Defence Science Journal, Vol. 59, No. 2, March 2009, pp.

106-112
2009, DESIDOC

Development of Eddy Current Test Procedure for Non-destructive Detection of

Fatigue Cracks and Corrosion in Rivets of Air-intake Structures
B. Sasi, B.P.C. Rao, T. Jayakumar, and Baldev Raj
Indira Gandhi Centre for Atomic Research, Kalpakkam603 102

ABSTRACT
Non-destructive detection of defects in countersunk of rivets in multi-layer air-intake structures is essential
for ensuring structural integrity and flight safety. This paper presents an eddy current test procedure developed
for reliable detection of simulated fatigue cracks and corrosion products in rivets of air-intake structures. This
procedure is capable of reliably detecting 0.25 mm deep defects in 4 mm dia rivets and 0.75 mm deep defects
in 5 mm dia rivets. Further, it is not influenced by thickness of the multilayers.
Keywords: Air-intake structures, rivets, eddy currents, fatigue crack, hidden corrosion

1. INTRODUCTION
Non-destructive testing (NDT) of critical components high stress concentration at the countersunk region of
of aircraft, especially airframes and aero engines, is essential the rivet heads and promote initiation of fatigue cracks.
for ensuring structural integrity and flight safety. The Small fatigue cracks grow with time radially through the
NDT of aircraft in service has assumed greater importance rivet head and cause the tear-off, resulting in catastrophic
after the Aloha accident of 1988 and special emphasis has failures. Typical portion of a riveted structure from a
been placed on detection of fatigue cracks and hidden Cat-E aircraft is shown in Fig. 1. Van der Walde 8, et al.,
corrosion in airframe structures and in rivet heads 1,2. Among studied the corrosion-assisted fatigue crack in 2024-T3
various NDT techniques, visual inspection using boroscopes aluminium sheets for different cyclic loadings, and found
and fibroscopes is routinely carried out. This technique that crack initiation was predominantly due to stress
has limitation for detection of sub-surface defects and amplification at corroded regions and joining of microcracks.
can not be used for sizing defects 3 . Eddy current (EC) Early and reliable detection of fatigue cracks and corrosion
testing is well-suited for testing airframes and aero engine in the countersunk region of the rivet heads through
discs and is routinely used during quality assurance, maintenance, periodic NDT is essential and is expected to ensure
and life extension of aircraft 4. EC technique is known for flight safety.
its ease, versatality, speed, and non-contact nature. Several Development of eddy current test procedure for detection
developments have taken place in the recent years for of fatigue cracks as well as corrosion in countersunk
applying this technique to riveted panels in aircraft structures. region of the rivet heads is desired. In this direction,
Sliding probes are developed for detection of fatigue studies have been focused on detection of cracks in rivet
cracks in the skins of the riveted panels 5 . For detection heads, and to examine the influence of corrosion products
of fatigue cracks around the rivet heads and measurement in defect regions. Further, studies have also been carried
of residual stresses, self-nulling rotating probes are developed6. out to study the effect of thickness of multilayers on the
To detect short-fatigue cracks under the skins of rivet eddy current procedure. In this direction, riveted multilayer
heads, as a part of damage tolerance assessment, Hegemaier7, aircraft structures with machined grooves in the countersunk
et al., studied the performance of various eddy current region of the rivet heads, simulating the fatigue cracks
systems and probes using electro-discharge machining and with a few grooves filled with aluminium powder in
(EDM) notches. a glue matrix (simulating corrosion) have been studied.
In the recent past, extensive investigations on failures This paper discusses details of these experimental studies,
occurred in MiG-21 aircraft revealed that the failures optimisation of test parameters, and results of the
were caused primarily due to the tear-off of rivet heads investigations towards development of an effective procedure
in the airframe structures, thus disturbing the structural for NDT of air-intake structures.
integrity of air-intake structures. The tear-off of rivet
heads was attributed to the presence of fatigue cracks 2. PRINCIPLES OF EDDY CURRENT TESTING
in the countersunk regions of the rivet heads rather than The EC testing works on the principles of electromagnetic
in the fuselage skins. The sharp edges in the skins provide induction and involves measurement of change in coil
Received 14 May 2008

106
 SASI, et al.: NON-DESTRUCTIVE DETECTION OF FATIGUE CRACKS AND CORROSION IN RIVETS

Figure 2. Principle of eddy current testing.

Figure 1. Photographs of a portion of riveted panels from a
failed aircraft (Cat-E).
etc also influence the impedance change. It is essential
impedance that arises due to perturbation of eddy currents to suppress the impedance changes due to these disturbing
at defect regions. In EC testing, an alternating current variables for successful detection and sizing of defects.
(frequency, f, in the range of 50 Hz 5 MHz) is made to Proper selection of probes and optimisation of test parameters
flow in a coil (also called probe) which in turn produces such as test frequency, phase angle, and gain is important.
an alternating magnetic field around it. When the coil is In many situations, multi-frequency methods are also
brought close to an electrically-conducting material, eddy adopted10 . Depth of penetration of eddy currents, in other
currents are induced in the material. These eddy currents words, the depth of inspection, is set following the classical
are generally parallel to the direction of coil winding as skin-effect phenomenon. For conducting materials such
shown in Fig. 2. Presence of any defect in the material as duralumin, lower excitation frequencies are used.
such as cracks, corrosion, wall thinning or any other
discontinuity which affects the electrical conductivity 3. DEVELOPMENT OF EDDY CURRENT TEST
(s) and magnetic permeability (m) of the material, causes PROCEDURE
distortion to eddy current flow and, in turn, changes the For developing the eddy current test procedure, EC
coil impedance which is a complex quantity with real (H) instrument (MIZ-20A, M/s Zetec Inc, USA) interfaced
and imaginary (V) components. to a personal computer with facility for data acquisition,
The locus of impedance change during the movement storage, and analysis was used (Fig.3). A shielded ring
of a probe coil over a test object is called an EC signal. type EC probe (6 mm inner dia) consisting of driver/
While the amplitude of EC signal provides information pick-up coils stacked one above another and having an
about the severity of the defect, the phase angle wrt to operating frequency in the range of 1 kHz-20 kHz was
lift-off (distance between coil and object surface) provides employed. This type of probe gives very good electromagnetic
information about the depth of the defect 9 . Phase angle coupling and minimises the operator-induced signals due
information of the signals is effectively used in EC testing to improper centering of probe over the rivet heads.
to discriminate desired and undesired variables. Reference During testing, the probe is kept over the rivet heads and
standard defects are used for assessing the EC test sensitivity. the imaginary (V) component of the impedance changes
Calibration graphs between EC signal parameters and defect during the probe lift-off are measured and analysed.
dimensions are used for sizing unknown defects.
Apart from defects, other variables such as wall 3.1 Specimen Details
thickness variations, surface roughness, edge-effect, For simulating the actual air-intake structures, six

107
 DEF SCI J, VOL. 59, NO. 2, MARCH 2009

Some of the grooves were filled with a mixture of aluminium

powder and glue (GA) simulating the corrosion. The
dimensional details of the panels and the grooves are
given in Table 1 and the nomenclature followed for the
grooves in the rivets is depicted in Fig. 4. In brief, 5 0.75
refers to a 0.75 mm deep groove in a 5 mm dia rivet. The
dimensions of the notches were arrived at based on the
expected defects and the design considerations.

3.2 Experimental Details

As discussed earlier, selection of test parameters is
important for minimising the interferences from undesired
Ring probe variables so that defects are detected reliably. Three important
(Reflection type) parameters are: test frequency, gain, and phase angle. In
general, test frequency is chosen such that maximum
Riveted panel amplitude signal is produced for defects, and at the same
time, with good phase separation from interfering signals.
In this study, lift-off signal was utilised for detection.
Probe was placed on a good (defect-free) rivet head and
the phase angle of lift-off signal was set along the x-axis
Figure 3. Reflection type ring probe connected to EC instrument (real component, H) on the instrument screen. The test
for detection of defects in countersunk regions of frequency, gain and phase angle in the EC instrument
rivets. were optimised such that:
sets of riveted duralumin sheets of different thicknesses (i) Lift-off signal amplitude due to grooves in rivets is
(2 or 3) were assembled. The rivets were of 4 mm and maximum wrt to defect-free rivets, and
5 mm dia. To simulate the fatigue cracks in rivets, artificial (ii) Signal phase angle between grooves in rivets and
grooves (G) of four different depths (0.25 mm width) defect-free rivets is maximum.
were machined in the countersunk regions of the rivets. The EC probe response at different frequencies for

Table 1. Details of machined grooves in rivet heads

Rivet Layer thickness Number of rivets Groove fill Panel
diameter, combinations, status number
(mm) (mm) *40.25 40.5 40.6 40.75
2 2 2 2 G
1.2+2.5+1.2 S1
2 2 2 2 GA
2 2 2 2 G
1.5+1.2+1.5 S2
2 2 2 2 GA
2 2 2 2 G
4 1.8+1.5+1.2 S3
2 2 2 2 GA
2 2 2 2 G
2.0+1.5+1.2 S4
2 2 2 2 GA
2 2 2 2 G
2.5+1.2 S5
2 2 2 2 GA
2 2 2 2 G
3.5+1.2 S6
2 2 2 2 GA
*50.75 51.0 51.1 51.25
2 2 2 2 G
1.2+2.5+1.2 S7
2 2 2 2 GA
2 2 2 2 G
1.5+1.2+1.5 S8
2 2 2 2 GA
2 2 2 2 G
5 1.8+1.5+1.2 S9
2 2 2 2 GA
2 2 2 2 G
2.0+1.5+1.2 S10
2 2 2 2 GA
2 2 2 2 G
2.5+1.2 S11
2 2 2 2 GA
2 2 2 2 G
3.5+1.2 S12
2 2 2 2 GA

108
 SASI, et al.: NON-DESTRUCTIVE DETECTION OF FATIGUE CRACKS AND CORROSION IN RIVETS

Figure 4. Nomenclature of artificial grooves in the countersunk regions of rivets in panel S7 filled with simulated corrosion product.

riveted panel S7 (5 mm dia rivet head) having grooves of the rivets. As it is preferred to use same probe for
of 0.75 mm, 1.0 mm, 1.1 mm and 1.25 mm depths is the actual inspection of 5 mm as well as 4 mm dia rivets
shown in Fig. 5. As can be seen, at 3 kHz, the response for field implementation point of view, different gain
from all the four grooves reaches a maximum. Similar settings that would give near-identical instrument response
response was observed for this probe placed over 4 mm were tried and 20 dB and 30 dB were found to be optimum,
dia rivets having grooves of 0.25 mm, 0.5 mm, 0.6 mm respectively. The vertical output was measured for all
and 0.75 mm depths. In view of this, 3 kHz was chosen the rivets in the panels at these optimised conditions and
as optimum test frequency. The depth of penetration of the results were analysed.
eddy currents in the rivets at this frequency is about
3 mm. The instrument phase was set such that the lift-off 4. RESULTS AND DISCUSSIONS
signal from defective rivets was dominant along the vertical Typical eddy current signals at optimum test conditions
axis (V). Further, V/H ratio of 2.5 was set for selective for defect-free rivets and for rivets under G (unfilled)
amplification of the vertical output from defective rivets. and GA (filled) conditions are shown in Fig. 6. As can
During EC testing of riveted panel, uniform near- be observed, good phase discrimination is noticed between
zero lift-off and proper probe centering were maintained. defect-free and defective rivets.
It was observed that the responses for 4 mm and 5 mm The difference in EC response for filled and unfilled
dia rivets were different. The response was higher for grooves is explained with the help of universal impedance
5mm dia rivets. This was expected and found to be due plane diagram of the EC probe shown in Fig. 7. The
to very good coupling between EC probe and outer edge impedance plane diagram depicts the impedance lift-off

3 .0
Fre q u en cy: 3 kH z U n filled -G 4 0.6
5 .0 2 .5
Panel- S 7 G a in : 3 0 d B
4 .5 R ivet D ia .G roove D epth 2 .0
V E R TIC A L O U T P U T, (Vo lts)

50.75
4 .0 1 .5
V ER T IC A L O U TP U T, (Vo lts)

51.0
3 .5 51.1 1 .0
D e fe ct-fre e rive t
3 .0 51.25 0 .5
2 .5 0 .0
A ir
2 .0 -0 .5
1 .5 -1 .0
1 .0 -1 .5
0 .5 -2 .0
Fille d -G A 4 0.6
0 .0 -2 .5

0 .0 0 .5 1 .0 1 .5 2 .0 2 .5 3 .0 3 .5 4 .0 4 .5 5 .0 -3 .0
-1 .5 -1 .0 -0 .5 0 .0
Tes t F req uen cy, k H z
H O R IZ O N TAL O U T PU T, (Volts)
Figure 5. Optimisation of test frequency. At 3 kHz (depth of
penetration of eddy currents ~ 2 mm), EC response from grooves Figure 6. Typical EC signals from 4 mm dia. defect-free rivet
shows a distinct peak. and rivets with filled (GA) and unfilled (G) grooves.

109
 DEF SCI J, VOL. 59, NO. 2, MARCH 2009

1.0 5
Filled (GA )
Im a gin ary com p onen t of induced voltag e, volts

0.5 U nfilled (G )
4

V e r t ic a l O u t p u t , V o lt s
0.0 Frequency: 3 kH z
Gain: 20 dB
SS 304 3
-0.5
Rivet-G
-1.0 Lift-off Defect-free rivet 2

Rivet-G A
-1.5
1
Al
-2.0
Cu
0
-2.5
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
R ea l com ponen t of induce d voltage, vo lts 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Groove D epth, m m
Figure 7. Universal impedance plane diagram for reflection type
ring EC probe showing the lift-off response for different Figure 9. EC response for 5 mm diameter rivets with unfilled
electrically conducting materials. and filled grooves.

5
G ro o ve G A
G ro o ve G G ro ove-G
4 3
G ro ove-G A
V e r t ic a l O u t p u t , V o lt s

F re q u e n cy: 3 k H z
2.5
G a in : 3 0 d B
Vert ical Vo lt ag e, Vo lts

3
2

1.5
2

1
0.5

0
0
S0 S1 S2 S3 S4 S5 S6
R iveted panel
0 .0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 0 .8
Figure 10. Detection performance of the EC procedure for 4 0.5
G ro ove D ep th , m m
riveted panels with various combinations of multi-
Figure 8. EC response for 4 mm diameter rivets with unfilled layers at optimum test conditions.
and filled grooves.
identified from the phase of the impedance plane signal.
locus of electrically-conducting materials, e.g., stainless The EC response for 4 mm dia rivets consisting of
steel, aluminium, duralumin, and copper. As the electrical grooves filled with aluminium powder (GA) and unfilled
conductivity increases, the impedance lift-off locus moves grooves (G) is shown in Fig.8. Similarly, response for 5
downwards in the impedance plane and vice-versa. In the mm dia rivet heads is given in Fig.9. As can be noted, the
case of unfilled grooves, the impedance lift-off locus EC response for filled grooves is higher as compared to
moves upwards due to reduction in effective electrical the unfilled grooves and this response is found to increase
conductivity. In contrast, the impedance lift-off locus with the groove depth, for both 4 mm and 5 mm dia
moves downwards in the case of grooves filled with aluminium rivets. By using a threshold of 0.5 V for the absolute
powder due to increase in effective electrical conductivity. vertical voltage, it is possible to reliably detect 0.25mm
The movement of impedance lift-off locus produces deep defect in 4 mm dia rivets and 0.75 mm deep defect
corresponding changes in EC response voltages along in 5 mm dia. rivets.
horizontal and vertical axes. Thus, at the optimum test Figure 10 shows the detection performance of the
conditions, it is possible to identify whether a rivet is EC procedure on various panels having same type of
defective at the countersunk region from the absolute defect i.e. 4 0.5 in different combinations of thickness
amplitude of the vertical signal. Similarly, whether the layers (Table 1). The response is found to be independent
defect is filled with some corrosion product, can be of thickness of layers. This is expected, essentially, because

110
 SASI, et al.: NON-DESTRUCTIVE DETECTION OF FATIGUE CRACKS AND CORROSION IN RIVETS

of the fact that the depth location of the groove is same 1997, 4(4), 237-57.
from top and the eddy currents are concentrated to the 2. Khan, M.A.U. Non-destructive testing applications
rivet head rather than in the multi-layers. in commercial aircraft maintenance.In Proceedings
In the studies reported here, corrosion product is of 7 th ECNDT, 1999, 4(6).
simulated using aluminium powder in glue matrix. However, 3. Anindya, G. & Heung, S. Damage tolerance management
the actual corrosion products are usually oxides, hydroxides, for aircraft and rotorcraft structural components. J.
etc. with poor electrical conductivity. Such products of Aircraft, 2006, 43(6), 1676-684.
would produce signals almost identical to the unfilled 4. Sasi, B.; Rao, B.P.C. & Jayakumar, T. Dual-frequency
groove signals and this is expected to enable accurate eddy current non-destructive detection of fatigue
sizing of defects, depending on the calibration reference cracks in compressor discs of aero engine. Def. Sci.
defects. This is an interesting observation made from J., 2004, 54(4), 563-70.
these experiments. This study clearly demonstrates that 5. David, G. M. & Floyd, W. S. Interlayer crack detection
proper selection of probe and systematic optimisation results using sliding probe eddy current procedures.
of test parameters enable reliable detection of potential In Proceedings of 10th APCNDT, 2001, Brisbane.
defects such as fatigue cracks and hidden corrosion in 6. Fulton, J.R; Winchesld, B.; Natge, S. & Min, Namkung.
the countersunk regions of rivets and also defect sizing. Development and testing of rotating probe method
The sensitivity of the procedure to detect much shallower for airframe rivet head inspection, In Review of
defects can be further enhanced using giant-magnetoresistive Progress in QNDE, 1995. 14B, Plenum Press, New
(GMR) sensors for receiving the magnetic fields of distorted York. pp 2317.
eddy currents at defect regions in the rivets. Also, in 7. Don, Hagemaier & Greg, Kark. Eddy current detection
some situations, apart from identification of defective of short cracks under installed fasteners. Materials
rivets, determination of circumferential location of the Evaluation, 1997, 5(1), 25-30.
defects is necessary. For such an application, the proposed 8. Van der Walde, K. & Hillberry, B. M. Initiation and
procedure is inadequate and it is necessary to use spot shape development of corrosion nucleated fatigue
type probes that are made to scan uniformly over the cracks. Int. J. Fatigue, 2007, 29, 1269-281.
rivet-heads. Preliminary studies carried out using reflection 9. Hagemaier, D. J. Fundamentals of eddy current testing,
type spot probe with a mechanical centring device have American Society for Non-destructive Testing (ASNT).
confirmed identification of circumferential location of Inc. ISBN 0-931403-90-1, 1990. pp. 61.
defects. 1 0 Rao, B.P.C. Introduction to eddy current testing.
Narosa Publishing, New Delhi, 2007.
5. CONCLUSIONS
An effective eddy current test procedure has been
developed for reliable detection of fatigue cracks in
countersunk regions of rivets in air-intake structures.
Lift-off signal has been utilised to distinguish defective
rivets from defect-free rivets. This procedure has resulted Contributors
in reliable detection of 0.25 mm deep defects in 4 mm
Ms B. Sasi obtained her MSc (Applied
and 0.75 mm deep defects in 5 mm dia rivets. Using the
Physics) from Gandhigram Rural Institute,
phase of impedance plane signals, it is possible to identify Deemed University. She has been working
if the defect is filled with corrosion product. The procedure in Indira Gandhi Centre for Atomic Research
is not affected by the thickness of the multilayers. Further, (IGCAR) in the field of development of
the procedure is amenable for field implementation. electromagnetic non-destructive evaluation
(NDE) techniques. Her areas of research
ACKNOWLEDGEMENTS include: Electromagnetic NDE, signal
processing, sensor design and material
Authors thank Dr A Venugopal Reddy, Director, RCMA
characterisation.
(Materials), Hyderabad, for useful discussions and Shri
S. L. Savrekar, Chief Manager (QC), Aircraft O/H and Dr B.P.C. Rao heads Electromagnetics,
AUR and DC, HAL, Bangalore, for providing the riveted Modelling, Sensors and Imaging (EMSI)
specimens and for useful discussions. Authors are also Section, NDE Division, IGCAR. He is
thankful to Shri P. Krishnaiah, EMSI Section, NDE Division, specialised in the area of NDE of metallic
IGCAR for help during the experiments. materials for detection and characterisation
REFERENCES of defects and microstructures. He has
1. Pitt, S. & Jones, R. Multiple-site and widespread published more than 100 research papers
in various international journals and
fatigue damage in ageing aircraft. Engg. Failure Analy.,
conferences.

111
 DEF SCI J, VOL. 59, NO. 2, MARCH 2009

Dr T. Jayakumar heads the NDE Division,

IGCAR. He is specialised in the area of
NDE for characterisation of defects, residual
stresses and microstructures and failure
analysis of engineering components. He
has published more than 200 research
papers in various international journals.

Dr Baldev Raj is Director, IGCAR,

Kalpakkam. He has published more than
650 papers in several leading national
and international journals in the area of
metallurgy, material science and NDE.
He has authored and edited 38 books.
He is a fellow of several national and
international academies and societies and
is a recipient of many national and
international awards.

112