外文原文-基于超声TOFD浅埋藏缺陷检测方法

J Nondestruct Eval (2013) 32:164171 DOI 10.1007/s10921-012-0169-1 Shallow Buried Defect Testing Method Based on Ultrasonic TOFD Dazhao Chi Tie Gang Received: 1 November 2011 / Accepted: 16 December 2012 / Published online: 3 January 2013 Springer Science+Business Media New York 2013 Abstract Ultrasonic time of fl ight diffraction (TOFD) suf- fers from the shortcoming of lack of near surface resolution, primarily owning to the superposition of the lateral wave and the shallow buried defect (SBD) wave, and the insen- sitive region beneath the inspection surface because of the restricted beam width of the probes. This paper presents a method for the detection of SBD based on conventional TOFD.ThemethodisnamedasTOFDWbecauseitemploys a three-fold refl ected longitudinal wave whose propagation path in the testing piece looks like the letter “W”. Based on Pythagorass theory, mathematical models are developed to locate SBD. In TOFDW mode, the time difference between the lateral wave and the SBD wave can be enlarged greatly so that they would never be superposed. In addition, through refl ecting the incident longitudinal wave by bottom surface, thenearsurface regionofthetestingpieceiscoveredwithan intenseacousticfi eld,makingTOFDWfreeoftherestriction of the probes beam width. These contribute to the improve- ment of the near surface resolution. The experiments show that with the proposed TOFDW and mathematical models, the artifi cial defect tip with the buried depth of 1 mm can be detected, and the tips can be measured within an error of 0.3 mm; the weld defect with the buried depth of 2 mm can be identifi ed effectively, and the location can be measured within an error of 0.5 mm. Keywords Time of fl ight diffraction Shallow buried defect Defect detection Defect location D. Chi (?) T. Gang State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin, 150001, P.R. China e-mail: 1 Introduction Ultrasonic time of fl ight diffraction (TOFD) has proved highly effective for the inspection of steel plates, tubular pipelines and tanks. The technique has a lot of advantages which make it the preferable non-destructive testing (NDT) method in material testing, especially for the detection and sizing of defects in thick wall weldment. Although TOFD has started to take its way to replace the other ultrasonic testing techniques, it suffers from several technological lim- itations. TOFDcanbeconductedeasilyfortestingthickwallcom- ponent with simple geometry, whereas, complex geometry and thin wall components are diffi cult to inspect. Through a manual ultrasonic TOFD inspection system, Nath S.K. et al. developed probability of detection (POD) and probabil- ity of sizing (POS) curves for the detection and sizing of sur- face breaking cracksin componentswith complexgeometry. Experiments show that the developed curves are useful for planning the risk-based inspection of components such as a steam turbine rotor shaft 1, 2. However, the scan proce- dures, calibration, data presentation and analysis processes involved are often as complex as the geometry itself. With miniature high frequency probes and shear wave TOFD, Baskaran G. et al. tried to overcome the limitation of thin sections. In these cases, the minimum thickness that could be examined was limited to 7 mm 3. Based on the combi- nation of TOFD and immersion technique, Subbaratnam R. et al. proposed a new method that extends the application of TOFD to thinner sections down to 3 mm 4. Although TOFD provides high speed inspection, high sizing reliability and low rate of false defect indications, the identifi cation and classifi cation of defects is still frequently questioned because it depends heavily on the knowledge and experience of the operator. Nowadays, even in state- J Nondestruct Eval (2013) 32:164171165 of-the-art TOFD scanning systems, the detection and siz- ing of a defect is an almost entirely manual process. In or- der to automatically interpret the data collected by TOFD, pattern recognition techniques, such as artifi cial neural net- work, are adopted in classifi cation of defect signals 5, 6. Digital image processing techniques are studied to automat- ically identify and locate the defects in the images collected by TOFD. Petcher P.A. and Dixon S. proposed a modifi ed Hough Transform method, with which the defect can be ex- tractedautomaticallyfromaTOFDB-scanimages7.Gang T. and Chi D.Z. presented an approach based on synthetic aperture focusing technique, with which TOFD B-scan im- ages can be enhanced and the defect tips can be located au- tomatically 8. The results obtained from the above studies are very promising and could give relevant contributions to the development of an automatic system of classifi cation, detection and localisation of welding defects inspected by TOFD technique. Lack of near surface resolution is by far the most com- monly quoted limitation of TOFD, and it has been substan- tiated by hard fact. An element of the beam in TOFD is de- tected which travels directly from the transmitter to the re- ceiver just beneath the material surface. This is referred to as “lateral wave”, which is a relatively short pulse and low amplitude signal occurring at a fi xed position. The fact that it is always present means that the very near surface will al- ways contain a signal which is often construed as a negative factor affecting near surface resolution. Moreover, there is an insensitive region beneath the inspection surface because of the restricted beam width of the dual probes. It means that SBD waves collected using TOFD are diffi cult to detect be- cause of the insuffi cient amplitude. Digital signal processing techniques are developed in some research works to over- come the limitation of the signals superposition. With the embedded signal identifi cation technique, Baskaran G. et al. developed an automated defect sizing algorithm by separat- ing partially superposed signals encountered in thin sections 9. Chi D.Z. et al. developed a background image removal algorithm, with which near surface defect signal in TOFD D-scan image can be separated from lateral wave effectively 10. The proposed methods are helpful in improving the identifi cation of SBD in TOFD mode, but they strongly de- pend on the degree of signal superposition. In order to detect SBD, this paper presents a new testing method, a TOFD based technique with a three-fold refl ected longitudinal wave employed. It is called TOFDW because the propagation path of the longitudinal wave in the testing piece looks like the letter “W”. Based on Pythagorass the- ory, simple mathematical models were developed to locate the defect in TOFDW mode. Artifi cial notches and weld de- fects were fabricated and tested using TOFDW, and the sen- sitivity and accuracy of the method were studied. (a) (b) Fig.1 TOFDWtestingmode.aArrangementtheprobesandthesound paths. b RF waveform 2 Principle of TOFDW Figure 1a shows the arrangement of TOFDW probes on a component with a point discontinuity. The corresponding A- scan line collected using TOFDW is shows in Fig. 1b. Five important sound paths from the transmitter to the receiver are described below. Path 1 is the lateral wave that travels justbeneaththeinspectionsurface.Path2isthedefectwave, with which the defect can be detected and sized by TOFD. When the defect is very shallow buried beneath the inspec- tion surface, the defect wave will superpose the lateral wave in TOFD mode, resulting in the so-called “dead zone”. Path 3 is the back-wall echo that refl ected by the bottom surface once. Path 5 is the three-fold refl ected wave, refl ected by the inspection surface once and the bottom surface twice. In Path 4, after being refl ected by the bottom surface, the lon- gitudinal wave propagates towards the inspection surface. Thentheone-foldrefl ectedwaveactswiththeSBD,forming a diffractive sound fi eld. After being refl ected by the bottom surface again, the defect wave fi nally reaches the receiver. In the A-scan line collected by TOFDW, Wave 1 and 4 are opposite to Wave 2, 3 and 5 in phase. The phase difference is helpful for the practitioner to identify defects in the data col- lected through TOFDW. Like the conventional TOFD mode, TOFDW can give 2-dimensional images called D and B- scan. 3 Defect location model According to the geometric relation between the defect tip and probes in TOFDW mode, simple mathematical models were developed to locate SBD. Pythagorass theory is ap- plied to calculate the buried depth of the defects, and the principle is described below in detail. In the models, both of the cases are taken into account, either the defect symmetri- cally and asymmetrically lies between the probes. 166J Nondestruct Eval (2013) 32:164171 (a) (b) Fig. 2 Locating model for the defect symmetrically between the probes. a Defect buried beneath the inspection surface. b Defect buried above the inspection surface 3.1 Defect symmetrically lies between the probes Assuming that the defect tip is centred between the dual probes, the buried depth of the defect can be measured ge- ometrically according to Pythagorass theory, as shown in Fig. 2a. O presents the point defect (or the defect tip), with the buried depth of d. In TOFDW mode, the sound path in which defect can be detected is ABOCD. Taking BC as the symmetryaxis, O?canbeobtainedasthesymmetricpointof O.Hence,thegeometricpathABO?CD hasthesamelength as path ABOCD. It looks like that O?is the true diffractive source. The “buried depth” of O?from the inspection sur- face can be calculated according to Pythagorass theory d?= 1 2 ? (TDC)2+4TDCS(1) where d?= the distance between O?and the inspection sur- face, TD= the arrival time difference between the defect wave and the lateral wave, C = the speed of ultrasonic lon- gitudinal wave, S = half of the probes centre separation. Then, the defect buried depth d can be calculated by d = 2hd?(2) where h = the thickness of the testing piece. It is the common case that the defects are buried beneath the inspection surface. However, when the defect lies in the weld crown, i.e. just in or above the inspection surface, sit- uations are changed. In this case, the defect can be located according to the model as shown in Fig. 2b. The calculated d?will undoubtedly be equal to or more than 2h. Then, Fig. 3 Locating model for the defect asymmetrically between the probes the buried depth of the defect calculated by Eq. (2) is non- positive. 3.2 Defect asymmetrically lies between the probes In general, the defect lies somewhere between the dual probes, but not necessarily midway between them. When the defect lies non-centred between the probes, B-scan can give the suggestion of the defect location. Unfortunately, B- scan testing is often unable to be carried out because the weld crown hinders the probes passing over the seam. In this case, the diffractive source can be located according to the geometric model shown in Fig. 3. The possible locations of the diffractive source are at cer- tain positions on the constant travelling time curve, which is composed of a half-ellipse with the two probes centres as foci. It is clear that some ambiguity exists in the measured depth with just two probes, although this error will be rela- tively small if the defect is close to the midway position. If another receiver is added, or the receiver is moved to another position, the ambiguity can be theoretically eliminated. A Cartesian coordinate is established in the defect con- taining cross-section. The x-axis lies in the inspection sur- face, and the y-axis is the perpendicular bisector of the weld seam. According to the y-axis, the probes are symmetrically placed at the position A and D respectively. For the hyper- bola L, all of its points are the constant distance of S1 S1= AO?+O?D(3) where S1can be calculated by S1= C(t1t0)(4) where, C = the speed of longitudinal wave, t1= the arrival time of defect wave, t0= the system delay time. Hence, the calculated S1 defi nes a half-ellipse, whose foci are the points A and D respectively. The major axis of the ellipse is 2a1= S1(5) The minor axis is 2b1= 2 ? a2 1 (AD/2)2(6) J Nondestruct Eval (2013) 32:164171167 Then the half-ellipse L is x2/a2 1+y 2/b2 1= 1 y 0(7) The transmitter remains unmoved at the position A, and the receiver is moved from the position D to D?. Hence, it de- fi nes another half-ellipse L?with the foci A and D?respec- tively, on which all the points are the constant distance of S2= AO?+O?D?. The major axis of L?is 2a2= S2(8) The minor axis is 2b2= 2 ? a2 2 ?AD?/2?2 (9) Then the half-ellipse L?is ?x DD?/2?2/a2 2+y 2/b2 2= 1 y 0(10) The intersecting point of the two half-ellipses L and L? would suggest the position of the defect tip, assuming there were no errors in the model. The solution O?(x1,y1) can be obtained from simultaneous Eqs. (7) and (10). x1is the lateral position of the defect tip in the coordinate, and the buried depth of the defect tip is y = 2h|y1|(11) 4 Artifi cial defects testing By means of electric discharge machining (EDM), six artifi - cial defects (bottom surface notches with length of 15.0 mm and width of 0.2 mm) were fabricated in a weld-free alu- minium plate (20.0 mm in thickness). The artifi cial defect block and TOFDW scanning directions are shown in Fig. 4. The buried depth values of the notch tips are shown in Table 1. The artifi cial defects were tested using dual high band- width probes (the crystal is 6 mm in diameter, the cen- tral frequency 5 MHz and the refraction angle 60in alu- minium). The TOFDW testing parameters are as follows: 2S = 120 mm, system gain 75 dB and scanning step length 0.2 mm. The D-scan images of Artifi cial Defect 3 and 4 are shown in the left part of Fig. 5, and the correspond- ing A-scan line of Artifi cial Defect 4 is shown in the right part. In the conventional TOFD mode, the defect detection re- gion lies between the lateral wave and the back-wall echo in the time domain. It can be seen that a fraction of lateral wave goes missing on the upper part of the D-scan image. This is the result of superposition of the defect wave and the lateral wave which are opposite in phase. The greater the value 2S, the more severe the waves would be superposed. In TOFDW mode, the back-wall echo and the three-fold refl ected wave are very important reference for defect detec- tion because the defect wave arrives between them. In addi- tion, the lateral wave is essential for defect locating calcula- tion. In the defect detection region, the defect stripe, with a regular texture feature, looks like a straight line. This feature appears corresponding to the geometric shape of the notch tip. The three-fold refl ected wave shows up as a continuous Table 1 Buried depth of the artifi cial defect tips Number of defect 123456 Buried depth d/mm Fig. 4 Artifi cial defect block and TOFDW test mode Fig. 5 TOFDW data of the artifi cial defects 168J Nondestruct Eval (2013) 32:164171 global component in the image if there is no discontinuity in the inspection surface of the block. It can be observed from the image that the phases of the lateral wave and the defect wave are negative, whereas those of the back-wall echo and the three-fold refl ected wave are positive. The defects can be observed clearly in the D-scan image and A-scan line be- cause they are sharp contrast to the background. For exam- ple, the wave amplitude of Artifi cial Defect 4 is nearly 80 % of full screen high (FSH), as shown in the A-scan line. The timedifferencebetweenthelateralwaveandthedefectwave isenlargedgreatly,whichmakestheSBDwavefreefromthe superposition of the lateral wave. Therefore, TOFDW mode can effectively improve near surface resolution. The artifi cial defects were measured with the model de- scribed in Sect. 3.1, with only the situation of notch tips ly- ing symmetrically between the probes considered. Accord- ing to Eqs. (1) and (2), the buried depth of the defect tips were calculated. The defects were tested three times, and mean buried depth values of the tips were obtained. The measured results are shown in Fig. 6. With the defect lo- cation model, the defect tips can be located with an error of less than 0.3 mm. Fig. 6 Measured result of the artifi cial defects 5 Weld defects testing Twenty weld pieces (aluminium alloy with thickness of 1820 mm) were made and tested through TOFDW mode, among which four SBD with buried depth of 2 8 mm were detected. The D-scan image and the corresponding A-scan line of Weld Defect 1 are shown in Fig. 7. In the conventional TOFD test region, disturbance of the lateral wave can be observed in the upper right part of the D-scan image. It shows that there may be continuous SBD in the weld piece. Unfortunately, it is diffi cult to measure the defect because of the interference of the lateral wave. In TOFDW test region, the defect stripe can be observed clearly in the lower right part of the D-scan image. The de- fect image shows the signature of signifi cant length along the weld (about 31 mm in length along the weld) but little through-wall size, from which the defect is considered to be a thread-like discontinuity. The defect was measured with the model described in Sect. 3.2. Five measurements were taken, and a mean value was obtained. The measured results of the defect are shown in Fig. 8. The calculated location of the defect is 7.5 mm from the inspection surface and 3.0 mm from the centreline of the weld seam. Destructive testing was carried out, and Fig. 8 Measured result of Weld Defect 1 Fig. 7 TOFDW data of Weld Defect 1 J Nondestruct Eval (2013) 32:164171169 the result shows that the defect is a continuous sidewall lack of fusion, as shown in Fig. 9. From the weld cross-section, the measured vertical height of the defect is 1.4 mm, and the centre of the defect is 7.7 mm from the inspection plane and 3.5 mm from the