Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
3703.pdf
Скачиваний:
9
Добавлен:
15.11.2022
Размер:
11.37 Mб
Скачать

Fractional Calculus in Mechanics of Solids

171

 

 

The fractional derivative Kelvin-Voigt and standard linear solid models were first proposed by Shermergor [76] and Meshkov [38] in 1966 and 1967, respectively, and then independently but at a somewhat later time by Caputo [11] and Caputo and Mainardi [14], respectively. The early applications of these models were made by Caputo [11] in 1967 and Caputo and Mainardi [14, 15] in 1971 for solving the problems dealing with geophysics. In 1974, the fractional derivative Kelvin-Voigt model was utilized by Caputo [12] to study vibrations of an infinite viscoelastic layer.

In the late 1970s, an investigation on fractional derivative models and their application to the problems of structural dynamics was initiated in the USA by Torvik and Bagley. Their first paper in the field [4] dated to 1979 was devoted to modeling an elastomer damper as a fractional derivative Kelvin-Voigt oscillator. During the next decade, these authors contributed significantly to incorporate the fractional derivative standard linear solid model into the numerical procedures for investigating viscoelastically damped structures [5, 83, 84].

Thus, from the papers cited in the forth column of Table 1 it is clearly evident what a tremendous work was carried out in the late 60’s and 70’s by Russian and Italian researchers in the application of fractional calculus viscoelastic models (formulated earlier or in the same period using the two approaches as shown in the first three columns of Table 1) for solving dynamic problems in the mechanics of solids and geophysics.

2.Rabotnov’s Fractional Operators and Main Formulas of Algebra of Fractional Operators

We shall proceed our presentation in this Section from the analysis of the rheological model, which is called as the fractional derivative standard linear solid model

σ + τεγ Dγ σ = E0(ε + τσγ Dγ ε),

(5)

in so doing

 

τεγ τσ−γ = E0E−1,

(6)

where σ is the stress, is the strain, τ and τσ are the relaxation and retardation times, respectively, E0 and Eare the relaxed (prolonged modulus of elasticity, or the rubbery modulus) and nonrelaxed (instantaneous modulus of elasticity, or the glassy modulus) magnitudes of the elastic modulus, respectively.

Meshkov [38] was the first to derive formula (5) in 1967.

Expressing σ in terms of ε or ε in terms of σ from (5) with due account for (6) yields

 

 

 

 

 

 

ε 1 + τεγ Dγ

 

 

 

 

 

 

 

 

 

σ = E

ε ν

1

 

ε

,

 

 

 

(7)

 

 

 

 

 

 

 

 

 

 

 

ε = Jσ + νσ 1 + τσγ Dγ

σ

,

 

 

 

(8)

 

 

 

 

 

 

 

 

1

 

 

 

 

 

 

 

where E0

= J−1

, E= J−1 ,

 

 

 

 

 

 

 

 

 

 

 

0

 

 

 

 

 

 

 

 

 

 

 

 

νε

=

E− E0

=

J0 − J

,

νσ =

E− E0

=

J0 − J

.

(9)

 

 

 

E

J0

 

E0

 

 

J

 

Complimentary Contributor Copy

172

Yury A. Rossikhin and Marina V. Shitikova

 

 

Relationships (7) and (8) involve one and the same operator which will be denoted further as 3γ iγ ), i.e.,

3γ

iγ ) =

1

(i = σ, ε).

(10)

1 + τiγ Dγ

Operator (10) was introduced into consideration by Yu. N. Rabotnov in a little bit another form [46].

In order to obtain operator (10) in another form, we multiply the numerator and denom-

inator of the fraction in (10) by I

γ

τ

−γ

, where

 

 

 

 

 

 

i

 

 

 

 

 

 

 

 

Z0

 

 

 

 

 

 

 

 

 

 

 

t

(γ)

 

 

 

Iγ x(t) =

(t − s)γ−1

x(s)ds

(11)

 

is the fractional integral.

 

 

 

 

 

 

 

 

 

 

 

Considering that Dγ Iγ = 1, we find

 

 

 

 

 

 

 

 

 

 

 

 

 

I

γ

−γ

 

 

 

 

 

γ

 

 

 

τi

 

 

 

3γ

i

) =

 

.

(12)

1 − −Iγ τi−γ

If we suppose that the right part of formula (12) is the sum of an infinite decreasing geometrical progression, the denominator of which is equal to d = −Iγ τi−γ , then 3γ iγ ) could be represented as

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

γ

 

X

 

 

n

−γ(n+1)

 

γ(n+1)

 

 

 

3γ

i

) =

 

(−1)

 

τi

 

I

 

,

(13)

 

 

 

 

 

n=0

 

 

 

 

 

 

 

 

 

 

or

 

 

 

 

 

Z0

3γ (−s/τi ) x(t − s)ds,

(14)

3γ iγ ) x(t) =

 

 

 

 

 

 

 

t

 

 

 

 

 

 

 

 

 

 

where

 

 

 

 

 

 

γ−1

n

 

γn

 

 

 

 

 

 

 

 

 

 

3γ

(

t/τ

) =

 

t

 

X

(−1) (t/τi)

.

(15)

 

τ γ

 

 

 

 

 

i

 

 

 

 

[γ(n + 1)]

 

 

 

 

 

 

 

 

 

 

i

 

n=0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Considering (14), formulas (7) and (8) take the form

 

 

 

 

 

σ = Eε − νε Z0t

3γ (−s/τε)ε(t − s)ds ,

(16)

ε = Jσ + νσ Z0t

3γ (−s/τσ)σ(t − s)ds .

(17)

Formulas (16) and (17) are Boltzmann-Volterra relationships with weakly singular kernels of heredity 3γ (−t/τi ), which attenuate at t → ∞, in so doing resolvent kernels occur to be the same. Only exponential kernels possess this feature, and the kernels (15) go over into exponential kernels at γ = 1, i.e.,

3γ (−t/τi) = τi−1 exp (−t/τi) .

(18)

Complimentary Contributor Copy

Fractional Calculus in Mechanics of Solids

173

 

 

These properties of 3γ (−t/τi)-function allowed Rabotnov to call it as a fractional

exponential function.

 

 

 

 

 

 

 

 

 

 

 

Relationships (16) and (17) sometimes are written in the form

 

 

 

 

 

 

 

 

3γ

γ

 

 

σ = Eε,

E = E

 

 

1

 

ν

 

γ ) ,

(19)

e

e

 

 

 

σ

γ

σ

 

 

e

e

 

 

 

ε

 

ε

 

 

 

 

 

 

 

3

(20)

ε = Jσ,

J = J

 

 

 

1 + ν

 

) .

In conclusion of this Section we present the important formula for the multiplication of the operators

 

 

 

 

 

 

γ

 

γ

γ

 

 

 

γ

 

 

 

 

 

γ

 

γ

 

τε

3γ ε ) − τσ

3γ

σ )

 

 

 

3γ ε ) 3γ

σ ) =

 

 

τεγ − τσγ

 

 

 

 

.

 

 

 

(21)

In fact

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3γ εγ ) 3γ σγ ) =

 

 

 

 

1

 

 

 

=

A1

 

+

 

A2

,

(22)

(1 + τεγ Dγ ) (1 + τσγ Dγ )

1 + τεγ Dγ

 

1 + τσγ Dγ

where

 

τεγ

 

 

 

 

 

 

 

τσγ

 

 

 

τσγ

 

 

 

A1 =

 

 

, A2 = −A1

 

 

 

 

 

 

 

 

 

= −

 

.

 

 

 

τεγ − τσγ

τεγ

τεγ − τσγ

 

 

It should be noted that formula (21) does not coincide with that presented in Rabotnov [48] (see formula (5.10) there), since we use the dimensionless operator 3γ iγ ), while the dimensional operator was utilized in [48].

3.Generalized Rabotnov Operators

Generalized Rabotnov operators are written in the form

 

 

E = E

1

j=1

mj γ

tj

 

,

(23)

 

 

3

 

 

 

 

 

 

n

 

 

 

 

 

 

X

 

γ

 

 

J = J

"1 +

ni 3γ

iγ )# ,

 

(24)

 

 

n

 

 

 

 

 

 

 

X

 

 

 

 

 

i=1

where mj , tj (j = 1, 2, ..., n) and ni , τiγ (i = 1, 2, ..., n) are constants.

Since operators (19) and (20) are reciprocal, i.e., E J = J E = 1, then the following

relationship should be valid:

 

ni

γ i

)

1

 

mj

 

tj

.

(25)

1 = 1 +

 

γ

"

n

3

 

#

 

3

 

 

 

 

 

X

 

 

 

 

 

X

 

 

 

 

 

 

 

γ

 

 

 

 

 

γ

 

 

 

i=1

 

 

 

 

 

j=1

 

 

 

Assume that the constants ni and τiγ (i = 1, ..., n) are known, and it is necessary to determine the constants mj and tγj (j = 1, ..., n). Using formula (21), let us rewrite (25) in the form

ni

γ i )

mj

γ

tj

nimj

 

3

 

γ

γ

3

 

= 0,

n

3

 

n

3

 

n n

γ

 

γ

 

γ

 

γ

 

i=1

γ

j=1

i=1 j=1

τi

 

τi

− tj

γ

tj

 

X

 

X

 

γ

X X

 

γ i

)

tj

 

Complimentary Contributor Copy

174

Yury A. Rossikhin and Marina V. Shitikova

 

 

whence it follows that

Xn

ni 3γ iγ ) 1 −

i=1

From (26) we find

 

 

 

i

 

j=1

mj

3

γ

tj

1

j=1

γ

 

γ

τi

− tj

 

 

 

"

n

mj τ

γ

 

n

 

 

 

γ

 

 

X

 

 

 

 

X

 

 

 

 

 

 

 

 

 

 

 

 

n

mj τiγ

 

 

 

 

 

 

 

1 −

X

 

 

 

 

= 0,

 

 

 

 

 

τiγ

tjγ

 

 

 

 

 

 

 

j=1

 

 

 

 

 

 

 

 

 

 

 

 

 

n

nitjγ

 

 

 

 

 

 

 

1 −

X

 

 

 

 

= 0,

 

 

 

 

 

τiγ

tjγ

 

 

 

 

 

 

 

i=1

 

 

 

 

 

 

 

 

Xn nitγj τiγ − tγj

#

= 0. (26)

i=1

(27)

(28)

or in another form

n

 

 

mj tj−γ

 

 

 

 

 

 

X

 

 

 

 

 

 

 

= 0,

(29)

 

−γ

−γ

1 +

τ

 

 

j=1

i

 

 

 

t

 

 

 

 

 

 

 

 

 

 

j

 

 

 

 

 

 

n

 

 

niτi−γ

 

 

 

 

 

 

X

 

 

 

 

 

 

 

= 0.

(30)

 

−γ

−γ

1 +

τ

 

 

i=1

i

 

 

 

t

 

 

 

 

 

 

 

 

 

 

j

 

 

 

 

 

 

From the n-th order Eqs. (28) or (30) we could define n magnitudes of tj±γ

(j =

1, ..., n), while knowing tj±γ from the set of n Eqs.

 

(27) or (29) we find the values mj

(j = 1, ..., n).

 

 

 

 

 

 

 

 

 

 

 

 

Suppose now that constants mj and tjγ (j

= 1, ..., n) are known, and it is a need to

determine constants ni and τiγ (i = 1, ..., n). In this case, from the n-th order Eqs. (27) or

(29) we could define n magnitudes of τi±γ , while knowing τi±γ , we could find the values of

ni from the set of n Eqs. (28) or (30).

 

 

 

 

 

 

 

 

 

 

 

 

The set of equalities (30) implies that the values tj−γ are the roots of the equation

 

 

 

 

n

 

 

ni τi−γ

 

F1(x) = 1 +

X

 

 

 

 

 

 

τ −γ

= 0.

(31)

 

 

i=1

 

 

 

x

 

 

 

 

i

 

 

 

 

 

 

Equation (31) possesses n real positive roots if τi−γ > 0 and ni > 0.

If we suppose that τk+1γ > τk−γ , then zeros of the function F1(x) are located between

the poles, thus

 

 

τk−γ < tk−γ < τk+1γ ,

τn−γ < tn−γ .

(32)

The plot of the function F1(x) for n = 2 is presented in Fig. 1a, where x = t−γ .

The set of equalities (29) represents a system of n linear equations for n unknown coefficients mj . Assign that

n

mj tj−γ

 

 

X

 

 

,

(33)

−γ

F2(y) = 1 +

 

j=1

y t

 

 

 

j

 

 

where y = τ −γ .

Function F2(y) (33) possesses n zeros at y = τi−γ , while at y → tj γ it tends to +∞ or −∞ depending on the sign of mj and the direction of the transition to the limit. Since the

Complimentary Contributor Copy

Fractional Calculus in Mechanics of Solids

175

 

 

Figure 1. Functions F1(x) and F2(y).

points τi−γ are located between the points tj γ , except the point τ1−γ (see formula (32)), then the plot for the function F2(y) for n = 2 should be as that in Fig. 1b, but this is possible only when all mj are positive.

At n = 1, from (27) and (28), (19) and (20) we obtain

 

1 + νε

 

τσγ

 

 

 

 

 

 

 

 

= 0,

 

 

 

 

(34)

 

 

τεγ − τσγ

 

 

 

 

 

1 + νσ

 

τεγ

 

 

 

 

 

 

 

 

= 0,

 

 

 

 

(35)

whence it follows

 

τεγ − τσγ

 

 

 

 

τσ

γ

, 1 + νσ =

τε

.

(36)

1 − νε =

 

 

τε

 

 

 

 

 

 

 

τσ

 

 

 

Considering (9), formulas (36) go over in relationships (6). At n = 2, operators J and E take the form

and

J = J

1 + n1 3γ 1γ ) + n2 3γ 2γ ) ,

 

 

E = E1 − m1 3γ (t1γ ) − m2 3γ (t2γ ) .

 

Multiplying reciprocal

operators (37) and (38) yields

 

 

 

 

 

 

 

 

 

1 +

n1τ1−γ

n2τ2−γ

−γ

 

 

 

 

+

 

= 0,

x = t

,

 

 

 

 

 

 

τ1−γ − x τ2−γ − x

i

 

(37)

(38)

(39)

Complimentary Contributor Copy

176

Yury A. Rossikhin and Marina V. Shitikova

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(b1m1

+ b2m2

= 1,

 

(40)

 

 

 

a1m1

+ a2m2

= 1,

 

 

 

 

where it is evident from Fig. 1 that

 

 

 

 

 

 

 

 

 

 

a1

=

 

t1−γ

 

a2

=

 

 

 

t2−γ

 

 

> 0,

 

 

 

 

> 0,

−γ

−γ

−γ

−γ

 

 

t1

− τ1

 

 

 

 

 

t2

 

− τ1

 

 

 

t1−γ

 

b2

=

 

 

t2−γ

b1

=

 

 

< 0,

 

 

 

> 0.

−γ

−γ

−γ

 

−γ

 

 

t1

− τ2

 

 

 

 

t2

 

− τ2

Note that the signs of the values a1, a2, b1, and b2 could be established with help of

Fig. 1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Equation (39) could be written in the form

 

 

 

 

 

 

 

x2 hτ1−γ (1 + n1) + τ2−γ (1 + n2)i x + τ1−γ τ2−γ (1 + n1 + n2) = 0,

(41)

and its roots are defined by formula

 

 

 

 

 

 

 

 

 

x1,2 =

t−γ

1,2 = 2 hτ1−γ (1 + n1) + τ2−γ (1 + n2)i

 

 

 

 

 

 

 

1

 

 

 

 

 

 

 

 

 

±

 

2 r

 

τ1−γ (1 + n1) − τ2−γ (1 + n2) + 4τ1−γ

τ2−γ n1n2 ,

(42)

 

1

 

h

 

 

 

i

2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

and both roots are positive.

 

 

 

 

 

 

 

 

 

 

 

 

Substituting the found roots (42) in (40), we have

 

 

 

 

 

 

m1

=

 

b2 − a2

 

> 0,

m2 =

a1 − b1

 

> 0,

 

(43)

 

 

 

a1b2 − a2b1

 

a1b2 − a2b1

 

 

 

which are positive as well.

If, vice versa, it is necessary to determine the values of τi−γ and ni, then using Eqs. (29) and (30), we have

 

 

 

m1t−γ

m2t−γ

 

 

x = τi−γ ,

 

 

1 −

 

 

1

 

 

 

 

2

 

 

= 0,

(44)

 

 

−γ

 

 

 

−γ

 

 

 

 

 

t1

− x t2

− x

 

 

 

 

 

 

τ1γ

 

 

 

 

 

 

τ2γ

 

 

 

 

 

γ

1

γ n1

+

γ

2

 

γ

n2

= 1.

 

 

 

 

γ

−τ

γ n1

+ tγ

−τ

γ

n2

= 1,

 

 

t1 τ γ1

 

1

τ γ2

 

 

(45)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

t2 −τ1

 

t2 −τ2

 

 

 

It is convenient to

rewrite (44) in the form

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

x2 ht1−γ (1 − m1) + t2−γ (1 − m2)i x + t1−γ t2−γ (1 − m1 − m2) = 0.

(46)

The solution of (46) gives us the values of τ1−γ and τ2−γ , and then substituting τ1−γ and τ2−γ in (45), we find n1 and n2.

Let us present here two more useful formulas, which are the immediate generalization of formulas (36).

Complimentary Contributor Copy

Соседние файлы в предмете [НЕСОРТИРОВАННОЕ]