EDDY CURRENT TECHNIQUE FOR DETECTING AND SIZING SURFACE CRACKS IN STEEL COMPONENTS

CANDU MAINTENANCE CONFERENCE 1 9 9 5

EDDY CURRENT TECHNIQUE FOR DETECTING AND SIZING

SURFACE CRACKS IN STEEL COMPONENTS

V.S. Cecco, J.R. Carter and S.P. Sullivan AECL

Chalk River Laboratories, Station 43

niver, untano, uanaoa i\UJ iJu iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiniiiiii

Phone {613) 584-3311 ext 3756 I III lilll ill 1 III II! Ill III III

CA9800517

INTRODUCTION

Cracking has occurred in pressure vessel nozzles and girth welds due to thermal fatigue. Pipe welds, welds in support structures, and welds in reactor vault liner panels in nuclear facilities have failed because of cracks. Cracking can also occur in turbine rotor bore surfaces due to high cycle fatigue. Dye penetrant, magnetic particle and other surface NDT methods are used to detect cracks but cannot be used for depth sizing. Crack depth can be measured with var-ious NDT methods such as ultrasonic time-of-flight diffraction (TOFD), potential drop, and eddy current. The TOFD technique can be difficult to implement on nozzle welds and is best suited for sizing deep cracks (>5 mm). The conventional eddy current method is easy to implement, but crack sizing is normally limit-ed to shallow cracks (<2 mm).

AECL has developed transmit-receive probes with directional sensitivity, capable of detecting and sizing cracks in plain and weld areas of carbon steel com-ponents. The G3 probe is used for sizing long, deep (>2 mm) cracks. Eddy current testing (ED tech-niques are readily amenable to remote/automatic inspections. These new probes could augment pre-sent magnetic particle (MT) and dye penetrant (FT) testing through provision of reliable defect depth information. Reliable crack sizing permits identifica-tion of critical cracks for plant life extension and licensing purposes. In addition, performing PT and MT generates low level radioactive waste in some inspection applications in nuclear facilities. Replacing these techniques with ET for some components will eliminate some of this radioactive waste.

with alternating current, and the receive coil is used to detect magnetic flux. With such spacing, the coils have little direct coupling. This transmit-receive (T/R) probe behaves as an absolute probe, producing one-sided signals from lift-off variations and defects. Similar probes have been used in the aircraft indus-try1 and are normally referred to as "Sliding Probes". The G3 probe only detects crack-like defects which guide the magnetic field from the transmit coil to the receive coil2. Unlike conventional eddy current probes, it is relatively insensitive to magnetic perme-ability variations. Long and deep defects increase the magnetic coupling between the coils thereby increas-ing the voltage induced in the receive coil. It has directional properties, being only sensitive to defects in-line with the T/R coils. Since the magnetic field in the peripheral region of the transmit coil is quite uni-form, even the small distortions originating from the bottom of deep defects are detected. This permits sizing of surface defects that are much deeper than 'one standard depth of eddy current penetration'. Figure 2 contains a block diagram of the eddy current system with a probe, test sample, ET instrument and strip chart recorder. Signals can be analyzed directly off the storage monitor or off a strip chart recorder trace. As shown in Figure 2(b), signals from cracks and probe wobble (lift-off) are separated by approxi-mately 90°. Since signal phase angle is essentially independent of defect depth, only the vertical (y) sig-nal component needs to be recorded on the strip chart as shown in Figure 2(c). With no signal phase rotation with defect depth, signal amplitude must be used to obtain size information.

G3 PROBE CHARACTERISTICS

The G3 probe consists of transmit and receive (T/R) coils laterally displaced by at least one coil diameter as shown in Figure 1. The transmit coil is excited

TEST RESULTS

G3 probes with large transmit-receive coil spacing and operating at high test frequencies MOO kHz) have excellent sensitivity to long surface breaking

CANDU MAINTENANCE CONFERENCE 1995

cracks. Figure 3(b) illustrates probe response to EDM (Electric Discharge Machined) notches 1 to 5 mm deep (20 mm long), lift-off, and 'weld noise', at 100 kHz. Since the probe tolerates a change in lift-off of almost 1 mm, it can be used in a non-contact mode. It can bounce over a rough weld crown without pro-ducing large background noise signals.

Figure 3(0 illustrates computer predicted probe response to cracks and lift-off, in addition to perme-ability and resistivity variations. The cracks are simu-lated as EDM notches of infinitesimal width. The computer simulated signals agree closely with the laboratory results of Figure 3(b). The probe appears almost insensitive to magnetic permeability (mr) and resistivity variations.

G3 probes have been used recently for inspecting steam drums, welds in support structures, PWR pressure vessels and turbine rotor bores3 for surface breaking cracks. Figure 4 illustrates the probe's defect detectability in welded samples. Because of the insensitivity to lift-off and permeability variations, the G3 probe can detect defects at welds. Figure 4(e) shows the undistorted signals from side-walMack-of-fusion and fatigue cracks at welds. Defects deeper than 2 mm can be readily detected, and because sig-nal amplitude is not significantly affected by the weld crown, depth can also be estimated.

Figure 5 illustrates similar results from a type 304 stainless steel weld sample. Although the 304 base material is nonferromagnetic, the 308 weld filler is strongly ferromagnetic, making conventional eddy current testing unreliable. The G3 probe clearly detects the 3 and 5 mm deep, by 20 mm long EDM, notches in the weld area. The lift-off signal is larger than that from the carbon steel weld sample, but defect detectability and sizing accuracy is similar. In another inspection application that illustrates the versatility of G3 probes, a manufacturing defect (cold shut) had to be sized on a type 304 valve casting. The coarse grain structure rendered ultrasonics unreliable and large ferromagnetic variations made

convention-al eddy current impossible. Clear signconvention-als with a G3 probe, illustrated in Figure 6, indicated a 75 mm long and 6 to 7 mm deep (50% through-wall) defect. The defect signal was so large that lift-off and permeabil-ity noise were insignificant.

SUMMARY

The transmit-receive eddy current probes described in this paper demonstrate how design optimization through computer modelling and laboratory valida-tion, overcome limitations in eddy current testing of ferromagnetic components. These probes signifi-cantly increase the scope and reliability of detecting and sizing long, deep cracks in as-manufactured and welded components.

G3 probes can be used with commercially available instruments, and the versatility and simplicity of their design and inspection procedure makes them very promising.

ACKNOWLEDGEMENTS

This work was supported by CANDU Owners Group under WPIR Numbers 1131 and 1121.

REFERENCES

1 Pellicer, J., "Sliding Probe Eddy Current System for Improved Fastener Hole Inspection", ATA Nondestructive Testing Forum, Kansas City, MO, August-September 1983.

2 Mayos, M. and J.L. Muller, "Geometrically Anisotropic Probes: an Improved Eddy Current Technique", Journal of Nondestructive Evaluation, Vol. 6, No. 2, 1987, pp. 109-116.

3 Krzywosz, K., L. Cagle and M. Richter, "Eddy Current Flaw Detection and Sizing of Cracks in Turbine Rotor Bore Surfaces", Paper presented at the EPRI Steam Turbine/Generator Workshop, July 16-19, 1991, Charlotte, N.C., USA.

CANDU MAINTENANCE CONFERENCE 1995

Transmit Coll

Receive Coil

Figure 1: Schematic of G3 probe and test sample illustrating the scanning direction.

G3 Probe Calibration Plate Lift-off Actual Air Cracks (b) Strip-Chart ET Instrument Recorder

t>

Test Sample with EDM Notch

i

Figure 2: (a) Block diagram of eddy current inspection system. (b) Experimental signals from a G3 probe with 10 mm coil spacing, (c) Computer simulated signals from a similar G3 probe.

CANDU MAINTENANCE CONFERENCE 1 9 9 5

EDM Notches 20 mm long

5 nun Deep (a) Lift-off, Plate (b) T00J • n. Air 5 0 (c)

Figure 3: (a) Calibration plate with EDM notches 1, 3 and 5 mm deep by 20 mm long, (b) Experimental signals from a G3 probe with 10 mm coil spacing, (c) Computer-simulated signals for a similar G3 probe.

— — 1 mm _ _ _ 3 mm ^ — 5 mm Deep 20 mm Long 3 mm D e e p / / " 3 mm Deep >*£* Side-Wall Laclc-of-Fusion / / F^gue Cracks

(a) 5 mm Deep f l3 m m LiH-olT_ i l l mm Air (b) (c) L.O. _{L.O. \}

Weld Noise Side-Wall Large Fatigue Small Fatigue (d) Lack-of-Fusion Crack Crack

(e) (f)

Figure 4: (a, b, c) Carbon steel calibration and welded test samples, (d) Calibration signals at 200 kHz. (e) Signals from side-wall lack-of-fusion. (f) Signals from fatigue cracks near welds.

CANDU MAINTENANCE CONFERENCE 1 9 9 5

7

/Tf

Figure 5: (a, b) Type 304 stainless steel calibration and test samples. (c) Calibration signals from EDM notches at 200 kHz. (d) Signals from EDM notches in weld areas at 200 kHz.

Flange

' / / 1 / 1 1 /11 / I / /11 f(

(b) " " %

j 3 5 mm Deep Flaw Signal x20mmLong (~ 6 to 7 mm Deep) Calibration ^ Signals Y-Signal Amplitude (c)

Distance along flaw length

Figure 6: (a) Plan view of a 75 mm long flaw in 304 stainless steel valve casting, (b) X-Y signals from calibration defects and a flaw, (c) Y-component signals along flaw length, with a G3 probe, with 15 mm coil spacing, at 100 kHz.