EDDY CURRENT TESTING, A SOLUTION TO DETECTING ROLLING

CONTACT FATIGUE IN RAIL?

Authors: John Hansen B.Sc. (Hons.), MBA, MIEE, FINDT, C. Eng [email protected]
Hocking NDT Ltd., Camp Rd., St. Albans, Herts., UK AL1 5HL
John Calvert MIEE, C. Eng. [email protected]
Hocking NDT Ltd., Camp Rd., St. Albans, Herts., UK AL1 5HL
ABSTRACT
This paper details work we have carried out into producing a practical solution to detecting rolling contact
fatigue in rail using the eddy current testing method.
Eddy current testing methods have been historically largely neglected in rail applications. However eddy
current testing is widely used in the aerospace industry to solve inspection problems that are not readily
solved using the other non-destructive testing methods. An example of one application in the aerospace
industry is used to illustrate how eddy current testing has been used in maintaining safety in the
transportation industry.
An outline of the theory of eddy current testing is given.
Eddy current and ultrasonic testing are largely complementary; here they are compared and contrasted
(including detail of each methods advantages and disadvantages).
A typical eddy current probe element covers at most 25 mm width but more typically 1 to 2 mm. To cover
the rail head multiple probes are generally used with multiplexed instrumentation. This leads to complex
data and large test units. In order to improve the practicality of applying eddy current testing Hocking
NDT Ltd have developed the WideScan Rail probe. This is a single probe that is capable of detecting
orientated defects in one pass of the rail head and of producing a signal output that is related to the
surface length and depth (size) of the flaw. The active face of the probe can easily be contoured to suit
different rail heads and indeed could be shaped to suit rolling stock applications.
Data will be presented of test results achieved with the WideScan Rail probe demonstrating sensitivity to
flaws (both natural and artificial). The WideScan Rail probe will be characterized using artificial flaws of
different sizes, orientation and probe to rail clearance necessary for practical application. This will be
used to determine sensitivity levels that may be achieved using this probe.
Modern eddy current instrumentation has led to the availability of high performance light weight flaw
detectors. These may be used with the WideScan Rail probe and a trolley, for rapid manual inspection,
offering advantages over the established ultrasonic rail inspection methods. The WideScan Rail probe
could be train mounted for high speed automatic inspection and may also be used for detecting surface
breaking flaws in rolling stock (e.g. wheels and axles)
INTRODUCTION
As a result of media and public pressure since the Hatfield derailment in October 2000, Hocking NDT
Ltd. have received numerous enquiries from organizations both in the UK and overseas concerning the
feasibility of using eddy current testing for detecting rolling contact fatigue in rail (Railtrack, 2001). This
paper intends to detail work we have carried out into producing a practicable solution.
Figure 1 shows the statistics for RailTrack (Sawley & Reiff, 2000), which like most statistics have
multiple, and contradicting interpretations. In recent history both defects and breaks have increased, with
roughly 10 defective rails for every 1 broken rail. Over a longer period defects have increased 8 fold
whilst breaks have remained fairly constant. However one certainty is that detecting defects is becoming
increasingly difficult, due not just to the basic statistics but also with rising rail network traffic. The time
available to perform inspection is therefore being steadily reduced.

It has been suggested (Sawley & Reiff, 2000) that the manual ultrasonic method be replaced by automatic
ultrasonic inspection methods with multiple probes. We believe that the existing inspection methods
should be evaluated against the use of other non-destructive testing techniques to determine the optimum
defect detection and assessment strategy. Our belief is that eddy current inspection is more appropriate to
detect surface breaking defects typical of rolling contact fatigue, either in parallel with ultrasonic
inspection (as is the case during rail manufacture) or as a filter to locate areas requiring more detailed
manual examination.
THEORY
This is not intended as a comprehensive theory of the eddy current method but mainly to give background
to help understand the key differences with the ultrasonic method.
By passing an alternating current in a coil, an associated magnetic field is produced. This in turn will
produce a current flow in an electrically conductive material. Changes in the properties of the material
(such as conductivity, permeability and path length) result in a change in the sensor impedance; these are
detected as changes in coil impedance.
For a crack to be detected it must cross the path of an eddy current (see Figure 2).

The formulae below illustrate the relationship between material electro-magnetic properties and the
standard depth of penetration (sometimes misleadingly referred to as the skin depth).

= 50
( . ) f r

f
r
depth of penetration
(mm)
electrical resistivity
(micro ohm . cm)
frequency (Hz)
relative permeability
Flaw parallel to eddy
currents - not detected
Flaw interrupts eddy
currents - detected
Figure 2
Some important points to note are:
Eddy current density is greatest at surface and reduces exponentially with depth
At the standard Depth of Penetration the current density is 37% of the value at the surface
The phase shifts as you move down through the material and lags that at the surface linearly with
depth (at 57
o
per standard Depth of Penetration).
Relative permeability of much greater than 1 (such as in ferromagnetic steel) ensures that the
signal stays close to the surface, less than 0.1mm. However surface breaking flaws will draw
current flow deeper (3-4 mm).
EXAMPLES OF EDDY CURRENT APPLICATIONS
Historically eddy current testing methods have been largely neglected in rail applications. This is a view
that also seems to be held in other countries (Thomas, 2000). They are used extensively in many
industries throughout the world, for example testing ferro-magnetic welds. To illustrate further, below are
shown two examples where eddy current non-destructive testing has proven to be a valuable tool.
ID tube inspection
In non-ferrous internal diameter tube testing (such as in heat exchangers) eddy current NDT
gives high quality results at high speed. Defects may be confirmed by slower methods such as
IRIS ultrasonic and visual using a fibre scope.
Fastener hole inspection
Fastener hole inspection, an example of one application in the aerospace industry, is used to
illustrate how eddy current testing has been used in maintaining safety air transportation. With
ageing aircraft this has led to a requirement to find small defects in the bore of many fastener
holes with the fasteners removed. Areas of application include fastener holes on aircraft
structures, undercarriage, wheels and holes in turbine discs.
Originally this was performed manually with a small diameter probe.
However it was realized that only a small part of the equipments response bandwidth was being
used. In order to exploit this, fully motorised rotary drive units were developed. These spun
typically at speeds of 3000 rpm. This resulted in greater sensitivity, improved inspection
coverage at a higher testing rate.
Demonstrated to be the most reliable NDT technique of all, aircraft inspection with eddy current
NDT is around 60% and still growing. This is probably because it seems more capable of solving
inspection problems on complex aircraft structures.
COMPARISON WITH ULTRASONIC INSPECTION
Eddy Current and Ultrasonic Testing are largely complementary. Here they are compared and contrasted
(including detail of each methods advantages and disadvantages).
Ultrasonic Eddy Current
1.Requires couplant, which introduces variability 1.No Couplant
2.Poor at surface defects 2.Good for surface defects
3.Near sub-surface defects difficult 3.Near sub-surface defects reasonable (impossible
on ferrous unless magnetised)
4.Deep sub-surface defects good 4.Deep sub-surface defects impossible
5.Strongly influenced by flaw orientation 5.Probes less sensitive to orientation
6.Blind to near surface defects and sub-surface
defects may be hidden by surface defects
6.Providing probe held correctly, relative to
surface, then signal is coupled
7.Defect must be on probe centre line 7.Probe can be made wide and profiled to cover
wear face
8.Slow due to PRF typical=1kHz limited by
physics
8.Fast 16 kHz as limited by test frequency and
electronics

Table 1
With Ultrasonics offering the potential for accurate flaw depth determination especially using the time of
flight method.

PROBE DEVELOPMENT
Alternative solutions researched to inspect the rail head (including multi-coil arrays)
Objectives were:
Uniform sensitivity across the test area
Sensitive to longitudinal and transverse defects
Good lift off characteristics to allow for variations in probe to rail clearance
Robust
Able to match rail profile.

A SOLUTION
After evaluating various strategies, a probe was devised that gave the characteristics required whilst being
extremely simple to use. We have named this probe WideScan Rail.
Typically small eddy current point probes cover 1 to 2 mm. A large eddy current probe element would
cover 25 mm width. To cover the rail head multiple probes are then generally used with multiplexers, this
leads to complex data and large test units. In order to improve the practicality of applying eddy current
testing Hocking NDT Ltd have developed the WideScan Rail probe. This is a single probe that is capable
of detecting orientated flaws in one pass of the rail head and producing a signal output that is related to
the surface length and depth (size) of the flaw. The active face of the probe can easily be contoured to suit
different rail head shape and indeed could be shaped to suit rolling stock applications.
Data will be presented of test results demonstrating sensitivity to flaws (both natural and artificial). The
WideScan Rail probe will be characterized using artificial flaws of different sizes, orientation and probe
to rail clearance necessary for practical application. This will be used to determine sensitivity levels that
may be achieved using this probe. It is a single probe, with an 80 mm scan width (not a limit) capable of
being made to fit a profile.
In order to ascertain the probe characteristics more efficiently a flat version of the probe was used for
experimentation.
CHARACTERISATION
In order to characterise the probe fully, a flat probe was developed and this was evaluated on artificial
flaws made in flat steel plates. This allowed the angle of the probe scan and flaw to be varied in order to
study the probe response.

Figure 3
Flaw size and length were chosen to fit closely the existing inspection procedures in the UK. Details of
the notches used are given in the table below. In addition to rectangular notches, segment notches were
used to more closely resemble natural flaws.
Ident
Depth
(mm)
Length
(mm)
Area
(mm2)
Rectangular Notch
a 1.39 4.0 5.6
b 2.69 8.0 21.5
c 6.02 14.0 84.3
d 1.40 10.0 14.0
e 2.71 20.0 54.2
f 2.72 30.0 81.6
g 5.98 10.0 59.8
h 6.00 20.0 120.0
i 5.99 30.0 179.7
Segment
Notch
c 1.00 14.3 8.2
d 2.00 20.0 23.0
e 4.00 27.7 63.3

Table 2
Results now follow for key characteristics of the probe.
Sensitivity to Flaws
The first observation that should be noted in the data is that there is a 150
o
phase separation in the eddy
current signal seen when passing over a longitudinal flaw (parallel with the scan axis) compared with a
transverse flaw (at right angles to the scan axis). This shows there are two modes of detection
Transverse where the field is interrupted
Longitudinal where the field is distorted
For transverse flaws the signal amplitude is proportional to surface length (up to sensing length of probe
80 mm).
Response from Transverse Notches
0
2
4
6
8
10
12
14
16
18
20
0 5 10 15 20 25 30
Surface Length (mm)
R
e
s
p
o
n
s
e

A
m
p
l
i
t
u
d
e

(
m
V
)

Rectangular Notch Segment Notch
Linear (Rectangular Notch) Linear (Segment Notch)

Figure 4
Figure 5
For longitudinal flaws the signal amplitude is proportional to flaw area (up to 14 mm length approx).


Response from Longitudinal Notches
0
5
10
15
20
25
30
35
40
45
50
0 20 40 60 80 100 120 140 160 180 200
Area (mm2)
R
e
s
p
o
n
s
e

A
m
p
l
i
t
u
d
e

(
m
V
)
Rectangular all points Rectangular 14 mm length and under
Segment Notch Linear (Rectangular all points)
Linear (Rectangular 14 mm length and under) Linear (Segment Notch)
Scan Width
Figure 6 below shows the response obtained to the segment notches at different positions across the probe
face. At - 12dB the width of the scan is equal to the physical width of the probe.
Scan Width
-14
-12
-10
-8
-6
-4
-2
0
-60 -40 -20 0 20 40 60
x (mm)
S
e
n
s
i
t
i
v
i
t
y

(
d
B
)
C Transverse C Longitudinal D Transverse D Longitudinal

Figure 6
Effect of probe to test piece separation
Increasing probe to test piece separation (termed lift-off) gives a reducing signal amplitude (sensitivity).
Here the concern is not variation in the overall height (this can be maintained by contacting rolls) but the
variations in profile due to wear and grinding.
The measured lift-off effect on gain is nearly identical and about 3dB per mm.
Compared with other eddy current probes for example:-
Pencil type probe is about 30dB per mm
WeldScan type probe is 8 dB per mm (usually considered to be very low)
So we can conclude that this probe is relatively insensitive to changes in profile.

Effect of Lift-off on sensitivity
y = 2.9341x
0
2
4
6
8
10
12
14
16
18
20
0 1 2 3 4 5 6 7
Height (mm)
A
t
t
e
n
u
a
t
i
o
n

(
d
B
)
Transverse Attenuation (dB) Longitudinal Attenuation (dB)
Linear (Transverse Attenuation (dB)) Linear (Longitudinal Attenuation (dB))

Figure 7
Orientation of flaw
There is a 160
o
phase change between flaws orientated 90
o
apart, with the majority of the phase shift and
amplitude change occurring between 20
o
and 40
o
.


Flat probe no 1
0
20
40
60
80
100
120
140
160
180
0 10 20 30 40 50 60 70 80 90
Angle of Scan
S
i
g
n
a
l

P
h
a
s
e

(
d
e
g
r
e
e
s
)

Figure 8
Flat probe no. 1
-14
-12
-10
-8
-6
-4
-2
0
2
4
6
0 10 20 30 40 50 60 70 80 90
Angle of Scan
S
e
n
s
i
t
i
v
i
t
y

(
d
B
)

Figure 9
Phase changes rapidly around minimum sensitivity point (a node). Node position is determined by surface
length of flaw and probe geometry.










Speed of response
The maximum speed of response of an eddy current system is determined by the following
1.Test Frequency which must be higher than the sample rate of the instrumentation (assuming no over
sampling)
2.Sample rate of test equipment which to reconstruct a pulse must be 10 times higher than the low pass
response.
3.Low pass response of test equipment
4.Design and size of coil (coil separation)

Figure 10
Shown here is the response which assuming a low pass response of 2kHz gives a maximum surface speed
of 32 m/s or 115 km/hr.
If the response of the equipment is improved to 4kHz this would improve by a factor of 2 which is
achievable.
APPLICATION
Recent developments by Hocking NDT Ltd. in eddy current instrumentation have led to the availability of
a battery operated high performance flaw detector weighing less than 1 kg and having a running time of
typically over 10 hours per charge. This coupled with the WideScan Rail probe and a trolley means rapid
manual inspection is readily achieved. The ability to detect surface breaking flaws, lightness, and lack of
consistency problems associated with ultrasonic couplant and higher surface speed capability. All offer
worthwhile advantages over the traditional manual rail inspection methods. Further, a data logging
system, with Global Positioning System (GPS) location and Text entry could ensure comprehensive
documentation of test results.
This method can also be applied to:
Automatic rail inspection systems
Trains in service
The WideScan probe is also equally applicable to wheels and other rolling stock areas such as axles.
WideScan Rail offers an extremely efficient and cost effective solution to detecting surface breaking
flaws typical of rolling contact fatigue.
ACKNOWLEDGEMENTS
The authors acknowledge the advice and assistance received from the personnel of the following
organisations:
Bombardier Adtrans, RailTrack, Irish Rail, Balfour Beatty, GTRM, Inspection Solutions (Cymru) and C J
Enterprises.

REFERENCES
Railtrack, Rolling contact fatigue in rails, Heron Press, 2001
Sawley & Reiff, Rail failure Assessment for the Office of the Rail Regulator, TCCI, October 2000
Thomas, H M Pioneering inspection of railroad rails with eddy currents, Proc. of 15
th
WCNDT, Rome
2000