Application of the “GUCHI” Technique in Radiographic Testingfor Flaw Sizing of Pipe

Toshibumi Kashiwa, Shinichi Higuchi+ and Norikazu Ooka+

Nuclear Systems QA Department, Hitachi-GE, Nuclear Energy, Ltd., Hitachi 319-1221, Japan

“GUCHI” (Geometric Unravel for Crack Height Image) technique is one of the stereo-radiography methods for flaw sizing in radiographictesting. The AN (Application of Nondestructive Testing techniques for industrial fields) committee in the JWES (The Japan WeldingEngineering Society) carried out the experiment on this technique. The results were reported at FENDT 2002 and APCNDT 2003 and 2006 withdata on the steel plates and pipes including fatigue cracks and EDM slits. The flaw sizing by RT-GUCHI was performed in a round robin test on6 inches stainless steel pipe which had circumferential semi-elliptical fatigue cracks with 0.2t, 0.4t and 0.6t height at the OD surface and on plateincluding a fatigue crack. A trial for flaw sizing of SCC was also performed.

This paper describes the fundamental of GUCHI technique, the discussion results of the accuracy for flaw sizing which was investigatedbefore 2006 and the results of the round robin tests on test specimens with fatigue cracks. [doi:10.2320/matertrans.I-M2011849]

(Received December 4, 2009; Accepted October 29, 2011; Published January 18, 2012)

Keywords: radiographic testing, flaw sizing, fatigue crack, crack height, digitizing system, digital image

1. Introduction

When the flaw is detected during the in-service inspectionof components, sizing should be performed to evaluate theflaw. ASME code Sec. XI, IWA-3000 provides requirementsfor flaw characterization and evaluation. The NondestructiveTesting (NDT) is expected not only for detecting but alsosizing for evaluation of the flaw.

Ultrasonic Testing (UT) is suitable technique for flawsizing. Many applications of flaw sizing have been reportedregarding UT-TOFD, Phased array systems, and the Crack-tiptechnique. However, sometimes, the component configura-tions may give difficulty to the scanning of search unit byUT. The inherently coarse-grained metallurgical structuresmay preclude meaningful examinations at austenitic anddissimilar metal welds. It is supposed that UT is not alwaysthe best way for flaw sizing in practical fields. Flaw sizingtechnique by RT may be effective as the substitution of thesizing technique by UT.

The AN committee carried out on the experiment on theGUCHI technique which is one of the stereo-radiographymethods for flaw sizing.

2. Fundamentals of GUCHI Technique

The fundamentals of this technique based on the geo-metrical calculation by two radiographic images which aretaken by different angles. Baseline markers such as a smalldiameter copper wire are placed on the surface of tested areaand X-ray or £-ray is exposed from two different positions.The sources positions are parallel to the film or imagingdevice. The width and length of the crack image that appearson each of the radiographs and the distances from baselineimage to the tips and/or the maximum density point of crackimage are measured by means of magnification. A digitizingsystem is used to characterize a flaw and to obtain itsdimensions. The size of the crack is estimated by several

theoretical calculations utilizing the dimensional data ob-tained from the geometrical arrangement of radiographs andmeasured from flaw positions on each of the radiographs.

This technique can be made with two types of formations.In this technique, the X-axis is defined as the directionparallel to the crack, Y-axis is the direction transverse to thecrack, and Z-axis is the direction normal to the surface of thefilm or imaging device. “Parallel motion with X-axis” meansthe sources are positioned parallel to the X-axis and “Parallelmotion with Y-axis” means the sources are positioned parallelto the Y-axis. The exposures per one motion from a differentsource position must be taken at least twice so that the imageof crack and the baseline marker represent simultaneously foreach of the radiograph. Figure 1 shows the arrangement andthe equation at the “Parallel motion with X-axis” formation.

When trying to flaw sizing by “Parallel motion withX-axis”, the length between baseline and maximum densitypoint at the flaw image on each of the radiographs have to bemeasured. If the input the data T, G, m, k and X2, X4 inFig. 1, the flaw height will be calculated as “a”. It depends onhow to measure precisely the flaw images.

The flaw image can be measured precisely due to filmdigitizing system. The tip points of the magnified flaw imageand the points of both side of high density region of the flaw

a=(T+G)-ZpWhere:

Zp=(m/A)Xp+m-(m/A)X4

Xp=(X4/A-X2/B)/(1/A-1/B)

A=(X4-X3), B=(X2-X1)X1=(D-t2), X3=(C-t4)t2=X2 t/E t/EE=m-t, t=T+G+0.5k

C,D: Length between Baselineand max. density point oneach of the radiographs

k: Diameter of Baseline marker

T

G

(X4)

t

m

(X2)

t4 (X1)

P(Xp,Zp)

(X3)

a

(+)(-)

(X)(Z)

t2

C D(+)(-) (+)(-)

t4=X4

Fig. 1 Basic arrangement and equation at “Parallel motion with X-axis”.

+Present address: The Japan Welding Engineering Society, Tokyo 101-0025, Japan

Materials Transactions, Vol. 53, No. 2 (2012) pp. 311 to 315Special Issue on APCNDT 2009©2012 The Japanese Society for Non-Destructive Inspection

image can be determined and the distance between baselineand the each point can be observed from digital data. Thenthe value of X5 to X12 in Fig. 2 can be obtained by thecalculations as same as the equation in Fig. 1. The flawheight “a” can be calculated as the height of a trapezoid.Personal computer (PC) software was developed for real-timecalculations of the flaw size. Figure 3 shows the flaw imageof fatigue crack which displayed on the monitor. The flawlength and distance between baseline and the edge of highdensity region can be also measured. Therefore the pointswhich are needed in GUCHI calculation as input data areobtained easily.

“Parallel motion with Y-axis” will be made as same waywith X-axis and will be considered to be suitable for sizing ofwide-gap flaws. In case of the flaw which is detected at curvedsection such as circumferential welded seams of piping orpressure vessels, it seems that the GUCHI technique will alsobe applicable with the polar coordinates analytic geometry.

3. Flaw Sizing Accuracy of GUCHI

Several experiments which applied GUCHI technique onsteel plates and pipes have performed. Fatigue cracks andElectric Discharge Machining (EDM) slits were installed inthe plates and pipes. The results compared with UT-TOFD,Phased Array and Crack Tip technique has been reported atFENDT 2002,1) APCNDT 2003 and 2006.2,3) Those wereinvestigation of the flaw sizing regarding the accuracy ofGUCHI.

A round robin test was carried out on 6 inches and 10inches stainless steel pipes including circumferential semi-elliptical EDM slits with 0.2t, 0.4t and 0.6t depth at inside ofthe pipe. Figure 4 shows the configuration and size of testpipes. The test pipes of No. 1 and No. 2 are 6 inches and 10inches in diameter respectively. At inside of each pipe,artificial flaws of “A”, “B” and “C” are produced by theEDM.

As conventional X-ray or £-ray equipment the radiationsource and 3 types of X-ray films (IX50, IX80, and IX100)were used and a type of imaging plate (ST-VI) was used asimaging devices. X-axis formation and Y-axis formation withdouble-wall technique, X-axis formation and Y-axis formationwith single-wall technique, and several source to objectdistance, offset of source position were provided for thegeometrical arrangement of taking radiograph. All radio-graphic films were digitized using a Film Digitizer.

Figure 5 shows the summary of sizing results for lengthand depth of the EDM slits. 100 sets of input data wasobtained through the round robin test. The value of horizontalaxis of left diagram indicates the actual length of EDM slits,right diagram shows actual depth. And, the vertical axisindicates the length and depth measured by GUCHI. A goodcoincidence at actual value and measured value by GUCHIhas been obtained. The digital deviations are summarizedon the table. The root mean square (RMS) error of theflaw length measurement was 0.98mm and the flaw depthmeasurement was 1.28mm. The accuracy of sizing byRT-GUCHI technique shown to satisfy the requirements forFlaw Sizing on Ultrasonic Testing in ASME Code Sec XIApp. VIII.4)

4. Round Robin Test for Fatigue Cracks

The AN Committee prepared a 10mm thickness test plateand a 6 inches stainless steel test pipe which include FatigueCracks. A round robin test which applied GUCHI techniqueon those test plate and pipe was carried out. And, fivedifferent company teams of the AN Committee membersparticipated in the round robin test.

Figure 6 shows the 10mm thickness stainless steel platespecimen and Fig. 7 shows the 6 inches OD stainless steelpipe specimen. Those specimens included semi-ellipticalfatigue clacks.

The outline of the test procedure is as followers; Thegeometrical arrangement of the stereo radiography is shownin Figs. 8 and 9. X and Y direction base line markers were puton the source side surface of fatigue cracks. Several case ofthe stereo radiography condition were tried in the round robintest that different type of conventional X-ray equipment andfilm types including imaging plate (IP) were used and sourceposition X0, Y0 and several X2 dimensions, several source tofilm distance were given.

The test parameters for the radiography of test specimenwere that 4 types of X-ray equipment, 3 types of film, 600 to825mm for J, ¹25, ¹30, ¹50, ¹100, ¹150, ¹200mm forX2, 25, 30, 50, 100, 150, 200mm for X4.

Table 1 shows the summary of the RT-GUCHI round robintest results. The crack with 0.2t height in the 12 cases of testparameter could not find out by RT. But GUCHI was appliedto crack in 9 cases with 0.4t height for the pipe and obtainedthe crack images. The deviation of estimated crack length andheight for the plate was 1.6mm in average length of 17.4mmand 0.7mm in average height of 4.4mm respectively. In thecase of pipe, the deviation of crack-height with 0.6t was0.7mm in average length of 25.2mm, and 1.16mm inaverage height of 6.1mm.

One company team carried out an additional experiment offlaw sizing by UT-TOFD technique for the round robin testspecimen to compare with the sizing results. Figure 10 showsthe TOFD data for the plate and pipe specimen. For the platespecimen, the flaw height “a” estimated by TOFD was4.5mm and “a” estimated by RT-GUCHI was 4.4mm. Forthe pipe specimen, ID No. “A” (fatigue crack height is 0.2t)was impossible to evaluate the flaw height by TOFD. In thesame way, that was impossible to sizing by GUCHI. At theID No. “B” and “C”, that crack height is 0.4t or 0.6t, couldbe evaluated by TOFD and the result was as shown in thetable in the Fig. 10. It indicates that the sizing result by RT-GUCHI was good correspondence to the UT-TOFD.

Furthermore, another one company team carried out theexperiment on SCC by RT-GUCHI and UT-Phased Arraytechnique. Figure 11 shows the test pipe which includes SCCat inside of the pipe. The outside diameter of the pipe was 6inches and the wall thickness was 11mm. The SCC couldnot be observed visually as shown in inside view of thephotograph. Figure 12 shows the geometrical arrangement ofthe radiography. The source position of Y was selected to geta longest flaw image. Figure 13 shows the sketch on theradiographic image obtained at the source position X0 andX2. Figure 14 shows the test results. The output data by RT-GUCHI is shown at top side and the output of the Phased

T. Kashiwa, S. Higuchi and N. Ooka312

0

10

20

30

40

50

0 10 20 30 40 50

Actual length of EDM Slit (a−lx) mm

Mea

sure

d le

ngth

by

GU

CH

I (Ix

) m

m

: Dγ −F−X

: Dγ −F−Y

: Dγ −F−X

: Dγ −F−Y

: SX−F−X

: SX−F−Y

: SX−I−Y

: DX−I−X

: DX−I−Y

0

5

10

15

20

0 5 10 15 20

Actual depth of EDM Slit (a−d) mm

Mea

sure

d de

pth

by G

UC

HI (

d) m

m

: SX−F−X

: SX−F−Y

: SX−I−Y

: DX−I−X

: DX−I−Y

Flaw Length (mm)Mean Deviation +2.87

Standard Deviation 2 2.10RMS Error 0.98

Flaw Depth (mm)Mean Deviation +0.63

Standard Deviation 2 2.56RMS Error 1.28

σσ

Fig. 5 Flaw sizing accuracy of RT-GUCHI.

Fatigue Crack

100

Material; Stainless Steel

60

10

Fig. 6 Test specimen ­ Plate.

Unit : mma

l

t

EDMSlit

a/t(%)

a/l

A 20 0.25B 40 0.25C 60 0.25

300

t

D

0.9B

A

C

No. D t

1 165.2 11.02 267.4 15.1

Fig. 4 Configuration and size of test pipes.

BASE LINE IMAGE

5.6 6.0

−8.4

17.0

FLAW IMAGE

Fig. 3 Typical flaw image on film digitizing system.

INPUT: X5-12,T,G,J

F(X5-12,T,G,J)=a

OUTPUT: a (Flaw Height)

ΔD

ΔD

T

G

J

ΔD:Density Levelof Crack Image

Z1 Z2

Z3 Z4

(X4)(X2)

X5X6 X7 X8

X9 X10 X11 X12

Fig. 2 Approximate technique (Trapezoid).

Application of the “GUCHI” Technique in Radiographic Testing for Flaw Sizing of Pipe 313

X2

X Base−line Marker

(−) Side

X4

(+) Side

Y Base−line Marker

Fatigue CrackT

J

X0

Film

Y0

Film

Fig. 8 Geometrical arrangement for plate test specimen.

X2

X Base−line Marker

(−) Side

Film

X4

(+) Side

Y Base−line Marker

FatigueCrack

R T

J

X0

Film

Y0

Fig. 9 Geometrical arrangement for pipe test specimen.

Table 1 Summary of the RT-GUCHI round robin test results.

TP

Fatigue CrackNos. of Test Caseand Case Nos. thatthe Crack find out

Crack Size by GUCHI (mm)

Length Height

ID No. HeightTestCase

Find-out

Average Deviation Average Deviation

Plate ® 0.4t 3333

(100%)17.4 1.6 4.4 0.7

Pipe

A 0.2t 12 0 ® ® ® ®

B 0.4t 219

(43%)17.0 0.6 3.6 0.5

C 0.6t 1919

(100%)25.2 0.7 6.1 3.2*

*2· = 1.16mm

400

Material; Stainless Steel

t=11.0

165.2

Fatigue Crack

Crack Height;A=0.2tB=0.4tC=0.6t

C

BA

Fig. 7 Test specimen ­ Pipe.

ID No. A (0.2t) B (0.4t) C (0.6t)

UT-TOFD

RT-GUCHI 3.6 mm 6.1 mm

d=4.5

FRONT BACKFlaw Height by RT−GUCHI 4.4 mm

Flaw Height by UT−TOFD 4.5 mm

Fatigue Crack

(a) TEST SPECIMEN−Plate

d=4.0 d=4.4Impossible

to Evaluate

Impossible

to Evaluate

(b) TEST SPECIMEN−Pipe

Fig. 10 Comparison with RT-GUCHI to UT-TOFD test results.

T. Kashiwa, S. Higuchi and N. Ooka314

Array UT system is shown at bottom side. The estimatedcrack height “a” by GUCHI is 5.1 and 5.3mm by PhasedArray. From those results, it is considered that the sizingcapability of GUCHI technique is comparable to that of UTunder the condition of this experiment.

5. Conclusion

The AN committee in the JWES carried out the experimenton application of RT-GUCHI technique which is one of thestereo-radiography methods for flaw sizing. The accuracy ofsizing by RT-GUCHI technique satisfied the requirementsof flaw sizing in the ASME Code Sec XI App. VIII. Thetechnique was applied and confirmed by the round robin test

for sizing of natural like fatigue cracks in the plate and thepipe. And, it has been also applied for sizing the SCC. Bothof the UT-TOFD for the fatigue cracks and the UT-PhasedArray for SCC have been performed to compare with theresults of flaw sizing by the RT-GUCHI. It was confirmedthat RT-GUCHI technique is applicable for flaw sizing forover 0.45t-height fatigue cracks or SCC in steel componentsas same as sizing by UT in practical fields.

Acknowledgement

Finally, authors would like to appreciate the members ofthe AN committee for their useful comments and support.

REFERENCES

1) S. Higuchi, M. Okudaira, Z. Makihara and N. Ooka: 6th FENDT, Tokyo,Japan, Oct., (2002) pp. 337­342.

2) S. Higuchi, Z. Makihara, Y. Nonaka and N. Ooka: Key Eng. Mater. 270­273 (2004) 1316­1323.

3) T. Kashiwa, S. Higuchi and N. Ooka: 12th APCNDT, Auckland, NewZealand, Nov., (2006) http://www.ndt.net/article/apcndt2006/papers/21.pdf

4) ASME Code 2004 Sec. XI, Division I, Appendix VIII, Sup. 2 Perform-ance Demonstrations for Ultrasonic Examination Systems.

Source Position : X0=0,Y=133Source Position : X2=−200,Y=133

X-BASE LINE

25.5

6.0

5.5

33.0

3.0

Y−BASE LINE

SCC

X-BASE LINE

Y−BASE LINE

4.9

3.0

25.4

30.3

5.6

SCC

Fig. 13 Sketch on the radiographic image of SCC.

Length, Angle, Heightlx

(mm)ld

(mm)2θθ(°°)

a(mm)

l1(mm)

26.4 28.2 22 5.1 8.4

OUTPUT DATA BY RT-GUCHI

OUTPUT DATA

BY UT-PHASED ARRAY

l1

a

ld

lx2θ

BOTTOM

TIP

d=5.3

Fig. 14 Sizing results of SCC estimated by RT-GUCHI and UT-PHA.

789

X2=−200 X0=0

(X)(Z)

X BASE LINE

MARKER

FILM SCC

Y=133

(Y)(Z)

Y BASE LINE

MARKER WELD

SCC

Fig. 12 Geometrical arrangement for sizing of SCC.

INSIDEVIEW

SCC

11

SCCTEST PIPE (OD=6B)

CircumferentialWeld

14.3

Fig. 11 Test pipe which is including SCC.

Application of the “GUCHI” Technique in Radiographic Testing for Flaw Sizing of Pipe 315