06_505_522TPs_18pgs 3/10/03 7:31 AM Page 505

Submitted October 2002

From Materials Evaluation, Vol. 61, No. 4, pp: 505-511.
Copyright 2003 The American Society for Nondestructive Testing, Inc.

Laser/Air Hybrid Ultrasonic Technique for

Railroad Wheel Testing
by Shant Kenderian,* B. Boro Djordjevic* and Robert E. Green, Jr.

ABSTRACT
Laser generation and air coupled detection of ultrasound were combined to produce a hybrid noncontact and remote ultrasonic technique. The
technique is suitable for general testing of a number of structural materials. Using this hybrid technique, guided surface waves were generated to
propagate along the flange, tread and rim edge of a railroad wheel. The
waves complete one revolution around the circumference of the railroad
wheel in 1 ms and interact with surface breaking discontinuities that are
present on the various parts of the wheel. The flexibility and remote nature
of this technique suggest that wayside testing of wheels on a moving train
can be made possible.
Keywords: laser ultrasound, air coupled ultrasound, railroad wheel,
cracks.

INTRODUCTION
The laser/air hybrid ultrasonic technique is a noncontact and remote technique that combines laser generation with air coupled detection of ultrasound. The noncontact and remote nature of this
technique provides it with the flexibility to detect discontinuities in
unfavorable positions and orientations. It has been shown that this
technique can be an effective and easy to use method for rail track
testing (Kenderian et al., 2001; Kenderian et al., 2002). For decades,
optical methods have been widely popular as noncontact and remote ultrasonic detection techniques. However, their efficiency relies heavily on the amount of light reflected back from the surface.
As a result, the curvature, roughness and cleanliness of the reflecting surface all have a negative influence on the amount of light reflecting back to the optical detector. This, unfortunately, renders
these techniques ineffective for many industrial applications including the railroad industry. A comparative study between air
coupled and interferometric detection (Kenderian et al., 2002) or
electromotive force (EMF) detection (Cerniglia and Djordjevic, in
review) of ultrasound demonstrates the superiority of the laser/air
hybrid ultrasonic technique over a purely laser based ultrasonic
technique, within the operating frequencies of the air coupled detector (that is, below 2.25 MHz).
The present paper presents a new technique that would enable
the railroad industry to perform noncontact and remote testing of
railroad wheels while in motion. No method is currently available
to the railroad industry to perform dynamic tests on railroad
wheels.

With a single pulse, a multimode signal was generated. The propagation properties of the signal are a function of the laser pulse
width and intensity, point of impact, surface condition and shape of
the pulse illuminating the surface, but are not affected by the angle
at which a pulse is delivered.
For the detection of ultrasound, a capacitive air coupled transducer (Schindel and Hutchins, 1994; Schindel and Hutchins, 1995)
capable of detecting frequencies between 50 kHz and 2 MHz was
used. The frequency range of this capacitive air coupled transducer
was adequate for the type of discontinuities that cause concern for
the railroad industry. These air coupled transducers are capable of
operating at remote distances exceeding 150 mm (5.9 in.) (Kenderian, 2002). Naturally, higher frequency components attenuate severely in air. Therefore, as the standoff distance between the air coupled transducer and the specimen is increased, the upper limit of
the frequencies retained by the detector is lowered.
Overall signal strength as a function of standoff distance follows
an exponential behavior. Good signals were readily available up to
40 mm (1.6 in.) and useful measurements were possible past 80 mm
(3.2 in.), as shown in Figure 1a. Figure 1b shows that the optimum
detection angle for a rayleigh wave propagating in steel is 6.5 degrees and angular variation of 2 degrees still retains good signal
intensity. The test configuration is that shown in Figure 1c. Using
the acoustic velocity in air Cair = 0.33 mm/s (0.01 in./s), the
rayleigh wave velocity in steel CR = 2.9 mm/s (0.1 in./s) and
90 degree propagation angle of the rayleigh wave in steel, Snells
law calculations confirmed our experimental observation of the
critical refraction angle at 6.5 degrees in air. This is presented in
Equations 1 and 2.

(1)

sin air
C
= air
sin steel Csteel

(2)

0.33
air = sin 1
= 6.5 degrees
2.9

RESULTS
EXPERIMENT
For the generation of ultrasound, a Q-switched Nd:YAG laser
operating at 1064 nm (4.2 10-5 in.) wavelength with pulse durations of 4 to 7 ns and maximum energy of 400 mJ/pulse was used.
* Center for Nondestructive Evaluation, The Johns Hopkins University,
810 Wyman Park Drive, Baltimore, MD 21211; (410) 516-0818; fax (410)
516-7249; e-mail <shant@jhu.edu>.
Materials Science and Engineering Department, The Johns Hopkins
University, 3400 N. Charles St., Maryland Hall 102, Baltimore, MD
21218.

Preliminary experiments performed on a small 32 kg (70 lb) section of the wheel showed that an acoustic signal generated with a
laser line source was more effective than that generated by a laser
point source for the detection of saw cuts. Similar results were reported for surface breaking cracks on rail head (Kenderian et al.,
2001; Kenderian et al., 2002), with a detailed discussion explaining
the underlying cause for the enhanced sensitivity of a line source to
surface discontinuities given in a separate paper (Kenderian et al.,
in publication). Based on these results, the laser source used in this
paper was focused to a line parallel to the direction of the slots,
which is the transverse direction on the wheel.
Materials Evaluation/April 2003 505

06_505_522TPs_18pgs 3/10/03 7:31 AM Page 506

A calibration railroad wheel, 910 mm (36 in.) in diameter and

weighing approximately 365 kg (800 lb), was used for test purposes. The wheel contained four manufactured discontinuities marked
with the letters A, B, C and D. A, B and C were slots made to the
wheel flange, tread and rim, respectively, and D was a hole drilled
in the side of the wheel.
The wheel was divided into 360 degrees. The location of 0 degrees was determined from the point where the acoustic signal was
generated, that is, the point at which the generation laser illuminated the surface of the wheel. The positions of the detector and the
discontinuity were referred to in terms of their location, in degrees,
along the wheel circumference with respect to the acoustic source.
The detector position was designated with the letter , while that of
the discontinuity was designated with . Figure 2 depicts the general setup for the wheel experiments performed in this study.
Surface Slots on Wheel Flange
In the detection of slot A, a 16 mm (0.6 in.) long, 1.5 mm (0.06 in.)
wide and 5.3 mm (0.2 in.) deep saw cut was made to represent a
surface breaking crack on the wheel flange. The detector was kept
at the = 90 degrees position. The wheel was rotated such that the
slot fell in various positions between = 10 and 180 degrees. As
noted previously, the optimum detection angle for a steel to air
leaky rayleigh wave is 6.5 degrees. Therefore, a detector positioned

(a)

between 0 and 180 degrees, say at 90 degrees, and inclined at 6.5 degrees from the surface normal, optimally detected a signal traveling
the short path, that is counterclockwise according to Figure 2, traveling one quarter of the wheel circumference. Similarly, when this
detector was inclined at 353.5 degrees from the surface normal, it
optimally detects a signal traveling the long path, that is clockwise
according to Figure 2, traveling three quarters of the circumference.
Short path and long path inclinations were denoted with the letters
S and L, respectively.
A 17 mm (0.7 in.) long laser line source was used to generate the
acoustic signal used for the detection of the 16 mm (0.6 in.) slot. Detector standoff distance was kept at 8 mm (0.3 in.) for long path detector inclination and 16 mm (0.6 in.) for short path inclination. Figure 3 shows a broad view, 0 to 1000 s, of a signal generated at
0 degrees and detected at 90 degrees with the detector in the long
path L inclination and slot A located at = 150 degrees. Interestingly, the wheel geometry is such that a rayleigh wave completes one
revolution around the wheel in approximately 1 ms. This is obvious
from the 3.1 and 2.9 m (122 and 115 in.) circumference of the wheel
flange and tread, respectively, and the nearly 3 mm/s (0.1 in./s)
rayleigh wave velocity in rail steel (Bray and Vezina, 1991). Direct,
reflected and transmitted waves are observed.
The first arrival, at 285 s, was a direct wave traveling 90 degrees counterclockwise to the detector, propagating 775 mm (31 in.)

(b)

(c)
Figure 1 Signal strength of capacitive air coupled detector: (a) as a function of standoff distance with optimum 6.5 degrees inclination to the
surface normal; (b) inclination to the surface normal; (c) test configuration under which measurements were taken.
506 Materials Evaluation/April 2003

06_505_522TPs_18pgs 3/10/03 7:31 AM Page 507

in steel (or one quarter the circumference of the flange) and 8 mm

(0.3 in.) in air. Although the detector was not optimized for short
path signal detection or counterclockwise propagation in Figure 2,
a fraction of the strong direct signal was still detected. This was due
to the large 10 mm (0.4 in.) aperture of the detector and small 8 mm
(0.3 in.) standoff distance of the detector. Other factors also played
an important role in the detection mechanism of the capacitive air
coupled transducer. Some of these factors include the direction of
particle displacement of the acoustic signal in steel and air and attenuation through air and impedance between air and the detector
membrane.
The second arrival, at 630 s, was a reflected wave traveling
from 0 degrees past the detector at = 90 degrees counterclockwise,
redirected upon reflection from slot A at = 150 degrees and propagating back to the detector at = 90 degrees in a clockwise direction. The total propagation distance is 1.8 m (71 in.) in steel and
8 mm (0.3 in.) in air. The amplitude of the reflected wave is greater
than that of the direct wave. The reason for this is that with its L inclination, the detector was optimized for the detection of signals
propagating in the clockwise rather than the counterclockwise direction, according to the setup shown in Figure 2.
The third arrival, observed at 800 s, was a transmitted wave
traveling clockwise from 0 degrees to slot A at = 150 degrees,
transmitting through the slot and continuing in a clockwise direction to the detector at = 90 degrees. The total propagation distance
is 2.3 m (92 in.) in steel, traveling three quarters of the circumference
of the flange, and 8 mm (0.3 in.) in air. Considering rayleigh wave
velocity CR 3 mm/s (0.1 in./s) in rail steel and wave velocity
CA = 330 m/s in air (0.01 in./s), the calculated arrival times of
all three waves agree well with the observed signals. The amplitude
of the transmitted wave was smaller than the direct and reflected
waves mainly because of the substantially longer propagation path
in steel, which would effectively increase signal attenuation.
While the acoustic source and receiver were stationary, the
wheel was rotated so that slot A changed its position with respect to
generation and detection points. The direct and the transmitted
waves did not change their position in the time domain. This is because their propagation distances did not change with the wheels
rotation. However, the arrival time of the reflected wave varied
with the rotation of the wheel for obvious reasons. Using this technique, successful detection of slot A was possible for positions between 0 and 180 degrees. Detection capacity was not limited to this
range, except that for positions between 180 and 360 degrees, a better signal was obtained using a detector in the short path S inclination to detect a reflected wave propagating in the counterclockwise
direction. Figure 4 shows results obtained with the detector in the S
inclination positioned at = 90 degrees and slot A positioned at
= 13 degrees.
Frequency analysis of the direct, reflected and transmitted ultrasonic waves is presented in Figure 5. Figure 5a shows that the
upper and lower boundaries of the frequency of the direct wave are
set by the 0.3 MHz high pass filter used in this experiment and the
2 MHz detection limit of the capacitive air couple transducer. As a
general rule, wavelengths smaller than the depth of a crack reflect
back from the crack while those with larger wavelengths transmit
through. At its deepest point, slot A is 5.3 mm (0.2 in.) deep. A
rayleigh wavelength of R = 5.3 mm (0.2 in.) corresponds with a frequency of 0.57 MHz in steel. Accordingly, frequencies higher than
0.57 MHz are expected to reflect back from slot A and those lower
than 0.57 MHz to transmit through. Figures 5b and 5c show the frequency spectrum of the reflected and transmitted waves, respectively. For the reflected wave, a sharp decline is observed at frequencies lower than 0.6 MHz, while the frequency of the
transmitted wave is confined between 0.3 MHz and 0.75 MHz, as
expected.
Surface Slots on Wheel Tread
A similar test configuration was used in the detection of slot B, a
26 mm (1.02 in.) long, 1.5 mm (0.06 in.) wide and 8 mm (0.3 in.)
deep saw cut made to represent a surface breaking crack on the center of the wheel tread. A 25 mm (0.98 in.) long laser line source was
used to generate the acoustic signal for the detection of the 26 mm

Figure 2 General configuration and nomenclature used in this

study: generation of acoustic source determines the reference point of
0 degrees ( = detector position angle; = crack position angle;
L = detector inclination for clockwise, or long path, signal;
S = detector inclination for counterclockwise, or short path, signal.

Figure 3 Broad view of a signal generated at 0 degrees and detected

at 90 degrees, with the detector in the long path inclination and a
machined slot located at = 150 degrees along the flange of a rail wheel.

Figure 4 Close up of a signal generated at 0 degrees and detected at

90 degrees, with the detector in the short path inclination and a
machined slot A located at = 13 degrees along the flange of a rail
wheel.
Materials Evaluation/April 2003 507

06_505_522TPs_18pgs 3/10/03 7:31 AM Page 508

(1.02 in.) slot. Detector standoff distance was kept at 25 mm

(0.98 in.) throughout this part of the experiment. Figure 6 shows a
broad view of a signal generated at 0 degrees with the detector in the
long path inclination positioned at = 90 degrees and slot B at = 10,
100 and 120 degrees. Similar to the results previously shown in Figure 3, three main arrivals are also observed in Figure 6. The direct
and transmitted waves travel one and three quarters of the circumference of the wheel tread, respectively. Therefore, their arrival time

can be predicted and was unchanged with the rotation of the

wheel. While these waves were stationary in the time domain, as
shown in Figure 6, the arrival time of the reflected wave would
change with slot position . Because the transducer was oriented in
the long path direction, it was not optimized for the detection of
any waves approaching from the short path direction. Accordingly,
the counterclockwise waves, which are the transmitted waves in
Figure 6a and the direct waves in Figures 6b and 6c, were absent or
weak. Note that the transmitted and direct wave determination is
interchangeable, based on the position of the slot with respect to the
generation and detection points. In Figure 6a, the arrival time of the
direct wave, approaching the detector from the long path direction,
is 833 s. The reflected wave from slot B at = 10 degrees propagates an additional distance of 163 mm (6.4 in.) to arrive 56 s later.

(a)

(a)

(b)
(b)

(c)
(c)
Figure 5 Frequency analysis with respect to a 5.3 mm (0.2 in.) slot
along a rail wheel: (a) direct waves; (b) reflected waves; (c) transmitted
waves.
508 Materials Evaluation/April 2003

Figure 6 Broad view of a signal generated at 0 degrees with the

detector in the long path inclination positioned at = 90 degrees and
slot B positioned at: (a) = 10 degrees; (b) = 100 degrees;
(c) = 120 degrees.

06_505_522TPs_18pgs 3/10/03 7:31 AM Page 509

Similarly, when the slot is positioned at = 100 degrees, the reflected wave propagates an additional distance of 163 mm (6.4 in.) to arrive 56 s behind the direct wave, as shown in Figure 6b. In Figure
6c, the corresponding delay associated with the reflected wave
from a slot positioned at = 120 degrees is 168 s, so that the arrival
time of the reflected wave would be 496 s.

Figure 7 Front and side views of transducer position with respect to

the rim edge of the wheel tread showing a limited interface between the
detector and the wheel.

Surface Slots on Wheel Rim

A similar procedure was followed for the detection of slot C, a
15 mm (0.6 in.) long, 2 mm wide (0.08 in.) and 15 mm (0.6 in.) deep
saw cut, which was made at the rim edge of the wheel tread. A
schematic drawing of the detector position with respect to the rim
edge, shown in Figure 7, demonstrates that, due to the roundness of
the rim edge and curvature of the wheel tread, the area of the wheel
surface available to the detector is limited to a large extent. Should a
contact transducer be used, this interface would be reduced to a
point. However, due to the noncontact nature of the capacitive air
coupled transducer and its 10 mm (0.4 in.) aperture, the transducer
was capable of detecting acoustic signals propagating along the rim
edge of the wheel tread, although some limitations were imposed.
That is, reflected and transmitted signals were not as distinct as
those presented earlier for the wheel tread and flange. In spite of the

(a)

(b)

(c)

(d)

(e)

(f)

Figure 8 The direct wave propagates counterclockwise along the rim edge of the wheel tread, detected with the air coupled transducer positioned at
= 90 degrees in the short path inclination, with 20 mm (0.8 in.) liftoff and crack positions at: (a) 10 degrees; (b) 5 degrees; (c) 10 degrees;
(d) 45 degrees; (e) 80 degrees; (f) 100 degrees.
Materials Evaluation/April 2003 509

06_505_522TPs_18pgs 3/10/03 7:31 AM Page 510

(a)

(a)

(b)

(b)

Figure 9 Transmitted, direct and reflected waves resulting from a

laser generated acoustic signal interacting with a 3.3 mm (0.1 in.) hole
and detected with an air coupled transducer positioned at
= 10 degrees: (a) hole positioned at = 3 degrees; (b) hole positioned
at = 3 degrees.

Figure 10 Direct and reflected waves resulting from a laser generated

acoustic signal interacting with a 3.3 mm (0.1 in.) hole and detected
with an air coupled transducer positioned at = 10 degrees: (a) hole
positioned at = 8 degrees; (b) hole positioned at = 13 degrees.

Table 1 Propagation distances and expected arrival times for the waveforms pertaining to Figures 9 and 10
Hole Position from
Acoustic Source
25 mm (1.0 in.)
25 mm (1.0 in.)
65 mm (2.6 in.)
100 mm (3.9 in.)

Figure
9
9
10
10

Detector
Standoff
21 mm (0.8 in.)
21 mm (0.8 in.)
16 mm (0.6 in.)
16 mm (0.6 in.)

fact that the wheel flange had a round edge similar to that of the
rim, the acoustic signal was somewhat confined to the body of the
flange and interacted more pronouncedly with slot A, which extended through most of the flange width. In contrast, along the rim
edge the acoustic signal was free to spread upward towards the
tread and downward towards the hub. In addition, due to the
geometry of the rim edge, slot C was deep in the center but was
quickly reduced to a shallow cut along the edges. The length of slot
C did not extend to a significant portion along the rim width the
way slot A did along the flange. These factors, acting collectively, reduced the detectability of slot C in comparison to the detectability
of slots A and B.
The results shown in Figure 8 demonstrate that the laser/air hybrid ultrasonic technique is capable of detecting saw cuts along the
rim edge of the wheel tread. However, the detected signals were
not as pronounced as those obtained for the wheel flange and tread,
as shown in Figures 4 and 6.
510 Materials Evaluation/April 2003

Direct
107 s
92 s
92 s

Transmitted
107 s

Reflected
123 s
135 s
158 s

Small Hole on the Side of the Wheel Flange

A 3.3 mm (0.1 in.) hole was drilled in the side of the flange and
marked with the letter D. The hole was too small to produce a considerable effect on an acoustic signal transmitting through the
hole. With the detector kept at = 10 degrees and standoff distance
of 21 mm (0.8 in.), the wheel was rotated so that the hole was
sometimes positioned between generation and detection, that is at
0 degrees < < 10 degrees, and sometimes outside this region, that
is at 10 degrees < < 0 degrees. When the hole was inside this region, between generation and detection the amplitude of the transmitted signal (Figure 9a) was not considerably different from that
of the direct signal (Figure 9b). However, Figure 9b shows that
when the hole was outside this region and with proper inclination
of the air coupled detector, a reflected component of the acoustic
signal could be detected to make possible the detection of a small
hole such as that represented by D. Figure 10 shows similar results
obtained with the hole positioned at = 8 and 13 degrees. Due to

06_505_522TPs_18pgs 3/10/03 7:31 AM Page 511

the curvature of the wheel, a detector fixed at = 10 degrees was capable of detecting the hole as far as = 15 degrees. Beyond this
range, the hole, generation source and detector fell out of alignment
and the hole was no longer within detection range of the current experimental setup. Table 1 gives signal propagation distances in steel
and air and expected arrival times for the waveforms shown in Figures 9 and 10.

ACKNOWLEDGMENTS

CONCLUSION

The authors thank Transportation Technology Center, Inc., for

providing wheel specimens and financial support.

The main characteristic of the laser/air hybrid ultrasonic technique is its noncontact and remote method of operation. The technique can be applied on many structural materials in their industrial field condition. That is, although thin layers of dirt, oxides, grease
and other contaminants can slightly dampen the signal detected by
the air coupled transducer, they have a more significant effect on
enhancing the laser generated signal. As a result, a stronger signal is
detected with the presence of such contaminants. The technique
can tolerate surface roughness reasonably well. However, pitting,
spalling and porosities open to the surface can attenuate the surface
wave to a great extent. Surface waves were generated on various regions of the railroad wheel and detected successfully without the
need for modifying the surface through cleaning or polishing. Furthermore, all signals shown in this paper represent single events,
that is averaging techniques were not necessary as a way to improve the signal to noise ratio.
The use of the laser/air hybrid ultrasonic technique introduces
a degree of flexibility to railroad wheel testing that would enable it
to be performed without the need to establish contact between the
probe and the tested part and at relatively remote distances. It was
demonstrated in this paper that 360 degree coverage is possible for
the flange and tread of the railroad wheel with an approximate
time of 1 ms elapsed between generation and detection. For the
rim edge, 90 degree coverage was possible with a single laser
pulse. These results strongly suggest that this technique may be

engineered to perform wayside ultrasonic testing of wheels on a

moving train. No such technique is available to the railroad industry today. In light of the increasing reports of wheel and track related rail accidents, the development of a wayside field testing technique has become increasingly critical.

REFERENCES
Bray, D.E. and G. Vezina, Ultrasonic Applications in the Railroad Industry,
Nondestructive Testing Handbook, second edition: Volume 7, Ultrasonic Testing, P. McIntire, ed., Columbus, Ohio, ASNT, 1991, pp. 594-634.
Cerniglia, D. and B.B. Djordjevic, Ultrasonic Detection by Laser-based Sensors and by Wideband Air-coupled Transducer, Research in Nondestructive Evaluation, in review.
Kenderian, S., B.B. Djordjevic and R.E. Green, Jr., Point and Line Source
Laser Generation of Ultrasound for Inspection of Internal and Surface
Flaws in Rail and Structural Materials, Research in Nondestructive Evaluation, Vol. 13, 2001, pp. 189-200.
Kenderian, S., Advanced Ultrasonic Techniques to Determine the Structural Integrity of Rail Steel, The Johns Hopkins University, Department of
Materials Science and Engineering, PhD dissertation, 2002.
Kenderian, S., B.B. Djordjevic and R.E. Green, Jr., Laser Based and Aircoupled Ultrasound as Noncontact and Remote Techniques for Testing
of Railroad Track, Materials Evaluation, Vol. 60, 2002, pp. 65-70.
Kenderian, S., B.B. Djordjevic and R.E. Green, Jr., Sensitivity of Point and
Line Source Laser-generated Acoustic Wave to Surface Flaws, IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, IEEE, in
publication.
Schindel, D.W. and D.A. Hutchins, Air Coupled Ultrasonic Transducer,
US Patent 5,287,331, 1994.
Schindel, D.W. and D.A. Hutchins, The Design and Characterization of Micromachined Air-coupled Capacitance Transducers, IEEE Transactions on
Ultrasonics, Ferroelectrics, and Frequency Control, Vol. 42, 1995, pp. 42-50.

Materials Evaluation/April 2003 511