COMPUTER MODELING OF EDDY CURRENT TRANSMIT-RECEIVE PROBES FOR. S.P. Sullivan, V.S. Cecco, L.S. Obrutsky, D. Humphrey, S.P. Smith and K.A.

Transcription

1 COMPUTER MODELING OF EDDY CURRENT TRANSMIT-RECEIVE PROBES FOR TUBE INSPECTION INTRODUCTION S.P. Sullivan, V.S. Cecco, L.S. Obrutsky, D. Humphrey, S.P. Smith and K.A. Emde Nondestructive Testing Development Branch Engineering Technologies Division Chalk River Laboratories, Atomic Energy of Canada Limited (AECL) Chalk River, Ontario, Canada KOJ lio Conventional eddy current bobbin probes are known to be ineffectual in detecting circumferential cracks in tubing. Rotating pancake and/or multi-pancake probes are required to detect circumferential cracks. However, it has been demonstrated in CANDU (CANada Deuterium Uranium nuclear reactor) steam generator tubes with deformation and ferromagnetic deposits that probes with surface transmit-receive (TIR) coils are superior to those using surface impedance coils. No subsurface cracks in this tubing were detected with rotating impedance pancake coil probes and multiplexed impedance probes. In the same tubing, circumferential cracks as shallow as 50% through-wall were detected with a single-pass C3 multiplexed TIR probe. Users of inspection technology are much more familiar with the capabilities of impedance probes. This paper illustrates the basic features of T IR surface probes calculated from computer models based on solutions to Maxwell's equations. Twodimensional voltage diagrams showing basic probe responses to frequency, lift-off, carbon steel supports, magnetite deposits and copper deposits are presented. Theoretical results show that TIR probes are able to detect defects in the presence of variable lift-off (due to tube-wall deformations) with several times the signal-to-noise ratio exhibited by comparable impedance probes. THEORY Background Information Detection of circumferential cracks is one of the most challenging aspects of eddy current tube inspection. Bobbin-type probes have very low sensitivity to circumferential Review of Progress in Quantitative Nondestructive Evaluation, Vol 17 Edited by D.O. Thompson and D.E. Chimenti, Plenum Press, New York,

2 cracks, because the eddy currents flow parallel to the circumferential coil windings, and do not interact with circumferential cracks [1]. The usual solution is to use rotating pancake probes or probes consisting of an array of small pancake coils. These probes are usually impedance probes that have good sensitivity to internal circumferential cracks. However, they have low sensitivity to external defects, and they generate large lift-off noise that complicates signal analysis and reduces sensitivity [2,3,4]. These probes are normally spring-loaded to minimize lift-off effects, but that in turn makes them prone to mechanical failure. Research at Chalk River Laboratories established that TIR probes have much lower sensitivity to lift-off noise than impedance probes. Computer simulations were performed using Dodd and Deeds solutions to Maxwell's equations [5] and the Burrows equations for modelling probe responses to small defects [6]. These computer models were used to calculate probe lift-off and responses to small defects for impedance and TIR probes with equivalent-size coils. The results are shown in Figure 1. For the same small defect, the signal-to-noise ratio with a TIR probe is several times that of an impedance probe. Normalized Voltage Diagram Normalized voltage diagrams can be used to display voltage changes oftir probes. Figure 2b shows responses of an absolute TIR probe to various test parameters shown in Figure 2a: lift-off (LO), internal (ID) and external (OD) wall thinning, internal and external magnetite deposits (ID Mag, OD Mag), external copper (Cu) deposits, and carbon steel tube supports (CS). The optimized test frequency of 250 khz was used to display an approximately 90-degree signal phase separation between shallow external defects and lift-off due to tube deformation, as shown in Figure 2c. The lowest frequency (70 khz) was used to detect 1m dance Probe Transmit-Receive Probe I mm Defect 6 mm 6 mm IXl lxllxi IXI lmm Defect Defect Signal Defect ignal Lift-off mm Lift-off; 0. 1 mm Impedance Probe ignals Transmit-Receive Probe Signals Figure 1. Computer-derived signal-to-noise comparison between an impedance probe and a transmit-receive probe. The "signal" is due to a small subsurface defect. The "noise" is due to a 0.1 mm variation in lift-off. 284

3 support plates and deposits. The highest frequency (450 khz) was used to detect tube deformation and deep cracks, and the last frequency of 160 khz was used to obtain complementary and supportive information for defect sizing. Figure 2d shows computerderived signals from internal and external cracks using a differential TIR probe. Probe response to conducting and ferromagnetic deposits and support plates increases with decreasing frequency, which is typical of all eddy current probes. Fortunately, unlike surface impedance probes, C3 probe signals from internal magnetite deposits are nearly in phase with lift-off signals at frequencies higher than 160 khz. Magnetite signals are therefore less likely to interfere with defect signals. This is an important advantage of TIR probes over surface impedance coils, since Canadian CANDU steam generator tubes have internal magnetite deposits with high permeability. Computer modeling permits paper studies of probe/tube responses, such as in Figure 2; it is very useful in optimizing probe design and is indispensable for signal analysis studies. 00 ID "!'!o O.2mm O.2mm t mm Ii\\\\\\\ Lift-Off Co P _ 1.7.cm J- O,02.5mm a Carbon S'eel p - U ll-cm ".. -so 1' ", - 2,- 1 mm 1.43, , Nonnalized In-Phase Voltage b Figure 2. (a) Tube with simulated defects and deposits, (b) computer-derived voltage display illustrating the effect of various test variables at 70 to 450 khz, (c) absolute probe signals at 250 khz, (d) differential probe signals of defects at 250 khz. 285

4 PROBE CHARACTERISTICS Probe Mechanical Design The C3 probe is a differential multi-coil TIR probe. Figure 3a illustrates a C3 probe with eight sets oftir coil units capable of operating at four multiplexed frequencies. Unlike multi-pancake array spring-loaded surface probes, it has no moving parts permitting single-pass inspections as fast as bobbin-type probes. The eight TIR units are contained in two separate bodies. Each probe body is encased in a metallic sleeve, with centering guides. The probe bodies are physically separated by a section of flexible cable, because a single solid body would be too long to go around tight U-bends. This is a robust design, but flexible enough to pass around U-bends as small as a 150 mm (6 inch) radius without difficulty. To obtain eight TIR units, the transmit coils adjacent to either side of the receive coil pair are active at different multiplexed times. This allows each receive coil pair to detect a signal from one transmit coil at a time [4], and results in eight separated 45-degree windows of sensitivity. Directional Properties The C3 probe shown in Figure 3 was optimized to detect circumferential cracks. A modified TIR array design, denoted as C5 (shown in Figure 4), was developed to detect axial and circumferential cracks with equal sensitivity. Figure 5 shows computer calculations of probe signals from axial and circumferential cracks. This figure shows a large signal from the circumferential crack, but a small signal from the axial crack with the C3 probe. The signals from the circumferential and axial cracks with the C5 probe are comparable. Area of Sensitivity Eight TR Units can Direction Back Probe Body <a> Front Probe Body Detection Windows TR TR6VTR TR4 (b) Detection Windows Figure 3. C3 probe showing TIR coil configuration and detection windows. 286

5 - Area of Sensitivity - Volumetric Defect Circumferential Crack Axial Crack k Figure 4. Coil layout of C5 probe for 16 mm (5/8 inch) OD steam generator tubes. This probe has 16 TIR units. o I C5 Probe Volumetric Flaw I Circumferential Crack I Axial Crack Figure 5. Computer-predicted TIR array probe voltage plane responses to a volumetric flaw, a short circumferential crack and a short axial crack. 287

6 25 Circumferential Crack Length = 29! CD '0 :I a E 4( iii c: a/ en O.-r--. o Circumferential Position (Degrees) Axial Crack Length 3.5 mm 5 O ;--.-r,- o Circumferential Position (Degrees) Figure 6. Computer simulated scans of circumferential and axial cracks in 16 mm (5/8 inch) diameter steam generator tubes. Circumferential Coverage Computer modeling and laboratory tests have been conducted to evaluate circumferential coverage with the array probes, to ensure that significant defects cannot be missed in axial scans. Figure 6 shows the computer-predicted amplitude as a function of circumferential position for a 16 channel C5 probe detecting circumferential and axial cracks. This figure shows a possible 25% discrepancy in signal amplitude with respect to the circumferential position of the defect. The signal amplitude of the probe's data channels is never less than 70% of the maximum, showing that this probe design completely covers the tube circumference for both circumferential and axial cracks. SUMMARY ICONCLUSIONS Computer modeling has shown that TIR probes can detect localized defects in deformed tubes (with lift-off noise) with better signal-to-noise ratios than pancake impedance coil probes. Computer modeling can simulate TIR probe responses to different types of defects, lift-off and deposits. This provides eddy current inspectors with valuable information for analyzing signals from tube inspections. 288

7 TIR probes can be used to detect both axial and circumferential cracks in heat exchanger and steam generator tubes. ACKNOWLEDGEMENTS The authors are indebted to J.P. Gravelle, W. Pantermoller, M.S. Addario and J.R. Carter for performing the required laboratory measurements to validate the computer modeling results presented in this paper. The valuable support of the CANDU Owners Group steam generator technical committee was particularly important. REFERENCES 1. Cecco V.S., Van Drunen G. and Sharp F.L., "Eddy Current Testing Manual, Vol. 1", AECL Report, AECL-7523, 1981, Chalk River, Ontario, Canada. 2. Cecco V.S. and Van Drunen G., "Recognizing the Scope of Eddy Current Testing", in Research Techniques in Nondestructive Testing. Vol. 8, 1985, pp Cecco V.S. and Carter lr., "Recent Advances in Eddy Current Testing of Heat Exchangers", in Proceedings of CNS Steam Generators and Heat Exchanger Conference, April 1990, Toronto, Canada. 4. Cecco V.S., "Eddy Current Inspection of Inconel600 Tubes with Circumferential Cracks", in Proceedings of 1976 ASNT Fall Conference, pp Dodd C.V. and Deeds W.E., "Analytical Solutions to Eddy-Current Probe-Coil Problems", Journal of Applied Physics, Vol. 39, No.6, May 1968, pp Burrows M.L., "A Theory of Eddy Current Flaw Detection", (University Microfilms, Inc., Ann Arbor, Michigan, 1964). 289