ApproximateSolutionofNonlinearDuffingOscillatorUsingTaylorExpansion10 5923 J Jmea 20160605 03
ApproximateSolutionofNonlinearDuffingOscillatorUsingTaylorExpansion10 5923 J Jmea 20160605 03
net/publication/319900354
CITATIONS READS
20 2,063
1 author:
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Abdelrady Okasha Elnady on 24 October 2017.
1
Mechatronics Department, Faculty of Engineering, O6University, Egypt
2
Mechanical Engineering Department, Shoubra faculty of Engineering, Banha University, Egypt
Abstract The Duffing oscillator is a common model for nonlinear phenomena in science and engineering. Its
mathematical model is a second order differential equation with nonlinear spring force used to describe the motion of a
damped oscillator with a more complicated potential than in simple harmonic motion. In the present paper, the Duffing
oscillator equation is solved using a new simple technique based on Taylor theory. The Duffing oscillator equation is
solved with different values of initial conditions and damping. The solution results are compared with Runge–Kutta 4th
order numerical solution method to investigate the accuracy and reliability of the suggested technique. Results show an
excellent agreement between the proposed technique and the Runge–Kutta method.
Keywords Duffing Oscillator, Nonlinear differential equation, Taylor expansion
expansion. The acceleration of the next point is obtained ẍ + αẋ + βx + γx 3 = F0 cos ωt (7)
from the equilibrium equation and so on. Results are With initial conditions
compared with that obtained by the fourth order
Runge–Kutta method. Results show a good agreement 𝑥𝑥(0) = 𝐴𝐴 (8)
between the proposed technique and the Runge–Kutta 𝑥𝑥̇ (0) = 𝐵𝐵 (9)
method. Equation (7) is a simple model that can show different
types of oscillations such as chaos and limit cycles. The
terms associated with this system represent:
2. Basic Fundamentals 𝑥𝑥̈ + 𝛽𝛽𝛽𝛽 Simple harmonic oscillator with angular
In this section the basic fundamentals of Taylor’s theorem frequency �𝛽𝛽
as well as the forward, backward and central difference 𝛼𝛼𝑥𝑥̇ Small damping
approximations of higher order derivative are reviewed. 𝛾𝛾𝑥𝑥 3 Small nonlinearity
Taylor’s Theorem: If f is a function continuous and n 𝐹𝐹0 cos 𝜔𝜔𝜔𝜔 Small periodic forcing term with angular
times differentiable in an interval [x, x + h], then there exists frequency 𝜔𝜔
some point in this interval, denoted by This is a forced oscillator with a nonlinear spring with a
x + λh for some λ є [0, 1], such that restoring force of F = −βx − α x 3 . Different values of α
ℎ2 ′′ can create either a hardening spring (where α > 0) or a
𝑓𝑓(𝑥𝑥 + ℎ) = 𝑓𝑓(𝑥𝑥) + ℎ𝑓𝑓 ′ (𝑥𝑥) + 𝑓𝑓 (𝑥𝑥) + ⋯
2 softening spring (where α < 0). Different values of β can
ℎ n −1 ℎ n −1 also change the dynamics of the system. For values of β less
+ (n−1)! 𝑓𝑓 𝑛𝑛−1 (𝑥𝑥) + 𝑓𝑓 𝑛𝑛 (𝑥𝑥 + 𝜆𝜆ℎ) (1)
n! than zero, the Duffing oscillator displays chaotic motion.
If f is a so-called analytic function of which the derivatives
of all orders exist, then one may consider increasing the 3.1. Methodology of the Proposed Technique
value of n indefinitely. Thus, if the condition holds that In this technique, the differential equation (7) is
ℎ 𝑛𝑛 𝑛𝑛 (𝑥𝑥) rearranged as follows:
limn→∞ 𝑓𝑓 =0 (2)
𝑛𝑛!
ẍ (t) = −αẋ (t) − βx(t) − γx(t)3 + F cos(ω t) (10)
which is to say that the terms of the series converge to zero as
By direct substitution of the initial conditions given in (8)
their order increases, then an infinite-order Taylor-series
and (9) the acceleration at starting point can be written as:
expansion is available in the form of
ℎ j ẍ (0) = −αẋ (0) − βx(0) − γx(0)3 + F cos(ω × 0) (11)
𝑓𝑓(𝑥𝑥 + ℎ) = ∑∞ 𝑓𝑓 𝑗𝑗 (𝑥𝑥)
(3)
𝑗𝑗 =0 j! The approximate displacement function at time t + ∆t is
This is obtained simply by extending indefinitely the obtained using Taylor expansion (1) up to the third term:
1
expression from Taylor’s Theorem. In interpreting the 𝑥𝑥(𝑡𝑡 + ∆𝑡𝑡) = 𝑥𝑥(𝑡𝑡) + 𝑥𝑥̇ (𝑡𝑡)∆𝑡𝑡 + 𝑥𝑥̈ (𝑡𝑡)∆𝑡𝑡 2 (12)
2
summary notation for the expansion, one must be aware of
the convention that 0! = 1. The approximate velocity function at time 𝒕𝒕 + ∆𝒕𝒕 is
obtained using the backward difference approximation of the
Forward Difference: first derivative (5):
The first derivative of a function f(x) can be 𝑥𝑥(𝑡𝑡+∆𝑡𝑡)−𝑥𝑥(𝑡𝑡)
approximated using forward difference: 𝑥𝑥̇ (𝑡𝑡 + ∆𝑡𝑡) = (13)
∆𝑡𝑡
𝑓𝑓 ′ (𝑥𝑥) = limℎ→0
𝑓𝑓(𝑥𝑥+ℎ)−𝑓𝑓(𝑥𝑥)
+ 𝑜𝑜(ℎ) (4) Then, the approximate acceleration function at time
ℎ 𝒕𝒕 + ∆𝒕𝒕 is obtained using equation (10) as:
Backward Difference: 𝑥𝑥̈ (𝑡𝑡 + ∆𝑡𝑡) = −𝛼𝛼𝑥𝑥̇ (𝑡𝑡 + ∆𝑡𝑡) − 𝛽𝛽𝛽𝛽(𝑡𝑡 + ∆𝑡𝑡) − 𝛾𝛾𝑥𝑥(𝑡𝑡 + ∆𝑡𝑡)3
The first derivative of a function f(x) can be +𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐(𝜔𝜔 × (𝑡𝑡 + ∆𝑡𝑡)) (14)
approximated using backward difference:
𝑓𝑓(𝑥𝑥)−𝑓𝑓(𝑥𝑥−ℎ)
So, the first iteration is obtained from equations (8)
𝑓𝑓 ′ (𝑥𝑥) = limℎ→0 + 𝑜𝑜(ℎ) (5) through (14) as:
ℎ
1
Central Difference: 𝑥𝑥(∆𝑡𝑡) = 𝑥𝑥(0) + 𝑥𝑥̇ (0)∆𝑡𝑡 + 𝑥𝑥̈ (0)∆𝑡𝑡 2 (15)
2
The first derivative of a function f(x) can be
approximated using central difference: ẋ (∆t) = (x(∆t) − x(0))/∆t (16)
3
𝑓𝑓(𝑥𝑥+ℎ)−𝑓𝑓(𝑥𝑥−ℎ) 𝑥𝑥̈ (∆𝑡𝑡) = −𝛼𝛼𝑥𝑥̇ (∆𝑡𝑡) − 𝛽𝛽𝛽𝛽(∆𝑡𝑡) − 𝛾𝛾𝑥𝑥(∆𝑡𝑡) + 𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐(𝜔𝜔∆𝑡𝑡) (17)
𝑓𝑓 ′ (𝑥𝑥) = limℎ→0 + 𝑜𝑜(ℎ2 ) (6)
ℎ The recurrence formula can be written as:
1
𝑥𝑥𝑛𝑛 = 𝑥𝑥𝑛𝑛−1 + 𝑥𝑥̇ 𝑛𝑛−1 ∆𝑡𝑡 + 𝑥𝑥̈ 𝑛𝑛−1 ∆𝑡𝑡 2 (18)
3. The Problem of Duffing Oscillator 2
𝑥𝑥̇ 𝑛𝑛 = (𝑥𝑥𝑛𝑛 − 𝑥𝑥𝑛𝑛−1 )/∆𝑡𝑡 (19)
The equation of motion of Duffing oscillator is normally 𝑥𝑥̈ 𝑛𝑛 = −𝛼𝛼𝑥𝑥̇ 𝑛𝑛 − 𝛽𝛽𝑥𝑥𝑛𝑛 − 𝛾𝛾𝑥𝑥𝑛𝑛3 + 𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐�𝜔𝜔((𝑛𝑛 − 1)∆𝑡𝑡)� (20)
written as
112 A. Okasha El-Nady et al.: Approximate Solution of Nonlinear Duffing Oscillator Using Taylor Expansion
3.2. Modified Technique Runge–Kutta numerical method since the error in version 1
It was noted that the results obtained by the previous is of order (∆𝑡𝑡) while in version 2 is of order (∆𝑡𝑡 2 ).
iteration formulae has a big difference with that obtained by Example 2
Runge–Kutta 4th order method. This because that the error α = 2, β = 1, γ = 25, A = 0.1, B = 0, F0=0 (31)
accompanied to the velocity is of order ∆t. The approximate
velocity formula is modified to be obtained using the central For the values given in Eq. (31), x(t) is obtained using the
difference approximation of the first derivative (6) as: recurrence formulae of the present technique (version 1) and
its modification (version 2).
𝑥𝑥̇ (∆𝑡𝑡) = (𝑥𝑥(2∆𝑡𝑡) − 𝑥𝑥(0))/(2∆𝑡𝑡) (21) Figure 2 shows the comparison between the results
The first iteration of the modified technique is written as: obtained using the present techniques (version 1) and its
1 modification (version 2) and the fourth-order Runge–Kutta
𝑥𝑥(∆𝑡𝑡) = 𝑥𝑥(0) + 𝑥𝑥̇ (0)∆𝑡𝑡 + 𝑥𝑥̈ (0)∆𝑡𝑡 2 (22) numerical method. It is clear that the results using the
2
𝑥𝑥(2∆𝑡𝑡) = 𝑥𝑥(0) + 2𝑥𝑥̇ (0)∆𝑡𝑡 + 2𝑥𝑥̈ (0)∆𝑡𝑡 2 (23) modified technique have good agreement with the results
obtained using the fourth-order Runge–Kutta numerical
𝑥𝑥̇ (∆𝑡𝑡) = (𝑥𝑥(2∆𝑡𝑡) − 𝑥𝑥(0))/(2∆𝑡𝑡) (24) method. The version 1 of the present technique has a big
3
𝑥𝑥̈ (∆𝑡𝑡) = −𝛼𝛼𝑥𝑥̇ (∆𝑡𝑡) − 𝛽𝛽𝛽𝛽(∆𝑡𝑡) − 𝛾𝛾𝑥𝑥(∆𝑡𝑡) + 𝑐𝑐𝑐𝑐𝑐𝑐(𝜔𝜔∆𝑡𝑡) (25) difference with that obtained by the fourth-order
The recurrence formula of the modified technique can be Runge–Kutta numerical method since the error in version 1
written as: is of order (∆𝑡𝑡) while in version 2 is of order (∆𝑡𝑡 2 ).
1 Example 3
𝑥𝑥𝑛𝑛 = 𝑥𝑥𝑛𝑛−1 + 𝑥𝑥̇ 𝑛𝑛−1 ∆𝑡𝑡 + 𝑥𝑥̈ 𝑛𝑛−1 ∆𝑡𝑡 2 (26)
2 α = 1, β = 20, γ = 2, A = −0.2, B=2, F0=0 (32)
2
𝑥𝑥𝑛𝑛+1 = 𝑥𝑥𝑛𝑛−1 + 2𝑥𝑥̇ 𝑛𝑛−1 ∆𝑡𝑡 + 2𝑥𝑥̈ 𝑛𝑛−1 ∆𝑡𝑡 (27)
For the values given in Eq. (32), x(t) is obtained using the
𝑥𝑥̇ 𝑛𝑛 = (𝑥𝑥𝑛𝑛+1 − 𝑥𝑥𝑛𝑛−1 )/(2∆𝑡𝑡) (28) recurrence formulae of the present technique (version 1) and
𝑥𝑥̈ 𝑛𝑛 = −𝛼𝛼𝑥𝑥̇ 𝑛𝑛 − 𝛽𝛽𝑥𝑥𝑛𝑛 − 𝛾𝛾𝑥𝑥𝑛𝑛3 + 𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐�𝜔𝜔((𝑛𝑛 − 1)∆𝑡𝑡)� (29) its modification (version 2).
Figure 3 shows the comparison between the results
obtained using the present techniques (version 1) and its
4. Results and Discussion modification (version 2) and the fourth-order Runge–Kutta
numerical method. It is clear that the results using the
4.1. Free Vibration modified technique have good agreement with the results
obtained using the fourth-order Runge–Kutta numerical
The recursive relations in sections 3.1 and 3.2 are applied method. The version 1 of the present technique has a big
to the duffing oscillator problem with non exciting force. difference with that obtained by the fourth-order
Three case studies [29] are resolved using the present Runge–Kutta numerical method since the error in version 1
technique. The first case study includes low damping is of order (∆𝑡𝑡) while in version 2 is of order (∆𝑡𝑡 2 ).
(periodic behavior), strong nonlinearity and initial As shown in Figures 1, 2, and 3 the results of the solution
displacement. The second case study includes critical of the Duffing equation by the proposed technique has
damping, strong nonlinearity and initial displacement and excellent agreement with that obtained by the Runge-Kutta
the third case study is a combination of initial displacement 4thorder (RK4) method.
and velocity with periodic behavior. They are considered as
follows: 4.2. Forced Vibration
Example 1 The responses of the forced duffing oscillators given in
α = 0.5, β = γ = 25, A = 0.1, B = 0, F0=0 (30) the previous examples are obtained using the modified
proposed technique. The amplitude of the exciting force
For the values given in Eq. (30), x(t) is obtained using the
F0 =1 and the exciting frequency ω =0.5.
recurrence formulae of the present technique (version 1) and
For the same data given in Example 1, Example 2 and
its modification (version 2).
Figure 1 shows the comparison between the results Example 3 with F0 =1 and ω =0.5, the responses x(t) of the
obtained using the present techniques (version 1) and its forced duffing oscillator are obtained using the recurrence
modification (version 2) and the fourth-order Runge–Kutta formulae of the present technique (version 2). The results are
numerical method. It is clear that the results using the compared with that obtained using the fourth-order
modified technique have good agreement with the results Runge–Kutta method and shown in figures 4, 5 and 6. The
obtained using the fourth-order Runge–Kutta numerical comparison shows an excellent agreement between the
method. The version 1 of the present technique has a big present technique and Runge–Kutta 4th order.
difference with that obtained by the fourth-order
Journal of Mechanical Engineering and Automation 2016, 6(5): 110-116 113
-0.05
-0.1
0 1 2 3 4 5 6 7
time t
Duffing Oscillator Example 1 Version 2
0.1
Present
0.05
Amplitude x(t)
RK4
-0.05
-0.1
0 1 2 3 4 5 6 7
time t
0.1
Amplitude x(t)
0.05
-0.05
0 1 2 3 4 5 6 7
time t
Duffing Oscillator Example 2 Version 2
0.1
Amplitude x(t)
0.05
0
0 1 2 3 4 5 6 7
time t
Amplitude x(t)
RK4
0.5
-0.5
0 1 2 3 4 5 6 7
time t
Duffing Oscillator Example 3 Version 2
0.4
Present
0.2
Amplitude x(t)
RK4
-0.2
-0.4
0 1 2 3 4 5 6 7
time t
0.06
0.04
Amplitude x(t)
0.02
-0.02
-0.04
-0.06
0 2 4 6 8 10 12 14 16
time t
0.3
0.2
0.1
Amplitude x(t)
-0.1
-0.2
-0.3 Present
RK4
-0.4
0 2 4 6 8 10 12 14 16
time t
0.3
0.2
Amplitude x(t)
0.1
-0.1
-0.2
-0.3
0 2 4 6 8 10 12 14 16
time t