See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262167836

ULTRASONIC WAVE PROPAGATION IN MATERIALS WITH MECHANICAL

STRESSES

Conference Paper · June 2010

DOI: 10.13140/2.1.2659.0401

CITATIONS READS

4 207

3 authors:

Yonka Ivanova Todor Partalin

Sofia University "St. Kliment Ohridski", Institute of Mechanics-BAS Sofia University "St. Kliment Ohridski"
35 PUBLICATIONS 72 CITATIONS 21 PUBLICATIONS 125 CITATIONS

SEE PROFILE SEE PROFILE

Mitko Mihovski
National Institute of Meteorology and Hydrology
29 PUBLICATIONS 52 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

INNOPIPES View project

Materials degradation by non-destructive methods View project

All content following this page was uploaded by Yonka Ivanova on 09 May 2014.

The user has requested enhancement of the downloaded file.

 ULTRASONIC WAVE PROPAGATION IN MATERIALS WITH MECHANICAL
STRESSES

Yonka IVANOVA, *Todor PARTALIN, Mitko MIHOVSKI,

INSTITUTE OF MECHANICS-BAS, Sofia, Bulgaria

e-mail: [email protected] , [email protected]
*SOFIA UNIVERSITY, Faculty of Mathematics and Informatics, Sofia,
Bulgaria, e-mail: [email protected]

ABSTRACT
The present paper concerns ultrasonic wave propagation in materials with mechanical
stresses. In order to evaluate the stresses by ultrasound the investigation is performed. The
variations of the velocities, phases and spectra of the ultrasonic impulses in stressed media
are studied.
KEY WORDS: ultrasonic surface waves, velocity, mechanical stresses
1. Introduction
The possibilities of application of ultrasonic surface waves for the evaluation of mechanical
stresses are studied in the papers [1,2]. The experiments on the metal sheet with a stress gradient as
transversal so longitudinal are carried out. That formulation makes the results difficult to be
compared and assessed. In order to find quantitative interrelations between the parameters of the
stress-state of the material and the parameters of surface ultrasonic waves it is necessary to create a
model in which only the transversal stress gradient is significant.
The aim of the work is to find the dependence between the parameters of the stress-state and
the ultrasonic surface wave.
The procedure of the experiment is realized by consecutive loading of metal sheets and
registration the ultrasonic wave parameters. A triangular shaped beam with constant strength is
subjected to bending so that desired stress distribution is obtained. The propagation of ultrasonic
surface waves in bent metal beam is evaluated depending on the stress gradient and the ratio λ/h,
where λ is ultrasonic waves length and h is thickness of metal sample.
The sample, made from low carbon steel, is loaded with force P as shown in fig. 1. The width of the
beam b(x) at the distance x from the application point of bending force P is defined by the
expression
α
(1.1) b( x) = 2 xtg = 2 xk
2
α
where α is the angle at the apex of the triangular beam and tg =k.
2
The normal stress σ is determined by the formula [3]
M y( x )
(1.2) =σ <σs ,
Wy ( x )
where M y is a bending moment toward axis y, W y is a resistance moment, σ s is a yield stress. The
axes y and z are principal axes of inertia. The maximum value of the normal stress is determined
following (1.2)
My Px
(1.3) σ( x ) = =
Wy b( x )h 2
6
where x is the arm of the applied force P [3,4]. Taking into account of (1.1- 1.3) we obtain the
normal stress that is independent of x
3P
(1.4) σ = 2 .
hk

Fig.1. Scheme of bending: metal beam with Fig.2. An experimental set-up for ultrasonic
constant height and variable width, loaded by investigation
bending force P

According the scheme in fig.1 the part of the beam where z<0 is compressed and the part
where z>0 is strained.
The samples have the same length, width (l=0.5 m, B=0.4142 m), and angle α=45o or
k =0.414 and different thicknesses h (5, 6, 8, 10 mm). The bending loads are chosen small enough
to prevent residual deformations and stresses. The stress gradient is constant and equal to the ratio
2σ / h .

2. Ultrasonic investigations
An experimental set-up is shown in fig. 2. A computerized ultrasonic instrument is used [8].
The ultrasonic instrument allows measuring the time of ultrasonic impulse with accuracy up to 1 ns
and 8 bits resolution at sampling rate of 100 MHz. In order to attain a higher accuracy of
investigations precautions for maintaining constant conditions for the acoustic signals generation
are undertaken.

3. Experiments and results

Surface Rayleigh ultrasonic waves are excited by an angular transducer with variable angle,
at an angle of refraction close to the second critical one of a system plexiglass-steel. The nominal
frequencies of the transducers are 2 and 4 MHz. The wave lengths λ R are approximately 0.75 mm
and 1.5 mm for the frequency 4 MHz and 2 MHz. The propagation depth of surface wave is of the
order of 1.7 λ R [5,6] which makes about 2.5 mm and 1.25 mm for 2 MHz and 4 MHz.
A digital form signals emitted by transducer E and received by transducer R are recorded by
a through transmission technique as shown in fig.2. The receiving transducer moves along the
acoustic path covering distance ∆L .Velocities of surface waves are measured for beams with
varying thickness h and loads till 500N, so the stress to be less than 50-60 MPa. Velocity of the
surface wave C R is estimated following (3.1) [7] and measured as shown in the fig. 2 .
2(L 2 − L1) ∆L
(3.1) CR = = ,
τ 2 − τ1 ∆τ
where ∆L is the distance between position R1 and R2 of the receiver R, τ1 and τ 2 are transit
times of the wave obtained at distances L1 and L2. The transit time is registered from the
beginning of the ultrasonic pulse. The distance L1 is approximately 30mm that makes about 20 λ R
for 2 MHz and 40 λ R for 4 MHz. The distance L2 is about 7 times higher. The relative change of
the surface wave velocity under stress is determined by (3.2)
(3.2) dCR = (C R − C R o ) / CR o
where C R and C Ro are the velocities in stressed and unstressed sample.
Measurements of the velocity in different directions without load show anisotropy of the
material. The anisotropy is assessed as a relative change in velocities in directions from 0 to 90 °
toward the x axis as shown in fig.3(a). Figure 3(b) gives a change of velocities that are measured in
different directions under loads. Anisotropy does not change the influence of the stress on the
propagation of ultrasonic wave. Additional studies are necessary to make clearer how significant
anisotropy is.

Fig.3. Ultrasonic investigation of the anisotropy: Fig.4. Dependence dCR-σ max: 1-h=5mm
(a) Scheme of ultrasonic measurement of the anisotropy f=2MHz;2-5mm,f=4MHz;3-h=8mm,f=2MHz,
(b) Dependence of C R on loads P 4-h=8mm,f=4MHz; 5-h=10mm,f=2MHz;
6-h=10mm,f=4Mhz
Figure 4 show the dependence of the relative changes dC R on the maximum stresses at the
surface, obtained for the frequencies 2 MHz and 4 MHz for different thicknesses h. The results
are obtained for the direction 0 degree according the fig.3a, in other directions are measured but not
presented. The use of the various frequencies allows the observing the stress gradient influence on
the surface wave velocity. The changes in velocity are different for different ratio λ/ h. The relative
changes of the surface wave velocity obtained at 2 MHz with deeper penetration of ultrasonic wave
are higher than these at 4 MHz.
Signals from ultrasonic surface waves before and after loading for samples are shown in fig.
5(a) - 5 (d). The signals when sample is loaded are received earlier than those of unloaded sample
and their shapes and lengths are changed.
The presence of the stress gradient cause a change of spectrum of the pulse wave [2].
Analysis of the dependence on the stress gradient is a complex task, which requires investigation at
the various frequencies and signal processing methods.
 Fig.5a) Signals before and after loading of the sample with h=5 mm, frequency 2 MHz

Fig.5b) Signals before and after loading of the sample with h=5 mm, frequency 4 MHz

Fig.5c) Signals before and after loading of the sample with h=8 mm, frequency 2 MHz

Fig.5d) Signals before and after loading of the sample beam with h=8 mm, frequency 4 MHz

Fig.5

4. Signal processing
The registered signals from surface acoustic ultrasonic waves are processed as described in
[7]. The procedure includes smoothing, filtering, applying a Fourier transformation and
determination of the spectrum parameter (real and imaginary part, phases and amplitudes,
respectively) [9]. To assess difference between signals coming from an unloaded specimens and
these ones with mechanical stresses, we use a cross-spectrum analysis [ 10,11]. It enables us to find
correlations between signals obtained for one and the same distance and shifted with respect to
time. We can also estimate the phase difference between those signals [10].
The complex frequency spectrum of signal x(t) is found after performing Fourier
transformation
t 0 +T

(4.1) G (ω ) = ∫ x(t )e
− iωt
dt = S (ω )e −iϕ (ω ) , where S (ω ) is the amplitude frequency spectrum given
t0

by (4.2)
(4.2) S (ω ) = G (ω ) = [Re G (ω )]2 + [Im G (ω )] 2 .
The phase frequency spectrum is expressed by
t 0 +T
Im G (ω )
(4.3) φ (ω ) = arg G (ω ) = arctg ( ) ,where Im(G (ω )) = ∫ x(t ) sin ωtdt and
Re G (ω ) t0
t 0 +T

Re(G (ω )) = ∫ x(t ) cos ωtdt .

t0

a) b)

c) d)
Fig.6. Phase spectra of ultrasonic signals
The phases of the ultrasonic signals obtained from thinner sample h=5mm with high stress gradient
are shown in fig. 6 a,b. at 2 and 4 MHz. The phases of the signals obtained for different loading are
well distinguished. Figure 6 c,d show the phase spectra from sample with small stress gradient and
thickness h=10 mm. The increasing of the loading causes the change in the phases curves. For the
materials with higher stress gradient the distances between phases are larger than these obtained in
lower stress gradient.

5. Cross-spectrum analysis

We look for change of the shape of signals, spectra and phase differences along each
material, using for that purpose a cross-correlation function (5.1).
The cross-correlation function CRF (τ ) characterizes the change of a specific signal x (t) as
compared to another one y (t + τ ) , shifted in time τ (5.1)
1 to +T
(5.1) CRFxy (τ ) = ∫ x(t ) y (t + τ )dt ,
T t
0

where x(t ) is the signal recorded at a distance х and time t, y (t ) is another signal recorded at the
same distance but shifted in time τ .
To assess the interrelation between the spectral components of the two signals, we use a
cross-spectral density or a cross-spectrum. Its is a Fourier image of the cross-correlation function
being expressed by the relation
∞
(5.2) CS xy ( f ) = ∫ CRFxy (t )e − j 2πfτ dτ
−∞
The cross-spectrum is found as follows (5.3) [11]
(5.3) CSxy ( f ) = F[x * (t) y(t + τ )] = S x * ( f )S y ( f ) , where S x ( f ) is the Fourier image of x(t ) ,
S x * ( f ) is an image complex-conjugated to S x ( f ) , and S y ( f ) is the Fourier image of y (t )
respectively.
(5.4) CS xy ( f ) = CO ( f ) + iQ ( f ) = CS xy exp{iΦ x }, where CO ( f ) is the real part, Q ( f ) is the
imaginary one, and Ф is the phase of the cross-spectral function.
The phase Ф xy of the cross-spectrum CS(f) is found as in [10,11] and it determines the
phase differences between the studied signals x(t ) and y (t ) in the specified frequency range.
 Im(CS xy ( f )) 
(5.5) Φ xy ( f ) = arctg  .
 Re(CS ( f )) 
 xy 
For comparing the ultrasonic signals at different loads the spectrum of signal of unloaded
materials S0 is correlated with the spectrum of loaded materials Si . The symbols are Ф0i, where
i=1,2...8 are corresponding loads 0N, 50N,90N,170,205N,330N,430N,520N. The phases differences
are determined in the frequency region 1-4 MHz and are shown in the figure 7 . The results from
cross spectral analysis show that the stress-state causes the variation in the cross-spectrum phase Ф.
The coefficient S is the slope of the cross-spectrum phase Ф and is determined by (5.6).
(5.6) Ф( f ) = S ( f − f o )
The dependences of the cross-spectrum phase Ф on the frequency f from samples with
h=5mm and h=10 mm are shown in the fig.7. The curves are different and the differences are
considerable for the increasing stress. The slope of Ф(f) can be used as the information parameter
for identification of different loads.
 a) b)

c) d)
Fig.7. Dependence of cross-spectrum phase Ф on the frequency f from sample with h=5mm
(a) - 2 MHz, (b) - 4 MHz and sample with h=10 mm (c) - 2 MHz, (d) - 4 MHz

The figure 8(a), (b) show Ф(f) for materials with the stress σ =50 MPa and stress gradient
ratio 2σ / h from 10 [MPa/mm] to 20 [MPa/mm]. The higher stress gradient causes greater slope.
This tendency is registered in the distances between emitting and receiving transducers about
20 λR and 40 λR and marked in red colour. The slopes of the curves are vastly higher at the 7 times
larger distances (marked in black colour). The changes are accumulated with the rising of the
acoustical path. The results at frequency 2 MHz where the penetration of ultrasonic wave is deeper
show the higher values of the slope.
 a) b)
Fig.8. Ф(f) for materials with the stress σ =50 MPa and stress gradient ratio 2σ / h 10, 12.5 , 20

The figures 9(a) and 9(b) are illustrative for the change in the slopes (S) of the Ф(f) in dependence
of the stress-state in materials with different stress gradient.

a) b)
Fig.9. The slope S vs maximum stress σ

Conclusion
The performed study shows the capability of the available ultrasonic apparatus and equipment
to estimate the acoustical properties of elastically deformed materials. The small changes of the
surface wave velocities in stressed materials are registered and measured. The use of a through
transmission technique with emitting and receiving transducers for Rayleigh waves and a digital
ultrasonic flaw detector is perspective approach for stress-state evaluation of the constructions.
The obtained results conform to the results of modeling in [2]. In the recent work it is
proved again that the spectral analysis of ultrasonic waves is exclusively suitable for analyzing
mechanical stresses by ultrasound. The mechanical stress causes a change of the shape and spectra
of the ultrasonic pulses. For recognition and identification of different loads a cross-spectra
technique and phase analyses are recommended. A new information parameter is pointed out
namely the slope of curve Ф(f).
The initial anisotropy of the metal sheets is of great importance for the stress and stress
gradient influence and ultrasonic waves propagation and their parameters. In order to achieve
 proper results it is necessary to study the situations with different configuration and mutual position
of the vectors of gradients (transversal and longitudinal), principal axes of the anisotropy and wave
directing vector. Such a study is a quite complex task that requires the stress to be with exactly
defined shape and suitable signal processing method.
The materials used in the present work are taken from the serial production. The samples made
from low carbon steel sheets have anisotropy and deviation of structure parameters. The cutting of
the samples is done with a water jet device so the thermal effects and hardening of the surface are
eliminated. The cleaning of the surface is carried out with a glass paper at low speed to prevent the
cold peening. Thus the study demonstrates the possibility of the method for stress-state evaluation
in conditions very close to the real in the practice and exploitation.
Authors should be very grateful to any researcher who carries out the similar study and shares
the results.

The research was performed as a part of Project № 083/2009 with the Sofia University “St.
Kliment Ohridski”.

REFERENCES

[1] Ivanova Y.,G.Telbizov. The Possibilities for Stress Evaluation Using the Ultrasonic Surface
Waves. Sci. Announces of Sci. Tech. Union Mech. Eng. (ISSN 1310-3946), Proc. XXIII Bulg.
Conf. Int. Part. NDT-days 2008, year XV, No. 2 (105) (2008), p.185 -190 (In Bulgarian).
[2] Partalin T., I. Mihovski, Y.Ivanova, Propagation of Ultrasonic Surface Waves in Materials
Subjected to Bending, Nat. Conference "Metal Science and New Materials' 2008",Institute of Metal
Sciences, 4-5 December 2008 , p.25-30 (in Bulgarian).
[3] Kissyov S. Strength of materials, Technique, S., (1978) (in Bulgarian).
[4] Physical quantities, Handbook, M., Atomizdat (1991). (in Russian).
[5] Viktorov A. Rayleigh and Lamb waves, Physical Theory and Application, Plenum Press, New
York , 154, 1967.
[6] Nondestructive testing, Handbook, Vol. 4, Moscow, Publ.House “Spectr” (2007) (in Russian).
[7] Ivanova Y., T. Partalin, M. Mihovski, Experimental study of the ultrasonic wave propagation in
materials with mechanical stresses, Journal of Theoretical and Applied Mechanics, 2010 (in print).
[8] Guide to OPKUD and OPBOX- Ultrasonic Testing Units-Technical Data, http://www.optel.pl.
[9].Ivanov R. , Digital signal processing of signals, Gabravo, Bulgaria, 1999. (In Bulgarian).
[10].Hartmann D., Cross Spectrum Analysis, ATM 552 Notes, Section 6c, p.157, Internet.
[11].Zywicki D., Advanced signal processing methods applied to engineering analysis of seismic
surface waves, PhD Dissertation, Georgia Institute of Technology, 1999.

View publication stats