A Pulsed Eddy Current Probe for Inspection of CANDU® Reactor

Steam Generator Tubes

J. Buck1,2, P. R. Underhill1, S. G. Mokros1,2, J. Morelli2, T. W. Krause1

V. Babbar1,3, B. Lepine3, J. Renaud3

1Department of Physics, Royal Military College of Canada, Kingston ON2Department of Physics, Engineering Physics and Astronomy, Queen’s University, Kingston ON3Atomic Energy Canada Limited, Chalk River Laboratories, Chalk River ON

j.buck@queensu.ca

Outline

• Description of Pulsed Eddy Current (PEC) testing

• Explanation of Principal Components Analysis (PCA)

• Overview of experimental setup

• Results:

– Isolation of fret signals at SS410 simulated support plate

– Fret depth sizing for various SS410 support plate hole IDs

– Shift detection and hole ID sizing without frets

• Continuing work

Motivation

• In-service inspection of CANDU steam generator

ferromagnetic drilled plate and broach support

structures:

– Gap measurements to identify corrosion of support

structures in the presence of tube fretting

– Detection of magnetite fouling at support structure

locations

– Development of a robust analysis technique to be used

for inspections

Pulsed Eddy Current Testing• A drive coil is excited using a square voltage pulse

• Transient eddy currents are produced via Faraday’s

Law of electromagnetic induction:

• Magnetization of ferromagnetic support structures under

approach to constant field conditions amplifies induced eddy

current response

• Transient signals are measured using pickup coils and analysed to

determine condition of steam generator tubes and surrounding

support structures

0.5 ms Transient Response

Conventional and Pulsed Eddy

CurrentConventional EC Pulsed EC

Sinusoidal excitation Square pulse excitation

More generic probe designs Probe tailored for inspection

Sensitive to small liftoff variations Less sensitive to small liftoff variations

Insensitive at large liftoffs (>5 mm) Effective at large liftoff (~10-30 mm)

Challenged in the presence of ferromagnetic

materials

Large magnetization component to the

signal

Depth of penetration limited by skin depth

and coil dimensions

Greater depth of penetration (approach to

DC) and field lines extend deeper

Pickup Coil Response

• Minimal change in time-domain signal variation in the

presence of different tube conditions

-0,5

0

0,5

1

1,5

2

2,5

3

0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4

Vo

lta

ge

[V

]

Time [ms]

Tube

Fret

Collar

Principal Components Analysis• Statistical method of turning highly correlated data into a

set of linearly uncorrelated variables called principal

components

• Each principal component accounts for the maximum

possible variance in the original data under the constraint of

orthogonality

• The dimensionality of the data can be reduced to a few

principal components and associated scores, simplifying the

analysis while maintaining most if not all of the original

information

0

1

2

0 0.05 0.10 0.15 0.20

Experimental DataVector 1Vectors 1, 2 and 3

file: e:\research\nsercdata.epfile: e:\research\nsercdata.ep

Time [ms]

Voltage [V

]

-0.2

-0.1

0

0.1

0.2

0 0.05 0.10 0.15 0.20file: e:\research\nsercvectors.epfile: e:\research\nsercvectors.ep

S3

S1

S2

Time [ms]

Eig

en

ve

cto

rs [

V]

• Transient Signal Response rebuilt using orthogonal PCA Eigenvectors

PCA vs. Fourier Analysis

0,01

0,1

1

10

100

1000

0 50 100 150 200 250

Am

pli

tud

e

Frequency Component

FFT 0.65

FFT 0.87

FFT 1.11

FFT Tube

Fast Fourier Transform (FFT) of various signals

PCA vs. Fourier Analysis

0,01

0,1

1

10

100

1000

0 5 10 15 20 25 30 35 40 45 50

Am

pli

tud

e

Component

PCA 0.65

PCA 0.87

PCA 1.11

PCA Tube

FFT 0.65

FFT 0.87

FFT 1.11

FFT Tube

Fast Fourier Transform (FFT) of signals

PCA of signals

Probe Design• Probe designed by NDT group at RMCC

• Drive coil wound coaxially with the probe body

• Pickup coils are arranged in sets of 4 at 90° intervals both in front

and behind the drive coil

• All pickup coil axes are perpendicular to the axis of the drive coil

and probe body

Front

Experimental Apparatus

SS410 Collars (with IDs) for

simulating varying degrees

of corrosion in support

plate structure

Micrometer apparatus

Sample fretted Alloy-800 SG tube

18.0 mm 18.8 mm

20.3 mm 21.8 mm

Collar and Fret Signal Separation• Ferromagnetic collar appears in S1 with a FWHM of 26 mm in both experiments

• Shape characteristic of a fret appears here in S3 in both experiments

• S2 appears to combine effects seen in S1 and S3

• PCA provides separation of feature/flaw signals at ferromagnetic structures

Fret Depth Sizing• Frets can be identified at ferromagnetic support structures using

translational measurements as shown previously

• Frets were located at the SS410 collar, with the tube centered

within the collar using the micrometer apparatus

• Five different fret depths were examined in combination with

four different collar ID sizesFret

Number

Fret Depth

[mm]

1 0.65

2 0.74

3 0.87

4 1.00

5 1.11

Collar

Number

Collar ID

[mm]

2 18.0

3 18.8

4 20.3

5 21.8

Fret Depth Sizing• Fret depth can be related to S2 with a centered tube

• The S2 fret depth relation appears to be independent of collar ID,

which simulates uniform corrosion of the drilled supports

-0,8

-0,6

-0,4

-0,2

1E-15

0,2

0,4

0,6

0 0,2 0,4 0,6 0,8 1 1,2

S2

[A

rbit

rary

]

Fret Depth [mm]

Collar 2

Collar 3

Collar 4

Collar 5

Mean

Poly. (Mean)

Fret Depth Sizing• Finite Element modelling data generated using a COMSOL model underwent

PCA to be validated by experimental results

• After normalization model provides excellent quantitative agreement with

experiment for fret depth measurement using S2

y = 1,9346x2 - 0,5183x - 0,9301

R² = 0,9896

y = 1,9759x2 - 0,4114x - 1,0054

R² = 0,9902

-1,5

-1

-0,5

0

0,5

1

1,5

0 0,2 0,4 0,6 0,8 1 1,2

S2

[A

rbit

rary

]

Fret depth [mm]

COMSOL

Experiment

Poly. (COMSOL)

Poly. (Experiment)

Tube Shift and Collar ID Sizing• Probe centered within the SS410 collar

• Nominal Alloy-800 tube shifted horizontally from one

side of the collar to the other in 0.25 mm increments

• Four different collar sizes used to simulate different

amounts of support structure corrosion

Tube Shift and Collar ID Sizing• Peak voltages for the different shift positions within the

SS410 collar were combined differentially

• Offset at the centered position stems from coil imbalances

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

0,4

0 0,5 1 1,5 2 2,5 3

Dif

fere

nti

al

Vo

lta

ge

[V

]

Tube Shift [mm]

Diff 3-7

Diff 4-8

Poly. (Diff 3-7)

Poly. (Diff 4-8)

• S1 vs. S2 for the four different collar IDs

• S2 provides a measure of the tube shift from a centered

position while the distance between curves can be used to

determine collar ID

Tube Shift and Collar ID Sizing

27,00

27,50

28,00

28,50

29,00

29,50

-4,00 -3,00 -2,00 -1,00 0,00 1,00 2,00 3,00 4,00

Sco

re [

S1

]

Score [S2]

C5c37

C4c37

C3c37

C2c37

Poly. (C5c37)

Poly. (C4c37)

Poly. (C3c37)

Poly. (C2c37)

Tube Shift and Collar ID Sizing• The minimums of the S1 vs. S2 curves have been plotted against

collar ID data sets collected using two different coil pairs

• Suggests that using calibration vectors would allow for ID sizing

25

25,5

26

26,5

27

27,5

28

28,5

29

29,5

17 18 19 20 21 22 23

S1

Min

ima

[A

rbit

rary

]

Collar ID [mm]

C37

C48

Poly. (C37)

Poly. (C48)

Summary• A Pulsed Eddy Current Probe has been developed for evaluation

of ferromagnetic support structures in CANDU® steam generators

• Principal Components Analysis (PCA), applied to acquired

transient signals, demonstrates capability to separately measure

gap and position (shift) of SG tube, which can be used to evaluate

condition of surrounding support structure

• PCA also shows capability to separate out and quantify the effect

of SG tube frets on signals obtained at support structures

• PCA provides a robust and powerful statistical tool for flaw

detection and characterization of transient signals

Continuing and Future Work

• Combine tube shift, collar sizing and fret depth

measurements in a single experiment

• Validate the use of calibration vectors with “blind” tests

• Investigate the effects of probe eccentricity and non-ideal

alignment of pickup coils with flaws

• Investigate the effects of magnetite fouling of the tube ID

and OD on transient signal response

• Examine trefoil broach supports using newly designed probe

Questions?