characterize crack-like defects, Journal of Nondestructive Evaluation, vol. 4, no. 29, pp , [15] C. C. Macduffee, Theory of equations, [ID108] 8