Ultrasound Phased Array Imaging on Curved Surface for Weld Inspection of Elbow Pipe as a Replacement for Radiographic Inspection

Transcription

1 19 th World Conference on Non-Destructive Testing 2016 Ultrasound Phased Array Imaging on Curved Surface for Weld Inspection of Elbow Pipe as a Replacement for Radiographic Inspection Choon-su PARK 1, Jin kyu PARK 2, Wonjae CHOI 1, Seunghyun CHO 1, Eul seok MOON 2, and Mon joo HUH 2 1 Center for Safety Measurements, KRISS, 267 Gajeong-ro, Yuseong-gu, Daejeon, South Korea 2 Industrial Application R&D Institute, DSME, 3370 Geoje-daero, Geoje-si, Gyongsangnam-do, South Korea Contact choonsu.park@kriss.re.kr Abstract. Ultrasound phased array (PA) imaging has been regarded as a safe replacement for radiographic inspection, even though they have different detectability depending on flaw types. There are innumerable demands for non-destructive testing on weld joints of various pipes which are indispensable to most of industrial structures. Diverse ultrasound testing has been used to detect flaws in welding joints, but some technical deficiencies still remain. For example, ultrasound PA imaging on weld of elbow pipe has many challenging issues due to its geometrically varying surface along the circumferential direction. Conventional PA imaging for weld inspection has particularly focused on ultrasonic wave propagation based on ray theory. This confines the incident angle to be narrowly fixed and the position of array transducers as well. Ultrasound PA imaging by total focusing method (TFM), however, can provide not only high resolution images but also flexibility that makes particular use of a transmitted wave with a specific direction from a single element in the array transducer. This leads us to develop a method to get TFM images of weld zone from an elbow part that curves. It is inevitable of each ultrasonic wave from the array transducer to transmit through different media and to be reflected from the boundary in the elbow part with an angle along the curved surface. To form a correct PA image, careful calculation is made to ensure that time delay of receive-after-transmit is correctly shifted and summed even under non-planar boundary condition. Here, a method to calculate wave paths for the zone of interest at weld joint of an elbow pipe is presented. A numerical simulation of wave propagation on an elbow pipe is made to verify the proposed method, and it clearly shows that the proposed method precisely localizes true source position. In addition, it is experimentally demonstrated that the proposed method is well applied to various actual pipes that contains artificial flaws in them. License: 1 More info about this article:

2 Introduction Radiographic testing (RT) has been a prevalent way to see flaws inside test objects. The x-ray images clearly visualize where the flaws are in projection view with high spatial resolution, even though it cannot give any information on crack depth. Ultrasonic testing (UT) is also a non-destructive way that can give flaw information in materials, but it has long been served as an indicator using ultrasound signal itself. The UT is, therefore, restricted to experts who have trained to read ultrasound signal. In recent, as an emerging non-destructive testing, phased array ultrasound testing (PAUT) has taken a successful role in image-based inspection. In addition, it has been increased to take into account the PAUT as a safe replacement for RT [1~3], even though they have different detectability depending on flaw types. There are innumerable demands for non-destructive testing on weld joints of various pipes which are indispensable to most of industrial structures. Diverse ultrasound testing has been used to detect flaws in weld zone, but some technical deficiencies still remain. For example, ultrasound PA imaging on weld of elbow pipe has many challenging issues due to its geometrically varying surface along the circumferential direction. Conventional PA imaging by linear or sectorial scan for weld inspection has particularly focused on ultrasonic wave propagation based on ray theory. This confines the incident angle to be narrowly fixed with respect to observation zone and the position of an array transducer as well, and therefore the image has nothing but have low spatial resolution and hard to focus beams to shadow zone. This leads us to develop a method to get TFM [4] images of weld zone from an elbow part that curves. It is inevitable of each ultrasonic wave from the array transducer to transmit through different media and to be reflected from the boundary in the elbow part with an angle along the curved surface. To form a correct PA image, careful calculation is made to ensure that time delay of receive-after-transmit is correctly shifted and summed even under non-planar boundary condition. Here, a method to calculate wave paths for the zone of interest at weld joint of an elbow pipe is presented. A numerical simulation of wave propagation on an elbow pipe is made to verify the proposed method, and it clearly shows that the proposed method precisely localizes true source position. In addition, it is experimentally demonstrated that the proposed method is well applied to various actual pipes that contains artificial flaws in them. 1. PA imaging algorithm on a surface with curvature 1.1 Imaging with TFM The PA imaging is a technique by which to make images by shifting (or delaying) signal phases and summing the shifted (or delayed) signals in time domain or by controlling phases in frequency domain. (e.g. see Ref. [5]) The delay and sum (DAS) is the oldest but one of the most robust array signal processing algorithm to form and steer beams [6] like as Time of Difference of Arrival (For example, see Ref. [7]). Most phased array imaging techniques for non-destructive testing (NDT) are based on the DAS algorithm, and the images are obtained by controlling phases and then summing the signals aligned. The TFM is a post-processing algorithm using FMC (Full Matrix Capture) data-set to get higher spatial resolution than other PA imaging algorithms. Fig. 1 shows a geometric coordinate for TFM and its simulation result. A point reflector is positioned at 19.8[mm] to x-axis and 25.2 [mm] to z-axis, and Fig. 2(b) is obtained with a virtual linear array that has 16 elements 5 [MHz] resonance frequency and 5-cycle tone burst signal with Gaussian window. 2

3 (a) Fig. 1. Simulation for TFM imaging with a point reflector at (19.8, 25.2) [mm]; (a) geometric coordinate for imaging, (b) a TFM imaging. (b) The white X mark represents the true reflector s position, and the maximum peak of TFM well indicates the position precisely. Fig. 2 shows a TFM image with experimental FMC data for a calibration block that has several side drill-holes. The holes we want to get as a TFM image are placed at 10 [mm] below from the top surface at which an array transducer was located, and the holes are 19.6[mm] apart from each other. The array transducer has the same number of elements and resonance frequency with that of simulation. From the image, the maxima quite well correspond to the holes, and even some peaks less than the maximum value also appear over the group of small holes near the bottom. Fig. 2. A TFM image for a calibration block that has several types of side drill-holes. 3

4 1.2 wave path calculation for a surface with curvature It is of particular importance to get variable wedge or transducer to securely contact with a surface with curvature. Many researches on flexible array transducer have continuously reported and developed. (for example, see Refs. [8~10]) However, the mechanically fabricated flexible arrays inherently have large pitch between elements due to limitation of mechanical moving parts and also some problems on coupling secureness of individual element. Sutcliffe et al. [11] introduced a real-time algorithm on a known geometry using a wedge that was customized to contact with the shape of specimen. Flexible wedges are another way to contact on non-planar surface. To flexibly contact on rigid test bodies, the wedges are generally filled with fluid that has low acoustic impedance than the test bodies. The impedance mismatch leads wave path change at the interface between wedge and the test bodies according to Snell s law as shown in Fig. 3(a). In general, elbow pipes have geometrically calculable shape and are easily formulated mathematically. In the other words, the thin wall of elbow pipes has a specific curvature and thickness, and these make us to formulate geometrical relation between contact points at interfaces and wave paths as shown in Fig. 3.(b). There are 5 unknown variables ( q 1, q 2, q 3, xbi, xbo ), and we can get 5 serial equations from the geometric relation and Snell s law for each observation point. z bi and z bo are dependant variables to x bi and x bo, respectively. They are all in the calculable equation of circle. That is, x + ( z - r ) = r, (1) bo( i) bo( i) o( i) o( i) where r o( i) is the radius of outer (or inner) for a given elbow pipe. In addition, the incident angle, q 1, from wedge to pipe wall could be obtained as zbo - ro - x t - z t + 2 z bo - r o xbo sin q1 = ( x - x ) + ( z - z ) 2 ( zbo - ro ) + 1 x 2 bo 2 2 t bo t bo (2) Fig. 3. Wave path change between different material for TFM imaging; (a) Wave path change on a flat surface, (b) Wave path calculation on a surface with curvature for weld zone inspection. 4

5 from the geometric relation. ( x, z ) represents the transmitters position. The other angles t t ( q 2, q 3 ) are also easily determined at each varying position on the curved wall for each observation position with respect to all of the transmitter-receiver pairs. 2. Experiments on weld 2.1 Experimental set-up We made use of a 5 MHz linear array transducer that has 64 elements to obtain full matrix capture data. The signals are generated and received from a pulser&receiver (OEMPA 64/64, Advanced OEM Solutions), and stored in a controller as shown in Fig.4. A spiky pulse is generated from the pulse and put into a transmitting element, and all the elements in the transducer get waves propagating and reflected through the wall of pipe. The pipes for experiments are made by carbon steel and composed of a straight pipe and 90 elbows whose nominal pipe size (NPS) is 150 [mm]. The pipes are produced and welded according to the KS/JIS standard. Wedges for the elbow pipe are, for now, made up of plexi-glass for observing specific points that contain artificial defects, and a flexible wedge filled with water is going to be used to get TFM images. The incident angles are initially designed by Snell s law for the element in the midst, and then specific points at which waves pass through between the wedge and the pipe are going to be calculated, which is explained in Sec. 1.2, with respect to every point in the zone of interest in Fig. 1. Fig. 4. Experimental set-up and a photo for the array transducer and a wedge on an elbow part. 5

6 2.2 Flat weld specimen Before get images from elbow pipes, we examined some flat weld specimens (made by Sonaspection) that include artificial defects for verifying the proposed algorithm as Jobst et al. did [12]. The specimens are made by carbon-steel and have various types of defects. A wedge for the flat specimen was made as shown in Fig. 5 (a). In addition, a signal generated and received at the centre (32nd) element when defect exists is in red, and also a signal measured at defect-free region is in black as depicted in Fig. 5(b). A TFM image for the defect case is in Fig. 5 (c), and we can clearly find the maximum peak on weld interface. On the other hand, Fig. 5 (d) is a TFM image on the defect-free region, and any specific peak does not appear in the image and the colour contrast is also small compared to the defect case. From Fig. 5 (c) and Fig. 5(d), the region below 15 [mm], which is made up with half-skip signals from piezo-electric element of the array transducer, give a clear indication on the defect, because the directivity of elements is more favourable to the half-skip signals than the signal of direct path whose depth is 5~15 [mm]. 2.3 Elbow pipes An elbow pipe of carbon steel was prepared as shown in Fig. 6 (a), and it has some artificial defects in weld zone around its circumference. In this paper, we only show a representative defect located as depicted in Fig. 6 (b), and a wedge fitting to the surface at the position was fabricated with plexi-glass as is in Fig. 6 (b). With the geometric information such as diameter, wave speeds, and thickness etc., points at which waves pass through between the wedge and the pipe were calculated using Matlab (Mathworks Inc.) as explained in Sec Fig. 5. Experiments on a weld specimen that has an artificial defect on weld interface; (a) Wedge designed for flat carbon steel specimen, (b) time data measured in 32 element in a region of defect (red) and defect-free(black), (c) A TFM image for the defect at the weld interface, (d) A TFM image for the defect-free case. 6

7 Fig. 6. Experiments on a carbon steel elbow pipe that has an artificial defect, a lack of fusion, on weld interface; (a) A carbon steel elbow pipe for experiment, (b) The position of the artificial defect and wedge designed for the elbow pipe, (c) A TFM image for the defect at the weld interface. In particular, the pipe has different thickness at the straight part and elbow part, and the elbow part has 7.1 [mm] and the straight part for 11[mm] as drawn with dotted line in Fig. 6 (c). The array transducer and the wedge are positioned in the elbow part. Therefore, the waves would pass through weld zone from thinner passage in the elbow part to the wider one in the straight pipe. On the other hand, it might be hard to propagate from the wider pipe to the narrower one due to the scattering around the corner. Fig. 6 (c) definitely shows a peak which appears at the weld interface in the straight part, and the pipe was known to have a Lack of Fusion at the interface as Fig. 6 (c) delineated. 3. Conclusions A TFM imaging technique for inspection to weld zone in elbow pipes is presented. TFM has been proved to have the best spatial resolution than other PA imaging methods by using FMC. The wave path formulation not only through dual media but also on surface with curvature is introduced and numerical calculation to solve five serial equations is explained briefly. The proposed technique has been applied to FMC data which is measured by an array transducer that has 64 elements and 5 [MHz] resonant frequency. A TFM images for a flat weld specimen definitely give a clear location of an artificial defects on a weld interface. In addition, low contrast and somewhat noisy image for defect-free region confirm the performance of the proposed imaging system experimentally. In the end, the TFM imaging to a carbon steel elbow pipe, which is particularly jointed with a thicker straight pipe, successfully shows the location of Lack of Fusion on the weld interface that was artificially made before the experiment. 7

8 Acknowledgements This work was partly supported by Korea Research Institute of Standards and Science (KRISS) under the project Intelligent safety diagnosis technology based on integrated of micro/macro behaviours of public facilities (KRISS ) and partly supported by Civil & Military technology cooperation program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (No.2014M3C1A ). References [1] N. S. Goujon, Replacement of radiography by ultrasonic inspection, HSE Books - Research report 301 (2005) [2] T.L. Moran et al., Replacement of radiography with ultrasonics for the nondestructive inspection of welds Evaluation of technical gaps- An interim report, PNNL (2010) [3] A. Bulavinov, R. Pinchuk, S. Pudovikov, and C. boller, Ultrasonic sampling phased array testing as a replacement for X-ray testing of weld joins in ship construction, International Journal on marine navigation and safety of sea transportation, 6(2), pp (2012) [4] C. Holmes, B. W. Drinkwater, P. D. Wilcox, Post-processing of the full matrix of ultrasonic transmit-receive array data for non-destructive evaluation, NDT&E international, 38, pp (2005) [5] C.-S. Park, J.-W. Kim, S. Cho, and D.-C. Seo, A high resolution approach for nonlinear sub-harmonic imaging, NDT&E international, 79, pp (2016) [6] C.-S. Park, J.-H. Jeon, Y.-H. Kim, and Y.-K. Kim, Display problem on acoustic source identification using beamforming method, Proceedings of the 38 th International Congress and Exposition on Noise Control Engineering; 2009 Aug ; Ottawa, Ontrario, Canada. Institute of Noise Control Engineering; IN (2009). [7] C.-S. Park, J.-H. Jeon, and Y.-H. Kim, Localization of a sound source in a noisy environment by hyperbolic curves in quefrency domain, Journal of Sound and Vibration, 333, pp (2014) [8] C. R. Bowen, L. R. Bradley, D. P. Almond, P. D. Wilcox, Flexible piezoelectric transducer for ultrasonic inspection of non-planar components, Ultrasonics, 48, pp (2008) [9] O. Casula, C. Poidevin, G. Cattiaux, and G. Fleury, A flexible phased array transducer for contact examination of components with complex geometry, Proceedings of the 16 th World Conference on Nondestructive Testing; 2004 Aug. 30~ Sep.3; Montreal, Canada. International Committee for Non-destructive Testing (2004) [10] C. J. L. Lane, The inspection of curved components using flexible ultrasonic arrays and shape sensing fibres, Case Studies in NDT&E, 1, pp (2014) [11] M Sutcliffe, M Weston, B Dutton, I Cooper, and K. Donne, Real-time full matrix capture with auto-focusing of known geometry through dual layered media, Proceedings of the NDT 2012 Conference in British Institute of Non-destructive Testing (2012) [12] M. Jobst and G. D. Connolly, Demonstration of the application of the total focusing method to the inspection of steel welds, Proceedings of the ECNDT 2010; June 7~11; Moscow, Russia (2010) 8