EDDY CURRENT NDT
AND DEEP FLAWS
F. THOLLON
CEGELY - UPRESA 5005
Ecole Centrale de Lyon
69131 Ecully Cedex-France
E-mail : thollon@trotek.ec-lyon.fr
Abstract Eddy current NDT and NDE are ever more
used for applications that need important sensitivity and resolution. The
design of eddy current devices by must be realized by using modelization.
The large number of electromagnetic and geometrical parameters of these
systems implies many simulations, and the quantity of obtained results
does not allow to conceive rapidly the most effective system. To achieve
such a design, we need in particular two tools, a 3D model of flaws and
an automatic optimization algorithm. We propose an experimental equipment
for validating a fast 3D flaw model.
I. INTRODUCTION
Fig. 1 shows an example of structure to control in
aeronautical industry. This is a screwed or riveted assembly.
The materials are aluminum alloy for the sheets and titanium for the
screws. Flaws can appear at great depth
(below the first sheet) and along a known direction. In this case,
ultrasonics cannot be used, because of a seal between
the two sheets. We have to use in this case eddy currents method to
control the assembly. Because of the great depth
of the flaw, the obtained signal is very low-level. The aim of this
problem is to be able to optimize the dimensions of the
coil and the excitation current to have the highest level of the signal
as possible.
A probe has been designed in our laboratory. It
has been demonstrated that pulsed eddy currents [1] must be used and
that the measured quantity must be the flux density at the surface
of the assembly. These results have been obtained
with the following methodology.
First, a 3D study has been necessary to understand
eddy currents distribution near the flaw and the influence of
this perturbation on sensor response [2]. It has allowed to conclude,
in particular, that flux induction measurement is
more effective than impedance measurement. Unfortunately, solving time
are very great, and only sinusoidal current
canbe modelized.
Fig. 1: eddy current NDT of screwed assemblies.
In a second step, an automatic design has been executed
[3]. Because of the two previous remarks, a genetic
algorithm has been implemented in a 2D FEM electromagnetic package.
The obtained results show that this
method is good adapted. Indeed, geometrical parameters of the sensors
and current shape are automatically obtained.
Moreover, one of the main advantage of this method
is its easy implementation.
This methodology has been validated experimentally.
The designed sensor is more effective than those already existing.
But, it has two main inconvenients:
1.The use of 3D only for the knowledge of the physical
phenomena for a sinusoidal current.
2· The use of automatic design only with
a 2D model.
To improve the efficiency of the sensor in particular:
· to be able to characterize the flaw (depth and sizes),
· to increase the detection depth, a 3D automatic optimization
must be envisage. Therefore,
we must have:
· An fast model of a 3D flaw with pulsed or sinusoidal current.
· An automatic optimization method.
The main objective is to have a tool fast enough
to design efficiently an eddy current probe. We propose a
test equipment for validating the flaw model with the aim of characterizing
it.
II. 3D FLAW MODEL.
The eddy currents are created by a cylindrical coil.
The screw and the coil are coaxial. During the control, the coil
doesn't move.
The measurement is made with two Hall effect sensors
(HES) located symmetrically in relation to the axis of the
screw. These HES measure the horizontal radial component of the flux
density. One HES is above a region without a
flaw, the other, above a region supposed with a flaw (the flaw and
the HES are in the same plane). The measured
quantity is the difference between the tensions provided by the two
HES. This quantity represents the horizontal
differential flux density (HDFD).
Fig. 2 : Coil effect with HES and aluminium cylinder with a flaw
The studied structure is simplified as shown Fig. 2: an eddy current
probe above an aluminum cylinder with an
artificial flaw. The screw is replaced with air. The cylinder sizes
are given table 1.
1) Eddy current probe.
This is "a pancake coil" [2]. The coil sizes and the position of the
HES are given table 2. The maximum value
for a sinusoidal current is 500mA RMS, and for a pulsed current, 5A
peak. The sensor has been calibrated:
signal (V)=27.74((23.61B1-0.0012)-(23.606B2-0.001))
where signal is the measured quantity, B1 the horizontal
radial flux density measured by one HES and B2 the horizontal radial
flux density measured by the other one.
2) Standard flaws.
In order to characterize the flaw, three standard flaws, at the same
depth, with the same volume but with different
dimensions are used. The lengths, heights and depths are given table
3. The width of each flaw is 0.3mm.
3) Coil supply.
The coil can be supplied either with a sinusoidal current, either with
a pulsed current. We have a current generator
with a 10kHz bandwidth and a maximal value of 7A. It is controlled
with a tension that has the same shape than the
desired current.
B. Item to compare.
We wish to compare HDFD for a current described
Fig. 3. In this case, only the current turn-off is interesting
(separation between excitation and material response).
Fig. 4, Fig. 5 and Fig. 6 give the experimental
responses obtained with the three flaws. They represent HDFD once
the current has been turned off. Numerical values are given table 4.
Only high frequencies have been rejected. No other
numerical treatments are performed.
C. Discussion.
The following elements are obvious points of discussion:
· Formulations used in the different regions.
· Interface conditions.
· Computational time.
· Geometrical description of the flaw (thickness).
· For impulse excitation, solving method.
III. REFERENCES.
[1] F. THOLLON, N. BURAIS: "Modeling and characterization of Pulsed
eddy currents. Application to detection of flaws in riveted assemblies
in aircraft
structures," Electric and Magnetic Fields From Numerical Models
to Industrials Applications, Ed. Nicolet and Belmans, Plenum Press,
pp. 261-264, 1995.
[2] J. L. RASOLONJANAHARY, F. THOLLON, N. BURAIS, X. BRUNOTTE, "Study
of Eddy Currents NDT Systems in RivetedAssemblies," IEEE Transactions
on Magnetics, Vol. 32 n° 3, may 1996, pp. 1585 -1588 .
[3] F. THOLLON, N. BURAIS: "Geometrical optimization of sensors for
eddy current NDT and NDE ," IEEE Transactions on Magnetics, Vol.
31
n° 3, may 1995, pp. 2026-2031.
Fig. 3: Excitation signal characteristics. 2
Fig.4 Response to a current turn-off
Flaw 3 characteristics (length*heigth) : 1*4mm
Fig.5 Response to a current turn-off
Flaw 2 characteristics (length*heigth) : 2*2mm
Fig.4 Response to a current turn-off
Flaw 3 characteristics (length*heigth) : 4*1mm
|
 |
|
|