Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

IEEE TRANSACTIONS ON MAGNETICS, VOL. 24, NO.

I, JANUARY 1988 435

A FINITE ELEMENT STUDY OF THE REMOTE FIELD EDDY CURRENT PHENOMENON

W. Lord, Y. S. Sun, S. S. Udpa and S. Nath

Electrical Engineering Department


Colorado State University
Fort Collins, Colorado, 80523 USA

Abstract - Conventional eddy current method s o f gives


defect detection in metals are limited t o surface
inspection or thin sections because of the skin J = J e -x/IJ sin (wt ·-x/Il) (2)
o
effect phenomenon. Such limit ions do not seem to
affect the low frequency , remote field eddy cur rent where Jp is the current desnity at the surface of
inspection of pipeline in which the phase difference the ha f spac e and is the skin depth or depth of
of widely spaced exciter and sensor coil signals penetration
a p p e ars to be equally sensitive to internal and
external pipeline corrosion.
This p a p e r desc ribes t h e d i f f e rences in
operation of conventional and remote field eddy at which the current density has dropped to 36.8% of
current nondestructive evaluation procedures , and JO a n d l a g s JO by 1 r a d i a n . Although t h i s
reports on a f in i t e e l ement study whose s o l e definition o f skin depth i s f o r an ideal geometry,
objective was t o provide physical understanding of the effect does tend to limit the application of
the remote field eddy current phenomenon. Finite conventional eddy current testing methods to the
element predictions of probe performance are shown detection of near surface inhomogeneities.
to agree very well with experimental observations In contrast , the remote f i e l d e d d y e f fect
and a l s o i nd i c a t e the p resence of an unusual ( RFEC) NDT p robe, whic h h a s been used for the
"potential valley" and corresponding "phase knot" in detection of corrosion and pitting in oil and gas
the pipeline wall which play a significant role in pipeline , a ppears to be equally sensitive to both
explaining the underlying physics of the method. internal diameter (I.D.) and outer diameter (O. D. )
flaws [2] thus apparently violating the skin effect
INTRODUCTION limitation associated with conventional EC NDT
methods. The RFEC probe does not measure impedance
Eddy current (EC) methods of nondestructive changes as in Fig. I, but r ather compares the steady
testing (NDT) are used extensively in industry for state A.C. phase difference between a low frequency
the determination of conductivity and the detection excitation coil and a sensor coil positioned several
of defects in conducting materials. The first such pipe diameters from the exciter as in Fig. 2.
use of electromagnetic induction was reported by
Hughes in 1876 [I] and apart from the use of more
sophisticated instrumentation, the basic operating
procedure has remained unchanged. Fig. 1 shows the
elements of a conventional single frequency eddy SENSOR
EXCilI R
c u rr e n t NDT S y s t e m i n w h i c h c h a n g e s i n t h e COlI. COIL
excitation coil impedance are used t o indicate the
presence of material inhomogeneities •

..c.
""""'"

IBM SYSTEM 9000


FREQUENC'( &: G.\IN SEl£CT 10-----'
STEPPER t.lOTOO CONTROL
DATA LOGGING

Fig. 2. R F EC probe operation involved a comparison


of exciter and sensor coil voltages. The
Fig. !. C o nv e ntio n a l e d d y current NDT method. relative steady state, A.C. phase difference
Variations in p r obe coil impedance due to is sensitive to both 0.0. and I . D . p i p e
the presence of material inhomogeneities wall defects.
unbalance the bridge and gi ve r i s e to a
s i g n a l c h a racteristic of the particular
defect type.
The sensOr coil voltage is very low, (microvolts) as
one might suspect, and the phase measurement i8
Considering the idealized situation of an infinite sensitive to both O.D. and I.D. flaws [2] . The
A.C. current sheet over a conducting hal f space, major purpose of this paper is to report on a finite
solution of the diffusion equation element study of the RFEC probe which sheds s ome
light on the underlying physics of operation.

UJ � a J (1)

OOI8-9464/88/0100-0435$01.00©1988 IEEE
436

FINITE ELEMENT MODEL RFEC SIMULATION

Electromagnetic induction phenomena associated Figure 3 s h o w s a comparison of the c o d e


with conventional and remote field eddy current NDT
predictions with experimental measurements for RFEC
methods are describable by the quasi-static form of p robe magnitu de, phase angle and defect signal,
Maxwell's equations because of the negligibly small
c o n f i r m i n g the v a l i d i t y of the finit e element
displacement current term. In differential operator
approach.
notation

A) -j A
Fig. 3
'V " (� 'V * = J W (5

where A is the magnetic vector potential in phasor
Similar exper imental observations have been
reported by Sch midt [2]. It should be noted from
v e ct or form. Two dimensional, axisymmetric and
Fig. 3a) that the sensor coil magnitude changes
t hree dimensional finite element s o l uti o ns of
conventional Ee NDT problems have been repor ted in
rapidly with distance from the ex citer coil until
approximately two pipe diameters, after which the
t h e l i t e r a t u r e [3-9) i n v o l v i n g the u s u a l
ch a n ge is mu ch slower . Sensor coil p h ase angle
discretization, energy functional minimization and
(relative to that of t he ex citer coil) c h a n g es
global ma trix inversion steps. Such code has been
abruptly in the same regions (Fig. 3b). The "double
u sed p r i m a r i l y f o r p r o b e design [6,7], t h e bump" phenomenon as sociated with defect detection
simulation o f test geometr ies t o o c o m p l ex for (Fig. 3c) has a spacing equal to that of the exciter
laboratory repli cation [9] , an d as a test bed for to senSOr coil distance.
t h e ge n eration o f p r obe signals used i n t h e
calibration of prospective signal processing schemes
(10).
Because both conventional EC NDT methods and
RFEC p robe beha vior ar� governed by t h e same
eq uation (4), it waS felt that a finite e lement
study of the RFEC phenomenon might shed some l ight
on the apparent violation of skin effect discussed
in the introdu ction. To this end, an axisymmetric
version of eddy current finite element code was
mod ified and used to study the basic exciter/sensor
geometry in a ferromagnetic pipe (see F ig . 2).
Space limitations preclude a full discussion of code
d eta i l s . In for mation is a vailable however in
references [4] to [8]. I'
...... I

--1.- /

1.0 zoo ._�_ -L�_ _

.S
1. Finite Element '50

2. Experimental
-.S 100

-l.0 (1)
.0
-1.5
----

-2.0
===========--
-2.5
-so

-'3.0

_s. s �----.:3,,-"( -100

-4.0
-ISO
-1.S

-200
-5.00 '0

'00
, ii i

00

(2)
00

(1)

-20 1. Finite Element

-40
2. Experimental

-00
Pipe 1. J. = 1. 5", \Jall Thickness � 0.2"

-00
f = 40 Hz, Jlr = 250,"= 0.7 x 10 0--m)

Fig. 4. Instantaneous f l ux d i s t r ibuted for w t


0 0
in 30
0
Fig. 3. Finite element predictions of RFEC probe a) values varyi n g from 0 to 150
magnitude, b) phase and c) defect signal increments. Pipe J.D. = 1.5", f = 40Hz,
compared with experimental measurements from w�IP th ic k!!iSS = 0.2", � r = 250, Cf = 0.7 x
the test rig of Fig 2. 10 (Q-m) •
437

Instantaneous flux lines for the RFEC geometry


are shgwu in Fig. 4 for six differenJ values of t J

Ht) = 1 sin wt
M
o
in 30 increments from 0 to 150 , as sum1ng the
exciter coil current

(5) ::::::::-.s..
/1 ,

The flux lines are plotted on a logarithmic basis in

\ \\\\
order to show details of the remote or far field
several pipe diameters away from the exciter coil.

I
For exam8le with maximum excitation coil current a t
t = 90 the four bands o f flux lines contain 90%,
9%, .9% a n d .09 % o f t h e t o t a l f l u x w i t h

I
cor responding contour intervals of 10%, 1%, .1% and
.01% respectively. The magnetic flux lines decay

a)
very rapidly close to the exciter (in the "near

field" region) and in a much s l ower, more uniform


manner several pipe diameters away from the exciter
(in the "remote" or "far field" region). In the
transition region one to two pipe diameters from the
exciter coil, the flux lines undergo rather dramatic
changes. Similar conclusions can also be drawn from
the RMS magnetic vector potential magnitude a n d
phase p l ots of Figs. 5 and 6 which clearly show the
existence of a "potential valley" (where RMS [A]
values are zero) and "phase knots" (regions in the
pipe wall where the A phase is undefined) one to two
pipe diameters from the exciter coil.

b)

Fig. 6. Phase Plots corresponding to Fig. 5 for a)


40Hz and b) 400Hz showing the existence of
"phase knots" in the transition region.

Fig. 7. F i nite e l e ment p r ed iction of P oynting


vector real energy directions in the RFEC
geometry.

A Poynting vector plot (Fig. n s h o w s that t h e


transition region occurs where outwardly directed
energy from the exciter coil meets inwardly directed
energy from the outer pipe wall region. This gives
rise to two skin effect zones. Close to the exciter
coil the electromagnetic fields decay exponentially
from LD. to O.D., in the remote field beyond the
transition region, the fields decay exponentially
Fig. 5. RMS magnetic vector potential lines for a) from O.D. to LD.. It is interesting to note that
40Hz and b) 400Hz showing the existence of a the "phase knots" of Fig. 6 are related to skin
"potential va11ey" in the transition region depth in the radial direction. Figures 8 and 9 show
o n e t o t w o p i p e d i a m e t e rs f r o m t h e instantaneous flux plots for O.D. and I.D. pipe
excitation coil. slots in the remote field regions, from whic h it can
be seen that a sensor coil passing through such
fields would give rise to similar signals.
438

which show the existen c e of un ique " potential


valley" and "phase knot" phenomena in the transition
region between RFEC exciter and sensor coils. The
study also confirms the experimental observation
that an RFEC probe is equally sensitive to both O. D.
and I.D. pipe wall defects.

AC KNOWLEDGEMENT

This work has been supported by a research


grant from the American Gas Assoc i a t i o n . Th e
authors are also grateful to Tom Schmidt for his
continued interest in the project work and his
insight into RFEC physic s. Y.S. Sun is on leave
frOm Nanjing Aeronautical Institute, PRC.

REFERENCES

[1] D . E . H u g h e s , "I n d u c t i o n - b a l a n c e a n d
expe r i m e n t a l r e s e a r c h e s t h e r e w i t h ,"
Phi losophical Magazine, Series 5, Vol. 8, 1879,
Fig. 8. RFEC instantaneous magnetic flux contours pp. 50-57.
around an 0.0. slot in a 24" diameter pipe
,wall 4.3 pipe diamters � om th_i. e xc i ter [2] T. R. Schmidt, "The remote field eddy current
coil. ]J = 80, cr = 1 x 10 (Q m) •
inspection techniques," Materials Eyaluation,
r Vol. 42, Feb. 1984, pp. 225-230.

[3] W. Lord, "Ap plication s of numer i c a l f ie l d


m o d e l i n g t o e l e c t r o m a g n e t i c methods o f
nondestructive testing," IEEE Transactions on
Magnetics, Vol. 19, 1983, pp. 2437-2442.

[4] R. Palanisamy and W. Lo r d , "Fi n i t e element


analysis of eddy current phenomena", Materials
Evaluation , Vol. 18, 1980, pp. 39-41.

[5] N. Ida and W. Lord, "Gra phical simulation of


electromagnetic NDT probe fields," IEEE Computer
�hics Applications, 1983, pp. 21-28.

[6] N. I da, K. Betzold and W. Lord, "Finite element


mo deling of a b s o lute e d d y current p r o b l e
signals", Journal o f Nondestructive Eyaluation,
Vol. 3, September 1983, pp. 147-154.

[7] N. Ida, R. Palanisamy and W. Lord, "Eddy current


probe design using finite el ement analysis,"
Materials Eyaluation, Vol. 41, 1983, pp. 138 9-
1394.

[8] N. Ida and W. Lord, "A finite element model for


3 -D e d d y c u r r en t N D T p h e n o m en a , " l.�
Transactions on Magnetics, Vol. 21, November
Fig. 9. Finite element prediction of instantaneous 1985, pp. 26355-2643.
flux contours around an 1.0. defect for the
same conditions as Fig. 8. [9] N. Ida, H. H o s h ikawa and W. Lord, "Finite
element prediction of differential eddy current
probe signals from Fe 0 deposits in PWR steam
� 4
From a defect characterization point of view Figs. 8 g enerators," NOT International, V o l . 18,
and 9 might indicate th at defects such as O.D. and December 1985, pp. 331-338.
I.D. corrosion would be difficult to differentiate.
Howev er, new signal processing techniques [11] show [10) S. S. Udpa and W. Lord, "A Fourier descriptor
promise for solving this problem. classification scheme for differential probe
signals," Materials Eyaluation, Vol. 42, August
CONCLUSIONS 1984, pp. 1136-1141.

Operation of the RFEC probe is governed by the [ll] S. S. Udpa et aI., "Signal processing for remote
quasi-static form of Maxwell's equations and as such field eddy current inspection technique," to
can be modelled by conventional eddy current finite appear in Review of Progress in Quantitative
e lement c o d e . In an attempt to understand the NDE, D. O. Thompson and D.E. Chimen t i Eds.
underlying physics of the remote field effect, this Phenum Press, New York, 1988.
pap e r pr esents r esults of a finite element study

You might also like