Вы находитесь на странице: 1из 8

ISSN 1063-7710, Acoustical Physics, 2009, Vol. 55, No. 1, pp. 100107. Pleiades Publishing, Ltd., 2009.

Original Russian Text V.E. Nazarov, A.B. Kolpakov, A.V. Radostin, 2009, published in Akusticheski Zhurnal, 2009, Vol. 55, No. 1, pp. 8290.

ACOUSTICS OF STRUCTURALLY INHOMOGENEOUS


SOLID MEDIA. GEOLOGICAL ACOUSTICS

Amplitude Dependent Internal Friction and Generation


of Harmonics in Granite Resonator
V. E. Nazarov, A. B. Kolpakov, and A. V. Radostin
Institute of Applied Physics, Russian Academy of Sciences, ul. Ulyanova 46, Nizhni Novgorod, 603950 Russia
Received January 22, 2008

AbstractThe results of experimental and theoretical studies of low-frequency effects of amplitude dependent
internal friction (nonlinear losses and resonance frequency shift) and generation of second and third harmonics
in an acoustic bar resonator made of Karelian granite are presented. The analytical description of the observed
effects is performed within the framework of phenomenological hysteretic equations of state containing elastic
and inelastic hysteresis. It was shown that elastic cubic hysteresis is manifested at small amplitudes of strains,
whereas superposition of elastic and inelastic quadratic hysteresises occurs at large amplitudes of strain. Relative weights of these hysteresises and effective parameters of acoustic nonlinearity of granite are determined.
PACS numbers: 43.25.Ba, 43.25.Dc
DOI: 10.1134/S1063771009010114

INTRODUCTION
At present, high (or strong) acoustic nonlinearity of
many micro-inhomogeneous (or mesoscopic) solid
bodies, in particular, crystalline mountainous rocks and
metals is a reliably established, non-disputable, and
commonly accepted fact [16]. However, experimental
and theoretical research of nonlinear wave processes in
such media will be always pressing and necessary (or at
least, the samples of not investigated natural and artificial materials would be available). This is related to the
facts that, first, acoustic nonlinearity of solid bodies
with micro-inhomogeneities exceeds strongly nonlinearity of a homogeneous material; second, acoustic
behavior of such media (whose number is practically
unlimited) is usually individually specific and does not
correspond to the canonical equations of the classical
theory of elasticity (with five or nine constants) [79];
and third, the concept about acoustic nonlinearity of a
specific solid body with micro-inhomogeneities can be
obtained only on the basis of the results of detailed
experimental investigations of nonlinear acoustic
effects (NAE) in this body. Analysis of the results of
such investigations allows us, in principle, to construct
a rheological model of the studied medium and reconstruct its equation of state as well as to determine the
effective nonlinear parameters of this equation, their
dependence on the frequencies of interacting waves,
etc., which is of course necessary for the development
of nonlinear methods of diagnosis. It is also worth noting that according to [10, 11], the investigation of
unusual nonlinear effects during the propagation of
waves in geophysical structures is related to one of the
five pressing problems of the modern nonlinear acoustics.

It is known [1216] that acoustic nonlinearity of


polycrystalline solid bodies is related to the presence of
different defects in their structure (vacancies, admixture and interstitial atoms, dislocations, grain boundaries, cracks, etc.) whose motion and interaction is the
cause of nonlinear amplitude-dependent internal friction (ADIF). The dislocation mechanism of nonlinearity in such crystals leads to the appearance of different
NAE, therefore investigation of these effects allows us
to understand deeper the physics of ADIF. In the relatively low frequency range (LF), acoustic nonlinearity
of such media usually manifests hysteresis properties,
therefore, one of the main problems in the analytical
description of nonlinear LF wave processes in such
media is the adequate choice of the hysteresis equation
of state for the investigated medium [1719]. Despite
the fact that application of hysteresis equations for the
description of nonlinear processes in the media with
irregular elasticity [20] does not cause any doubts
already, the justification, unambiguity, and adequacy of
such choice is still quite a complex and pressing problem. For example, the author of [18] suggested using
parameter r, which is equal to the ratio of amplitudedependent defect of the elasticity modulus to the damping decrement, as a characteristic indicator facilitating
distinguishing different (symmetric) types of acoustic
hysteresises. In different models of hysteresises (friction and break-off [17] or elastic and inelastic [19]), this
ratio does not depend on the amplitude of strain [12],
while in the symmetrical models, parameter r varies
within a few units, which could seemingly be an informative identifier of the hysteresis form. However,
experimental studies of ADIF effects in many polycrystalline solid bodies as well as investigation of fine

100

AMPLITUDE DEPENDENT INTERNAL FRICTION

effects of hysteresis nonlinearity (for example, generation of higher harmonics) evidence that hysteresis
equations of state for these materials are usually asymmetric, and the variation range of parameter r remains
indefinite, thus the criterion suggested in [18] is frequently not effective. The authors of [19] suggested
investigating amplitude dependence of other NAE, in
particular, generation of higher harmonics to determine
the type of hysteresis besides the traditionally studied
ADIF effects: defects of the modulus and damping decrement. The authors of recent paper [21] demonstrated
on the basis of analyzing the amplitude dependences of
NAE in limestone that elastic hysteresis gives a selfconsistent description of the experimental results,
while inelastic hysteresis is not applicable for their
description.
This work is dedicated to determination of the equation of state for a solid body with hysteresis nonlinearity. This problem is solved on the basis of the analysis
of the results of experimental and theoretical investigations of LF wave processes related to the manifestation
of amplitude dependent internal friction and generation
of higher harmonics in an acoustic bar resonator made
of Karelian granite Pitkyarant. The analytical description of observed nonlinear effects is made within the
framework of phenomenological equations of state
containing elastic and inelastic hysteresises [19]. Comparison of theoretical calculations and experimental
results demonstrates that the equation of state for granite is determined by a superposition of these hysteresises. Their relative weights and effective parameters of
granite nonlinearity were determined.
CONDITIONS OF THE EXPERIMENT
The experiment was carried out with a bar resonator
made of granite: the length of the bar was L = 35 cm,
the section of the bar was a square with a side of 1.6 cm.
The density of granite was approximately 2.75 g/cm3
(1%). The measurements were made at room temperature according to the standard scheme described in
detail, for example in [13, 21]. The investigated bar is
an acoustic resonator with almost rigid and soft boundaries. Its first natural frequencies are determined by
relation Fp (2p 1)C0/4L, where C0 is phase velocity
of LF longitudinal wave in the bar; p is the mode number. The errors of measurements of frequencies and
amplitudes of LF acoustic waves were 5 101 Hz and
5 102 dB, respectively. Resonance frequencies Fp
and q-factors Qp for the first three longitudinal modes
of the resonator at low amplitude of its excitation, when
nonlinear effects were not observed, were: F1 3820 Hz,
F2 10220 Hz, F3 17200 Hz, and Q1 255, Q2 238,
Q3 273. Resonance frequency of the first mode of
such resonator corresponds to phase velocity C0 5.3
105 cm/s.
ACOUSTICAL PHYSICS

Vol. 55

No. 1

2009

101

m
104

105

p=1
p=2
p=3

106

107

108
80

120

160
U

Fig. 1. Wave amplitude in resonator versus voltage at the


pumping emitter (in dB relative to 1 V). The straight line
corresponds to dependence m ~ U.

EFFECTS OF ADIF AND GENERATION OF HIGH


HARMONICS OF LOW FREQUENCY WAVE
Low frequency ADIF effects caused by hysteresis
nonlinearity of granite were studied in this experiment:
nonlinear losses and resonance frequency shift as well
as generation of the second and third harmonics of the
pumping wave [13, 21]. In order to do this, LF acoustic oscillations were generated in the resonator at frequency F close to frequency Fp of one of its three longitudinal modes, and amplitude dependencies of nonlinear resonance frequency shift Fnl = F Fp < 0 and
dumping coefficient nl as well as amplitudes of the
second and third harmonics were measured. Figure 1
presents the dependences of the bar strain amplitudes
m (during the resonance) on the amplitude of voltage U
at the pumping emitter when the resonator is excited at
three longitudinal waves. It is seen from the figure that
when the amplitude of resonator excitation increases,
dependences m = m(U) become nonlinear. This indicates the existence of amplitude dependent losses. Figures 2a and 2b present the graphs of dependences of the
relative absolute value of resonance frequency |Fnl /Fp |
and coefficient of nonlinear decay nl /p on m, where
p = 2Fp, p = (pQp)1, p = 1, 2, 3. The relative
coefficient of nonlinear decay was determined from
expression:
nl m1 U
------ = ------- ------ 1,
p
m U 1
which describes the deviation of the observed dependence m = m(U) shown in Fig. 1 from the linear one

102

NAZAROV et al.

20log(Fnl /Fp)
30

3
107

U2, cm
106
(a)

40

50

p=1
p=2
p=3

107

108

II
2

60

108

109

I
70
109
106

80
20log(nl /p)
10

Fig. 3. Amplitude of displacement U2 of free bar butt-end at


the frequency of the second harmonic (1) and amplitude of
strain 3 at the frequency of the third harmonic (2) versus
amplitude m. The straight lines correspond to dependences:

(b)
0

10

1010
104

105
m

U2, 3 ~ m .

p=1
p=2
p=3

II

106, where Fnl /Fp ~ m , nl /p ~ m , and second (II)


m > *, where Fnl /Fp ~ m, nl /p ~ m. It is clear that
in each range, the ratio of nonlinear defects of the elasticity modulus and damping decrement (i.e. parameter
F nl /F p
does not depend on amplitude m,
r1, 2 = -------------------- 2F p nl 1, 2
and r1 4.7, while r2 3.3. The existence of the same
amplitude dependences of the resonance frequency
shift and nonlinear losses (in each range) evidences the
fact of manifestation of hysteresis nonlinearity of the
investigated sample [1217]. In the first range, this nonlinearity should be cubic, and in the second range, it
should be quadratic.
After a sufficiently strong excitation of the resonator
(at m > 3 106, i.e. in the second range), generations
of the second and third harmonics of the pumping frequency were observed. Figure 3 shows a graph of the
dependence of the displacement amplitude of the free
butt-end of the bar U2(L) at the frequency of the second
harmonic and the amplitude of strain 3 at the frequency
of the third harmonics on the amplitude of deformation
m at the first mode (during the resonance).
It is seen from this figure that dependences of U2(L)
on 3 and m are close to quadratic function: U2(L), 3 ~
2

20
I
30

40
107

106

105

104

Fig. 2. Relative nonlinear shift of resonance frequency (a)


and attenuation coefficient (b) versus wave amplitude in
resonance. The straight lines correspond to dependences:
2

(I) Fnl /Fp ~ m , nl /p ~ m ; (II) Fnl /Fp ~ m, nl /p ~ m.

plotted at initial experimental dots (U1, m1) corresponding to small amplitudes of resonator excitation,
when the ADIF effects are negligibly small and m ~ U
[2, 3]. It follows from Fig. 2 that, first, relations Fnl /Fp
and nl /p as functions of m do not depend on the frequency of resonator excitation Fp (for the first three
modes), and second, two amplitude ranges can be distinguished in the dependences of these parameters on
the excitation amplitude: first (I) range m < * 2

m . This also evidences the manifestation (in the second range) of quadratic hysteresis nonlinearity of the
investigated sample of granite [1].
2

ACOUSTICAL PHYSICS

Vol. 55

No. 1

2009

AMPLITUDE DEPENDENT INTERNAL FRICTION

ANALYTICAL DESCRIPTION OF NAE


WITHIN THE ELASTIC HYSTERESISES
Let us try to give an analytical description and
explanation of the found amplitude dependences of LF
ADIF effects and generation of the second and third
harmonics first of all within the equation of state containing elastic hysteresis [19], which is qualitatively
similar to the dislocation Granato-Lucke hysteresis [12,
17, 18].
( , sgn , ) = E [ f ( , sgn ) ] + ,

103

II
m

I
*

(1)

f ( , sgn )
n
1 , > 0, > 0;

n1
n
(2)
1 ( 1 + 2 ) m 2 , > 0, < 0;
= ---
n
n , < 0, < 0;
3

( 1 ) n ( 3 + 4 ) nm 1 + 4 n , < 0, > 0,

where , , and are longitudinal stress, strain, and


strain rate of the bar, E is Youngs modulus, f = f (,
sgn ) is hysteresis function, | f(, sgn )|  1, 14 are
parameters of hysteresis nonlinearity, m = m(x) is local
strain amplitude, m < |th |, th is yield strength limit
(when it is exceeded irreversible plastic strains appear in
the solid body and it is broken; for many materials |th | >
104103), is viscosity coefficient ( = C 0 /pQp),
is density, n = 3 for the first range and n = 2 for the secn1
ond range, |14  m |  1, |14|  1. (Here, we use the
same indexation of parameters 14 at m < * and m >
*, but of course the values of these parameters are different in different ranges.) In the quasi-static condition
when | |  E | f (, sgn )| we get for such hysteresis: (, sgn , ) = 0 at = 0 (Fig. 4). Therefore, this
hysteresis is called elastic.
We note that hysteresis equations of state written in
the form (1), (2) are characteristic of many micro-inhomogeneous media with irregular elasticity in the range
of sufficiently low frequencies when relaxation (and
even more resonance) properties of the medium defects
are not pronounced and at the same time, nonlinear
effects occur without inertia.
These equations have a greater scope of application
that the commonly accepted equation of five-constant
theory of elasticity [8, 9] because the equation of fiveconstant theory can be obtained (as a particular case)
from equations (1) and (2) assuming that = 0 (i.e.,
considering the medium as an ideal one) and 1 = 2 =
3 = 4 = . In this case, hysteresis disappears and only
quadratic elastic nonlinearity remains. The opposite
statement is not correct, i.e. it is impossible to obtain
equations (1) and (2) from the equation of five-constant
theory of elasticity.
2

ACOUSTICAL PHYSICS

Vol. 55

No. 1

2009

Fig. 4. Quasi-static dependences = (, sgn , ) for elastic hysteresis (1), (2) at small (m < * (I) and large (m > *)
(II) strain amplitudes.

Equations (1) and (2) take into account the fact that
the medium is not ideal (i.e. linear dissipation) and its
hysteresis is nonlinear. In these equations we assume
only the restrictions on the degree of nonlinearity (| f(,
sgn )|  1) and strain amplitude (m < |th |), which
agrees with small nonlinear acoustic effects observed in
the experiment. These restrictions are used to get the
analytical expressions for nonlinear losses, shift of resonance frequencies and amplitudes of higher harmonics. From the mathematical point of view (i.e. during
the theoretical description of nonlinear effects) there
are no other restrictions for the application of these
equations, which follows from the literature [1, 5, 6,
1721]. During the description of nonlinear effects
observed for a specific solid body, we have to compare
each time the results of the analytical calculations
based on equations (1) and (2) with the results of the
specific experiment. If they agree, it is possible to use
these equations (applied to this specific solid body).
Finally, all parameters (constants in equations (1), (2)
or any their combinations) as well as nonlinear modules K, , and A, B, C in the five-constant theory of elasticity [79] are determined from the experiment.
Equations of state (1), (2) together with the equation
of motion Wtt = x(, sgn , ) and boundary conditions at the butt-ends of the resonator: W(x = 0, t) =
A0 cost, Wx(x = L, t) = 0 (where W is displacement, =
Wx, A0 and are amplitude and frequency of the pumping oscillator, respectively) describe nonlinear LF wave
processes in such resonator [13]. Their calculation
was performed on the basis of the perturbation method
because the following conditions were satisfied in the
experiment: |Fnl /Fp |  1, 2, 3  m (2, 3 are strain
amplitudes at frequencies 2 and 3).

104

NAZAROV et al.

The frequency response of the resonator is determined by relation [3]:


A0 p
1
-,
- --------------------------------------------------------------------------- m = -----------2
L [ ( + ) + ( + ) 2 4 /4 ] 1/2
nl

nl

(3)

2
b 1 = --------2 ( 1 + 2 + 3 + 4 ) > 0,
9
1
1
a 2 = ------ ( 1 2 3 + 4 ) + --------- ( 1 + 2 3 4 ),
24
64

where nonlinear shift of the resonance frequency nl =


2Fnl and losses nl in the first and second ranges
depend on parameters i of the hysteresis (2), = p,
||  p.
In the first range (m < *), expressions for Fnl /Fp
and nl /p are written as [3]:
F nl /F p =

2
a1 m ,

nl / p =

2
b1 Q p m ,

1 + 2 3 4
- > 0.
b1 = ------------------------------------16
Ratio r1 = a1/b1 for such hysteresis depends on the
relations between parameters 14. Without knowing
them (or their linear combinations 1 3 and 2 = 4) it
is impossible to determine the value of r1, but it is clear
that in the first range, r1 4.7. We find coefficients a1
and b1 from the comparison of experimental results
(Figs. 2a and 2b) and relations (4), and knowing them
we find parameters 1 3, 2 4: a1 = 5.6 108, b1 =
1.2 108, 1 3 = 1.1 1010, 2 4 = 4.9 109.
Using relation (4) one can determine effective parameter 3 of elastic cubic nonlinearity of granite (at small
amplitudes m < *). Assuming in equation (2) n = 3,
1 + 2 = 0, 3 + 4 = 0, and 1 = 3 = 3 we get equation
of state (1) with elastic cubic nonlinearity f(, sgn ) =
33/3, in which 3 = 32a1/3 = 6 109. Thus, the ADIF
effects in the first range can be described by elastic hysteresis (1), (2) without any contradictions.
In the second range (m > *), expressions for
Fnl /Fp and nl /p as well as for amplitudes U2(L), 3,
and phases 2, 3 of the fluctuations at the frequencies of
the second and third harmonics are written as [2, 3]:
nl / p = b 1 Q p m ,

U2( L) =

a 2 + b 2 m L,
2

2 2

(5)
(6)

4 a3 + b3 p m
3 = -----------------------------------------------------------------------------------------------,
2
2 1/2
15 [ ( p /Q 3 p 1 ) + 4 [ nl p /3 ] ]

(7)

2, 3 = arctan ( b 2, 3 /a 2, 3 ),

(8)

2
a 3 = --------- ( 1 2 + 3 4 ),
15
1
b 3 = --------- ( 1 + 2 + 3 + 4 ),
30

(4)

1
3
where a1 = ------ (1 + 2 3 4) + --- (1 2 3 + 4) ,
32
4

F nl /F p = a 1 m ,

1
b 2 = --------- ( 1 + 2 3 4 ),
48

where
1
4
a 1 = --------2 ( 1 2 + 3 4 ) + ------ ( 1 + 2 + 3 + 4 ),
6
9

p = 3p 3p 1 is dispersion difference of frequency 3p relative to the resonance frequency of


mode number 3p 1.
In the first range, hysteresis nonlinearity was
neglected for obtaining relations (5)(8) because the
amplitude and duration of the phase (i.e. the time when
point (, ) is located on the diagram = (, sgn , )
in the first range is much smaller than the corresponding values in the second range (the ratio of the maximum amplitudes for the second and first ranges is
approximately 10; the ratio of the duration of phases of
the second and first ranges is close to this value).
Here, in the general case, ratio r2 = a1/b1 also
depends on the relation between parameters 1 4. It is
also not defined while these coefficients are not determined (or linear combinations of parameters 1 + 3 and
2 + 4), but naturally in the second range r2 3.3. From
the comparison of experimental results (Figs. 3, 4) and
relations (5), (6), (7) we find coefficients a1, b1,
a 2 + b 2 , a 3 + b 3 , (at p = 1, 1 2 1250 s1) and
parameters 1 + 3 and 2 + 4: a1 = 9 102, b1 = 2.7
2

102, 1 + 3 = 8.9 103, 2 + 4 = 3.3 103,


2

a2 + b2 =

30, a 3 + b 3 = 60 corresponding to coefficients a1 and


b1. Effective parameter 2 of quadratic elastic nonlinearity of granite (at m > *) can be determined from
relation (6). Assuming in equation (2) n = 2, 1 + 2 = 0,
3 + 4 = 0 and 1 = 3 = 2 we can get equation of
state (1) with elastic quadratic nonlinearity f(, sgn ) =
22/2, in which parameter 2 = 16 a 2 + b 2 = 4.8 102.
As was expected, the value of quadratic parameter 2
for the investigated sample of granite appeared much
greater (by a factor of 102) than the similar parameter
for relatively rigid solid bodies (steel, glass) [8, 9].
2

The value of coefficient a 3 + b 3 , which is responsible for the amplitude of the third harmonic, can also
be calculated unambiguously from parameters determined above 1 + 3, 2 + 4: a 3 + b 3 = 2.7 102 (using
2

ACOUSTICAL PHYSICS

Vol. 55

No. 1

2009

AMPLITUDE DEPENDENT INTERNAL FRICTION

nonlinear losses and frequency shift). It is seen that


2

determination of coefficient

a 3 + b 3 by two indepen-

dent methods (based on the third harmonic

a3 + b3 =

60 and from nonlinear frequency shift a 3 + b 3 = 2.7


102) gives strongly different results (the difference is
2.7 102/60 4.5 times), which of course requires
explanation.
2

ANALYTICAL DESCRIPTION OF NAE


WITHIN NONELASTIC HYSTERESISES
We note the following important fact that nonlinear
ADIF effects can be described, generally speaking, not
only by elastic hysteresis (2) but also by inelastic hysteresis [17]. In this case, the amplitude dependences of
NAE will be the same as for the elastic hysteresis. For
example, in the first range, the ADIF effects could be
called symmetric cubic inelastic hysteresis (with one
parameter of nonlinearity  1), for which nonlinear
hysteresis function is written as:
f ( , sgn ) = ( 3 m + )
2

 m , > 0,
+ 3 m
2
2
+  m , < 0.
2

(9)

The qualitative shape of quasi-static (when | | 


E|f(, sgn )|) inelastic hysteresis (1), (9) is shown in
Fig. 5. It is seen from this figure that (, sgn , ) 0
at = 0. In this case, the relations for Fnl /Fp and nl /p
coincide with relations (4), in which: a1 = 45/64, b1 =
3/. However, we immediately get a contradiction
because ratio r1 = a1/b1 does not depend on parameter :
r1 = 15/64 0.74, while the measurements showed that
r1 = 4.7. We find coefficients a1 and b1 from experimental results (Fig. 2): a1 = 5.6 108, b1 = 1.2 108, and
then knowing these coefficients we find the only parameter .
Now, two different values of the same parameter
correspond to these two coefficients (a1 and b1): = 8
108 based on the resonance frequency shift and = 1.3
108 based on the losses. This is also a contradiction;
hence inelastic cubic hysteresis (9) is not applicable for
the description of ADIF effects (in the first range).
In the second range, similarly to the cubic hysteresis (9), NAE could be described by inelastic quadratic
hysteresis consisting of two branches (quadratic parabolas), for which the hysteresis function has the following form [20]:
f ( , sgn ) =  m
1 1 ( 1 + 2 ) m /2, > 0,
+ ---
2 2 + ( + ) 2 /2, < 0.
2
1
2 m
2

ACOUSTICAL PHYSICS

105

Vol. 55

No. 1

(10)

2009

m
m

Fig. 5. Quasi-static dependences = (, sgn , ) for


inelastic cubic hysteresis (1), (9).

This equation has three parameters of nonlinearity


 1, 1, 2, |1, 2 |  1 and the additive not related to
hysteresis m determines only the defect of elasticity
modulus (or resonance frequency shift), while the hysteresis parameters (i.e. all other additives) determine
nonlinear losses and generation of higher harmonics. It
is interesting to note that function (10) can be presented
as the sum of even and odd parts:
f ( , sgn ) =  m
1
2
2
2
+ --- [ ( 1 2 ) + ( 1 + 2 ) (  m ) sgn ].
4

(11)

It is clear that the even (quadratic) additive (1


2)2/4 is responsible for the generation of the second
harmonic only (in the first approximation), while the
2
odd additive (1 + 2)(2  m ) sgn /4 is responsible
for nonlinear losses (1 + 2 > 0) and generation of odd
harmonics. Here, similarly to the elastic hysteresis (2)
the resonance frequency shift and damping decrement
would also be proportional to m, and the amplitudes of
the higher harmonics would be proportional to m
(however, among the even harmonics only the second
would have such dependence). Thus, from the point of
view of nonlinear acoustics, the media described by
quadratic elastic (2) and inelastic (10) hysteresises are
similar, however, they differ principally by the fact that
during the generation of harmonic wave in the media
with inelastic hysteresis there will be no waves of even
harmonics (higher than the second) and they will
appear in the media with elastic hysteresis [19]. This
distinguishing indicator can be used, in principle, for
the determination of the hysteresis type in a microinhomogeneous medium.
For the resonator with hysteresis nonlinearity (10)
the values of Fnl /Fp, nl /p, U2(L), 3, and 2, 3 are
2

106

NAZAROV et al.

determined by relations (5)(8), in which a1 = c1 =


2/3, b1 = d1 = (8/92)(1 + 2), a2 = c2 = (1
2)/16, a3 = c3 = 4(1 + 2)/45, b2, 3 = d2, 3 = 0; hence
inelastic hysteresis (10) results in the same amplitude
dependences of ADIF effects and amplitudes of higher
harmonics as elastic hysteresis (2) at n = 2:
F nl /F p = c 1 m ,

nl / p = d 1 Q p m ,

U2( L) =

c2 + d 2 m L ,
2

2 2

(12)

(15)

DISCUSSION OF THE RESULTS AND CHOICE


OF HYSTERESIS
2

and c 3 + d 3 determined by different methods can be


explained if we suppose that granite contains two types
of defects, one of which (their relative concentration is )
is responsible to the action of elastic hysteresis (2)
while the other with relative concentration (1 ) is
responsible for the inelastic hysteresis (10) so that
finally NAE is determined by the superposition of these
hysteresises with weights and 1 , respectively:

The value of 1 + 2 = 3.1 103 found from nonlinear losses corresponds to the value of coefficient
2

c 3 + d 3 determining the amplitude of the third har-

F nl /F p = [ a 1 + ( 1 )c 1 ] m ,

(16)

nl / p = [ b 1 + ( 1 )d 1 ]Q p m ,
U2( L) = =

[ a 2 + ( 1 )c 2 ] + [ b 2 + ( 1 )d 2 ] m L,
2

2 2

(17)

4 [ a 3 + ( 1 ] )c 3 ] + [ b 3 + ( 1 )d 3 ] p m
-.
3 = ----------------------------------------------------------------------------------------------------------------------2
2 1/2
15 [ ( p /Q 3 p 1 ) + 4 [ nl p /3 ] ]
2

Generally speaking, the nature of such hysteresis


defects in granite is not known, however, we can note
that they can be, for example, edge or screw dislocations, their aggregations, boundaries of grains, microcracks, etc. In this case, as seen from relations (16)
(18), the amplitude of the third harmonics is determined by less dependent combination of coefficients
a1, b1 or c1, d1 (i.e. parameters of nonlinearity of elastic or inelastic hysteresises, due to the interference
and compensation of its components generated by
elastic and inelastic hysteresises (they are in different
phases) and owing to the free parameter: concentration . As a result, the above mentioned quantitative

The found difference between coefficients a 3 + b 3

the second harmonic: c 2 + d 2 = 30 or |1 2 | = 4.8


102, which naturally coincides exactly with the previously introduced effective parameter of quadratic nonlinearity 2.

a 3 + b 3 and c 3 + d 3 determined from nonlinear


shift of the resonance frequency (and losses) and amplitude of the third harmonic (or limit ourselves with the
second harmonic during measurements of NAE), the
ADIF effects in the studied sample of the Karelian
granite (at m > * 2 106) can be described either by
elastic (2) or inelastic (10) hysteresises.

(13)

Calculations show that if we determine coefficients


c1, d1 and , 1 + 2 using traditionally measured ADIF
effects: nonlinear shift of resonance frequency and
losses, we obtain that c1 = 9 102, d1 = 2.7 102, =
4.2 103, 1 + 2 = 3.1 103, and of course, r2 = c1/d2
3.3. We find from the measurement of the amplitude of
2

from its amplitude a 3 + b 3 = c 3 + d 3 = 60. This contradiction also requires explanation (as for the elastic
hysteresis). However, if we neglect the noted quantitative discrepancies between the values of coefficients

2, 3 = arctan ( d 2, 3 /c 2, 3 ).

86/60 1.4) exceeds the values of c 3 + d 3 determined

4 c3 + d 3 p m
- , (14)
3 = ----------------------------------------------------------------------------------------------2
2 1/2
15 [ ( p /Q 3 p 1 ) + 4 [ nl p /3 ] ]
2

c 3 + d 3 = 86, which notably (by a factor of

monic:

(18)

contradictions (in coefficients a 3 + b 3 and c 3 + d 3 )


are eliminated. For example, at relative concentration
0.38 and at the following values of the parameters
of elastic and inelastic hysteresises: 1 + 3 = 8.9 103,
2 + 4 = 3.3 103, a3 = 2.4 102, b3 = 1.3 102, =
4.2 103, 1 + 2 = 3.1 103, c3 = 2.8 102, d3 = 0
we get that the effective parameter of nonlinearity
[ a 3 + ( 1 )c 3 ] + [ b 3 + ( 1 )d 3 ] responsible
for the generation of the third harmonic calculated from
nonlinear shift of resonance frequency (and losses) is equal
to 60 (as it should be!) and all other effective coefficients
responsible for the resonance frequency shift losses, and
2

ACOUSTICAL PHYSICS

Vol. 55

No. 1

2009

AMPLITUDE DEPENDENT INTERNAL FRICTION

amplitude of the second harmonic remain unchanged:


a1 + (1 )c1 = 9 102, b1 + (1 )d1 = 2.7 102,
[ a 2 + ( 1 )c 2 ] + [ b 2 + ( 1 )d 2 ] = 30.
2

CONCLUSION
Experimental and theoretical investigations of low
frequency ADIF effects and generation of higher harmonics in a resonator of Karelian granite demonstrated
that:
(1) In the LF range (at least up to 17 kHz), acoustic
nonlinearity of granite is frequency independent hysteresis nonlinearity. At small amplitudes of strain (m < * =
2 106) elastic cubic hysteresis is manifested, while at
large amplitudes (* < m < 3 105) superposition of
quadratic elastic and inelastic hysteresises is observed.
(2) Effective parameters of cubic (at m < * 2
106) and quadratic (at * < m < 3 105) elastic nonlinearity of granite exceed the similar parameters of
homogeneous materials, for example glass and steel [8,
9] by several orders of magnitude (quadratic nonlinearity exceeds by a factor of 102). They are equal to 3 =
6 109 and 2 = 4.8 102.
(3) Strong hysteresis nonlinearity of mountainous
rocks similar to Karelian granite makes possible using
different nonlinear seismic acoustic effects for the diagnosis of mountainous rocks in the Earths interior and
controlling its stressed state.
In conclusion, we note that besides LF hysteresis,
many poly-crystalline mountainous rocks possess
strong high-frequency (HF) dissipative (or inelastic)
nonlinearity, which also can be used for their acoustic
diagnosis. The results of investigation of nonlinear LF
effects related to the manifestation of dissipative nonlinearity in the same resonator of Karelian granite will
be reported in the next article.
ACKNOWLEDGMENTS
This study was supported by the Russian Foundation for Basic Research (project no. N08-02-97039r_povolzhe_a).
REFERENCES
1. V. E. Nazarov, L. A. Ostrovsky, I. A. Soustova, and
A. M. Sutin, Phys. Earth Planet. Inter. 50 (1), 65 (1988).
2. V. E. Nazarov, Fiz. Met. Metalloved., No. 3, 172 (1991).

ACOUSTICAL PHYSICS

Vol. 55

No. 1

2009

107

3. S. V. Zimenkov and V. E. Nazarov, Fiz. Zemli, No. 1, 13


(1993).
4. R. A. Guyer and P. A. Johnson, Phys. Today, No. 4, 30
(1999).
5. L. A. Ostrovsky and P. A. Johnson, Riv. Nuovo Cim. 24,
146 (2001).
6. O. V. Rudenko, Usp. Fiz. Nauk 176, 77 (2006) [Phys.
Usp. 49, 69 (2006)].
7. L. D. Landau and E. M. Lifshitz, Course of Theoretical
Physics, Vol. 7: Theory of Elasticity (Nauka, Moscow,
1987; Pergamon, Oxford, 1986).
8. L. K. Zarembo and V. A. Krasilnikov, Usp. Fiz. Nauk
102, 549 (1970) [Sov. Phys. Usp. 13, 778 (1970)].
9. L. K. Zarembo and V. A. Krasilnikov, Introduction to
Nonlinear Physical Acoustics (Nauka, Moscow, 1966)
[in Russian].
10. O. V. Rudenko, Actual Problems in Nonlinear Acoustics, in Nonlinear Waves-2006, Ed. by A. V. Gaponov
and V. I. Nekorkin (IPF RAN, Nizhni Novgorod, 2007)
[in Russian].
11. O. V. Rudenko, in Acoustics of Heterogeneous Mediums,
Proceedings of the Seminar of Science School of
S. A. Rybak (2007), pp. 1428.
12. A. Granato and K. Lucke, J. Appl. Phys. 27 (5), 583
(1956).
13. D. Niblett and J. Wilkes, Usp. Fiz. Nauk 80, 125 (1963).
14. Ultrasounds Methods of Dislocation Study, Collected
Vol., Ed. by L. G. Merkulov (Inostr. Literatura, Moscow,
1963) [in Russian].
15. Physical Acoustics, Vol. 4: Applications to Quantum and
Solid State Physics, Ed. by U. Mason (Academic Press,
New York, 1975; Mir, Moscow, 1969).
16. R. Truell, C. Elbaum, and B. B. Chick, Ultrasonic Methods in Solid State Physics (Mir, Moscow, 1972; Academic Press, New York, 1969).
17. S. Asano, J. Phys. Soc. Jpn. 29 (4), 952 (1970).
18. A. B. Lebedev, Fiz. Tverd. Tela 41, 1214 (1999) [Phys.
Solid State 41, 1105 (1999)].
19. V. E. Nazarov, A. V. Radostin, L. A. Ostrovski, and
I. A. Soustova, Akust. Zh. 49, 405 (2003) [Acoust. Phys.
49, 344 (2003)].
20. G. S. Pisarenko, Vibrations of Mechanical Systems with
Allowance for the Imperfect Elasticity of the Material
(Naukova Dumka, Kiev, 1970) [in Russian].
21. V. E. Nazarov, A. B. Kolpakov, and A. V. Radostin,
Akust. Zh. 53, 254 (2007) [Acoust. Phys. 53, 217
(2007)].

Translated by E. Morozov

Вам также может понравиться