THICK-WALLED ALUMINUM PLATE INSPECTION USING REMOTE FIELD EDDY
CURRENT TECHNIQUES
Y.S. Sun, W. Lord, L. Udpa, S. Udpa, S.K. Lua and K.H. Ng
Material Characterization Research Group
Department of Electrical and Computer Engineering
Iowa State University, Ames, lA 50011
S. Nath
Amtak, Inc.
ISU Research Park, Suite 800
2501 N. Loop Drive, Ames, lA 50010
INTRODUCTION
The detection of defects that are located deep in thick walled ( >12 mm) alumirrum
plates is of interest to both the aircraft and space industries. Conventional eddy current
(EC) techniques are limited to the inspection of surface and subsurface anomalies. Newly
developed high sensitivity magnetic sensors, such as magnetoresistive e1ements and
superconducting quantum interface devices (SQUIDs) have enhanced the EC technique's
capability. Suchsensors can be used to detect flaws that are located deep in alumirrum
plates. However, inspection of a defect located 12 mm to 25 mm below the surface of an
alumirrum plate is beyond the ability of conventional single frequency EC techniques.
The remote field eddy current (RFEC) technique, that is currently being used for
metallic tubular product inspection, is characterized by equal sensitivity to a flaw
irrespective of its location in the tube wall. In recent few years, it has been successfully
extended to the inspection of metallic plates[1], [2]. A prototype of such a probe designed
for inspecting thick alumirrum plate has been designed and evaluated using finite element
models. Preliminary experimental results obtained support the validity of the approach.
This paper presents some of the finite element (FE) modeling and experimental
results obtained to date. The experimental results were obtained during efforts to inspect a
12.7 mm thick alumirrum plate with a defect machined on the other side ofthe plate
(related to the probe).
Revzew of Progress m Quantitative Nondestructzve Evaluatwn, Val 16
Edtted by D.O. Thompson and D.E. Chtmentt, Plenum Press, New York, 1997
1005
PROBE DESIGN
The key objective in designing the RFEC probe is to ensure that the energy flows
through the plate wall twice.
In the case of tubes, the eddy current induced inside the tube wall restricts the flux
pattem from expanding axially resulting in rapid attenuation of the directly coupled field. In
the case of metal plates the following four steps have been taken to force the energy to
traverse the plate twice [2]:
I. Using a special designed magnetic circuit consisting of a pot-core and an excitation coil.
2. Employing a magnetic circuit to improve the sensitivity ofthe pick-up coil.
3. Using an auxiliary coil to help guide the signal path.
4. Excitation and pick-up coils are shielded to minimize direct coupling between them.
It has been shown by both through FE modeling studies and experimental
measurements that in the case of an RFEC probe for aluminum plate inspection the auxiliary
coil is not necessary, because of the low permeability value of aluminum typical better
surface conditions. Consequently, the size of the probe can be reduced significantly relative to
probe for inspecting steel plate.
A schematic of the probe is given in Fig. I. The probe consists of an excitation coil
wounded inside a pot-core that provides a magnetic circuit for the flux. The parameters of
both the magnetic circuit and the metallic cover are chosen carefully to enable the
electromagnetic energy released from the coils to penetrate downward into the plate. The
pick-up coil senses the electromagnetic field that traverses the plate from the bottom to the
upper surface. The magnitude and phase of the signal are sensitive to the condition (the
thickness, permeability and conductivity) of the aluminum plate.
With the help of a FE model, a prototype has been designed and built for inspecting
aluminum plates up to 25 mm of thickness. The excitation unit is about 100 mm in diameter
and 65 mm in height. Two sensor units, one is with an E-shaped ferrite core the other with a
U-shaped core, similar to those used for inspecting steel plate, are used.
EXCITATIO
aluminum plate
IT
RECEIVER UNIT
d fcct
Fig. 1 Schematic of the probe prototype for thick walled aluminum plate inspection.
1006
NUMERICAL MODEL
A linear, 2-D (axisymmetric) finite elementcodewas used to estimate the optimal
design parameters ofthe probe. A simple 2-D code, capable of providing reasonably good
estimates quickly, was used to arrive at the initial design ofthe probe. The performance of
the probe was evaluated after the probe parameters were established. Figs. 2, 3 show the
magnetic field distributions, equi- ln1(flux magnitude) lines and equi-phase Iines, around
the probe inspecting a 25 mm thick aluminum plate. Fig. 2 shows a case without any defect,
while Fig. 3 shows results obtained with a 20 mm wide, 40% deep slot on the plate on the
far side. The Iift-off for both cases was 2 mm.
A cursory review of the data reveals little difference between the two plots in the
near field (R<60 mm) region. However, if one carefully compares the two plots at the
remote field region undemeath the receiver unit ( 130 mm<R> 170 mm), a noticeable
difference between the two cases can be observed. The difference becomes even more
evident in Fig. 4 where the phase distributions on the upper surface of the plate are
compared. The phase in the 'defect' case Ieads the phase values observed in the 'no-defect'
case in the remote field region. Note that there is a small area around R=160 mm where the
phase is lagging in the case when the defect is present. The lag is due to the special shape of
the core of the pick-up coil.
Equi-ln(magmtudc) plot. no dcfcct. f=400Ht
Equo-phasc plot, no dcfcct. f=400111
ontcrval bctwccn contours = 18 dcgrees
120
E
E
20
40
60
80
100
120
140
R: Oi;tancc from cxcilallon coil ccntcr, mm
160
180
Fig. 2 Magnetic field distribution around the probe used for inspecting a 25 mm thick
aluminum plate without defect.
1
Naturallogarithm.
1007
Equi-ln(magnitude) plol. 40% dcfcct, f=400HL
E
E
Equi- phasc plol, 40'* dcfccl. f=4001 1z
intcrval bet\\cen contours = 18 dcgrccs
E
E
100
R. Di>tancc from cx itauon co•l ccntcr. mm
Fig. 3 Magnetic field distribution around the probe used for inspecting a 25 rnrn thick
alurninum plate containing a 20 rnrn wide, 40% deep defect.
Phase dl\tnbuuon on the upper surfa c of the 25mm plate
404defect
- 200
~
<>
~ -400
()
"0
~-600
no defect
"E.
-00
defe
140
120
100
0
Duance from eAcitauon cotl center, mm
- 1~ ~----~----~----~-----L-----L~--~~--~----~----~
20
40
60
160
10
200
Fig. 4 Comparison of flux-phase distributions right at the upper surface of the plate for the
'no-defect' and the 40% deep 'defect' cases.
In theory, it should not be possible to obtain any signal response to a defect if an
axisyrnrnetric FE code is used, since the distance between the probe and the defect is fixed.
Neither the probe, Jocated along the R axis, nor the defect, which is a circular slot, vary in
location. A 3-D code is required to simulate the geometry accurately. However, for an
initial probe design the reduction of computation time represents a higher priority over
accuracy, Consequently following approximationwas used in arriving at the solution.
1008
If we assume that the scan length LlR is much smaller than the average radius of the
defect, Rter, we can ignore the change in defect radius with the radial distance. The
assumption allows movement of a defect within LlR and enables the estimation of the probe
responses.
The responses to two defects in a 15 mm thick aluminum plate areshownon Fig 5.
The scanning was performed directly underneath the receiver unit ( 130 mm < R > 170 mm
). Their signal phase to ln(signal magnitude) trajectories are compared in Fig 6.
EXPERIMENTAL MEASUREMENTS
A 2 mm wide, 25 mm lang and 50% deep slot was machined on the bottarn side of
a 12.7 mm thick a1uminum plate. The probe excitation unit and receiver unit are
mechanical1y coupled, with a 25 mm gap in between. Note, the probe is scanned along the
length of the defect, while the defects in the FE simulation were perpendicular to the scan
direction. An 200Hz, 0.5 Ampere AC current was applied to the excitation coil. A lock-in
amplifier, Model ITHACO 3981A, was used for measuring the signal magnitude and phase.
Fig. 7 shows the results when an E-shaped core is used in the receiver unit. Fig. 8. shows
the results when a U-shaped core is used.
The phase to ln(magnitude) trajectories of the signals obtained using E-shaped and
U-shaped cores are compared in Fig. 9.
ln( ignnl mngnitudc) in E- hnpc fcrritc cotl. 40% dcfcct nnd O'l dcfcct, f=400 11t
- 24r----.----,-----,----,-----,----.----,-----.----.---- .
-24.5
:g
-25
2
-~-25.5
~
-26
- 26.5
_____ L_ _ _ _L __ __ J_ _ _ _
-27L---~
140
142
-L----~--~~---L----~--~
144
146
148
150
152
ignnl phasc m E-shape ferrne cotl. 40% dcfect and 8
154
156
15
160
defcct. f=400H7
50r----.----,-----,---~-----r----~---,-----.----.----.
0
"~
".;
-50
"Q
~
5:
-100
142
144
146
14
150
152
Dcfcct ccntcr locatton, mm
154
160
Fig. 5 Simulated signal responses of the probe inspecting a 15 mm thick plate to 40% and
80% deep defects.
1009
ignal to ln(signal magnitude) trajectoric for the "40% defect" and "80% defect"
ca~es
50.------.------.-------.------.------.------.
0% defecr
0
"' -SO
~
'-'
"0
.;
:(!
s: -100
-150
______
-200L-----~------_L
-27
-26.5
_ L_ __ _ _ _~------~----~
-26
-25.5
ln( ·ignal magnitude)
-2
- 24.5
-24
Fig. 6 Comparison of signal phase vs. ln(signal magnitude) trajectories for the '40% defect'
and '80% defect' cases.
E - SHAPE S
)( 10"'
NSOR
150
100
~
~
5C
~
50
0
I eC
I
50
100
- 200
0
2
3
a
9
10
Fig. 7 Signal obtained by scanning the specimen with a receiver using an E-shaped core.
1010
U - SHAPE SENSOR
)( 10 ..
4 .2
4
- 10 r-----r-----r-----r-----r-----r-----~----~----~----~----.
- 15
DOIUC::t
- 45o
~----L-----.,~----3~----~----~s----~--~~7----~a----~----_J
,o
dlstanco. c:rn
Fig. 8 Signal obtained by scanning the specimen with a receiver using a U-shaped core.
E- HAPE SENSOR
150
100
50
e
.,
CO
0
-o
...;
..."'
E:
-50
- 100
- ISO
-200
-11.5
- II
- 10.5
- 10
(magnitude), log
- 9.5
-9
Fig. 9 Comparison of phase vs. ln(magnitude) trajectories from signals obtained from a
receiver using an E-shaped core. Signals obtained when the probe scans over the defect.
!Oll
-15
-20
C)
~
CO
C)
"'0
,..:.-25
C)
"'"'
..<:
&
Defecl
- 0
-35
~ L---------~--------~--------~------~
- 7.95
-7.9
(magnilude), log
-7.85
-7.8
Fig. 10 Comparison ofphase vs.ln(magnitude) trajectories from signals obtained from a
receiver using a U-shaped core. Signals obtained when the probe scans over the defect.
CONCLUSIONS
A novel RFEC probe has been designed and built for inspecting thick aluminum
plate employing a finite element model. Modeling results obtained to date show that it can
be used for inspecting a plate with thicknesses up to 25 mm. Preliminary experimental
measurements carried out on a 12.7 mm aluminum plate demonstrated excellent
performance.
REFERENCES
1. Y. S. Sun, S. Udpa, William Lord, and D. Cooley, "A Remote Field Eddy Current NDT
Probe for Inspection of Metallic Plates", Materials Evaluation, Vol. 54, No. 4, April
1996, pp. 510-512.
2. Y. S. Sun, S. Udpa, William Lord, and D. Cooley, "Inspection ofMetallic Plates Using
A Novel Remote Field Eddy Current NDT Probe", Review of Progress in Quantitative
Nondestructive Evaluation, Vol. 15A, Edited by Donald 0. Thompson and Dale E.
Chimenti, Plenum, 1996, pp. 1137-1141.
1012