17th World Conference on Nondestructive Testing, 25-28 Oct 2008, Shanghai, China Development of Automated Electromagnetic Techniques

for Inspecting Inner Cracks of LPG Tanks

Yung-How WU 1, Feng-Ju HSIEH 1, Mu-Chung PENG 1, Hsing-Jeng YANG 1, Vicent LIN 2 1Industrial Technology Research Institute, Chinese Taipei

2China Petroleum Corporation, Chinese Taipei Tel: +886-(0)933792939, Fax: +886-(0)3-5910214; E-mail: yhwu01@itri.org.tw,

fongru@msn.com, MuChungPeng@itri.org.tw, henryang@itri.org.tw, 224171@cpc.com.tw

Abstract

LPG tanks are essential storage device in refineries and were frequently subjected to SCC on inner surface during service and must be inspected from inside regularly. MT or PT are two conventional methods used for the inspection but is very costly and toilsome because of the requirement of scaffolding and the extreme hot working condition inside the tank. Practical experience had shown that epoxy layer may be applied on the inner surface of the LPG tanks to alleviate SCC however extra work of epoxy layer removal was required before MT or PT. In this study, an automated inspection method based on multi-channel electromagnetic technique was developed in lieu of conventional MT or PT. A remote operated scanner with magnetic rollers carrying inspection probe tracking by a video camera were incorporated in this method. Experimental results and field trials had shown that this innovational design may detect real cracks down to 10mmL×1mmD in dimension and to assure the full coverage of inspection area. Meanwhile, with this method, scaffolding and epoxy removal may be also relieved to reduce to maintenance cost and, the inspection work is no longer a toilsome one.

Keywords: LPG tanks, Electromagnetic, Eddy current, Automation, Video 1. Introduction

LPG tanks were frequently subjected to Stress Corrosion Cracking (SCC) on inner surface due to loading/unloading of corrosive LPG products and temperature variation during service. Practical experience had shown that epoxy layer may be applied on the inner surface of the LPG tanks to alleviate SCC however, inner surface cracks were still found while the epoxy coating became deteriorated after some years. MT or PT has been two conventional methods used for SCC inspection and is very costly and toilsome because of the requirement of scaffolding and the extreme hot working condition inside the tank in summer. To inspect the tank with epoxy cladding, extra work of epoxy layer removal was required before routine MT or PT. Though UT scanning from outer surface had also been used, it is practical only to cover the weld zone rather than the plates where SCC appeared frequently. Former inspections showed most cracks ranges from 3 to 10 mm in length. The current regulation in Taiwan requires cracks less than 3mm in depth to be ground flush before returning to service. Aside from PT, MT and UT, a current perturbation based electromagnetic array techniques (EMA) or more commonly called ACFM, are routinely used both for sub-sea and topside for structural weld inspections. Such techniques are capable of both detecting and sizing surface breaking cracks especially under corrosion insulation coating. The technique has now been widely used on various structures in the Oil and Gas Industry for in service inspection beneficial for the following.

� Capable of measuring both crack depth and length.

� Inspecting cracks without removal of surface coating up to several millimeters and hence reducing cost and time.

� Inspection results may be kept as digital records for better review and monitor. � Various signal displays are helpful in defect discrimination.

In this study, an automated inspection method based on multi-channel electromagnetic technique was developed in lieu of conventional MT or PT. This method is expected to inspect inner surface cracks from inside LPG tanks without scaffolding.

2. Current perturbation based EC Method

The current perturbation based EMA or ACFM methods were originally derived from ACPD principle (Fig. 1(a)) which utilizes current potential drop to measure crack depth by contact probes. Alternatively, EMA or ACFM methods may inspect by non-contact electromagnetic probe which offers the flexibility of coil array design to meet various requirements[1-6] and characterized as follows:

� Using large exciting coil to induce uniform surface current to flow across open crack as shown in Fig. 1(b). The current beneath the probe may go toward the crack and then flow downward around the crack tip to return to sample surface beyond the crack which induced magnetic field with favorable component parallel to sample surface. The current flow may also take the shortest route around crack ends with magnetic field normal to the surface. Those two magnetic components may be picked up by integral probe with properly laid absolute and differential coils for measuring crack depth and length respectively.

� Unlike conventional EC method, EMA or ACFM techniques used array probe with one or more large excitation coils with axes perpendicular to small receiver coils. With such probe design, the induced eddy current beneath the probe may be uniformly distributed and direct couple between excitation and receiver coils somewhat minimized. Hence, inspection may be done with lift-off up to several millimeters and less sensitive to lift-off.

� Though the signal display is based on impedance as conventional eddy current technique, the hardware design incorporating absolute and differential coil array offers various displays to enhance signal analysis as shown in Figure 2. In which, a closed loop (or butterfly shape) plot correlating signals from absolute (Abs) and differential (or current perturbation, CP) coils was mainly used to characterize an open crack.

Figure 1. (a) ACPD method, crack depth is measured in terms of potential drop by contact probe.

(b) EMA/ACFM method, surface current is induced and detected by electromagnetic coils and disturbed by open crack.

V2/V1 = (2d + △) / △

∴ d =△/2((V2/V1) – 1) △(Probe span)

Current perturbation Surface current

Crack end

V1 V2

d (Crack depth) Surface current Open crack

Figure 2. Signal displays of EMA or ACFM method.

3. Evaluation of Crack Detection and Sizing

As shown in Figure 3, NEWT Lizard EMA system and a hand held LP600 EMA probe were first utilized to determine the probe specification and basic parameters. Based on the evaluation results, a 32-channel probe mainly for detection purpose was then designed according to the field application conditions of LPG tank inspection. During lab test, an automatic X-Y scanner was also used for moving the 32-channel probe to simulate the probe scanning considering orientation and speed factors that are essential to cover cracks on welds and plates of LPG tank during ISI. Transparencies were stacked between probe and sample to create different lift-off during the tests as in Figure 4. A series of tests considering crack dimension, orientation and lift-off were conducted for the evaluation of crack detection and sizing on plate and weld samples with EDM notches and built-in cracks, as shown in Figure 5 and Figure 6.

Sample plate with cracks

Probe

Auto scanner

Transparencies

Figure 3. Lizard EMA system and probe. Figure 4. Probe scanning setup.

Crack

Lift -off

CP Imped. Abs Imped. Time based

Abs CP

Butterfly plot

Correlation plot

SDDPLizard Solo

32-channel probe Hand held probe LP600

Motor control unit

Notches(mm) Sample No.

Type/Location Depth Length Width Note

2 40 0.25 2 20 0.25 Notches/lined up(0°)

A group 2 10 0.25

0.5 5 0.25 1 5 0.25 Notches/lined up(90°)

B group 0.5 10 0.25

Independent notch 1.5 5 0.25 1.5 10 0.25

Plate-1

Notches/lined up(90°) C group 1 10 0.25

� Plate thickness: 10mm； � Distribution of notches as

shown above.

5 0.35 0.5

10 0.40 5 0.35

Notches/lined up(0°) A group

1 10 0.40

2 10 0.50 4 20 0.50 Notches/lined up(0°)

B group 6 20 0.50

0.5 8 0.21 1 12 0.21 Notches/lined up(0°)

C group 2 18 0.21

2+4+2＊ 39 0.21

6 30 0.21

Plate-1

Notches/lined up(90°) D group

4 25 0.21

� Plate thickness: 8mm； � Distribution of notches as

shown above.

� ＊notch of varying depth

Figure 5. Steel plate No. 1 & 2 containing various notches.

41210

150

Plate-1

A group

B group C group

40000

300

60

60

80

50

50

Plate-2

D group

B group

A group

C group

Cracks (mm) Sample No.

Type Depth Length Location Cross-section

Weld-1 Toe Crack 3.9 12 30

Weld-2 Centerline

Crack 5.8 13 38

HAZ Crack 1 4.3 12 20 Weld-3

HAZ Crack 2 4.3 12 91 HAZ Crack 1 5.8 15 46

Weld-4 HAZ Crack 2 5.8 15 110

Figure 6. Weld samples with artificial cracks which depth were confirmed by TOFD method.

Results showed that notches larger than 5mm×0.5mm were all detectable regardless of scanning orientation (Figure 7) when the lift-off was zero. As in Figure 8, defects appeared with a phase angle other than lift-off by a difference greater than 90°. One may also notify a single crack creates signals in successive three channel pairs in water fall plot in most cases. And, notches or cracks may be distinguished by the correlation plot while showing as closed loops.

Scanning direction

Notch axial

Crack axial

Scanning direction

Figure 7. Series of scans were done on samples with various orientations referring to a certain

defect.

Figure 8. Notch signals (arrows) appeared in 3 successive channel pairs in water fall plot (Left) and may be identified more clearly together with other plots especially as closed loops.

Figure 9 shows the effect of scanning orientation and lift-off on the detection sensitivity. For a notch (10mmL×2mmD) tested by the 32-channel probe, the signal amplitude fell to a minimum at 38° and reduced 6 times exponentially from zero lift-off to 2mm.

02468

0 20 38 50 90Scanning orientation in degree

Amp(

Max

/Min

)

1張5張15張

10mmx2mm notch

02468

0 10 20 30Numbers of transparency

Amp(

Max

/Min

)

數列1

Figure 9. The effect of scanning orientation (Left) and lift-off (Right) on a notch (L×D =10mm×2mm) tested by 32 channel probe design for automated inspection. Lift off was varied by stacking different number of transparency per 0.1mm in thickness.

Crack depth measurement was further evaluated on notches of different depth (0.5mm~6mm) with LP600 probe. With proper calibration, the sizing error varied in 10%~50% as shown in Figure 10. The weld toe cracks were also confirmed detectable though the signal affected by weld crown (Figure 11).

Figure 10. Depth sizing of notches on plate-1

& plate-2 referring to 2mm deep notch.

Figure 11. Crack at weld toe was detectable though affected by weld crown.

crack crack

Plate1 & Plate2

0

1

2

3

4

5

6

0 1 2 3 4 5 6 7

Actual depth(mm)

Mea

sure

d d

epth

(mm

)

4. Automated Inspection System and Field Trials

As shown in Figure 12, an automated inspection method based on multi-channel electromagnetic technique was developed for ISI of LPG tank in lieu of conventional MT or PT. The method was designed to incorporate a remote operated scanner with magnetic rollers carrying the 32-channel probe which detectability had been confirmed in the lab test. A video camera installed at the center of the LPG tank was used to monitor and track the scanner moving on the inner surface. A video tracking software was designed to capture the scanner which image was in turn used as a feedback to manipulate the scanner automatically. Crack detection was then monitored by a Lizard solo system through 64 channel SDDPU interface. In our inspection scheme, the surface of LPG tank will be inspected zone by zone per which may be several square meters. Welds and mother plates of the tank are inspected respectively. The scanning pattern of each zone together with the scanner image grabbed by image processor may be shown on the PC panel for real time monitoring as shown in Figure 13.

CCD Camera

NTSC StandardPosition

Sense

Frame GrabberCard

LPG Tank Steel Plate

Field of View

Real Scan Path

Ideal Trajectory

Figure 12. Configuration of automated EMA

inspection system of LPG tanks. Figure 13. Scanning pattern and video tracking

of moving scanner.

Several field trials were then carried out on a retired and in service LPG tanks. The inspection was monitored with EMA system, motion control set up under the tank as shown in Figure 14. And, Figure 15 shows the scanner moving on the inner surface of the tank during field trial.

Figure 14. Field trial of automated LPG tank

inspection. Figure 15. Video tracking scanner moving on

the inner surface of LPG tank.

5. Conclusions

(1) Lab test results showed that notches larger than 5mm×0.5mm were all detectable regardless of scanning orientation when the lift-off was zero. And, with proper calibration, the sizing error varied in 10%~50% for notches of different depth varied between 0.5mm and 6mm.

(2) Experimental results and field trials had shown that this innovational design with video tracking of the scanner may detect real cracks down to 10mmL×1mmD in dimension and to assure the full coverage of inspection area.

(3) Considering the safe operation of scanner mounted by magnetic rolls, the lower two third of the tank surface may be covered with this method which is sufficient in most ISI. In this case, scaffolding and epoxy removal may be relieved to reduce to maintenance cost and, the inspection work is no longer a toilsome one.

Acknowledgement

The research and development on this project were founded by China Petroleum Corporation. The authors also gratefully acknowledge CPC Taoyuan and Kaoshiung refineries in supporting this program. References

[1] D. A Topp, “Quantitative In-Service Inspection using the Alternating Current Field Measurement Method”, International Symposium on NDT Contribution to the Infrastructure Safety Systems, Torres, Santa Maria, RS, Brazil, Nov. 22~26, 1999.

[2] G. L. Burkhardt, et al., “Electric Current Perturbation NDE”, Metal Handbook Ninth Edition Vol. 17.

[3] M. N. Bentley, et al., Typical Application of Structural Survey Using the LIZARD Defect Detection, Sizing, Imaging & Reporting System”, 10th Conference & Exhibition World Trade Center, Singapore, Dec. 6~9, 1994.

[4] Raine, et al., “NDT of Pipework and Process Plant using the ACFM Technique”, EPRI Topical Workshop, Charlotte, NC, Aug. 21~23, 1995.

[5] S. H. H. Sadeghi and D. M-Syahkal, “A Method for Sizing Long Surface Cracks in Metals Based on the Measurement of the Surface Magnetic Field”, Review of Progress in QNDE, Plenum Press, N.Y., Vol. 9, pp. 351~358, 1990.

[6] A. M. Lewis, et al., “The Thin-skin Electromagnetic Field Near a Surface Crack in a Ferromagnetic Metal”, Review of Progress in QNDE, Plenum Press, N.Y., Vol. 9, pp. 273~280, 1990.