Defect Detection in Thick Weld Structure Using Welding In-ProcessLaser Ultrasonic Testing System

Setsu Yamamoto1,+, Takeshi Hoshi1, Takahiro Miura1, Jun Semboshi1,Makoto Ochiai1, Yoshihiro Fujita2, Tsuyoshi Ogawa2 and Satoru Asai2

1Power and Industrial Systems Research and Development Center, Toshiba Corporation, Yokohama 235-8523, Japan2Keihin Product Operations, Toshiba Corporation, Yokohama 230-0045, Japan

A new approach of non-destructive testing for thick welded structural materials based on laser-ultrasonic technique is investigated. In thisstudy, weld part of structural materials, which should be conventionally inspected after welding, is inspected during welding process in order tosave time and cost of manufacturing. The laser-ultrasonic is a method to generate and detect ultrasonic signals by laser beams and has potentialto be applied to remote inspection/monitoring of materials under welding at elevated temperature. Bulk longitudinal acoustic wave generated bya Q-switched Nd:YAG laser irradiation and detected as surface vibration by laser interferometer coupled with a long pulse detection laser is usedto detect defects around the weld. To overcome the lack of sensitivity of laser-ultrasonic testing on thick welded part having a thickness of morethan 100mm at higher temperature, we have originally developed a modified synthesis aperture focus signal processing technique (m-SAFT).The in-process testing with actual piping weld having a thickness of 150mm with high temperature more than 200 degrees C. was demonstrated.By using m-SAFT, an actual weld defect of 1.5mm in diameter at 106mm depth in the specimen was clearly observed. The measurement resultwell agreed with the result of conventional ultrasonic testing conducted after weld process and also the cross-sectional observation of thespecimen. [doi:10.2320/matertrans.I-M2014809]

(Received August 12, 2013; Accepted January 23, 2014; Published April 18, 2014)

Keywords: lase ultrasonic testing, thick welding, welding in-process, synthesis aperture focus technique (SAFT)

1. Introduction

Laser-ultrasonic testing (LUT) that uses lasers as analternate use of conventional contact transducers has beenestablished.1­3) Thermal expansion or ablation induced byshort pulse laser irradiation to target surface are used astransmitters. As receivers, laser Doppler vibration measure-ment is used. This combination was originally limited fornon-contact measurement in laboratory due to low sensitivity.A long pulsed laser with enhanced peak power and aninterferometer applicable to rough surface (ex. Fabry-Perotinterferometer) have brought LUT into industrial applica-tion.4­6) LUT has two significant advantages over conven-tional ultrasonic testing (UT) with contact piezoelectrictransducers. First one is applicability to inspection at limitednarrow space. To inspect a target, LUT requires only laserirradiation area that sizes are about º1mm. Using thatadvantage, authors have developed surface inspectiontechnique using surface acoustic waves which can detectand measure cracks at narrow space. The technique has beenapplied to structure inspection of inside of the narrow tubewith º10mm internal diameter in nuclear power plants.7­9)

The other LUT advantage is applicability to inspection ofhigh temperature target. LUT has been adopted as a thicknessmeasurement method in a process of manufacturing steelsor pipes.6) Furthermore, LUT has been applied to weldinspection. However, its use is limited only to thin welding,e.g., spot welding, buildup welding, etc.10,11) because of lackof sensitivity. Weld defects such as blowholes, incompletefusions, slag inclusion, detection technique for targets thickerthan 100mm has not been established.

On the other hand, in the manufacturing domain,remarkable progress of welding technique enables thick

structure fabrication.12) However, welding thick structureshas high probability of defect generation. A schematicdiagram of defect generation in thick welding is shown inFig. 1. Figures 1(a), 1(b) and 1(c) show normally weldedarea of thick welding at a narrow U-shaped groove, a defectimage of blowhole and an incomplete fusion respectively. Ifthese defects were detected after weld, even full re-welding,rather than partial repairing, would be required at worst. Re-working cost would increase depending on the thickness ofthe welded area. Thus, the inspection technique which candetect defects at early phase and apply to thick weld parts hasbeen strongly required.

Filler metalwire

Tungstentorch

Penetration bead

Arc U-Shapedgroove

Welded metal

Base metal

BlowholeIncompletefusion

(a)

(b) (c)

Fig. 1 Schematic diagrams of the thick welding. (a) Normally welded area.(b) Blowhole image. (c) Incomplete fusion image.

+Corresponding author, E-mail: setsu.yamamoto@toshiba.co.jp

Materials Transactions, Vol. 55, No. 7 (2014) pp. 998 to 1002Special Issue on Laser Ultrasonics and Advanced Sensing 2013©2014 The Japanese Society for Non-Destructive Inspection

Thus authors have been proposing welding in-processtesting technique.13) In-process testing means conductingwelding and testing simultaneously. Using in-processtesting technique, defects can be detected at early phase inwelding. To apply the in-process testing to thick target,synthesis aperture focusing technique (SAFT)14,15) have beenadopted for improving the signal to noise ratio (SNR) ofLUT.

In this study, principles of in-process testing will be offeredand experimental results of defect detection which wasconducted using 150mm thick specimen at over 200 degreesC. will be also shown.

2. Principles of In-Process Testing

2.1 Experimental setupsFigure 2 shows a schematic diagram of an in-process

testing system. A welding machine for thick welding andlasers are depicted. LUT inspection area was moving behindthe welding torch. Giant pulsed Nd:YAG laser (­ = 1064 nm)was used as a generation laser that generates ultrasonicwaves. The pulse width was shorter than 10 ns and the energywas 700mJ/pulse. Long pulsed Nd:YAG laser (­ = 1064nm) was adopted as a detection laser. The pulse width wasabout 100 µs and the energy was 15mJ/pulse. The detectionlaser was transmitted by optical fibers and irradiated usingoptical probe which could conduct irradiation and receptionwith a same object lens. A scanning stage which connected toan optical probe enabled multi point signal detection. Forultrasonic signal extraction, confocal Fabry-Perot interfer-ometer (CFPI) was adopted.1,2) Combination of a long pulsedlaser with CFPI gave high sensitivity and was applicable torough surfaces. Obtained signals were recorded and used forsignal processing by a personal computer.

Irradiated generation laser induced ablation at the targetsurface and ultrasonic waves were generated. The generatedultrasonic wave, especially bulk wave, propagated inside thewelded area and diffracted from defects. Diffracted waveswere detected at surface points which were also detectionlaser irradiation points.

2.2 Principle of signal processingEcho intensity which was diffracted from defects was

inferior to that diffracted from the bottom for use in thickmeasurement. Simple application of conventional LUT is not

sufficient to achieve required detectability. Several improve-ment methods for SNR of LUT have been proposed, forexample, averaging, frequency filtering and SAFT.14,15) Toreduce effects of mode conversion echo and enhance spatialresolution, modified-SAFT was adopted in this system.13)

Basic SAFT technique for LUT used scanned signals whosegeneration and detection point was the same. In this system,the generation and detection points were separated like thetime of flight diffraction (TOFD).16,17) The generation pointwould be defined flexibly. For example, surface of matrix andwelding bead. In this paper the surface of welding bead wasdefined as the generation point. It could excite ultrasonic ata defect which was located just below the bead surface.The detection point was set away from the welding beadwhere detection laser reflection at the surface was likely to bestable.

In this paragraph, the coordinates for our SAFT system isdefined. Scanning direction of detection laser is x-axis whichis perpendicular to the welding direction. The weldingdirection is defined as y-axis and depth direction is definedas z-axis, respectively. Figure 3 shows coordinates for ourSAFT system. Illustrated horizontal axis means orthogonaldirection of welding and vertical axis means depth direction.The original position of the x-axis is the groove center, y isthe welding start point and z = 0 is the surface of the target.

The generation laser irradiation point is T ðXt; ZtÞ, thedetection laser irradiation point is RiðXi; ZiÞ (i = 1, 2, 3,+,n), where i means the index of the detection point. In thissystem, the irradiation point of the generation laser is fixed atthe welding bead surface, meaning that T ð0; ZtÞ ¼ T ðZtÞ.And the detection laser is irradiated only on the target surfacegiving, in the same analogy, RiðXi; 0Þ ¼ RiðXiÞ. If thecoordinates of the meshes assumed in the xz section, wherethe ultrasonic wave propagates, is written by ðX;ZÞ, thepropagation time from the generation point T, to the detectionpoint Ri, is given by

ti ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiX2 þ ðZ� ZtÞ2

pþ

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðX�XiÞ2 þ Z2

pVL

ð1Þ

Here, VL is the velocity of the longitudinal wave in the testingtarget. The received signal at Ri is siðtÞ. Using eq. (1), signalintensities of each mesh are given

Weldingtorch

Conveyingdirection

Welding machineConfocal Fabry-Perotinterferometer

Optical fibers

Personal computer-control-acquisition-signal processing

Detection laser

Scanning stage

Opticalprobe

Arc

45° mirror

Generation laser

Fig. 2 Experimental setup of an in-process testing system.

),( tt ZXT Ri RnR1

zDefect

),( ii ZXRi

Ri

Rn

R1

x

Fig. 3 Coordinates of SAFT system. Horizontal axis means orthogonaldirection of welding and vertical axis means depth direction.

Defect Detection in Thick Weld Structure Using Welding In-Process Laser Ultrasonic Testing System 999

UiðX;ZÞ ¼Z tiþ�t

ti

SiðtÞdt: ð2Þ

Here, ¦t is the width of the time window for signal intensityextraction from siðtÞ. According to eq. (2), an echo from ascattering source is plotted as an ellipse of which the focalpoints are T and Ri. In Fig. 3, ellipse with solid line meansintensity plot of a siðtÞ. Applying eq. (2) to all siðtÞ (i = 1, 2,3,+ , n), conclusive summation is given

UðX;ZÞ ¼Xni¼1

UiðX;ZÞ: ð3Þ

According to eq. (3), all signals are plotted and each ellipse iscrossed at a point of scattering source. In Fig. 3, ellipse withdotted line means intensity plot of s1ðtÞ and dashed lineshows intensity plot of snðtÞ. Signal intensity of mesh isstrengthened at the intersection and weakened at other area.So, the localized signal shows the location and the size ofdefect.

Here, right-hand side of eq. (3) is equal to the phasematching summation where, according to each delay time ¸i,signals siðtÞ are shifted and summed at each XZ position.Using variable transformation, t ¼ ti þ ¸ and dt ¼ d¸, andconsidering the integral range, eq. (2) is transformed to

UiðX;ZÞ ¼Z �t

0

Sið¸ þ tiÞd¸ ð4Þ

Substituting eq. (4) into the eq. (3)

UðX;ZÞ ¼Z �t

0

Xni¼1

Sið¸ þ tiÞd¸ ð5Þ

is obtained. SAFT result at xz plane given by eq. (3) or (5) isnamed B-Scope. By obtaining B-Scope along y-axis on thewelding groove, welded area is three-dimensionally visual-ized. SAFT of yz region is named D-Scope.

2.3 Inspection arrangement at welding in-processFigure 4 means inspection arrangement at welding in-

process. Here, welding target was thick walled pipes and twopipes were welded by butt welding. Material of the pipes wasCrMoV steel and the thickness was 150mm. Figure 4(a)shows a cross-section image of the welded area at the originalpoint of axial direction of pipes. Figure 4(b) is a cross-sectional diagram of the welded area at the original pointof circumferential direction and shows an arrangement ofinspection. Figures 4(a) and 4(b) mean midstream of the buttwelding. The groove was U-shaped narrow and the gap wasnarrower than 10mm. To conduct welding, the torch positionwas fixed and the pipes were rotated. The temperature of thetarget was above 300 degrees C. so the applied wave velocity(5750m/s) was measured using well-known bottom echopropagation time. Here, the original position of y was thewelding start point. Generation and detection laser werelocated about y = 240 degrees. Scanning the detection laseralong the axial direction, B-Scope image was obtained. Herehorizontal axis, x, represents the axial direction and thevertical axis, z, was the depth direction. In a D-Scope image,horizontal axis, y, is in the circumferential direction andvertical axis, z, the depth direction. Generation laser wasirradiated at the surface of bead and detection laser was

scanned from x = 25mm to x = 40mm at 0.1mm pitch. Thenumber of the detection points was 150 and all waveformsobtained were used for SAFT processing.

3. Experimental Results

Using above condition, LUT was conducted at welding in-process. To confirm a detectability of this system, weldingcondition was inappropriately selected. Figure 5 showstesting result of the defect which was located at the deepestarea of all detected defects. Figure 5(a) is D-Scope; Fig. 5(b)is B-Scope. Significant echo was observed at 106mm indepth; 170 degree. At this time, bead surface was about95mm in depth, so it means that defect was detected at11mm below from the welding surface. Intensity analysisshowed obvious echo from the defect. Figure 5 shows arelationship between the equivalent diameter and the signalintensity of LUT. The black marks represent measured echovalue and the line is approximated sensitivity. Comparingdefect echo intensity of Fig. 5 with sensitivity line of Fig. 6,defect size was evaluated as about º1.0mm.

10

(b)

25

55

150

Bottom

Generation laser Detection laser

15

....

Axial direction: xDepth direction: z

[mm]

(a)Welding torch

Welded layer

RotationdirectionGeneration

laserDetection laser

zy

Fig. 4 Measurement arrangement of thick weld pipe. (a) Cross-sectionalimage of the welded area at original point of axial direction. (b) Cross-sectional image of the welded area at original point of circumferentialdirection. Generation and detection laser were located about y = 240degrees. Detection laser was scanned from x = 25mm to x = 40mm at0.1mm pitch.

S. Yamamoto et al.1000

4. Verification of Inspection Result

4.1 Result comparison with conventional UTFirstly, to verify the detectability of in-process testing,

the result was compared with result of conventional after-welding UT. Inspection method was TOFD which used 2piezoelectric probes.17) The two probes were placed at the

opposite regions of target across the weld area. The probeswere moved along the circumferential direction with keepingrelative position. This inspection was conducted afterwelding so the target was fully welded and cooled to theroom temperature (20 degrees C.). Figure 7 presentsmeasurement results of D-Scope. Figure 7(a) is result ofLUT at in-process testing and Fig. 7(b) is result of conven-tional UT after welding. Obvious echo was observed in eachresult and the locations of the defect so measured were almostthe same. Especially in TOFD result, locus curve was clearlyobserved. As a result of signal intensity analysis, estimatedequivalent diameter was º1.0mm at in-process testing andº1.1mm at after welding testing, respectively.

4.2 Result comparison with cross-sectional observationSecondly, verification of in-process testing was conducted

by comparison with cross-sectional macroscopic observationof the welded area. The welded target was cut at 175 degreesof circumferential direction. After cutting, the specimen waspolished and etched. Photos of the specimen are shown inFig. 8. Figure 8(a) is the overall view and Fig. 8(b) is adetailed view of the specimen. The horizontal axis means

155 205

105

115

95

85

180

Bead surface

Dep

th d

irec

tion:

z [

mm

]

Circumferential direction: y [deg]

Defect

105

115

95

85

Dep

th d

irec

tion:

z [

mm

]

Axial direction:x [mm]

-20 0 20Intensity [A.U.]

0 100

Defect echo

Defect

Bead surface

(a) (b) (c)

Fig. 5 Measurement result of a blowhole in thick weld. (a) D-Scope image. (b) B-Scope image. (c) Intensity analysis of B-Scope image.Solid lines in each figure mean welding bead surface of the specimen.

0.0 0.5 1.0 1.5 2.0

20

40

60

80

100

120

140

0

Equivalent diameter [mm]

Sign

al in

tens

ity [

A. U

.]

Fig. 6 Relationship between equivalent diameter and signal intensity. Theblack marks represent measured echo value and the line is approximatedsensitivity.

155 205

105

115

95

85

180

Bead Surface

Dep

th d

irec

tion:

z [

mm

]

Circumferentialdirection: y [°]

Defect (φ 1.0mm)

155 205180Circumferentialdirection: y [°]

Defect (φ 1.1mm)

105

115

95

85

Dep

th d

irec

tion:

z [

mm

]

(a) (b)

Fig. 7 Measurement results of D-Scope. (a) Result of LUT at in-process testing. (b) Result of conventional UT at after-welding. (a) and (b)show same region of the specimen.

Defect Detection in Thick Weld Structure Using Welding In-Process Laser Ultrasonic Testing System 1001

axial direction, x, of the pipes and the vertical axis means thedepth direction, z. Blowhole was observed at 105.5mm indepth and the size was almost º1.5mm. the observed locationand size were almost the same as in-process and after weldingtesting.

The actual blowhole size was slightly bigger than in-process measurement. The scale between the equivalentdiameter and echo intensity was obtained by measurementsusing flat bottom holes. However, an actual defect has aspherical shape; so echo from an actual defect was highlylikely to be smaller than that from flat bottom holes, giving apossibility of underestimation.

5. Conclusions

In this paper, in-process testing technique which candrastically save a re-work cost at manufacturing process wasreported. Sensitivity of LUT is inferior to conventional UT.However, this problem was improved by using SAFTprocessing, and an in-process testing system was developed.The system is applicable to 150mm thickness welded areaand over 200 degrees C. target. Using the system, abilityconfirmation test was conducted under the in-processcondition and º1.5mm blowhole which was smaller thanoriginal criteria (º1.6mm) was successfully detected.

This technique is applicable to general purpose ofmanufacturing process and can save a large amount of re-work costs. Furthermore it ultimately plays an important roleat fully automated welding sequence which requires weld-test-analyze-repair cycles. In the future, improving quantita-tive defect evaluation and automated defect screeningtechniques, the use of this system will be more expanded.

Acknowledgement

The authors acknowledge that at writing of this paper, wewere technically advised by Prof. Kazushi Yamanaka ofDepartment of Materials Processing, Tohoku University.

REFERENCES

1) C. B. Scruby and L. E. Drain: Laser Ultrasonic: Techniques andApplications, (Adam Hilger, Bristol, 1990).

2) J.-P. Monchalin: IEEE Trans. Ultrason. Ferroelectr. Freq. Control 33(1986) 485­499.

3) M. Hercher: Appl. Opt. 7 (1968) 951­966.4) J.-P. Monchalin: Appl. Phys. Lett. 47 (1985) 14­16.5) D. Levesque, M. Choquet, C. Padioleau, C. Neron, C. Corbeil, R.

Talbot, A. Bendada, J.-P. Monchalin, R. V. Kolarik, II and G. V. Jeskey:J. Acoust. Soc. Am. 112 (2002) 2350.

6) J.-P. Monchalin et al.: Rev. Progr. Quant. Nondestr. Eval. 22 (2003)264­272.

7) M. Ochiai: J. Phys.: Conf. Ser. 278 (2011) 012009.8) M. Ochiai, T. Miura and S. Yamamoto: AIP Conference Proc. 975

(2008) 231­238.9) M. Ochiai, T. Butsuen, T. Miura, H. Kuroda, S. Soramoto and S.

Kanemoto: J. At. Energy Soc. Japan 43 (2001) 275­281.10) M. Ochiai, T. Butsuen, J. Senboshi, K. Kubo and S. Asai: Q. J. Japan

Weld. Soc. 17 (1999) 426­431.11) M. Kinoshita, M. Ebina, M. Ohsawa, R. Hibino, H. Kadoura and K.

Akihama: Trans. Japan Soc. Mech. Eng. Part C 78 (2012) 2824­2836.12) A. Kitagawa, S. Asai and Y. Makino: J. Japan Weld. Soc. 72 (2003) 44­

48.13) Y. Fujita, T. Ogawa, S. Asai, S. Yamamoto, T. Ohdake and M. Ochiai:

Weld. World 56 (2012) 15­25.14) A. Blouin, D. Lévesque, C. Néron, D. Drolet and J.-P. Monchalin: Opt.

Express 2 (1998) 531­539.15) M. Ochiai, D. Lévesque, R. Talbot, A. Blouin, A. Fukumoto and J.-P.

Monchalin: Materials Evaluation 62 (2004) 450­459.16) T. Mihara, Y. Otsuka, H. Cho and K. Yamanaka: Key Eng. Mater. 261­

263 (2004) 987­992.17) J. P. Charlesworth and J. A. G. Temple: Engineering Applications

of Ultrasonic Time-of-Flight Diffraction, Second Edition, (ResearchStudies Press Ltd., 2001).

Dep

th d

irec

tion:

z

Blowhole

106

Axial direction: x

1.5

1.91.7

[mm]

Dep

th d

irec

tion:

z

Axial direction: x

(a)

(b)

Fig. 8 Photos of cross-sectional macroscopic test. (a) Overall view.(b) Detailed view.

S. Yamamoto et al.1002