Effect of Viscosity On Ultrasound Wave Reflection

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

ELSEVIER Ultrasonics 34 (1996) 817-824

Effect of viscosity on ultrasound wave reflection from a


solid/liquid interface
Vimal Shah, Krishnan Balasubramaniam *
Department of Aerospace Engineering and Mechanics, Mississippi State University, Mississippi State, MS 39762, USA

Received 4 April 1996; revised 7 October 1996

Abstract

A study of the simulated reflection of a wideband ultrasound shear wave from the solid/viscous fluid interface is presented.
Various parameters affecting reflection factors including the material properties of the solid, fluid properties like density and
viscosity, and the operating frequency are discussed. Simulated ultrasonic response waveforms are compared with the experimentally
obtained data for NIST traceable calibration standards of viscosity. A good agreement was observed between the simulated and
experimental waveforms at various viscosities and for different solid substrates.

Keywords: Viscosity; Ultrasound reflection; Simulation

1. Introduction predicting viscosity using only the magnitude of a


reflected shear ultrasound RF pulse has been demon-
Measurement of material properties using ultrasound strated by Sheen et al. [ 51. Since the ultrasound response
is a well-researched area of study. Ultrasound has been to viscous loading depends on frequency, most measure-
used to measure the viscosity of fluids as early as 1949 ments were made assuming that the operating frequency
[I]. It has since been explored as a device for non- was the resonant central frequency of a wideband ultra-
invasive, on-line property and process monitoring [ 2-51. sonic transducer. The aim of this paper is to understand
Early research has concentrated on the conventional the effect of broadbanded nature of the transducer on
means of producing and detecting ultrasound [6]. This the solid-viscous fluid interface.
method of employing piezoelectric transducers to gener-
ate and receive ultrasound is, by far, the simplest
approach. Piezoelectric transducers have many distinct
advantages [7], which have so far outweighed their 2. Theoretical background
disadvantages. Recently, a hybrid system using laser-
based generation and piezoelectric transducer-based The basis of shear reflectance method was obtained
detection has been used to measure viscosity [8]. by Moore and McSkimin [6] using stress equality at
The traditional ultrasound-based approach to obtain- the surface of contact between the viscous fluid and the
ing viscosity has been to correlate it with ultrasound solid base (Fig. 1). This relation is further modified [ 111
properties like attenuation and reflection factors. A to obtain a relation between the viscous impedance of
viscosity sensor based on shear horizontal acoustic sur- the fluid and complex ultrasound reflection factor for a
face wave mode has been developed [9] where the given set of substrate material properties. Since the
propagation loss and the phase shift between the refer- complex reflection factor can be represented in an expo-
ence and the loaded waveforms were used to predict the nential form, the amplitude is referred to as the absolute
viscosity. Approaches based on continuity of stresses reflection factor and phase as the phase difference in
and displacements [lo] at the solid/liquid interface have this paper.
also been used to derive the reflection/transmission The complex acoustic impedance (2) of the fluid in
coefficients. Based on these derivations, a method of terms of its viscosity and density is given as Z = (ipqo)“’
[ 123, where p is the density and q is the shear viscosity
* Corresponding author. Department of Aerospace Engineering and of the fluid, and o is the circular frequency of the
Mechanics, Mail Stop 9549, Mississippi State University, MS 39762, propagating wave. If we assume the complex impedance
USA; fax: + l-601-325-7730; e-mail: KBI@ra.msstate.edu. of the fluid as Z = R + ix, a relation between the phase

0041-624X/96/$15.00 Copyright 0 1996 Elsevier Science B.V. All rights reserved


PZZ SOO41-624X(96)00082-0
818 Vimal Shah, Krishnan Balasubramaniam 1 Ultrasonics 34 (1996) 817-824

(obtained from Cannon Instruments Inc.) were used to


study their relative effects on the reflection factors. A
typical variation in fluid density is 891 to 872 kg/m3 due
to a temperature change from 25 to 60°C and the
corresponding variation in viscosity is from 416 to
26 poise. Density change due to temperature variation
Viscous Liquid is an order of magnitude less than corresponding viscos-
ity change. Hence, the effect of density on the reflection
factors is also an order of magnitude less than the effect
of viscosity change (Fig. 3).
The operating frequency governs the sensitivity of
Fig. 1. Schematic of plane polarized shear horizontal wave
reflection factors to viscosity (Fig. 4). As the operating
propagation at a solid-fluid interface.
frequency (f) increases, the slope of the viscosity versus
reflection factors curve increases, leading to greater
difference (40) and the absolute reflection factor (r) is sensitivity to small viscosity changes. This is true for
obtained by equating the real parts: both phase difference and absolute reflection factor. The
1 - 2r2( 1 + 2 sin2(@)) + r4 = 0. solid substrate material properties come into effect by
(1)
the virtue of an impedance mismatch. As the impedance
The imaginary part gives the relation between viscosity, mismatch decreases (Fig. 5), the total energy transferred
phase difference and the absolute reflection factor as, to the fluid increases and consequently the reflected
2pzci cos2(8,) I sin(@)( 1 - r2) energy and phase are more sensitive to a change in
~=xpfcos2(8,)(1+rZ+2rCos(@))2’
(2) viscosity of the fluid.

where ps and c, are the density and shear wave speed in


the substrate respectively, f is the frequency of propagat- 2.2. Simulation of rejected ultrasonic wave response
ing wave and, 8, and 0r are the angles of incidence and
transmission of the ultrasound wave at the interface For the simulation of the ultrasonic shear wave
defined by Snell’s law and the shear wave velocities c, response from a solid-fluid interface, normal incidence
and cf as, situation (0, = Bf= 0) were considered. Material proper-
ties of three solid substrates (Plexiglas, graphite and
of = sin-’ [sin(B,)cf/c,]. (3) aluminum) and various calibration fluids were used
In these relations, we assume the fluid media to be in the simulation. To simulate the fluid loading, the
perfectly viscous (Newtonian fluid), and therefore neglect actual reference signals had to be obtained for each
the elastic component of the shear modulus of the fluid. transducer/substrate combination. The reference signals
The reflection factor relationship with viscosity has been were obtained at zero load (stress free state), from the
decoupled to a relation between viscosity and either the back-wall reflection of a solid substrate (Fig. 6).
absolute reflection factor or the phase difference. Wideband ultrasound waves were generated and
received by Y-cut broadband piezoelectric transducers
2.1. Parameters affecting rejection factors (Panametrics V153 and V154, with center frequencies
1 and 2.25 MHz respectively), pulsed using a pulser/
The effect of viscosity on the absolute reflection factor receiver (Panametrics PR5052). The received ultrasonic
and the phase difference are shown in Fig. 2. With an signals were averaged at 256 samples per signal to reduce
increase in the viscosity of the fluid media, the absolute random noise. These signals were digitized using an 8 bit
reflection factor decreases exponentially (Fig. 2a). The digital oscilloscope (Tektronix TDS 320). The digitized
phase on the other hand increases (Fig. 2b). This beha- signals (x(t)) were gated and windowed using a rectangu-
vior holds until the phase reaches a discontinuity lar window to avoid pseudo phase changes associated
(r x 0.41). In the physical problem, other effects such as with other windows. The amplitude (AR(@) and phase
relaxation [9] become important and this state is not (&(w)) information was obtained using standard Fourier
achieved for Fig. 2. As the angle of incidence of the transforms [ 13,141 represented as follows:
ultrasound wave increases, effect of viscous loading on
the shear wave becomes more prominent, leading to an x(t) c, A&) exp(i&&)). (4)
increase in sensitivity of the reflection factors to change
in viscosity. A -6 dB bandwidth of the ultrasound signal was
The fluid properties p and q both have a similar assumed to avoid portions of the spectrum with a low
relationship with the reflection factors [Eq. (2)]. The signal/noise ratio (SNR). The reflection factors (r&q u))
fluid properties of NIST traceable calibration standards for the substrate-fluid interface were theoretically com-
Vimal Shah, Krishnan Balasubramaniam J Ultrasonics 34 (1996) 817-824 819

2500
$
._
: 2ooa
.r 1
- 1500

1
C

::
$ 1000
._
>
500 -J
OZ.4
I
0

(b)

2500.
0)
.E

2 2000-

.r
a 1500-
.il

I:
y) lOOO-
._
>
soo-

01
0

Fig. 2. Typical plots (theoretical) showing the variation of (a) the absolute reflection factor and (b) the phase difference, with viscosity and angle
of incidence of plane polarized shear wave at the solid (Plexiglas)/fluid (oil) interface.

puted using Eqs. ( 1) and (2) as: expected response spectrum using standard algorithms:

r&4 44 = Ir&, 491exp(iW4 r)). (9 4(e4 r) exp(iMc4 rl))++ As(t). (8)


A lookup table was used, since the relationship
The real part of this inverse transform gives the simulated
between viscosity and reflection factors is indirect. The
RF response. This simulated RF response was archived
absolute reflection factors and phase differences corre-
for comparison with experiments.
sponding to each frequency were factored from the
The experimental RF response was also obtained
frequency spectrum using the relations:
using the experimental setup illustrated in Fig. 6. After
A&J, 4 = AR64 I~A%49Iv (6) the reference signal required for the simulation was
obtained, the substrate was loaded with the calibration
MJ.4 44 = #R(W)- 4(% ?>* (7) fluid. The fluid was maintained at a constant temper-
An inverse Fourier transform was performed on the ature, and the experimental RF response (&(t)), due to
820 Vimal Shah, Krishnan Balasubramaniam / Ultrasonics 34 (1996) 817-824

Density of Fluid

0 200 400 600 800 1000


Viscosity in Poise

Fig. 3. Variation of phase difference with change in fluid density and viscosity.

frequency

1 MHz

4MHz

0 SO 100 150
Viscosity in Poise

Fig. 4. Effect of operating frequency on the relation between the phase difference and viscosity.

the viscous loading assumed in the simulation, was 891.3 kg/m3 at 25°C) interface. Fig. 7(a) shows the refer-
obtained. ence signal (x(t)) for a Plexiglas/air interface. The ampli-
tude spectra (A&B)) of this reference signal and response
due to the assumed viscous loading ( Irt(q ?)I), within
3. Results and discussion the bandwidth of the RF signal are shown in Fig. 7(b).
Corresponding phase spectra (&(o)) and response due
Plexiglas, graphite and aluminum substrates were to the assumed viscous loading (@&o, q)) for the same
used to obtain the experimental data. Fig. 7 shows the RF signal are illustrated in Fig. 7(c). the simulated RF
experimental and simulated waveforms obtained for response (As(t)) is plotted along with the reference (AR(t))
Plexiglas/fluid N15000 (q = 416.4 poise and p = and the experiment (Au(t)) in Fig. 7(d). The experimental
Vimal Shah, KrtihhnanBalasubramaniam/ Ultrasonics 34 (1996) 817-824 821

fj 0.6
a
Substrate Solid
d
.a Os ----m--- Plexiglas
--+--- Graphite
--bk- Aluminum
- -a- - Steel
+_-~a- _ 4-
-*--

0.1

0.0
0 5000 10000 15000 20000 25000 30000
Viscosity in Poise

Fig. 5. Effect of substrate material properties on the relation between the phase difference and viscosity.

may be contributing to the disagreement in comparison


are discussed next.
In the derivation of Eq. (2), it is assumed that the
media is semi-infinite and the incident wave is a plane
wave. In the experiment, however the substrate and
source were both finite. The plane wave assumption is
not strictly valid and the generated waves were spherical
in nature. Also, the transducer does not generate a
perfect shear wave and due to the Poisson’s effect shear
wave generation is always associated with simultaneous
generation of compressional waves. Compressional
f
waves normally travel almost twice as fast as shear
waves in solids and are affected by the fluid bulk
modulus and viscosity. Depending on the substrate
Fig. 6. Block diagram showing experimental setup. characteristics, these waves could be received at the
same time as shear waves and so may interfere. It is
difficult to eliminate the effects of these compressional
and the simulated responses are similar and as expected, waves entirely.
have a reduced amplitude and a phase delay when A maximum error of +2.5% is expected in all the
compared with the reference. Figs. 8,9 and 10 show the measurements. Temporal averaging was used to reduce
correlation of experimental and simulated data for the instrument noise. Trigger jittering delay may have
different substrates. Data, obtained using Plexiglas as been averaged and could have led to experimental errors
the substrate, show the most sensitivity to change in in the measurement of the phase.
viscosity (Fig. 8). The response of aluminum/viscous Accounting for phase difference in the reflected wave
fluid interface to change in viscosity, shows the least is not in complete agreement with the experimental
sensitivity (Fig. 10). phase difference. In the simulations, the phase difference
In general, we observed that the comparison between relation is assumed to be valid only within the bandwidth
the experimental data and the simulation showed a good of the received ultrasound signal. Physically, the relation
agreement at the onset of the waveform until the first is valid for the low and high frequency regions beyond
peak. After the first rise (or the second rise when the the bandwidth. So, the fluid responds to the entire
1 MHz transducer is used), the correlation began to frequency range, whereas the simulation response is only
degrade. It was observed that the degradation became with respect to the bandwidth. Hence, a small difference
severe with an increase in viscosity of the fluid. This was in the simulated signal compared with the experimental
true for all three types of substrates. Some factors that data can be expected.
822 Vimal Shah, Krishnan Balasubramaniam / Ultrasonics 34 (1996) 817-824

-0.15
0 100 200 20 40 60 80
Time in 20 * nanosec. Frequency in 12.2 * KHz

-6
20 40 60 80 0 100 200
Frequency in 12.2 * KHz Time in 20 * nanosec.

Fig. 7. Simulation of reflection from Plexiglas/fluid NlSOOO (416.4 poise at 25°C) interface for 1 MHz transducer. (a) Reference signal (@),
(b) amplitude spectrum (-) and viscous loading response (x) to amplitide, (c) phase spectrum (-) and viscous loading response (x) to phase, and
(d) comparison between reference (@I),experimental (0) and simulated (-) waveforms.

p”
z -0.05
E”
4
-0.20
0 50 100 0 50 100
Time in 5 * nanosec. Time in 5 * nanosec.

m
% 0.10
> 0.10 (cl
E
.f .9
d 0.00 . 2 0.00
; : .%
%
=
z
*c 4
-0.10 -0.10
0 50 100 150 0 50 100
Time in 5 * nanosec. Time in 5* nanosec.

Fig. 8. Comparison of simulated (-) and experimental (0) reflections from a Plexiglas/viscous fluid interface. Simulations for fluids (a) NlOOO
(19.94 poise) (b) N2000 (52.49 poise), (c) N4000 (104.5 poise), and (d) all three simulations shown together (+ NlOOO, + N2000, x N4000), for
2.25 MHz transducer at 25°C.

4. Conclusions experiments. Simulations compare well with the experi-


mental measurements. The comparison is dependent on
Simulations based on the impedance model of reflec- the acoustic impedance mismatch between the substrate
tion of plane polarized ultrasound shear waves from a and the viscous fluid, as well as the viscosity of the fluid.
solid/fluid substrate were presented and validated by The sensitivity of the shear wave reflection response to
Vimal Shah, Krishnan Balambramaniam / Ultrasonics 34 (1996) 817-824 823

Time in 10 * nanosec. Time in 10 * nanosec.

k 0.2 (cl
E

0.0

-0.2
w

0 150 300 0 150 300


Time in 10 * nanosec. Time in 10 * nanosec.

-.
__-
--
Fig. 9. Comparison of simulated (-) and experimental (0) reflections from a graphite/viscous fluid interface. Simulations for fluids (a) NlOOO
(19.94 poise), (b) N2000 (52.49 poise), (c) N4000 (104.5 poise) and (d) all three simulations shown together (+ NlOOO, l N2000, x N4000), for
1 MHz transducer at 25°C.

2 (a)

lIii!il
.9 0.4
4

@
4
5 iO.2
-0.8
0 150 0 150
Time in 10 * nanosec. Time in 10 * nanosecs.

>
.: 0.4 IIfC!EI (c)
%
.E, 0.4

5 4
5 -0.2 3 -0.2
3 %
4
-0.8 -0.8
0 150 0 150
Time in 10 * nanosecs. Time in 10 * nanosecs

Fig. 10. Comparison of simulated (-) and experimental (0) reflections from an aluminum/viscous fluid interface. Simulations for fluids (a) NlOOO
(19.94 poise), (b) N2000 (52.49 poise), (c) N4000 (104.5 poise) and (d) all three simulations shown together (+ NlOOO, l N2000, x NW) for
1 MHz transducer at 25°C.

changes in viscosity was obtained for the substrate with Acknowledgements


best impedance match. Some reasons for discrepancies
between the impedance model-based simulations and This work is supported by the United States
the experiments were discussed in the context of experi- Department of Energy, through grant number
mental and signal processing procedures. DE-FG02-93CH-10575 to the Diagnostic Instru-
824 Vimal Shah, Krishnan Balasubramaniam / Ultrasonics 34 (1996) 817-824

mentation and Analysis Laboratory at the Mississippi [7] J. Krautkramer and H. Krautkramer, Ultrasonic Testing of
State University. Materials (Springer, Berlin, 1990).
[8] R.D. Costley Jr. et al., Rev. Prog. QNDE eds. D.O. Thompson
and D.E. Chimenti, 15 (1996).
[9] T. Sato et al., Jpn. J. Appl. Phys. 32 (5B 1) (1993) 2392.
References [lo] H.T. Chien (1995) Private communication.
[11] V.V. Shah et al., Rev. Prog. QNDE, eds. D.O. Thompson and
[l] W.P. Mason et al., Phys. Rev. 75(6) (1949) 936. D.E. Chimenti, 15 (1996) 2067.
[2] W.P. Mason and H.J. McSkimin, Bell Systems Tech. J. 3( 1) [12] K.F. Herzfeld and T.A. Litovitz, Absorption and Dispersion of
(1952) 122. Ultrasonic Waves (Academic Press, New York, 1959).
[ 31 W. Roth and S.R. Rich, J. Appl. Phys. 24(7) (1953) 940. [13] E. Oran Brigham, The Fast Fourier Transform and its
[4] L.C. Lynnworth, IEEE Trans. Sonics and Ultrasonics 22(2) Applications (Prentice-Hall, New Jersey, 1988).
(1974) 71. [ 141 A. Papoulis, Signal Analysis (McGraw-Hill, Singapore, 1984).
[S] S.H. Sheen et al., Rev. Prog. QNDE, eds. D.O. Thompson and [15] T.A. Litovitz and C.M. Davis, Physical Acoustics, Vol. 2, ed.
D.E. Chimenti, 14(A) (1995) 1151. W.P. Mason (Academic Press New York, 1964) pp. 218-349.
[6] R.S. Moore and H.J. McSkimin, 1970 Physical Acoustics, Vol. 6, [ 161 M.C. Davis, J. Acoust. Sot. Amer. 64(2) (1978) 406.
ed. W.P. Mason (Academic Press, New York, 1970) pp. 167-243. [ 171 S.H. Sheen et al., IEEE Symp. Ultrasonics (1988) 537.

You might also like