Inspection of thick welded joints using laser-ultrasonic SAFT.pdf

Inspection of thick welded joints using laser-ultrasonic SAFT D. Lvesque a, Y. Asaumib, M. Lorda, C. Besconda, H. Hatanakab, M. Tagamib, J.-P. Monchalina aNational Research Council Canada, 75 de Mortagne Blvd., Boucherville, Quebec J4B 6Y4, Canada bIHI Corporation, 1 Shin-nakahara-cho, Isogo-ku, Yokohama 235-8501, Japan a r t i c l ei n f o Article history: Received 18 November 2015 Received in revised form 1 April 2016 Accepted 1 April 2016 Available online 2 April 2016 Keywords: Thick weld inspection Laser ultrasonics Synthetic aperture focusing technique a b s t r a c t The detection of defects in thick butt joints in the early phase of multi-pass arc welding would be very valuable to reduce cost and time in the necessity of reworking. As a non-contact method, the laser- ultrasonic technique (LUT) has the potential for the automated inspection of welds, ultimately online during manufacturing. In this study, testing has been carried out using LUT combined with the synthetic aperture focusing technique (SAFT) on 25 and 50 mm thick butt welded joints of steel both completed and partially welded. EDM slits of 2 or 3 mm height were inserted at different depths in the multi- pass welding process to simulate a lack of fusion. Line scans transverse to the weld are performed with the generation and detection laser spots superimposed directly on the surface of the weld bead. A CCD line camera is used to simultaneously acquire the surface profi le for correction in the SAFT processing. All artifi cial defects but also real defects are visualized in the investigated thick butt weld specimens, either completed or partially welded after a given number of passes. The results obtained clearly show the potential of using the LUT with SAFT for the automated inspection of arc welds or hybrid laser-arc welds during manufacturing. Crown Copyright ? 2016 Published by Elsevier B.V. All rights reserved. 1. Introduction Arc welding thick metal plates in a butt joint confi guration requires in practice several passes to fi ll completely the gap between the plates. Defects, like porosity and lack of fusion, may occur at any depth during the welding process. Ultimately, the detection and visualization of such defects in the early phase after a small number of passes may reduce cost and time in the neces- sity of reworking. As a non-contact method, the laser-ultrasonic technique (LUT) has the potential for the automated inspection of welds, ultimately online during manufacturing. LUT uses lasers for the generation and detection of ultrasound 1. It has the capa- bility for measurement at high temperature and the advantage of broadband ultrasound generation and detection, as well as easily modifi able laser spots to achieve adequate resolution 2. More- over, LUT can be combined with the synthetic aperture focusing technique (SAFT) to allow detection of small discontinuities such as cracks, inclusions and porosities 36. Recently, LUT combined with SAFT was investigated for the defect detection in friction stir welds (FSW) made of aluminum sheets of thickness less than 3 mm thick in the lap joint and butt joint confi gurations 7,8. More recently, the technique was employed to examine the possibility of inspecting internal fl aws located within the 9 mm thick gauge section of the hybrid laser- arc welded (HLAW) steel plates 9. Preliminary results indicated that the LUT-SAFT inspection can successfully detect and visualize the presence of porosity, lack of fusion and internal crack defects. However in this study, the weld root back surface was used for inspection with the basic SAFT technique with generation and detection superimposed. More recently, LUT with a modifi ed ver- sion of SAFT (m-SAFT) was employed to inspect 150 mm thick welded pipe during welding process with temperature of more than 200 ?C 10. Ultrasound generation was performed on the weld bead while scanning the detection in the transverse direction away from the weld on the parent materials surface. By using m- SAFT, an actual weld defect of 1.5 mm in diameter at 106 mm depth in the specimen was observed. However, this approach relies on detecting small diffracted signals far from the defect and the depth of the weld bead at the generation point needs to be known for SAFT reconstruction. In this work, testing is carried out using LUT still with the SAFT approach on 25 and 50 mm thick butt welded joints of steel both completed and partially welded. Line scans transverse to the weld are performed with the generation and detection superimposed directly on the surface of the weld bead. A CCD line camera is used to simultaneously acquire the surface profi le for correction in the SAFT processing. With generation and detection superimposed, /10.1016/j.ultras.2016.04.001 0041-624X/Crown Copyright ? 2016 Published by Elsevier B.V. All rights reserved. Corresponding author. E-mail address: daniel.levesquecnrc-nrc.gc.ca (D. Lvesque). Ultrasonics 69 (2016) 236242 Contents lists available at ScienceDirect Ultrasonics journal homepage: there is no need of surface preparation even in the presence of oxide. EDM slits are inserted at different depths in the multi-pass welding process to simulate a lack of fusion. The specimens and the laser-ultrasonic setup used are fi rst described along with the SAFT approach. Then the results are presented and discussed. All artifi cial defects but also real defects are visualized in the investi- gated thick butt weld specimens, either completed or partially welded after a given number of passes. 2. Specimens and inspection technique Four thick butt welded joints of steel, with thicknesses T = 25 and 50 mm, both complete and partially welded (underway), were prepared. In each specimen, EDM slits of 2 or 3 mm height were inserted at different locations and depths in the multi-pass weld- ing process to simulate a lack of fusion. Figs. 1 and 2 show sche- matic diagrams of the weld specimens with the identifi cation number for each slit. The slits have the height and length dimen- sions indicated on the fi gures. Fig. 3 shows photos of all four weld specimens. It is noted that for thick specimens the multi-pass welding process may involve fi lling the V-grooves present on both sides. In these situations, access to one or two sides for inspection may be acceptable or desirable. Laser-ultrasonic setup used for non-contact off-line inspection of the test specimens was made with the intent of future online implementation in the process. Line scans transverse to the weld was performed with the generation and detection superimposed directly on the surface of the weld bead. Ultrasound generation was made in the slight ablation regime (less than 1lm) with a short pulse Nd:YAG laser in its 2nd harmonic (532 nm wavelength, 6 ns pulse duration) to achieve high frequencies. For detection, a long pulse Nd:YAG laser (1064 nm wavelength, 60ls pulse dura- tion) and a small spot size of about 0.5 mm were used. The phase demodulator was a 1-m long confocal Fabry-Perot interferometer in refl ection mode. Frequency content up to 30 MHz in steel was successfully generated and detected in the above test specimens. Also it is noted that with generation and detection superimposed, there is no need of any surface preparation even in the presence of oxide. Mechanical scanning along single line up to 100 mm long was performed for data acquisition of the waveforms with a step size of 0.25 mm. Also, a CCD line camera was used to simultane- ously acquire the surface profi le for correction in the SAFT processing. For numerical focusing, SAFT processing is used for synchro- nization of the ultrasonic signals scattered back in different direc- tions from each point in the weld region. While maintaining the depth resolution, SAFT reconstruction improves the lateral resolu- tion as well as the signal-to-noise ratio (SNR) 4. The depth and lateral resolutions are estimated using: Dx ? z avDt ? v 2tghfmax Dz ? 1 2vDt ? v 4fmax 1 wherevis the ultrasonic wave velocity,Dt is the ultrasonic pulse duration, a is the dimension of the synthetic aperture, h is the aper- ture angle from surface normal and fmaxis the maximum frequency detected. To evaluate the SNR, an amplitude profi le is extracted from the B-scan image and the SNR is evaluated thereafter using: SNR 20log10 A ?l?r 2r ? 2 where A is the maximum amplitude at the defect location,lis the background level andris the noise standard deviation near the defect. When scanning on the weld bead, a SAFT algorithm that accounts for the surface variations on an irregular surface should be used 5. For synchronization in the time-domain SAFT version, the time-of-fl ight of a signal acquired at location (xi, zi) for a defect to be imaged at location (x, z) is simply: ti 2 ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffiffi x ? xi2 z ? zi2 q =vL3 where zi 0 stands for the surface height at the generationdetec- tion point with reference to the parent materials surface at z = 0, andvLis the velocity of the longitudinal mode. SAFT reconstruction Fig. 1. Schematic diagram of the 25 mm thick weld specimens, (a) complete and (b) underway. Dimensions are in mm. D. Lvesque et al./Ultrasonics 69 (2016) 236242237 Fig. 2. Schematic diagram of the 50 mm thick weld specimens, (a) complete and (b) underway. Dimensions are in mm. Fig. 3. Photos of the weld specimens complete and underway of thicknesses (a) and (b) 25 mm, (c) and (d) 50 mm. 238D. Lvesque et al./Ultrasonics 69 (2016) 236242 was performed with an aperture angle of 30? and a longitudinal wave velocity of 6.0 mm/ls. With the above setup, the spatial res- olution is estimated to be better than 0.2 mm. Fig. 4 shows the performance of SAFT reconstruction for a single line scan over 100 mm on a 50 mm thick reference block of steel with three side-drilled holes one above the other at different depths on a vertical line. The synthetic aperture with such angular range allows reconstruction of the top of each hole to be recon- structed from the hyperbolas observed in the raw signals. SAFT processing clearly improves the SNR for defect detection and reso- lution for defect sizing. From top to bottom, the SNR for the 3 holes after SAFT reconstruction are 34, 27 and 24 dB, respectively. 3. Laser-ultrasonic inspection results The thick butt welded joints described in previous section, with thicknesses T = 25 and 50 mm, both complete and partially welded (underway), were tested with the LUT-SAFT approach. In each case, a few line scans transverse to the weld are performed directly on the surface of the weld bead in the region where EDM slits are expected to be present. 3.1. Specimens of thickness T = 25 mm Fig. 5 shows the top surface profi le acquired using the CCD line camera during laser-ultrasonic testing of the 25 mm thick butt welded joint complete. Such surface profi le is used for correction in the SAFT processing using Eq. (3). As an illustration, Fig. 6 shows the B-scan images on the weld specimen complete without correc- tion for the top profi le for a line scan over 50 mm. While the slit 32C1 is detected, no improvement in resolution is obtained after SAFT reconstruction, in addition to a prominent indication on the back wall partly due to the curvature on top surface. Fig. 7 shows the results after correction with slits 32C1 and 32D1 well detected and visualized with a SNR of 21 and 13 dB respectively, at a depth of about 11 mm and 22 mm from the fl at surface on top (see Fig. 1a). Not shown here, repeatability was demonstrated with similar results found for slits 32C2 and 32D2. Regarding the weld specimen T = 25 mm underway, Fig. 8 shows the B-scan images with correction for the top profi le and a line scan of 13 mm for both slits 42C1 and 42D1, respectively at a depth of about 11 mm and 22 mm with respect to the fl at surface on top (see Fig. 1b). Simulating a lack of fusion, both EDM slits 42C1 and 42D1 are clearly visualized with a SNR of 16 and 34 dB respectively, using LUT-SAFT testing directly on the weld bead. Not shown here, repeatability was demonstrated with similar results found for slits 42C2 and 42D2. These results show the potential for online inspec- tion during welding after a given number of passes. 3.2. Specimens of thickness T = 50 mm Fig. 9 shows the surface profi les acquired using the CCD line camera during laser-ultrasonic testing on top and bottom of the 50 mm thick butt welded joint complete. Again it is noted that for such thick specimen, inspection from either side may be acceptable or desirable since the multi-pass welding process involves two V-grooves. Again such surface profi les are used for correction in the SAFT processing using Eq. (3). Fig. 10 shows the results after correction over a line of 50 mm for both slit 5E1 and 5F1 with a SNR of 17 and 20 dB respectively, at a depth of about 22 mm and 35 mm with respect to the fl at surface on top (see Fig. 2a). In Fig. 10a, a spurious linear indication is present near the back wall which is attributable to refl ection of surface wave on top of the weld specimen of fi nite dimensions. Fig. 11 also shows the results from testing on the back surface visualizing both edges of the EDM slit 5F1 with a SNR of 21 dB, as well as a real defect about 7 mm along the weld away from the center of the slit on the upper left of the B-scan image in Fig. 11b. Regarding the weld specimen T = 50 mm underway, Fig. 12 shows the B-scan images with correction for the top profi le for a line scan of 13 mm for both slit 62E1 and 62F1 respectively at a depth of about 22 mm and 35 mm with respect to the fl at surface on top (see Fig. 2b). Simulating a lack of fusion, both EDM slits 62E1 and 62F1 are clearly visualized with a SNR of 21 and 20 dB respectively, using LUT-SAFT testing directly on the weld bead. This is in addition to a real defect present on top of the B-scan image in Fig. 12b. Not shown here, the slit 62F1 can also be visual- ized from inspection on the back surface. Again, repeatability was demonstrated with the similar results found for slits 62E2 and 62F2. These results show the potential for online inspection during welding after a given number of passes. Fig. 4. B-scans on a 50 mm thick reference block of steel with three side-drilled holes, (a) raw data and (b) after SAFT reconstruction. Fig. 5. Top surface profi le for weld specimen T = 25 mm complete. D. Lvesque et al./Ultrasonics 69 (2016) 236242239 Fig. 8. B-scans on weld specimen T = 25 mm underway over 13 mm after SAFT reconstruction with correction for top profi le at (a) slit 42C1 (nominal depth of 11 mm from top) and (b) slit 42D1 (nominal depth of 22 mm from top). Fig. 9. Surface profi le for weld specimen T = 50 mm complete for inspection from (a) top and (b) bottom. Fig. 7. B-scans on weld specimen T = 25 mm complete over 50 mm after SAFT reconstruction with correction for top profi le at (a) slit 32C1 (nominal depth of 11 mm) and (b) slit 32D1 (nominal depth of 22 mm). Fig. 6. B-scans on weld specimen T = 25 mm complete over 50 mm at slit 32C1 (nominal depth of 11 mm from top); (a) raw data and (b) after SAFT reconstruction without correction for top profi le. 240D. Lvesque et al./Ultrasonics 69 (2016) 236242 4. Conclusion Laser-ultrasonic inspection combined with SAFT was performed on butt weld specimens of thicknesses up to 50 mm. Line scans transverse to the weld are performed with the generation and detection laser spots superimposed directly on the surface of the weld bead without any surface preparation. A CCD line camera is used to simultaneously acquire the surface profi le for correction in the SAFT processing. EDM slits of 2 or 3 mm height were inserted at different locations and depths in the multi-pass welding process to simulate a lack of fusion. All artifi cial defects but also real defects are Fig. 10. B-scans on weld specimen T = 50 mm complete over 50 mm after SAFT reconstruction with correction for profi le at (a) slit 5E1 (nominal depth of 22 mm) and (b) slit 5F1 (nominal depth of 35 mm). Fig. 11. B-scans on weld specimen T = 50 mm complete from back surface over 50 mm after SAFT reconstruction with correction at (a) center of slit 5F1 and (b) 7 mm beyond along the weld. Fig. 12. B-scans on weld specimen T = 50 mm underway over 13 mm after SAFT reconstruction with correction for top profi le at (a) slit 62E1 (nominal depth of 22 mm from top) and (b) slit 62F1 (nominal depth of 35 mm from top). D. Lvesque et al./Ultrasonics 69 (2016) 236242241 visualized in the investigated thick butt weld specimens, either completed or partially welded after a given number of passes. The results obtained clearly show the potential of using the LUT with SAFT for the automated inspection of arc welds or hybrid laser-arc welds during manufacturing. A simple calculation shows that a 1 kHz laser repetition rate would allow real time inspection at a speed of typically 1 cm/s along the weld. References 1 C.B. Scruby, L.E. Drain, Laser ultrasonics: Techniques and Applications, Adam Hilger, Bristol, 1990. 2 J.-P. Monchalin, Laser-ultrasonics: from the labo