CH 2 - Wave Propagation in Viscous Fluid PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Chapter 2

Wave Propagation in Viscous Fluid

This chapter summarizes with the derivation of the mathematical form of the acoustic
wave propagation in the fluid. Before we derive the final form of the wave propagation
equation in viscous fluid, we first look at two conservation (mass and momentum) of
equations and state equation in the fluid. Detailed derivations can be found in the
literatures [4, 6, 27, 28]. We limit our discussion only on the lossy 1-dimensional plane
wave.

2.1 One-dimensional Viscous Wave Equation


2.1.1 One-dimensional Continuity Equation (Conversation of Mass)
Consider the flow of a compressible fluid through a duct of arbitrary cross section (area
S) in one dimension. The control volume (CV), is the segment between x and x + x .

We want to know the rate at which the inside mass changes. First, we made two
assumptions:
1.

The CV is fixed in space

2.

The flow is one-dimensional, so the mass flow only depends on t and x.

Mass flow in
= uS | x

Mass flow out


= uS | x + x
x + x

, where , u are the average mass flow density and average mass flow speed,
respectively. For x 0 , and u will become a true point function.

The time rate of increase of mass inside the CV is equal to net mass flow into the CV
through the CV surfaces or in mathematical terms,

( S x) = uS | x uS | x + x
t

(1)

Since S is a constant and x is not a function of time, Eq. (1) may be rearranged as
following:

u | x u | x + x
=
t
x
Note that, to simplify the notation, we use the subscript (t & x) to represent the derivative
of the function respective to time and/or distance. In the limit as x 0 , the right-hand
side becomes ( u ) x , and we obtain

t + ( u) x = 0

(2)

This is the equation of continuity.

2.1.2 Conversation of Momentum in One Dimension


The same control volume (CV) is again used to derive the conservation of momentum
relation. The momentum per unit volume is u , and the momentum flux (momentum
per unit area per unit time) is u 2 S .

PS | x

PS | x + x

Momentum outflow
= u 2 S | x + x

Momentum inflow
= u 2 S |x
x + x

Applying Newtons second law, and taking into account the momentum inflow across the
CV boundaries, the time rate of increase of momentum inside the CV equals to the sum
of the net momentum inflow across boundaries and the sum of the forces acting on the
CV.
Forces are generally classified into two kinds: body forces and surface forces. In this
case, we make some assumptions about the forces in addition to the assumptions listed
for the derivation of continuity equation.
1.

We could neglect body forces. The most common body force is gravity, since we

are only considering a short distance; the effect of gravity is not significant.
2.

The fluid is inviscid, (we will later discuss the effect of viscosity). So the only

significant surface force is that due to the pressure P at the CV end surface.

Given these four assumptions, the mathematical form of the conservation of momentum
statement is

( u S x) = u 2 S | x u 2 S | x + x + PS | x PS | x + x
t
Divide the above equation by S x , and take the limit as x 0 :

( u ) t + ( u 2 ) x + Px = 0

(3)

Expanding the first two terms, we have

t u + u t + ( u ) x u + u u x + Px = 0
From the continuity equation (eq (2)), the first and third term is canceled:

(u t + u u x ) + Px = 0

(4)

,which is simpler than eq (3). Thus, it is the final form of the conservation of momentum.
To include the viscous effect in momentum equation, the surface force is therefore
present. From Navier-Stokes equation, we could rewrite the conservation of momentum
as:

(ut + u u x ) + Px = ( + 2 )u zz

(5)

, where is the shear viscosity coefficient and is the dilatational viscosity coefficient.

2.1.3 Equation of State

For fluids or gases, the equation of state relates to pressure can be expressed as a function
of density and entropy. When losses are negligible, the entropy remains constant, and
then the pressure is a function of density alone:
P = P( )

For gases, the isentropic equation of state (so called adiabatic gas law) is often used:

(P

p0

) = (

0 )

(6)

where is the specific heats ratio and p0 and 0 are the static values of P and .
In the case of liquid, the equation of state can be expressed as the series of the general
isentropic equation of state from eq. (6):

P = p0 + A

0 B 0 2 C 0 3
+ (
) + (
) +L
2! 0
3! 0
0

(7)

The coefficient A, B, C are determined from experiments or from other analyses.


From some derivation coefficient A is equal to 0 c02 . 0 is the static value of the fluids
density and c 0 is the static value of the velocity of the wave, respectively.
If we take a further step, simplify eq. (7) by introducing the excess pressure
p P p0
and the excess density is

0
and incorporating these into coefficient A, then the isentropic state of equation for fluid
(eq (7)) becomes

p = c02 [1 +

B
C
(
) L]
+
2! A 0 3! A 0

(8)

2.1.4 Linearization / Small-signal approximation


Eq. (2), (5), (8) are nonlinear equations, we could obtain the linear version by introducing
the so-called small-signal approximation. Most sound waves disturb the status quo in the
fluid only around a small region, the variables associated with sound, excess pressure,
excess density, and particle velocity may be assumed to be small quantities of first order.
We start by assuming

<< 0
Given the expression of excess density, we can expand the equation of continuity (eq (2))
as

t + u x + 0 u x + u x = 0
The second and fourth terms are second-order terms; each contains the products of two
small quantities.

The first and third terms are first-order term. If the first-order terms

are small, then the second-order can be neglected. Dropping them, we obtain

t + 0 u x = 0

(9)

10

The momentum equation may be linearized similarly. Expanded eq. (5) becomes

0 ut + 0 uu x + ut + uu x + p x = ( + 2 )u xx

( 10 )

Again, we can drop the higher order terms. The second, third, fourth terms are higher
than first order, so we can neglect them. Then the conservation of momentum equation
becomes

0 ut + p x = ( + 2 )u xx

( 11 )

Lastly, the isentropic equation of state (eq. (8)) can be linearized by inspection. Eq. 8
becomes
p = c02

( 12 )

2.1.5 Linear Viscous Wave Equation


Partial Discharge (PD) is a point charge emitted from the windings [19]. Since it is a
sudden release of energy, it generates some acoustic wave propagating around its
neighborhood. Besides, considering the point charge is released in an infinitely large
amount of medium, the acoustic wave could be considered as a plane wave.
The three fundamental acoustic plane wave equations are derived in the previous section.
The linearized equations are summarized as following:
Continuity: t + 0 u x = 0,

(9)

Momentum: 0 u t + p x = ( + 2 ) u xx

(10)

State: p = c02

(11)

11

is the density of the fluid, p is the excessive pressure of the fluid, u is the velocity of
the fluid. The subscript (t & x) represents the derivative of the function respective to the
variable. , ,and are dilatational and shear viscosity coefficients, respectively.
In this section, we combine eq. (9), (10) and (11) to find the final form of the viscous
wave equation.
Since the shear viscosity coefficient has been measured for many fluids, the
dilatational viscosity coefficient will not be found. By Stokes assumption, we could
approximate = 2 3 , which leads to replacement of + 2 by 4 3 . For fluids
that do not follow Stokess assumption, modern practice is to replace by 2 3 + B ,
~
where B is called the bulk viscosity coefficient. Therefore, + 2 = 43 + B = V ,

~
where V = 4 3 + B . In our study, we assume that the fluids follow Stokes
assumption. The momentum equation becomes

0 ut + p x =

4
u xx
3

( 13 )

take the derivative respective to x on (9) and substitute u xx into (12) and it becomes:

0 ut + p x =

4
tx
3 0

( 14 )

from (11), we take the derivative respective to t and x, and we get ptx = c02 tx , and
substitute tx into (13),

0 ut + p x =

4
ptx
3 0 c02

and we take the derivative respect to x of the above equation

12

0 u tx + p xx =

4
ptxx
3 0 c02

( 15 )

From (9) and take the derivative respect to t, and we get tt = 0 u xt ; also from (11)
and take the 2nd derivative respective to t, and we get tt = 1 c02 ptt , therefore
u xt = 1 0 tt = 1 c 02 0 p tt . Finally we substitute u xt into (14), and then we have

4
1
p xxt + p xx 2 ptt = 0
2
3c0
c0

( 16 )

where = 0 is the kinematic viscosity coefficient. The three-dimensional version of


this equation is shown as follows:

4 2
1
pt + 2 p 2 ptt = 0
2
3c0
c0

( 17 )

13

2.2 Time-Harmonic Analytical Solutions to the Viscous


Wave Equation
2.2.1 Solution
To solve eq(16), it is customary to assume a general solution of the partial differential
equation and find the particular solution for the equation. Also, to take the various
mechanism of dispersion in the fluid one at a time, we can determine their effect on
sound propagation by following the procedure outlines in Fig 1.

Wave Equation
j ( t kx )
Assume
Assume solution
solution pp =
= pp 00 ee j ( t kx )

Dispersion relation
k = k ( )
k = j

Attenuation
coefficient

Phase velocity
c ph = /

Fig 1. Dispersion relation algorithm

Since we let p = p 0 e j ( t kx ) , then p xxt = p 0 ( k 2 j ) e j ( t kx ) , p xx = p 0 (k 2 ) e j ( t kx ) ,


and ptt = p 0 ( 2 ) e j ( t kx ) .

Insert the above three derivatives into eq(15) and

eliminating the constant and the common exponential terms, then we get

14

4 2 2
(1 + j 2 )k 2 = 0
3 c0
c0
or

c0

k =

4
1+ j v
3

= ( j )

, where v is the dimensionless coefficient

v =

c02

If we only consider the wave propagates in positive direction, which k= - j , the timeharmonic solution of the excess pressure becomes
p = p 0 e j ( t x + j x )
= p 0 e x e j ( t x )
= p 0 e x e

j ( t

(18)
)

The magnitude of the attenuation coefficient determines how fast the peak amplitude of
e

j ( t

decays. Also, the variation of wave propagation speed c ph = will be

determined by .
Notice that the form of above excess pressure is ideal for time-harmonic forced waves. A
source emitting a signal p = p0 e j t at x=0 is implied. Also, the above derivation is done
in Cartesian coordinate; it can be easily extended to spherical or cylindrical coordinate.

15

2.2.2 Ideal 1D Acoustic Wave Propagation at Different Frequencies in


Water and Oil
To study the wave propagation phenomenon in different mediums and under different
frequencies, we can apply the theoretical acoustic viscous wave equation solution from
section 5 with the coefficients associated with different fluid medium.
Recap assumptions on the acoustic wave theoretical derivation:

Plane wave (1-dimension)

Adiabatic process

No internal molecular process

Only viscous effect considered

The fluid follows Stokes assumption, for which the dilatational viscosity
coefficient equals to minus two-third of viscosity coefficient.

Using caster oil as the medium, the parameters are given as following:
Density: 0 = 950 (kg

m3

Propagation Velocity: c0 = 1540 (m )


s
Viscosity coefficient: 0 = 0.96 ( N s )
m
Let the wave peak amplitude be 1 at the source (x=0), we simulate the wave peak
amplitude, which is the term p0 e x in eq. (17) with p0 to be 1, vs. the distance from the
source at various frequencies. The results are shown in the following two figures (in dB
and linear scale):

16

Caster Oil
0

Excess Pressure Amplitude (dB)

-5

-10

-15

-20

-25
0.2

freq =
freq =
freq =
freq =
freq =
0.4

50000 Hz
100000 Hz
200000 Hz
300000 Hz
400000 Hz
0.6

0.8
1
1.2
1.4
Distance from Origin (m)

1.6

1.8

Fig 2. Excess pressure amplitude (dB) vs. the distance from the origin. The wave peak amplitude at
the origin is 1.
Caster Oil
1
0.9

Excess Pressure Amplitude

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.2

freq =
freq =
freq =
freq =
freq =
0.4

50000 Hz
100000 Hz
200000 Hz
300000 Hz
400000 Hz
0.6

0.8
1
1.2
1.4
Distance from Origin (m)

1.6

1.8

Fig 3. Excess pressure amplitude vs. the distance from the origin. The wave peak amplitude at the
origin is 1.

As expected, the peak wave amplitude becomes smaller as we move further from the
source. We can also notice that as the wave frequency goes higher, more attenuation can
be observed at any given location. In other word, the attenuation factor depends on both
the location and frequency.
Similarly, we simulate the wave peak amplitude vs. distance of water using the following
parameters:
17

Density: 0 = 998 (kg

m3

Velocity: c0 = 1481 (m )
s
Viscosity coefficient: 0 = 0.001 ( N s )
m
The major difference between caster oil and water, in terms of parameters, is viscosity.
Water is much less viscous than oil, about a thousand times less. So we can expect that
the wave peak amplitude attenuation in water as further away from the source is much
less the attenuation in caster oil.
Fig. 4 and 5 shows the wave peak amplitude vs. distance from the origin in water

Water
0

Excess Pressure Amplitude (dB)

-0.005

-0.01

-0.015

-0.02

-0.025
0.2

freq =
freq =
freq =
freq =
freq =
0.4

50000 Hz
100000 Hz
200000 Hz
300000 Hz
400000 Hz
0.6

0.8
1
1.2
1.4
Distance from Origin (m)

1.6

1.8

Fig 4. Excess pressure amplitude (dB) vs. the distance from the origin. The wave peak amplitude at
the origin is 1.

18

Water
1.0005

Excess Pressure Amplitude

0.9995

0.999

0.9985

0.998

0.9975

0.997
0.2

freq =
freq =
freq =
freq =
freq =
0.4

50000 Hz
100000 Hz
200000 Hz
300000 Hz
400000 Hz
0.6

0.8
1
1.2
1.4
Distance from Origin (m)

1.6

1.8

Fig 5. Excess pressure amplitude vs. the distance from the origin. The wave peak amplitude at the
origin is 1.

Note that, the acoustic wave peak amplitude in water gets very minor attenuation across
frequencies in this calculation, even if we are at 2 meter away from the source. In the
caster oil case, we could observe large attenuation of wave peak amplitude, due to the
larger coefficient of viscosity. This result shows that if we measure the acoustic wave in
water at any location, assuming we are inline with the source, we will get a comparable
signal density as the source, however in the caster oil case, we will need a more sensitive
sensor to catch up the signal and perhaps a need for more complicated signal processing
units to filter out the noises, since the signal level might be low.

2.3 Experiment
We measured the acoustic wave with a container filled with water inside. Two PZTs, one
acting as the source emitter and one as the detector are then immersed inside the water.
The experiment setting is shown in Fig 6.

19

Fig 6. Experiment setup. Left: top view. Right view: side view

The source and detector are inline and both are lifted up to a same height by some steel
fixture. A typical acoustic waveform is shown in fig. 7.
Acoustic w aveform at 120mm from emitting source
0.4
0.3
0.2
0.1
Amplitude 0
-0.1
-0.2
-0.3
-0.4
1

1.5

2.5
3
Time in sec

3.5

4.5

5
-4

x 10

Fig 7. Typical acoustic waveform

We recorded the waveform three times at each location and did an average in time
domain among the three. Then we performed Discrete Fourier Transform to get the
spectrum of the waveforms. Most of the energy concentrates in the region of 300kHz to
400kHz.

The following two plots show the zoom spectrum between 100kHz and

600kHz; and 280kHz and 400kHz, respectively.

20

FFT

20

10mm
20mm
30mm
40mm
50mm
60mm
80mm
100mm
120mm

18
16
14
12
10
8
6
4
2
0
1

1.5

2.5

3
3.5
4
Frequency (Hz)

4.5

5.5

6
5

x 10

Fig 8. Zoomed FFT of the averaged acoustic wave between 100kHz and 600kHz. The legend indicates
the distance between the source and measurement point
FFT

20

10mm
20mm
30mm
40mm
50mm
60mm
80mm
100mm
120mm

18
16
14
12
10
8
6
4
2
0
2.8

3.2

3.4
Frequency (Hz)

3.6

3.8

4
5

x 10

Fig 9. Zoomed FFT of the averaged acoustic wave between 280kHz and 400kHz. The legend indicates
the distance between the source and measurement point

21

As shown in fig. 8 and fig. 9, the maximum amplitude occurs in the region of 320kHz
and 340kHz across different locations.

The maximum peak between 300kHz and

400kHz region (in fig. 10) shows that there is not much amplitude variation.
Maximum Peak Amplitude from FFT between 300kHz and 399kHz
16
14
12
10
8
6
4
2
0

10

20

30

40

50 60
80
100
Distance from source (mm)

120

Fig 10. Maximum amplitude between 300kHz and 400kHz.

Also, we compare the peak amplitudes at 320 kHz. The peak amplitudes stay within the
same level across the distances from the origin.
Amplitude from FFT at 320kHz
16
14
12
10
8
6
4
2
0

10

20

30

40

50 60
80
100
Distance from source (mm)

Fig 11. Amplitude at 320kHz.

22

120

The experiment results in water coincide with the findings from theoretical derivation,
which the acoustic wave peak amplitude does not change with various distances from the
origin.

2.4 Simulation
A computer simulation using ABAQUS is set up using the geometry of the experimental
container; the acoustic wave propagation damping-rate in the experimental container can
be estimated. Fig 12 shows the normalized wave amplitude at the locations with different
distances from the source.

Peak Amplitude
( Normalized)

Acoustic Wave Propagation Damping Rate in


the Experimental Container from Computer
Simulation
1
0.8
0.6
0.4
0.2

80
10
0
12
0
14
0
15
0
16
0
17
0
18
0
19
0

50
60

30
40

0
10
20

Distance from the Source (mm)

Fig 12. Propagation Damping Rate in the Experimental Container

Comparing these two figures (Fig 10 and Fig 12) obtained from the experiment and the
simulation; it is found the wave propagation damping-rates are similar. The wave peak
amplitude does not change much with various distances from source in water, which
confirms the above conclusion when comparing the experimental result with the
theoretical analysis result. But when distance to the source is larger than 120mm, the
damping rate from the experiment and the computer simulation are different from that of
the theoretical analysis. In the theoretical analysis, acoustic wave propagates in the

23

infinite medium while it is not the case in the experimental container. So the discrepancy
was due to the size limitation of the experimental container (same happened to the
computer simulation model), and influence of its inner structures. The wave would be
reflected at the rear surface of the experimental container to fortify the wave amplitude it
to some extent. In this situation, the wave propagation may show different damping
features from that in infinite medium in the theoretical analysis.

24

You might also like