By
Manal M. Al-Ali
At
Faculty of Graduate Studies
Jordan University of Science and Technology
August, 2009
EVOLUTION OF THE CURRENT AS A FUNCTION OF TIME IN A
SIMPLE ELECTRIC CIRCUIT CONSISTS OF AN ELECTRIC
CURRENT SOURCE AND A RESISTOR
by
Manal M. Al-Ali
August, 2009
DEDICATIONS
To Mum, Dad, Sister Amani, Little Sweety Malak, and
My fiancé Fawaz Hrahsheh
i
ACKNOWLEDGMENTS
A big thanks to Dr . Akram Rousan for his helpful discussions and guidance.
I would also like to thank the member of my master degree committee Prof.
Nabil Ayoub for his help and encouragement.
I would like to heartily thank my friend Yara Al-Moghrabi for here extreme
motivation and support.
ii
TABLE OF CONTENTS
DEDICATIONS……………………………….. i
AKCNOLODGEMENTS…………………..….. ii
TABLE OF CONTENTS……………………… iii
LIST OF FIGURES……………….…………… iv
ABSTRACT…………………………………… v
CHAPTER:
i INTRODUCTION…………………………... 1
1.1 History………...……………………………………. 1
1.2 Previous Work……………………………………... 9
1.3 This Work………………………………………….. 14
ii THEORY……………………………………… 17
2.1 Riemann – Liouville Definition ………………….. 17
2.2 Mittag-Leffler Function…………………………... 18
2.3 Laplace Transform………………...……………… 20
iii ANALYTICAL SOLUTION………………… 22
3.1 Analytical Solution of Fractional RL Circuit …. 22
iv NUMERICAL SOLUTION…………………. 26
4.1 Numerical Solution of RL Circuit Fractional
Differential Equation…………………..…………….. 26
4.2 Results of Constant Voltage……………………… 27
4.3 Results of Alternating Current Source………….. 32
4.5 Conclusion……………………………………........ 34
REFERENCES…………………………………… 36
APPENDICES ……………………………………….. 41
A. Fractional RL Circuit ……………………………. 41
B. Fortran Program …………………………………... 43
Arabic Abstract ……………………………… 51
iii
LIST OF FIGURES
L = 10 −2 , m = 100 , and ε 0 = 3 .
4.2 (a) The initial current versus α . (b) The maximum current versus α . 31
4.3 3 (a, b, and c) Current versus time for large, intermediate, and 33
small values of α , respectively. R = 1 , h = 0.1 , L = 10 −5 , and
m = 100 .
4.4 The initial current versus α for the alternating current case. 34
iv
ABSTRAC
By
Manal M . Al-Ali
Chairman: Dr. Abdalla Obeidat
Co-Advisor: Dr . Maen Gharaibeh
The phase transition between the inductance and the resistance behavior of the
resistor was solved using the numerical scheme. Our results met our expectations of
developing the current gradually before it reaches the saturation value. The change in the
current signal with changing the value of α and its amplitude dependence on α
supported our conclusion about the current evolution.
v
CHAPTER I
INTRODUCTION
1.1. History
The original question that led to create a new mathematics named
dny
fractional calculus was "what if n is 1/2?" in the notation by
dx n
L'Hôpital in 1695 , "This is an apparent paradox for which ,one day ,useful
consequences will be drawn" Leibniz replied , together with " d1/2x will be
proportional to x " [1]. By these few words, a new branch in
mathematics, namely, fractional calculus was born.
dnp
In 1730 L. Euler [4] wrote "When n is a positive integer, the ratio ,
dx n
p a function of x , can be expressed algebraically. Now the question is what
1
Lagrange [5] has his place in the history of fractional calculus, he
contributed indirectly by his low of exponents (indices) for differential
operator of integer order. In 1772, he wrote
dn dm d n+m
n
. m y = n+m y (1.1)
dx dx dx
dn Γ( m + 1)
y= x m−n (1.2)
dx n
Γ (m − n + 1)
For special case y = x , he replaced n by the fraction 1/2, and used the
d 1/ 2 2 x
y= (1.3)
dx 1/ 2
π
2
1 ∞ ∞
f ( x) =
2π ∫
−∞
f (α )dα ∫ cos p ( x − α )dp
−∞ (1.4)
dn πn
Now cos p( x − α ) = p n
cos[p( x − α ) + ] , for n an integer.
dxn 2
Formally, replacing n with υ (υ arbitrary), he obtained the derivative version
dυ 1 ∞ ∞ υπ
dxυ
f ( x) =
2π ∫
−∞
f (α )dα ∫ pυ cos[p( x − α ) +
−∞ 2
]dp (1.5)
Fourier stated: "the number that appears in the above can be any quantity
whatever, positive or negative"
Hôpital, Leibniz, Euler, Lacroix, Fourier have talk about the idea of
derivative of arbitrary order, but the first application of fractional operator was
made by Niels Henric Abel in 1823 [8], he used the derivative of arbitrary
order to solve the tautochrone problem (i.e., the problem of determining the
shape of the curve such that the time of descent of a frictionless point mass
sliding down the curve under the action of gravity is independent of the
starting point). He worked with the integral
x
∫ (x − t)
−1 / 2
f (t ) dt (1.6)
0
d −1 / 2
Abel wrote the integral as π [ −1 / 2 ] f ( x ) , which is precisely of the form
dx
that Riemann used to define fractional operator.
3
Fourier's integral formula and Abel's solution had attracted Liouville's
attention, who perhaps has the first serious woke to give a logical definition of
a fractional operator, he published nine papers on the subject between 1832
and 1837.
dn
In 1832 Liouville [10] applied the differential operator D n
=
dx n
4
∞
I = ∫ u a −1e − xu du , (for Re a 〉 0 , x 〉 0 ) (1.11)
0
∞
∫ t α −1e−t dt = x− aΓ(a)
−a
I=x , ( Re a〉0 ) (1.12)
0
Or
1
x −a = I
Γ (a ) (1.13)
Liouville [12] was the first to suggest a solution for differential equation
involving fractional operator. In 1834, he wrote "The ordinary differential
dn
equation y = 0 has the complementary (general) solution
dx n
n −1 dυ
yc = c0 + c1 x + c2 x + ...... + cn−1 x
2
, thus υ y = 0 , (υ arbitrary) should
dx
have a corresponding complementary solution".
Any one versed in calculus never be surprised when he face operators like
d −1 d −2 d −n
y , y or y , it's nothing but indefinite integrals in disguise. But
dx −1 dx − 2 dx − n
what if n is fraction???
Riemann [13] maybe thought as the founder of what is called fractional
integral, in 1847, Riemann states the following definition:
d −r 1 x
∫
− r −1
−r
u ( x ) = ( x − k ) u ( k )dk (1.14)
dx Γ(r ) c
5
Nowadays, this formula is used as a definition of fractional integral with
zero lower limits.
d −( p+1)
fractional operator". This operator was .
dx−( p+1)
A serious problem was born in 1848, when W.Center [15] tried to find
0
the fractional derivative of constants. Using x to denote a constant, unity, he
0
observed that the fractional derivative of x according to Lacroix formula
(1.2) was not equal to zero. The Semiderivative of x0 using Lacroix is 1 .
πx
dυ x0
Center stated: "The whole question is plainly reduced to what is ,
dxυ
for when this is determined we shall determine at the same time which is the
correct system ".
6
In this controversy, Augustus De Morgan stood aside in his judgment, he
stated that: ”Both these systems may very possibly be parts of a more general
system” [2].
D 1 / 2 e mx = m 1 / 2 e mx , ( Re m 〉 0 ), and obtain
1
D 1 / 2 sin x = (cos x + sin x ) (1.15)
2
1
D 1 / 2 cos x = (cos x − sin x ) . (1.16)
2
7
Oldham and Spanier [22], after joining a collaboration that began in
1968, published the first book in fractional calculus in 1974. Since that time
and on; list of texts and proceedings have been devoted to fractional calculus
[1,2,23-34].
The first international conference on fractional calculus was organized by
Bertran Ross in 1974, at the University of New Haven, Connecticut, and was
sponsored by the National Science Foundation.
It had an exceptional turnout of 94 mathematicians, included a group of
distinguished luminaries like T. Osier, R. Askey, M. Mikolas, S.Samko and
many mentioned above.
The covered topics included papers on the fractional calculus and generalized
functions, inequalities obtained by use of the fractional calculus, and
applications of the fractional calculus to probability theory.
8
Recently, a considerable interest in fractional calculus has been motivated
by the application of this kind of calculus in numerical analysis, physics and
engineering, for more details, see [,2,22].
In spite of the problem of understanding the definitions of the fractional
calculus, and grasping the physical meaning of the non-integer order of
derivative/integral, but some relatively easy functions to grasp make some insight
into fractional calculus, such as, Gamma function, Beta function, Mettag-Leffler
function, and Laplace transform.
9
conditions to solve his fractional order equations. Also he improved a
fractional equation to describe the movement of a viscose fluid between two
moving surfaces.
After that, more general formulations of the law of deformation were
improved to study the viscoelastic behavior of real materials [37-40].
Koh and Kelly [41] used a numerical method in applying semi-derivative
in the formulation of stress – strain relationship of elastomers. One can say
that viscoelasticity is the field of the most extensive applications of fractional
calculus operators, see [42-44].
Liouville used his second approach in his investigations of potential
theory [11]. Many scientists have paid attention to this field in physics, since
not all forces in nature are conservative. And none of them can confirm that
these dissipative forces are an integer derivative of some physical quantity.
For this reason, many modeled have been proposed to understand the nature of
the potentials that reside behind such forces fractionally [45-47].
10
Understanding the simple harmonic oscillator is crucial in many branches
in physics, such as; classical mechanics (motion of pendulum), quantum
mechanics (second quantization), solid state (vibration of atoms and
molecules), Ferromagnetism (domain wall motion), etc,. For this reason,
solving the equation of motion that governs the simple harmonic oscillator is
demandable. This equation of motion has been investigated from many
researchers [50-58] to understand the myth behind such important equation.
The equation of motion that covers the simple harmonic oscillator is
usually given by the following equation
d 2 x (t ) dx
m + b + kx = 0
dt 2 dt (1.16)
The second term which represents the contribution of friction, or viscosity,
shows a linear dependence on the first derivative of the displacement, but it may
not be.
A. Rousan, etal [58] suggested a modified equation to cover the problem of
simple harmonic oscillator, using fractionalized second term as follows
d 2 x (t ) dx α
m + b α α + kx = 0
dt 2 dt (1.17)
Now the damping term is proportional to a fractional order derivative of the
displacement (fractional velocity).
The value of α varies from 0 to 1, and α = 0 corresponds to an undamped
simple harmonic oscillator, and α = 1 corresponds to a damped simple harmonic
oscillator.
They suggested that the system goes through an evolution process as the
fractional order goes from zero to one letting it pass through intermediate stages,
where the system can have a damping characteristic.
11
For the same problem, a numerical scheme was improved by Lixia Yuan and
P. Agrawal [59], using that scheme, the fractional differential equation was
transformed into a set of differential equations with no fractional derivative
terms, they have used Laguerre integral formula to convert this set to a set of first
order ordinary differential equations, and integrated them using a appropriate
numerical scheme.
They have found that the solution converges as the numbers of Laguerre node
points are increased, they obtained a numerical solution agrees with the solution
obtained using analytical scheme.
d 1+α Q
1+α
+ ω (α ) 1 + α Q = 0 (1.18)
dt
where 0 ≤ α ≤ 1 , and ω(α) = ω1αω21-α .
one can immediately see that when α = 0, equation (1.18) converts to the
equation of RC circuit
dQ ( t )
+ ω 2 Q (t ) = 0 where ω2 = 1/RC (1.19)
dt
d 2 Q (t )
2
+ ω 12 Q ( t ) = 0 where ω1 = 1/LC (1.20)
dt
Between these canonical cases (i.e. 0 < α < 1), they got a general solution,
which shows how the oscillatory behavior of the LC – circuit goes over to a
decay behavior of the RC circuit as the memory parameter (α) decreases from 1
12
to 0. In other words, it shows how the inductor in LC – circuit looses some of its
character as an inductor, and accepts a resistive behavior.
The concept of the intermediate stages ( 0 < α <1) has been introduced by many
authors [give some].
A fractional differential equation was suggested by S.Westerlund [ref]
in1991, to describe propagation of plan electromagnetic wave in an isotropic
and homogeneous, lossy dielectric. This equation was:
∂2E (υ ) ∂2E
μ 0ε 0 2 + μ 0ε 0 χ 0 E + 2 = 0 , (1 〈 υ 〈 2 ) (1.21)
∂t ∂t
We end this section by naming the most applications that the fractional
calculus intrigues; diffusion processes [22,64-66], anomalous diffusion [67-73],
viscoelastic materials [74-76 and more], signal processing [77], vescoplastic
materials [78], relaxation in filled polymer networks [79,80], heat propagation
[81], splines and wavelets [82,83], control theory [84]. For more information the
reader can refer to Mainadri [85,86] and Podlubny [ref].
13
1.3. This Work
14
d 1−α dQ * d
α −1
dQ
*
L 1−α + R α −1 = ε (t )
dt dt dt dt
(1.23)
Where L* = (1 − α ) L1−α R α and R* = αR2−α Lα −1 , L* and R* are chosen in this way
to take care of dimensions for all values of α (the memory parameter ) between
zero and one . One can choose different expressions for L* and R*, and the
proposed expressions are just an example.
One can see from the definition of L* and R* that when α equals zero L* is
maximum, and R* is minimum and equals zero, that is if we have a circuit
consists only of an inductor and a power supply. For small values of α the
inductor is the dominant, conversely, when α closer to one the resistor is the
dominant, until it becomes one the circuit consists of a resistor and a power
supply only.
Some helpful relations and definitions are mentioned in chapter two, these
relations will help when we try to give an analytical solution of our equation
using Mettag-Liffler function for two cases, constant voltage and alternating
current source as we will do in chapter three.
The main part of this work will be shown in chapter four. The computational
results of the current evolution with time will be explained for the same two cases
at several values of α . Also the effect of the power source frequency on the
current amplitude will be studied for small and large values of α .
15
HAPTER II
THEORY
−υ υ d n n−υ
I
a x f ( x)= D
a x f ( x) = ( ) a Ix f (x) Reυ > 0, n −1 ≤ υ ≤ n (2.3)
dx
16
x
d 1
∫
−υ υ
I f ( x)= a Dx f ( x) = ( ) m ( x − u ) n−υ −1 f (u )du (2.4)
dx Γ(n − υ ) a
a x
One can unify the definitions of integrals and derivatives of arbitrary order υ,
(Re υ ≠ 0) by:
⎧ d m n −υ
⎪ ( dx ) a I x f ( x )....... Re υ 〉 0 .... n − 1 ≤ Re υ ≤ n
υ ⎪ (2.5)
a D x f ( x) = ⎨ x
⎪ 1
∫
⎪⎩ Γ ( −υ ) a
( x − u ) −υ −1 f ( u ) du ..... Re υ 〈 0
17
Example (1): Take α =1, β = 1, then we get
∞
zk
E1,1 ( z) = ∑ = ez
k =0 Γ(k + 1)
(2.8)
∞
zk ∞
zk 1 ∞ z k +1 e z −1
E1, 2 ( z ) = ∑ =∑ = ∑ = (2.9)
k = 0 Γ ( k + 2) k = 0 ( k + 1)! z k =0 ( k + 1)! z
∞
zk ∞
zk 1 ∞
z k +2 ez −1− z
E1,3 ( z ) = ∑ =∑ = 2 ∑ = (2.9)
k = 0 Γ ( k + 3) k =0 ( k + 2 )! z k =0 ( k + 2 )! z2
In general
1 ⎧ z m−2
zk ⎫
E 1,m ( z ) =
z m −1
⎨e − ∑ ⎬
Γ ( k + 1) ⎭
(2.10)
⎩ k =0
Sine, cosine, hyperbolic sine and cosine functions are also particular cases of
Mittag-Leffler function, see the following examples
Example (4): Take α =2, β = 1, and z = −z 2 then we get
(−1)k z 2k
∞ ∞
(−1)k z 2k
E2,1 (−z ) = ∑
2
=∑ = cos(z)
k =0 Γ(2k + 1 ) k =0 (2k )! (2.11)
Example (5): Take α =2, β = 2, and z = −z 2 then we get
∞
( − 1k ) z 2 k 1 ∞ ( − 1) k z 2 k +1 sin( z )
E 2,2 (− z ) = ∑
2
= ∑ = (2.12)
k =0 Γ ( 2 k + 2 ) z k = 0 ( 2 k + 1)! z
The fractional sine and cosine can also be written in terms of this function as
follows
18
(−1)k z2k+1
∞
sinλ,μ (z) = ∑ = zE2μ,2μ−λ+1(−z2 ) (2.12)
k=0 Γ(2μk + 2μ − λ +1)
(−1) k z 2k
∞
cosλ ,μ ( z) = ∑ = E2 μ , 2 μ −λ +1 (− z 2 ) (2.13)
k =0 Γ(2μk + μ − λ + 1)
(2.14)
Laplace transform of Mittag-Leffler function will help to write the solutions
of fractional differential equations in terms of this function. The transform is
given by the following equation:
∞ k! p α − β
∫
− pt αk + β −1 α
e (t (k )
E
α ,β (± at ))dt = α , (Re(p) > a׀1/α) (2.15)
0 ( p ∓ a ) k +1
dk
where E
(k )
ε ,β ( z ) = k Eα ,β ( z ) .
dz
A particular case of (2.15),which is useful for solving the semidiffrential
1
equation (i.e α = β = ) is .
2
k −1
∞ k!
∫
− pt
e (t 2
E1(/k2),1/ 2 (±a t ))dt = k +1 , (Re (p) > ׀a׀2) (2.16)
0
( p ∓ a)
19
CHAPTER III
ANALYTICAL SOLUTION
3.1 Analytical Solution of Fractional RL Circuit
d 2−α Q α
* d Q
L *
+R = ε (t ) (3.2)
dt 2−α dt α
Where L* = (1 − α ) L1−α R α and R* = αR2−α Lα −1
d 2−α Q α
* d Q
L *
+R = ε0 (3.3)
dt 2−α dt α
Apply Laplace transform to eq. (3.3), and use eq. (2.14)
ε0
L*{s 2−α Q(s) − A − sB} +R*{sα Q(s) − c} = (3.4)
s
d 1−α Q(t ) d −α Q(t ) d α −1Q(t )
Where A = , B = and C= .
dt 1−α t =0
dt −α t =0
dt α −1 t =0
Rearranging eq.(3.4)
ε01 A+ω2C sB
Q(s) = * 2−α 2 α + 2−α 2 α + 2−α 2 α (3.5)
L s(s +ω s ) (s +ω s ) (s +ω s )
R*
where ω 2
=
L*
20
Using Taylor Expansion, Eq.(3.5) can be written as
s −1−α ε0
Q(s) = 2−2α { + ( A + ω 2C)S + BS 2 }
(s +ω ) L
2 *
(3.6)
Comparing the last equation with the Laplace transform of Mittag-Leffler
function eq. (2.15), we get
ε0
Q(t ) = t −α { *
t 2 Eα ′,β1 ( z ) + ( A + ω 2 C )tEα ′,β 2 ( z ) + BEα ′,β3 ( z )} (3.7)
L
Where β1 = 3 − α , β 2 = 2 − α , β 3 = 1 − α , α ′ = 2 − 2α , and z = − ω 2 t 2 − 2 α .
γ αk + β −1 k
Appling the definition 0 Dt (t Eα ,β (λt α )) = t αk + β −γ −1 Eαk , β −γ (λt α ) to eq.(3.7)
ε0
I (t ) = t −α { *
tEα ′,β1′ ( z ) + ( A + ω 2C ) Eα ′,β2′ ( z ) + Bt −1Eα ′,β3′ ( z )} (3.8)
L
where β1′ = β1 − 1 , β 2′ = β 2 − 1 , β 3′ = β 3 − 1
One can see that β 3′ is negative for all n and α, and for Mittag-Leffler
function the parameters must be larger than zero, so we can choose B to be zero.
Thus, Eq.(3.8) becomes
ε0
I (t ) = *
t1−α E2−2α ,2−α (Z ) + ( A + ω 2C)t −α E2−2α ,1−α (Z ) (3.9)
L
d
When α=0, eq (3.3) becomes L I (t ) = ε 0 , and the solution of it is
dt
ε0
I (t ) = I 0 + t (3.10)
L
Going back to eq(3.9),using the same special case, the current is given by
ε0
I (t ) = t1E2,2 (−ω 2t 2 ) + AE2,1 (−ω 2t 2 ) (3.11)
L
Or
21
ε0
I (t ) = sin(ωt ) + Acos(ωt ) (3.12)
Lω
Comparing eq(3.10) to eq(3.12), one can conclude that (ωt ) << 1 , sin(ωt ) → ωt ,
and cos(ωt ) → 1 . Then
ε0
I (t ) = t+A (3.13)
L
Where, A α =0
= I0 .
Case 2 : ε (t ) = ε 0 cos ω 0t
d 2−α Q * d
α
Q
L* 2−α
+ R = ε 0 cos ω 0 t (3.14)
dt dt α
Apply Laplace transform to eq. (3.14), and use eq. (2.14)
s
L*{s2−α Q(s) − A − sB}+R*{sα Q(s) − c} = ε 0 (3.15)
(s2 + ω02 )
Rearranging eq.(3.15)
ε0 s A + ω 2C sB
Q(s) = + + 2−α (3.16)
L (s + ω0 )(s + ω s ) (s + ω s ) (s + ω2 sα )
* 2 2 2−α 2 α 2−α 2 α
ε0 ω 02 k ω 2 ( k + q ) ( k + q )! sβ′
Q (t ) =
L*
∑ (−1) k +q
k ,q k!q! ( s 2 + ω 02 ) k + q +1
(3.17)
s −1−α
+ 2 − 2α {( A + ω 2 C ) S + BS 2 }
(s +ω )2
22
ε0 ω 02 k ω 2 ( k + q )
Q (t ) =
L * ∑ (−1)
k ,q
k +q
k! q!
t 2 ( k + q )+ β −1 E 2( k,β+ q ) ( −ω 02 t 2 )
(3.18)
+ t −α {( A + ω 2 C )tEα ′,β 2 ( z ) + BEα ′,β 3 ( z )}
ε0 ω02kω 2( k +q )
I (t ) =
L*∑ (−1)
k ,q
k +q
k!q!
t 2( k +q )+β −2 E2( ,nβ+−k1+m+q ) (−ω02t 2 )
ε0
I (t ) = I 0 + sin ω0t (3.20)
Lω 0
And in eq.(3.19), put α=0 and take the first term only
ε0
I (t ) = tE 2, 2 (−ω 02 t 2 ) + AE 2,1 (−ω 2 t 2 ) (3.21)
L
ε0
Or I (t ) = sin ω 0 t + A cos ωt (3.22)
Lω 0
(3.24)
23
CHAPTER IV
NUMERICAL SOLUTION
24
In this chapter we will use the weighted average method that described in
[ref], the algorithm was checked in studying the motion of domain walls with
series method [94].
The numerical Expression of fractional derivative is defined as
m
α
Dt I = h −α
∑ωα I
j =0
j m− j (4.1)
d −1 I
Using the definition of Q = , Eq(3.3) can be written as
dt −1
d 1−α I * d
α −1
I
L* 1−α
+ R α −1
+ Rex I = ε (t ) (4.2)
dt dt
Where I is the current, L is the inductivity, R is the resistance, and ε (t ) is the
power source, t is the time.
Applying the expression in eq(4.1) to eq(4.2) gives the following numerically
solvable equation
m m
L* h −(1−α ) ∑ ω 1j−α I m − j + R * h −(α −1) ∑ ω αj −1 I m − j + Rex I m = ε (t ) (4.3)
j =0 j =0
Where t = jh , m is the index of the current at the last limit of the time interval,
t
and h = is the time increment.
m
4.2. Results of Constant Voltage
In this section, the fractional differential equation for RL circuit will be solved
in case of constant voltage. Our concern in this part is to study the behavior of
the current with changing the value of α (The geometry of the inductor), so we
25
will scale the current values for all α s to compare between them in one graph.
Scaling process was simply considering the primitive current equals the same
constant value for all values of α .
From equ (4.3), when α equals one, the solution of the current is given as
ε0
I (t ) = (4.5)
( R + Rex )
the current as α goes to one, the linearity of the lines increases. That means the
expression of the current getting more closer to equ (4.5). This behavior of the
current with changing α means that the inductor is converting step by step to a
resistor.
26
Figure 4.1. Current vs time for several values of α at R = 1 , Rex = 2 ,
h = 0.001 , L = 10 −5 , m = 100 , and ε 0 = 3 .
d 1−α I * d
α −1
I
L *
1−α
+R α −1
= 3 cos(ω 0 t ) (4.7)
dt dt
When α =0 , the solution of equ (4.7) is
3
I (t ) = I 0 − sin(ω 0 t ) (sine signal) (4.8)
Lω 0
One can notes immediately that the figures (4.2, 4.3, and 4.4) show how the
sine signal (solution of equ (4.7) at α equals zero) converts to cosine signal
(solution of equ (4.7) at α equals one) passing through intermediate cases
( 0 < α < 1). This result supports our expectation about converting the temporary
induction property of the resistor into stable resistance.
27
Figure(4.2) shows the development of the current at small values of α . The
amplitude of the current is decreasing to smaller values as α increases.
28
Figure 4.3. Current vs Time at intermediate values of α .
Figure(4.4) shows the behavior of the current at high values of α . There is a
clear increasing in the amplitude of the current with increasing α . The signal of
the current at high values of α is similar to the signal of the power source which
means that the solution of the current is function of cosine as in equ (4.9).
29
Figure 4.4. Current vs Time at large values of α .
The figures 4.2 and 4.4 draw our attention to a critical point of α at which the
current amplitude reverses it behavior from decreasing to increasing when α
departs from zero to one.
Figure(4.5) shows the approximate value of that critical point at α ≈ 0.0755 .
4.4. The Effect of the Frequency on the Amplitude of the Current in the case
of alternative source
The formula of the current at α very close to zero, equ (4.8) shows that the
negative part of the current decreases with increasing the frequency which means
that the current amplitude increases with increasing the frequency whereas the
current formula at α very close to one, equ (4.9), shows that there is no
dependence between the current amplitude and electric source frequency.
Figure (4.6) shows how the amplitude of the current grows up with increasing
the frequency at very small α .
30
Figure 4.6. The effect of the frequency on the current amplitude.
The same behavior appears in figure (4.7) for α = 0.5 but with smaller change
in the amplitude. This decreasing in the frequency effect on the current amplitude
happens because of increasing the resistant behavior of the resistor as α going
closer to one.
31
Figure (4.8) shows that there is no change on the current amplitude for several
values of the frequency when α = 0.9999 , because the amplitude does not depend
on the frequency of the source at large α , see equ (4.9) .
Figure 4.8. The behavior of the current for several values of ω at high α .
All the values of the frequencies in figure (4.8) were taken for integers and
half integers of ω 0 . The constructive interference happens at time equals 8
because cos(ωt ) in equ (4.9) always equals positive one. Figure (4.9) shows that at
the same value of α but for quarters of ω 0 destructive interference at the same
time occurs because cos(ωt ) in equ (4.9) always equals negative one.
32
Figure 4.9. Destructive interference at quarters of ω 0 .
4.5 Conclusion
In view of the results above for a circuit contains a resistor, an inductor, and a
constant voltage resource, we notice that the indictor, as α departs from 0, starts
to develop a resistive behavior a pseudo-RL circuit starts to develop until α
becomes one the circuit will be a very simple circuit includes a resistor and
power supplier. This converting of the inductor can be explained as stretching the
inductor to form a simple wire with normal resistant.
The results of the alternative current source and zero external resistance
circuit show that the resistance starts with inductor behavior before ending with
its regular property as a stable resistor.
The results of the alternative current source showed that there is a critical
point of α to get the minimum current amplitude.
33
The frequency of the power supply contributes in the amplitude of the current
when α very small and it starts loosing this action when α become close to one.
Reference:
1. R.Hilfer, application of fractional calculus in physics, WORLD
SCIENTIFIC, 2000.
2. B.Roos, an introduction to the fractional calculus and fractional
differential equations, JOHN WILEY AND SONS, INC, 1993.
3. K.Oldham, and J.Spanier, The factional calculus, ADADEMIC PRESS
, New York, 1974.
34
4. Baleanu, Dumitru (2009) Fractional Electromagnetic Equations
Using Fractional Forms. International Journal of Theoretical
Physics
5. Baleanu, Dumitru (2009) Fractional Newtonian mechanics.
Central European Journal of Physics
35
17. R. L. Bagley and P.J. Trovic, Fractional Calculus – A different
approach to the analysis viscoelastically damped structure. AIAA
Journal 21, No. 5 (1983), 741-748.
18. R. L. Bagley and P.J. Trovic, Fractional Calculus in the transient
analysis of viscoelastically damped structure. AIAA Journal 23, No. 6
(1983), 918-925.
19. N. Engheta, On fractional Calculus and Fractional Multipoles in
Electromagnetism. IEEE Trans. Antennas Propagate. 44, No, 4 (1996)
554-566.
20. N. Engheta, Fractional Paradigm in Electromagnetic Theory. In:
Frontiers in Electromagnetic (Eds. D. H. Werner and R. Mittra), IEEE
Press (2000), Ch. 12, 523-552.
21. N. Engheta, on fractional paradigm and intermediatezones in
electromagnetism ,II: Cylindrical and spherical observation. Microwave
Opt .tech .lett. 23, No.2 (1999), 100-103.
22. Akram A. Rousan, E. Mlkawi, E. Rabei and H. Widyan, Application of
fractional calculus to gravity. Fractional Calculus and Applied Analysis
5, No.2 (2000), 155-168.
23. F, Mainardi, Fractional relaxation-oscillation and fractional diffusion-
wave phenomena. Chaos, Solitons and Fractal 7,No.9 (1996), 1461-
1477.
36
Appendix A
d 1−α I * dI
α −1
L 1−α + R
*
Rex I = ε (t ) (A.2)
dt dt α −1
As it is known the current is defined as the first derivative of the charge with
dQ
respect to time ( I =
dt )
Thus eq. (A.2) can be written as:
d 1−α dQ * d
α −1
dQ dQ
*
L 1−α + R α −1 + Rex = ε (t ) (A.3)
dt dt dt dt dt
Eq. (A.3) is more suitable, that is to avoid the negative derivative that appears
in the second term in equation (A.2).
Q (t ) = Q1 (t ) + Q 2 (t ) + Q 3 (t ) (A.5)
where
ε0 1
Q1(s) = (A.6.1)
L* s(s2−α + ω2sα + ωex2 s)
37
sB
Q2 ( s ) = (A.6.2)
( s 2 −α + ω 2 s α + ωex2 s)
And
A + ω2C
Q3 (s) = 2−α (A.6.3)
(s + ω2 sα + ωex2 s)
Now, for Q1
ε0 1 ε0 s −2
Q1 ( s ) = = (A.7)
L* s ( s 2 − α + ω 2 s α + ω ex2 s ) L* ( s1− α + ω 2 s α −1 + ω ex2 )
ε0 s −2
Q1 ( s ) =
L* 1−α 2 α −1 ω ex2 (A.8)
{s +ω s }{1 + }
( s 1−α + ω 2 s α −1 )
∞
1
Using
(1 + x )
= ∑ ( −1)
k =0
k
xn
ε0 s n−1−αn−α
∞ n
Q1 (s) = * ∑ (−1) ω 2n
(A.9)
(s 2−2α + ω 2 )n+1
ex
L n=0
Comparing the last equation with eq (2.15) , one can note that k= 0 , α = 2-
2α, a = ω2 and α-β = n-1-αn-α , thus Q(t) can be written in terms of Mittag-
Leffler Function as follows
s n−αn−α
∞ n
38
(−1)n ωex2n n+1−αn−α (n)
∞
Q2 (t) = ( A + ω C)∑2
t E2−2α ,αn+2−n−α (−ω2t 2−2α ) (A.12)
n=0 n!
And
s n+1−αn−α
∞ n
Q3 (s) = B∑ (−1) ω 2n
(A.13)
(s 2−2α + ω 2 )n+1
ex
n=0
Now Q (t ) = Q1 (t ) + Q 2 (t ) + Q 3 (t ) (A.15)
∞
(−1)n ωex2n n−αn−α ε 0 2 (n)
Q(t) = ∑ t { * t Eα′,β1 (z) + ( A + ω2C)tEα(n′,)β2 (z) + BEα(n′,)β3 (Z )}
n=0 n! L
(A.16)
Where β1 = αn + 3 − n − α , β 2 = αn + 2 − n − α , β 3 = αn + 1 − n − α , α ′ = 2 − 2α , and
Z = − ω 2 t 2 − 2α .
39
APPENDIX B
FORTRAN PROGRAMS
B.1.Numerical solution
program current
implicit none
real(8)::alpha,sum
integer::j,k,ii,jj,kk,gstat
character(70)::filename
real(8)::Ls,Rs,R,L,eps,I0,Re,dom,h,del,I,beta
real(8)::s1,s2,s,t,omega
OPEN(12,File="const_current.txt",STATUS="REPLACE",ACTION="WRITE")!
k=0
jj=9
open(12,file="cosine.txt",status="replace",action="write",position="rewind",
iostat=gstat)
do kk=1,1!
ii=int(kk/1)
jj=(kk-ii*10)
alpha=dfloat(kk)/100.0d0
'filename='Current'//achar(ii+48)//'_'//achar(jj+48)//'.txt
!
open(12,file=filename,status="replace",action="write",position="rewind",iost
at=gstat)
'if(gstat==0) print*,trim(filename),' opened!
if(jj==0) jj=9
jj=jj-1
alpha=0.3d0
del=1.0d0-alpha
h=1.0d-1
Re=0.0d0
L=1.0d-5
R=1.0d0
omega=2.0d0*22.0d0/7.0d0*R/L*4.0d0/4.0d0
Ls=(-del)*L**(del)*R**alpha/h**(del)
Rs=alpha*R**(1.0d0+del)*L**(-del)/h**(-del)
beta=Ls+Rs
dom=(beta+Re)
do j=1,100,1
t=dfloat(j)*h
eps=3.0d0*dcos(omega*t)
I0=eps/dom
I0=0!
print*,i0,3.0d0/L/omega!
pause!
40
if(j==1) I=I0
s1=Ls*weight(del,j)
s2=Rs*weight(-del,j)
s=s1+s2
sum=sum+s*I
I=(eps-sum)/dom!
I=eps/dom-sum/dom
print*,eps,(i),alpha!
write(12,*) j*h,i,alpha
write(*,*) j*h,i,alpha
write(*,*) alpha,sum!
if(i<=1.0d-3) exit!
end do
k=k+1
if(k==3) k=0
(close(12
==============================!
contains
real(8), dimension(5)::w,x,y,v
integer::i,j,k
real(8)::sum,product,q
real(8)::a,b
real(8)::m,max
x(1)=0.906179845938664d0
x(2)=-0.906179845938664d0
x(3)=0.538469310105683d0
x(4)=-0.538469310105683d0
x(5)=0.0d0
w(1)=0.236926885056189d0
w(2)=0.236926885056189d0
w(3)=0.478628670499367d0
w(4)=0.478628670499367d0
w(5)=0.568888888888889d0
max=100
m=1000
k=0
product=1
do while(p+k<1)
product=product*(p+k)
k=k+1
end do
q=p+k
sum=0
41
do j=1,m
a=(max/m)*(j-1)
b=(max/m)*j
do i=1,5
integration over [a,b]*****!
y(i)=(b+a)/2.+(b-a)*x(i)/2
v(i)=(b-a)*w(i)/2
sum=sum+v(i)*F(y(i),q)
end do
end do
Gamma=sum/product
F=x**(p-1)*exp(-x)
end function F
*********************************************************!
(real(8) function weight(alpha,j
real(8), intent(in)::alpha
integer, intent(in)::j
weight=(-1.0d0)**(j)*gamma(alpha+1.0d0)/gamma(dfloat(j)+1.0d0)&
&/gamma(alpha-dfloat(j)+1.0d0)
end function weight
***********************************************************!
B.2.Analytical solution :
program current
implicit none
double precision::alpha,t
integer(4)::j
OPEN(12,File="const_current.txt",STATUS="REPLACE",ACTION="WRITE")
alpha=-0.5d0
do j=1,100
print*,weight(alpha,j)
end do
pause
do alpha=0.00d0,1.d0,.1d0
if (alpha==0.d0 .or. alpha==1.d0) cycle
print*,alpha
do t=0.0d0,1.d-3,1.d-4
if (t==0) cycle
write(*,*) alpha,t,current_const(alpha,t)
write(12,*) alpha,t,current_const(alpha,t)
write(*,*) !
current_const(alpha,t),current_const_mettag(alpha,t)
end do
end do
42
CLOSE(12)
!alpha=0.7d0
!t=.2d0
!omega=alpha/(1.d0-alpha)
!call mettaq(2.d0-2.d0*alpha,2.d0-alpha,-omega*t**(2.d0-2.d0*alpha),E)
!print*,E
contains
integer::i,j,k
double precision::sum,product,q
double precision::a,b
double precision::m,max
x(1)=0.906179845938664d0
x(2)=-0.906179845938664d0
x(3)=0.538469310105683d0
x(4)=-0.538469310105683d0
x(5)=0.0d0
w(1)=0.236926885056189d0
w(2)=0.236926885056189d0
w(3)=0.478628670499367d0
w(4)=0.478628670499367d0
w(5)=0.568888888888889d0
max=100
m=100000
k=0
product=1
do while(p+k<1)
product=product*(p+k)
k=k+1
end do
q=p+k
sum=0
do j=1,m
a=(max/m)*(j-1)
b=(max/m)*j
do i=1,5
!*****integration over [a,b]
y(i)=(b+a)/2.+(b-a)*x(i)/2
v(i)=(b-a)*w(i)/2
sum=sum+v(i)*F(y(i),q)
end do
end do
Gamma=sum/product
43
double precision function F(x,p)
double precision, intent(in)::x,p
F=x**(p-1)*exp(-x)
end function F
subroutine Mettaq(alpha,beta,x,E)
double precision,intent(in)::alpha,beta,x
double precision,intent(out)::E
double precision::sum
integer::k
sum=0
do k=0,10
sum=sum+x**k/Gamma(alpha*k+beta)
end do
E=sum
subroutine Mettaqp(alpha,beta,n,p,a,x,E)
double precision,intent(in)::alpha,beta,x,n,a
double precision,intent(out)::E
integer,intent(in)::p
double precision::sum
integer::k
sum=0
do k=0,20
if (int(n*k)<p) cycle
sum=sum+gamma(n*k+1.d0)*a**k*x**(n*k-p)/(gamma(n*k-
p+1.d0)*Gamma(alpha*k+beta))
end do
E=sum
subroutine Mettaqpold(alpha,beta,p,x,E)
double precision,intent(in)::alpha,beta,x
double precision,intent(out)::E
integer,intent(in)::p
double precision::sum
integer::k
sum=0
do k=0,20
sum=sum+gamma(p+k+1.d0)*x**k/(Gamma(k+1.d0)*Gamma(alpha*(k+p)+beta))
end do
E=sum
44
weight=((-1)**j)*gamma(alpha+1.d0)/(gamma(j+1)*gamma(alpha-j+1.d0))
FUN=1.D0
f0=Fun(omega,t)
L=1.d-3
R=1.d0
Rs=alpha*R**alpha
Ls=(alpha-1.d0)*L**(1.d0-alpha)
omega=Rs/Ls
x=omega**2*t**(2.d0-2.d0*alpha)
alphaa=2.d0-2.d0*alpha
betaa=2.d0-alpha
a=0.d0
call mettaq(alphaa,betaa,-x,E1)
call mettaq(alphaa,betaa-1.d0,-x,E2)
I1=f0*t**(1.d0-alpha)*E1/Ls
I2=omega**2*a*E2/t**alpha
current_const_mettag=I1+I2
f0=Fun(omega,t)
L=1.d-3
R=1.d0
Rs=alpha*R**alpha
45
Ls=(alpha-1.d0)*L**(1.d0-alpha)
omega=Rs/Ls
a=0.d0
sum1=0.d0
sum2=0.d0
do p=0,10
sum1=sum1+(-1)**p*omega**(2*p)*t**(1.d0-alpha+2*p-
2*p*alpha)/gamma(2.d0-alpha+2*p-2*p*alpha)
end do
do p=0,10
sum2=sum2+(-1)**p*omega**(2*p)*t**(-alpha+2*p-
2*p*alpha)/gamma(1.d0-alpha+2*p-2*p*alpha)
end do
I1=f0*sum1/Ls
I2=omega**2*a*sum2
current_const=I1+I2
ﺗﻄﻮر اﻟﺘﻴﺎر اﻟﻜﻬﺮﺑﺎﺋﻲ ﺑﺪﻻﻟﺔ اﻟﺰﻣﻦ ﻓﻲ داﺋﺮة ﺑﺴﻴﻄﺔ ﺗﺤﺘﻮي ﻋﻠﻰ ﻣﺼﺪر ﺗﻴﺎر وﻣﻘﺎوﻣﺔ
46
اﻟﻤﻠﺨﺺ :
ﺗﻢ اﻗﺘﺮاح ﻣﻌﺎدﻟﺔ ﺗﻔﺎﺿﻠﻴﺔ آﺴﺮﻳﺔ ﻟﺪراﺳﺔ ﺗﻄﻮر اﻟﺘﻴﺎر ﻓﻲ دارة آﻬﺮﺑﺎﺋﻴﺔ ﺑﺴﻴﻄﺔ ﺑﺎﺳﺘﺨﺪام اﻟﻄﺮﻳﻘﺔ
. Mittag-Liffler functionوﻋﻨﺪﻣﺎ ﻳﻜﻮن ﻣﻘﺪار اﻟﺪرﺟﺔ اﻟﺘﺤﻠﻴﻠﻴﺔ ,وآﺘﺒﺖ اﻟﻨﺘﺎﺋﺞ ﺑﺪﻻﻟﺔ
اﻟﻜﺴﺮﻳﺔ αﻣﺴﺎوﻳﺔ ﻟﻠﺼﻔﺮ ﻧﺤﺼﻞ ﻋﻠﻰ ﺣﻞ ﻟﻤﻌﺎدﻟﺔ دارة ﺑﺴﻴﻄﺔ ﺗﺤﺘﻮي ﻋﻠﻰ ﻣﺤﺚ ﻓﻘﻂ وﻋﻨﺪﻣﺎ
ﺗﻜﻮن ﻣﺴﺎوﻳﺔ ﻟﻠﻮاﺣﺪ ﻧﺤﺼﻞ ﻋﻠﻰ ﺣﻞ ﻟﻤﻌﺎدﻟﺔ دارة ﺗﺤﺘﻮي ﻋﻠﻰ ﻣﻘﺎوﻣﺔ ﻓﻘﻂ.
وﻟﻘﻴﻢ اﻋﺘﺒﺎﻃﻴﺔ ﻟﻠﺪرﺟﺔ اﻟﻜﺴﺮﻳﺔ ﻣﺎ ﺑﻴﻦ اﻟﻮاﺣﺪ واﻟﺼﻔﺮ ﺣﺼﻠﻨﺎ ﻋﻠﻰ ﺣﻞ ﻋﺎم ﻳﻤﺜﻞ آﻴﻔﻴﺔ ﺗﺼﺮف
اﻟﺪارة آﻤﺤﺚ ﻓﻲ اﻟﺒﺪاﻳﺔ وﻓﻘﺪاﻧﻬﺎ ﻟﺨﺎﺻﻴﺔ اﻻﺳﺘﺤﺜﺎث ﺗﺪرﻳﺠﻴﺎ ﻟﺘﺼﺒﺢ ﻣﻘﺎوﻣﺔ ﻣﻊ ﺗﻐﻴﺮ ﻗﻴﻢ اﻟﺪرﺟﺔ
اﻟﻜﺴﺮﻳﺔ ﻣﻦ اﻟﺼﻔﺮ اﻟﻰ اﻟﻮاﺣﺪ ,وذﻟﻚ ﺑﺎﺳﺘﺨﺪام ﻣﺼﺪر ﺗﻴﺎر ﻣﺘﺮدد.
اﻣﺎ ﻓﻲ ﺣﺎل اﺳﺘﺨﺪام ﻣﺼﺪر ﺗﻴﺎر ﺛﺎﺑﺖ وﻣﻘﺎوﻣﺔ ﺧﺎرﺟﻴﺔ ﻓﻘﺪ ﺗﻢ دراﺳﺔ ﺗﺤﻮل اﻟﻤﺤﺚ اﻟﻰ ﻣﻘﺎوﻣﺔ
ﻋﻦ ﻃﺮﻳﻖ ﻣﺪﻩ اﻟﻰ ﺳﻠﻚ .ودﻋﻤﺖ هﺬﻩ اﻟﺘﻮﻗﻌﺎت ﺑﺤﻞ اﻟﻤﻌﺎدﻟﺔ ﺑﺎﺳﺘﺨﺪام ﻃﺮﻳﻘﺔ اﻟﺘﺤﻠﻴﻞ اﻟﻌﺪدي.
47