• Aucun résultat trouvé

Energy harvesting from quasi-periodic vibrations

N/A
N/A
Protected

Academic year: 2021

Partager "Energy harvesting from quasi-periodic vibrations"

Copied!
13
0
0

Texte intégral

(1)

DOI 10.1007/s11071-016-2668-6

O R I G I NA L PA P E R

Energy harvesting from quasi-periodic vibrations

Mohamed Belhaq · Mustapha Hamdi

Received: 27 September 2015 / Accepted: 2 February 2016

© Springer Science+Business Media Dordrecht 2016

Abstract Quasi-periodic (QP) vibration-based ene- rgy harvesting is studied in this paper. The energy har- vesting system consists in a delayed van der Pol oscil- lator with time-varying delay amplitude coupled to an electromagnetic energy harvesting device in which the vibration source is due to self-excitations. We consider the case of delay parametric resonance for which the frequency of the delay modulation is near twice the natural frequency of the oscillator. Application of the double-step perturbation method enables the approx- imation of the amplitude of the QP vibrations of the oscillator far from the resonance. This amplitude is used to extract the maximum and average QP pow- ers from the harvester device. The influence of dif- ferent system parameters on the performance of the QP vibration-based energy harvesting is reported and discussed. Results show that for appropriate values of parameters, QP vibrations can be more efficient for energy harvesting in pure self-excited systems not only in terms of power extraction, but also in terms of broad- ening the parameter range of energy extraction. Numer- ical simulation is systematically conducted to support the analytical predictions.

M. Belhaq (

B

)

Laboratory of Mechanics, University Hassan II of Casablanca, Casablanca, Morocco

e-mail: mbelhaq@yahoo.fr M. Hamdi

Faculty of Science and Technology-Al Hoceima, University Mohammed I, Oujda, Morocco

Keywords Quasi-periodic vibrations · Energy harvesting · Delayed van der Pol oscillator · Electro- mechanical coupling · Time-varying delay amplitude

1 Introduction

The design aspect and performance improvement of energy harvesting systems deeply depend on the opti- mal choice of mechanical and harvester system charac- teristics. To efficiently harvest energy from the ambi- ent environment, the ultimate goal is to produce large- amplitude oscillations with broadband performance characteristics, keeping the inherent mechanical damp- ing low [1].

When energy harvesting systems are subject to forced excitations, the resulting vibration responses usually have the disadvantage to strongly depend on the unknown frequency and amplitude of excitations.

In such circumstances, the efficiency of the energy harvesting operating in the linear resonance regime is achieved in a narrow region around the resonance peak limiting considerably its performance in the sense that the fundamental frequency of the harvester must always match the frequency of the forcing [2–5]. To overcome this limitation, nonlinearities were used to improve the energy harvesting capability by extend- ing the bandwidth of the harvester over broad range of excitation frequencies. In such, the harvesting system may be designed with hardening characteristic [6–10]

or with a bistable double well potential; see for instance

(2)

[9,11–15] and references therein. However, bistability can reduce the system performance when it is attracted to the low-amplitude motions [16] or when it suffers jump phenomena. Likewise, in forced excitations the dependence of the response on the unknown excitation parameters may result in a complicated design realiza- tion of the excitation [17–19].

In the case where energy harvesting systems are self- excited, the dependence of the system response on the excitation parameters can be remedied. The induced limit-cycle (LC) oscillations produced by a destabiliza- tion of the system trivial solution through supercriti- cal Hopf bifurcation may indeed be exploited in har- vesting power in the range of large-amplitude oscilla- tions. Under certain circumstances (presence of forced excitations and bistability), the steady-state LC oscilla- tions may lose stability via a secondary Hopf bifurca- tion producing QP vibrations and causing a substantial reduction in the harvested power away from the reso- nance [20–22]. To avoid this undesirable effect, focus was placed toward adjusting the parameter control to shift the secondary Hopf bifurcation to higher values of the control parameter [21]. As an example, in an energy harvester system under combined aerodynamic and base excitations it was observed that beyond the flutter speed, the response of the harvester is QP far from the resonance, leading also to a substantial drop of the output power [22].

In energy harvesting systems operating only in a pure self-induced oscillations regime (absence of forced excitations and bistability), LC oscillations born by destabilization of the system trivial solution can be exploited efficiently to enhance the power output using the largest vibrational amplitude. An open question is whether the QP vibrations born by destabilization of the LC may also be exploited to enhance the energy har- vesting performance. A recent work reported on this issue in terms of vibrational amplitude by considering a delayed pure van der Pol oscillator with modulated delay amplitude [23]. More precisely, it was shown analytically that modulating the delay amplitude in the pure delayed van der Pol oscillator may produce large- amplitude QP vibrations (larger than the amplitude of the periodic response) performing in broader range of parameters.

Motivated by this finding, the present paper seeks to take advantage of such large-amplitude QP vibrations to enhance the power output in the QP vibration range

near and far from a delay resonance. The device under consideration consists in a delayed van der Pol equa- tion with time-varying delay amplitude [23] coupled to an electromagnetic energy harvesting system. Atten- tion is paid to the effect of different system parameters on the performance of the QP vibration-based energy harvesting.

The organization of the paper is as follows: In Sect. 2, the electromagnetic vibration-based harvesting system is presented. The steady-state periodic response and the harvested power near the delay resonance are derived using averaging method. The influence of dif- ferent parameters of the harvesting system on the peri- odic response and the corresponding power output is examined. In Sect. 3, we approximate the QP response applying the second-step perturbation method. The har- vested power extracted from the QP vibration is ana- lyzed and compared to the periodic-based power out- put. A summary of the results is provided in the con- cluding section.

2 Equation of motion and periodic energy harvesting

The energy harvester under consideration consists of a pure delayed van der Pol oscillator coupled to an electric circuit through an electromagnetic coupling, as shown in the schematic presented in Fig. 1. The quantities F

k

, F

d

and F

vd P

are, respectively, the lin- ear restoring force, the time delay in the position and the van der Pol damping component. The governing equations can be written as

Fig. 1 Schematic of the energy harvesting model

(3)

m ¨ z α − ¯ β z

2

) z ˙ + kz = λ

t¯

z ( t ¯ − τ) + ˆ i (1a) L di

d t ¯ + R

eq

i = − ˆ z ˙ (1b) where the overdot represents a derivative with respect to time t. The variable ¯ z represents the relative displace- ment of the mass m, α ¯ and β ¯ are damping coefficients, k is the linear stiffness of the restoring force, ˆ is the electromagnetic coupling coefficient, λ

t¯

is the delay amplitude assumed to be time-varying, τ is the time delay, while i and

dit

have been substituted for electri- cal charge coordinate q (i = ˙ q). The quantities R

eq

and L are the equivalent resistance load and the inductance of the harvesting coil, respectively.

To obtain a dimensionless form of Eq. (1), we introduce the following nondimensional parameters:

t = ω

n

t, ¯ z =

zz0

, i =

ˆLz

0

i, γ

1

=

k Lˆ2

, γ

2

=

LRωeqn

, α =

mα¯ωn

, β =

mβ¯zω20n

and λ

t

=

mλωt¯2

n

where z

0

is the initial elongation of the compressed spring charged by proof mass m and ω

n

= √

k / m is the natural frequency of the mechanical oscillator. Introducing a bookkeep- ing parameter ε and scalling the electromechanical cou- pling, γ

1

, the damping components, α, β and the delay amplitude λ

t

, Eqs. (1) can be rewritten in terms of the nondimensional parameters as

¨

zβ z

2

z + z = λ

t

z ( tτ) + γ

1

i (2a) di

dt + γ

2

i = −˙ z (2b)

where an overdot denotes now differentiation with respect to time t .

In a recent work [23], the mechanical oscillator part of the considered energy harvester system presented in Fig. 1 (i.e., the pure delayed van der Pol equation) has been studied in the case where the delay amplitude λ

t

is modulated harmonically around a mean value, such that

λ

t

= λ

1

+ λ

2

cos ωt (3)

where λ

1

is referred to as the unmodulated delay ampli- tude, while λ

2

and ω are the amplitude and frequency of the delay amplitude modulation. Specifically, it was shown [23] that modulating the delay amplitude in the delayed van der Pol oscillator induces larger QP vibration amplitude than the periodic one in broadband

of parameters. Here, we shall take advantage of such large-amplitude QP vibrations to improve energy har- vesting performance in the QP vibration domain. In this study, one considers the case of primary delay para- metric resonance for which the frequency of the delay modulation ω is near twice the natural frequency of the oscillator ( ω ≈ 2), such that the resonance condition reads 1 = (

ω2

)

2

+ εσ where σ is a detuning parameter.

Neglecting higher-order harmonics, the steady-state solution of Eqs. (2) can be sought in the form

z(t ) = R cos ω

2 tθ

(4a) i (t ) = I cos

ω 2 tϕ

(4b) where R, I , θ and φ are the amplitudes and the phases of the responses.

To approximate the steady-state solutions near the delay parametric resonance, we apply the averaging method [24]. To this end, we rewrite Eq. (2a) as

¨ z + ω

2

4 z = H (z, z, ˙ i, t) (5)

where

H ( z , ˙ z , i , t ) = −σ z + β z

2

z

+(λ

1

+ λ

2

cos ω t ) z ( tτ) + γ

1

i (6) The approximate solution of Eq. (5) is supposed to take from (4a) and (4b) where the amplitudes R and I and the phases θ and ϕ are governed by the equations R ˙ = −

2 π

4π

ω

0

sin ω

2 tθ

H(z, z, ˙ i, t )dt (7a)

R θ ˙ =

4π ω

0

cos ω

2 tθ

H (z, z, ˙ i, t)dt (7b) I ˙ + ˙ R cos(θ − ϕ)R θ ˙ sin(θ − ϕ)

= −γ

2

I + ω

2 R sin(θ − ϕ) (7c)

I ϕ ˙ + ˙ R sin ϕ) + R θ ˙ cos ϕ)

= − ω 2 Iω

2 R cos(θ − ϕ) (7d)

Integrating and truncating the Taylor series expansions

for those terms containing time delay [25] by assum-

ing the product τ is small, one obtains the following

modulation equations

(4)

R ˙ = k

1

R + k

2

R

3

+ k

3

R cos 2 θ + k

4

R sin 2 θ + γ

1

ω I sin(θ − ϕ) (8a)

R θ ˙ = k

5

R + k

4

R cos 2θ − k

3

R sin 2θ + γ

1

ω I cos ϕ) (8b)

I ˙ + ˙ R cos(θ − ϕ)R θ ˙ sin(θ − ϕ)

= −γ

2

I + ω

2 R sin(θ − ϕ) (8c)

I ϕ ˙ + ˙ R sin(θ − ϕ) + R θ ˙ cos(θ − ϕ)

= − ω 2 Iω

2 R cos(θ − ϕ) (8d)

where k

1

=

α2

λω1

sin(ωτ/2), k

2

= −

β8

, k

3

=

λ2

2ω

sin(ωτ/2), k

4

=

2λω2

cos(ωτ/2) and k

5

= −

σω

+

λ1

ω

cos(ωτ/2). Equilibria of system (8), corresponding to periodic oscillations of Eq. (5), are determined by setting R ˙ = ˙ θ = ˙ I = ˙ φ = 0. Squaring, adding and eliminating θ and ϕ and define ρ = R

2

, we obtain the following system of equations

ρ( k

1

+ k

2

ρ)

2

+ ρ( k

5

+ k

6

ρ)

2

= ρ( k

32

+ k

42

) (9a) I = ω

22

+ ω

2

R (9b)

sin(θ − ϕ) = 2γ

2

22

+ ω

2

(9c) cos(θ − ϕ) = − ω

22

+ ω

2

(9d)

where k

1

= k

1

+

42γγ21γ2

22

and k

5

= k

5

4γ2γ21ω 22

. The first equation of this system, Eq. (9a), represents the amplitude–frequency response which has two solutions given by

R

1,2

=

B ± √

B

2

AC

A (10)

where A = k

22

, B = −k

1

k

2

and C = k

21

+k

25

−k

32

−k

42

. To extract the maximum and average output powers, we consider the nondimensional instantaneous power given by

P ( t ) = γ

2

i ( t )

2

(11)

It follows that the average power can be obtained by averaging over a single delay modulation period. This leads to

P

av

= ω

4π ω

0

P ( t ) dt (12)

Using Eqs. (4b), (9b), (11) and the maximization pro- cedure, one obtains the maximum power response as P

max

= γ

2

ω

2

4 γ

22

+ ω

2

R

2

(13)

and using Eqs. (4b), (11) and (12) the average power response reads

P

av

= γ

2

ω

2

2 ( 4 γ

22

+ ω

2

) R

2

(14)

Equations (9a), (13) and (14) are used to examine the influence of different system parameters on the steady- state response and on the maximum and average out- put powers of the harvester. In what follows, we fix the damping parameters α = 0 . 1, β = 0 . 7 and the electromechanical coupling coefficient γ

1

= 0.13.

The frequency–response curve, as given by Eq. (9a), is depicted in Fig. 2a near the delay parametric res- onance for two different values of the time delay.

Figure 2b, c illustrates, respectively, the effect of the time delay on the maximum and average output pow- ers, P

max

, P

av

, versus the frequency indicating that the periodic output powers increase by increasing τ and can be extracted only in a small range of the frequency ω near the resonance.

Figure 3a–c depicts, respectively, the variation of the amplitude of the steady-state response, the max- imum power response P

max

and the average power response P

av

versus λ

1

. For validation, analytical pre- dictions [solid (dot) lines for stable (unstable)] are compared to results obtained by numerical simula- tions (circles) using dde23 [26] algorithm. Inset in Fig. 3a, b are shown, respectively, time histories of the steady state and of the harvested power responses obtained by integrating numerically Eq. (2) and using Eq. (13).

The variation of the amplitude of the steady-state

response, the P

max

and P

av

, versus the modulated delay

(5)

0 0.5 1 1.5 2 2.5 3 3.5 4 0

0.2 0.4 0.6 0.8 1 1.2

ω

R

(a)

τ=1.58

τ=π

0 0.5 1 1.5 2 2.5 3 3.5 4

0 0.2 0.4 0.6 0.8 1 1.2

ω Pmax

(b)

τ=1.58

τ=π

0 0.5 1 1.5 2 2.5 3 3.5 4

0 0.2 0.4 0.6 0.8 1 1.2

ω Pav

(c)

τ=1.58

τ=π

Fig. 2 Vibration and powers amplitude versusωforλ1 = −0.2,λ2 = 0.2,γ2 =1.33 and for different values ofτ;afrequency response,bmaximum power response,caverage power response.Solid linefor stable,dotted linefor unstable

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 0

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

λ1 λ1 λ1

R 50 100 150 200

−1 0 1

Time, t

z(t)

(a)

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 0

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Pmax

0 50 100 150 200

0 0.5 1

Time, t

P(t)

(b)

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 0

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Pav

(c)

Fig. 3 Vibration and powers amplitude versusλ1forω=2.1,τ=1.58,λ2=0.2 andγ2=1.33;avibration amplitude,bmaximum power response,caverage power response.Solid linefor stable,dotted linefor unstable andcirclefor numerical solution

amplitude λ

2

is presented, respectively, in Fig. 4a–c for different values of the unmodulated delay amplitude λ

1

. The figures show clearly an increase of the output powers by increasing λ

2

and negative λ

1

in the range where the stable steady-state solution emerges.

Figure 5 depicts the variation of the amplitude of the periodic response and the output power amplitudes P

max

and P

av

versus the coupling term γ

2

for two values of λ

1

. It can be clearly seen that the amplitude response and the output powers increase substantially by increas- ing negative λ

1

and the powers can be extracted only in a certain range of the coupling coefficient γ

2

.

Similar plots are presented in Fig. 6 showing also an increase of the output powers by increasing nega- tive λ

1

and indicating that the energy harvesting can be generated only in small alternate ranges of time delay.

3 Quasi-periodic energy harvesting

Next, we shall approximate the QP response, its mod- ulation envelope as well as the QP vibration-based energy harvesting near and far from the delay para- metric resonance using the second-step perturbation method [27–29]. Since the slow flow (8) is invariant under the transformation θ → −θ +

π2

, φ → −φ +

π2

, ω → −ω, k

3

→ −k

3

and k

5

→ −k

5

, system (8) can be rewritten in the form

R ˙ = k

1

R + k

2

R

3

+ sk

3

R cos 2θ + k

4

R sin 2θ + γ

1

I sin ϕ) (15a)

R θ ˙ = sk

5

R + sk

6

R

3

+ k

4

R cos 2θ − sk

3

R sin 2θ + γ

1

I cos(θ − ϕ) (15b)

(6)

0 0.5 1 1.5 0

0.5 1 1.5 2 2.5 3

λ2 λ2 λ2

R

(a)

λ1=−0.2 λ1=−0.9

0 0.5 1 1.5

0 0.5 1 1.5 2 2.5 3

Pmax

(b)

0 0.5 1 1.5

0 0.5 1 1.5 2 2.5 3

Pav

(c)

Fig. 4 Vibration and powers amplitude versusλ2forω=2.1,τ=1.58,γ2=1.33 and for different values ofλ1;avibration amplitude, bmaximum power response,caverage power response

0 0.5 1 1.5 2 2.5

0 0.5 1 1.5 2 2.5

γ2 γ

2 γ

2

R

λ1=−0.9

λ1=−0.2

(a)

0 0.5 1 1.5 2 2.5

0 0.5 1 1.5 2 2.5

Pmax

(b)

0 0.5 1 1.5 2 2.5

0 0.5 1 1.5 2 2.5

Pav

(c)

Fig. 5 Vibration and powers amplitude versusγ2forω =2.1,τ =1.58 andλ2=0.2;avibration amplitude,bmaximum power response,caverage power response

0 1 2 3 4 5 6 7 8 9

0 0.5 1 1.5 2 2.5 3

Time delay, τ

R

(a)

λ1=−0.2 λ1=−0.9

0 1 2 3 4 5 6 7 8 9

0 0.5 1 1.5 2 2.5 3

Time delay, τ Pmax

(b)

0 1 2 3 4 5 6 7 8 9

0 0.5 1 1.5 2 2.5 3

Time delay, τ Pav

(c)

Fig. 6 Vibration and power amplitudes versusτforω =2.1,λ2=0.2 andγ2 =1.33;avibration amplitude,bmaximum power response,caverage power response

(7)

I ˙ + ˙ R cos ϕ)R θ ˙ sin ϕ)

= −γ

2

I +

2 R sin(θ − ϕ) (15c)

I ϕ ˙ + ˙ R sin(θ − ϕ) + R θ ˙ cos(θ − ϕ)

= − 2 I

2 R cos(θ − ϕ) (15d)

where s = ±1. This invariance property will be exploited later to determine the boundaries of the mod- ulation envelope of the QP vibrations. To investigate the QP response, it is convenient to transform the polar form (15) to the following Cartesian system using the variable change u = R cos θ , v = −R sin θ, w = I cos ϕ, y = − I sin ϕ

du

dT

0

= (sk

5

k

4

)v + η

(k

1

+ sk

3

)u +(k

2

u + sk

6

v)(u

2

+ v

2

) + γ

1

y

(16a) dv

dT

0

= −(sk

5

+ k

4

)u + η

(k

1

sk

3

)v +(k

2

vsk

6

u)(u

2

+ v

2

)γ

1

w

(16b) dw

dT

0

= −γ

2

w − du

dT

0

η 2 v +

2 y

(16c) d y

dT

0

= −γ

2

y − dv dT

0

+ η

2 u +

2 w

(16d)

where η is a new bookkeeping parameter introduced to implement a second perturbation analysis [27]. As usual, the parameter η is introduced in damping and nonlinearity so that the unperturbed system of Eq. (16) admits a basic solution.

The periodic solution of the slow flow (16), corre- sponding to the QP response of the original system (2), is obtained using the multiple scales method [24]. This can be done by expanding the solution as

u(T

0

, T

1

) = u

0

(T

0

, T

1

) + ηu

1

(T

0

, T

1

) + O(η

2

) v(T

0

, T

1

) = v

0

(T

0

, T

1

) + ηv

1

(T

0

, T

1

) + O(η

2

) (17) where T

0

= t and T

1

= ηt. Introducing D

i

=

T

i

yields

dtd

= D

0

+ ηD

1

+ O(η

2

), substituting Eqs. (17) into Eqs. (16) and collecting terms, we get the following hierarchy of systems

D

02

u

0

+ ν

2

u

0

= 0 , (18) ( sk

5

k

4

)v

0

= D

0

u

0

(19) D

0

w

0

+ γ

2

w

0

= −D

0

u

0

(20) D

0

y

0

+ γ

2

y

0

= −D

0

v

0

(21) D

02

u

1

+ ν

2

u

1

= (sk

5

k

4

)

D

1

v

0

+(k

1

sk

3

)v

0

+(k

2

v

0

sk

6

u

0

)(u

20

+ v

02

)

γ

1

s ω w

0

D

0

D

1

u

0

+ D

0

(k

1

+ sk

3

)u

0

+ (k

2

u

0

+ sk

6

v

0

) (u

20

+ v

20

) + γ

1

s ω y

0

(22) (sk

5

k

4

)v

1

= D

0

u

1

+ D

1

u

0

(k

1

+ sk

3

)u

0

−(k

2

u

0

+ sk

6

v

0

)(u

20

+ v

02

)

γ

1

y

0

(23)

D

0

w

1

+ γ

2

w

1

= −D

1

w

0

(D

0

u

1

+ D

1

u

0

)

2 v

0

2 y

0

(24)

D

0

y

1

+ γ

2

y

1

= − D

1

z

0

( D

0

v

1

+ D

1

v

0

)

2 u

0

+

2 w

0

(25)

where ν =

k

52

k

42

(26)

is the frequency of the periodic solution corresponding to the frequency of the QP modulation. Figure 7a, b shows the variation of this frequency modulation as function of τ and λ

1

, respectively. The plots indicate that the modulation occurs in certain alternate ranges of time delay, as depicted in Fig. 7a, while Fig. 7b determines the ranges of λ

1

where the modulation of the QP vibrations is more intense.

The solution to the first order is given by

u

0

(T

0

, T

1

) = r(T

1

) cos(νT

0

+ φ(T

1

)) (27a) v

0

( T

0

, T

1

) = − ν

(sk

5

k

4

) r ( T

1

) sin T

0

+ φ( T

1

)) (27b) w

0

(T

0

, T

1

) = S r (T

1

)

γ

2

sin(νT

0

+ φ(T

1

))

−ν cos(νT

0

+ φ(T

1

))

(27c) y

0

(T

0

, T

1

) = H r(T

1

)

γ

2

cos(νT

0

+ φ(T

1

)) +ν sin(νT

0

+ φ(T

1

))

(27d) where S =

γ2ν

22

, H =

2 ν2

22)(sk5k4)

, and r and

φ are, respectively, the amplitude and the phase of

(8)

0 1 2 3 4 5 6 7 8 0

0.05 0.1 0.15 0.2 0.25

τ

Frequency,ν

(a)

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1

0 0.05 0.1 0.15 0.2 0.25

Frequency,ν

λ1

(b)

Fig. 7 Variation of the QP frequency modulation versusaτwithλ1= −0.2 and versusbλ1withτ =1;ω =2.1,λ2 =0.2 and γ2=1.33

the slow flow limit cycle. Substituting (27) into (22)–

(25) and removing secular terms give the following autonomous slow-slow flow system on r and φ

dr

dT

1

= ( k

1

+ S

1

+ H

1

) r + sk

2

k

5

sk

5

k

4

r

3

r

dT

1

= − ν

γ

2

(S

1

+ H

1

)r (28)

where S

1

=

γ1γ22s(sk5νωk4)

S and H

1

=

γ2s1γω2

H . Equilibria in Eqs. (28) are obtained by setting d d

Tr1

= 0 and given by

r = 0, r =

−(k

1

+ S

1

+ H

1

) sk

5

k

4

sk

2

k

5

(29) I = ν

γ

22

+ ν

2

r (30)

with the condition λ

1

=

cosσ+χωτ

2

. Note that the expres- sion for I in Eq. (30) is obtained using Eqs. (24), (27a) and maximization procedure. The nontrivial equilib- rium in (29)–(30) corresponds to the amplitude of the slow flow limit cycle and consequently to the amplitude of the QP vibrations in the original Eq. (2).

On the other hand, the stability and bifurcation of the nontrivial QP solutions are obtained by calculat- ing the eigenvalues of the Jacobian matrix J of sys- tem (28) evaluated in the sth (s = 1, −1) branch (29).

The curves delimiting the regions of existence of the QP oscillations and the domains of their stability are given, respectively, by the conditions of saddle-node (Det(J) = 0 and T r (J) = −2k

1

< 0) and Hopf bifur-

cations given in terms of λ

1

as λ

1

< ω

sin

ωτ2

α

2 + κ

(31a) λ

= ( 4 − ω

2

) + 4 χ ± 4 ω k

4

4 cos

ωτ

2

(31b)

where κ =

4γ2γ21γ2

22

and χ =

4γγ12ω2

22

. The condition (31a) is written in terms of γ

2

as

γ

2

> −γ

1

+

γ

12

+ 4 ω

2

α

2 − λ

1

ω sin ωτ 2

(32)

Thus, the explicit expression of the QP response of the original Eq. (2) is written as

z(t ) = u (t ) cos ω

2 t

+ v(t ) sin ω

2 t

(33) and the QP solution of the current i (t), obtained by inserting the expression of z(t ) given by Eq. (33) into the second equation of system (2), can be extracted via a convolution integral with the boundary condition i(t) = i (0). This leads to

i(t) = −e

−γ2t

t

0

˙

z(τ)e

γ2τ

dτ (34)

Consequently, the power, the average and the maximum

powers output in the QP modulation region are given,

respectively, by

(9)

P(t) = γ

2

e

−γ2t

t 0

˙

z(τ)e

γ2τ

2

(35) P

avQP

= γ

2

ν

2

2(γ

22

+ ν

2

) r

2

(36)

P

maxQP

= γ

2

ν

2

γ

22

+ ν

2

r

2

(37)

Figure 8a presents the bifurcation curves delimiting the regions of existence and stability of the QP vibrations in the parameter plane ( λ

1

, τ ), as given by Eqs. (31). The first condition, Eq. (31a), is presented by the dashed lines, while the second one, Eq. (31b), is presented by the solid lines. The regions in which the QP vibrations are stable are indicated by SQP (gray regions), while those where QP vibrations are unstable are designated by UQP (aqua regions). The limit-cycle oscillations exist in the domain indicated by LC (white regions).

Figure 8b shows time histories and the correspond- ing power output responses related to the different regions of Fig. 8a, and the transitions of solutions are shown by moving between the crosses 1, 2, 3 and 4 in Fig. 8a. Between cross 1 and cross 2 the response of the system undergoes a transition between SQP vibra- tion and no oscillations which describes a bifurcation between a stable equilibrium point and the SQP solu- tion. From cross 2 to 3 Hopf bifurcation occurs giving rise to LC oscillations. From cross 3 to 4 the system changes its behavior from LC to SQP oscillation hav- ing, as expected, a slight modulation just beyond the saddle-node bifurcation of periodic orbits.

Figure 8c depicts analytical results of the amplitude response and the maximum powers P

maxP

, P

maxQP

as function of λ

1

for different values of τ (= 0.2, 1.58, 2.8) picked from Fig. 8a. The upper left and lower right boundaries of the QP modulation envelopes are given by Eq. (29) for s = −1, while the lower left and upper right ones are obtained for s = 1. Results obtained by numerical simulations are indicated by circles and cor- respond exactly to cross 1 (τ = 0.2, λ

1

= 0.1), cross 4 (τ = 1.58, λ

1

= −0.9) and cross 3 (τ = 2.8, λ

1

= 0 . 05) in Fig. 8a. The plots in Fig. 8c show clearly that for negative values of λ

1

, the maximum power output in the QP domain, P

maxQP

, is larger than that of the periodic response, P

maxP

. Instead, for positive λ

1

and values of τ taken inside the small gray area in Fig. 8a, P

maxQP

is lower than P

maxP

.

The variation of the stability domains of the QP solu- tions in the plane

1

, τ) is presented in Fig. 9 (gray regions) for three different values of λ

2

. The domains

located in the positive range of λ

1

provide small- amplitude QP vibrations, while the large domains sup- ply large-amplitude QP oscillations.

Figure 8c middle left is redrawn in Fig. 10a for clar- ity. The curve labeled M

+

corresponds to s = +1 in Eq.

(29) and that labeled M

corresponds to s = −1. The analytical predictions of the periodic response and the QP modulation envelope (solid lines) are compared to results obtained by numerical simulations (circles and time history inset in the figure) for validation. It can be clearly observed from Fig. 10a that the amplitude of the QP modulation reaches larger values comparing to the amplitude of the periodic response [23].

The variation of the maximum power output for periodic and QP responses versus λ

1

is illustrated in Fig. 10b. The corresponding power amplitudes are shown inset in the figure demonstrating clearly improvement of the energy harvesting performance by the QP vibrations.

Figure 11a shows the frequency response and the QP modulation envelope for fixed values of λ

1

, λ

2

, τ and for a critical value of the coupling coefficient γ

2cr

given by the condition γ

2

γ

2cr

= ν

2 +

| ν

2

− 4 |) (38) obtained from (37) by imposing the coefficient

γ2ν

γ222

≥ 1 which is required to ensure optimal per- formance of the QP vibration-based energy harvest- ing. Under such a condition, it can be observed from Fig. 11a that the amplitude of the QP envelope is larger than that of the periodic response for a certain range of frequencies. In this range of frequencies, a time his- tory corresponding to ω = 1 . 5 is presented inset in the figure. The connected circles in the figure reported for ω = 1.5 and for other values of ω are obtained by numerical simulation for validation. The area around the periodic frequency response is enlarged for clarity.

Figure 11b shows the maximum power response cor-

responding to periodic and QP vibrations. As expected,

the energy harvesting performance is achieved in

broadband of the QP response zone (aqua regions)

away from the resonance. Inset in the figure is shown

time history of the power response. To appreciate the

response in the area around the periodic output power,

an enlargement is shown inset in the figure. The average

power response is also reported in Fig. 11c illustrating

similar results.

(10)

τ

λ1

0 1 2 3 4 5 6 7 8

−1

−0.8

−0.6

−0.4

−0.2 0 0.2 0.4 0.6 0.8 1

2 UQP

LC 1

3

4 SQP

(a)

100 150 200 250 300

−0.5 0 0.5

z(t)

100 150 200 250 300

0 0.2 0.4

P(t)

0 10 20 30

−0.4

−0.2 0 0.2

z(t)

0 10 20 30

0 0.2 0.4

P(t)

100 150 200 250 300

−1 0 1

z(t)

100 150 200 250 300

0 0.5 1

P(t)

100 150 200 250 300

−2 0 2

Time, t

z(t)

100 150 200 250 300

0 1 2 3

P(t)

λ1=0.1 τ=0.2 at cross 1

λ1=0.4 τ=1.58 at cross 2

λ1=0.05 τ=2.8 at cross 3

λ1=−0.9 τ=1.58 at cross 4

(b)

Time, t

−0.50 0 0.5

0.5 1

R, r

conected points at cross 1

−0.50 0 0.5

0.5 1

PmaxP, PmaxQP

−1 −0.5 0 0.5 1

0 1 2 3

R, r

conected points at cross 4

−10 −0.5 0 0.5 1

1 2 3

PmaxP, PmaxQP

−0.50 0 0.5

0.5 1

R, r

circle point at cross 3

λ1 −0.5 0 0.5

0 0.5 1

PmaxP, PmaxQP

λ1

τ=0.2

τ=1.58 τ=1.58

τ=2.8 τ=2.8

(c)

τ=0.2

Fig. 8 aBifurcation curves in the plane1, τ),btime histories corresponding to different regions,cvibration amplitudes ver- susλ1;λ2 =0.2,ω=2.1 andγ2 =1.33.UQPunstable QP

solutions,SQPstable QP solutions,LClimit-cycle oscillations (color online)

Finally, the variation of periodic and QP responses as well as the corresponding maximum powers ampli- tude P

max

, P

maxQP

versus γ

2

is depicted, respectively,

in Fig. 12a, b for fixed values of ω and delay para-

meters. Again, the analytical prediction (solid lines) is

compared to numerical simulations (circles) for valida-

(11)

τ

λ1 λ1 λ1

0 1 2 3 4 5

−1

−0.8

−0.6

−0.4

−0.2 0 0.2 0.4 0.6 0.8 1

(a)

0 1 2 τ3 4 5

−1

−0.8

−0.6

−0.4

−0.2 0 0.2 0.4 0.6 0.8 1

(b)

0 1 2 τ3 4 5

−1

−0.8

−0.6

−0.4

−0.2 0 0.2 0.4 0.6 0.8 1

(c)

Fig. 9 Domains of stability (gray regions) of QP vibrations in the plane1, τ)forω=2.1,γ2=1.33 andaλ2=0.4,bλ2=0.2,c λ2=0.02

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1

0 0.5 1 1.5 2 2.5 3

λ1

R, r

50 100 150 200

−2

−1 0 1 2

50 100 150 200

−2

−1 0 1 2

Time, t

z(t)

(a)

M+

M−

at cross 4

z(t)

Time, t

−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1

0 0.5 1 1.5 2 2.5 3

λ1 P maxP, P maxQP

0 50 100 150 200

0 1 2 3

Time, t

P(t)

0 50 100 150 200

0 1 2 3

(b)

at cross 4

P(t)

Time, t

Fig. 10 Vibration (a) and power (b) amplitudes versusλ1forτ=1.58,ω=2.1 andγ2=1.33; analytical (solid lines) and numerical (circles) approximations

tion. The boxes inset in the figures show time histories for the QP response (Fig. 12a) and for the variation of the maximum QP power output (Fig. 12b). It can be observed from Fig. 12a that for the chosen set of para- meters, the periodic and the QP vibrations have com- parable amplitudes, while Fig. 12b shows that there exists optimal values of the parameters for which the QP response may supply a broadband energy harvest- ing with good energy harvesting performance.

4 Conclusions

We have studied the QP vibration-based energy har- vesting in a delayed van der Pol oscillator with time-

periodic delay amplitude coupled to an electromagnetic

energy harvesting device. The only vibration source

arises in such a system can be produced by self-induced

vibrations. By modulating the delay amplitude around

a nominal value near a delay parametric resonance, a

QP behavior may occur. Using the second-step pertur-

bation method, analytical expression of the QP vibra-

tions amplitude was obtained and exploited to obtain

the corresponding maximum and average output pow-

ers. The influence of the system parameters on the

performance of the QP vibration-based energy har-

vesting was reported and discussed. In particular, it

was shown that the modulation of the delay amplitude

near the delay parametric resonance gives rise to large-

amplitude QP vibrations in broadband of parameters

(12)

0 0.5 1 1.5 2 2.5 3 3.5 4 0

0.5 1 1.5 2 2.5 3

ω

R, r

1.9 1.95 2 2.05 2.1

1.7 1.75 1.8 1.85 1.9 1.95 2 2.05 2.1

50 100 150 200

−2

−1 0 1 2

Time, t

z(t)

zoom

(a)

0 0.5 1 1.5 2 2.5 3 3.5 4

0 0.5 1 1.5 2 2.5 3

ω

P maxP, P maxQP 1.9 1.95 2 2.05 2.1

0 0.2 0.4 0.6 0.8 1

50 100 150 200

−1 0 1

Time, t

P(t)

zoom

(b)

0 0.5 1 1.5 2 2.5 3 3.5 4

0 0.5 1 1.5 2 2.5 3

ω P avP, P avQP

1.9 1.95 2 2.05 2.1

0 0.1 0.2 0.3 0.4 0.5

(c)

zoom

Fig. 11 Frequency responses (a) and powers output (b) andcforτ=1.58,λ1= −0.53,λ2=0.2 andγ2=γ2cr(color online)

0 0.5 1 1.5 2 2.5 3

0 0.5 1 1.5 2 2.5 3

R, r

γ2

50 100 150 200

−2

−1 0 1 2

Time, t

z(t)

(a) (b)

0 0.5 1 1.5 2 2.5 3

0 0.5 1 1.5 2 2.5 3

γ2

P maxP, P maxQP

50 100 150 200

0 1 2 3

Time, t

P(t)

Fig. 12 Vibration (a) and power (b) amplitudes versusγ2forω=2.1,τ=1.58,λ1= −0.6 andλ2=0.2. harvesting

(13)

near and far from the resonance. For negative values of the unmodulated delay amplitude, such stable large- amplitude QP vibrations are found to be more efficient to improve energy harvesting performance than the periodic vibration. More importantly, it was reported that in the presence of linear stiffness only, the QP vibration-based energy harvesting may be extracted in a broad range of system parameters providing a reli- able alternative to extend the bandwidth of the har- vester.

It is clear from this study that in the absence of forced excitations and bistable double well potential, QP vibrations induced by delay amplitude modulation in pure delayed self-excited systems can improve the energy harvesting performance in broadband of para- meters. This result is not limited only to van der Pol- type equation, but can also be obtained in general for other pure self-excited systems, as Froude pendulum or others.

References

1. duToit, N., Wardle, B.: Experimental verification of mod- els for microfabricated piezoelectric vibration energy har- vesters. AIAA J.45, 1126–1137 (2007)

2. Stephen, N.G.: On energy harvesting from ambient vibra- tion. J. Sound Vib.293, 409–425 (2006)

3. Lesieutre, G.A., Ottman, G.K., Hofmann, H.F.: Damping as a result of piezoelectric energy harvesting. J. Sound Vib.

269(3–5), 991–1001 (2004)

4. Sodano, H.A., Inman, D.J., Park, G.: Generation and storage of electricity from power harvesting devices. J. Intell. Mater.

Syst. Struct.16, 67–75 (2005)

5. Sodano, H.A., Inman, D.J., Park, G.: Comparison of piezo- electric energy harvesting devices for recharging batteries.

J. Intell. Mater. Syst. Struct.16, 799–807 (2005)

6. Peano, F., Coppa, G., Serazio, C., Peinetti, F., D’angola, A.:

Nonlinear oscillations in a MEMS energy scavenger. Math.

Comput. Model.43, 1412–1423 (2006)

7. Mann, B.P., Sims, N.D.: Energy harvesting from the non- linear oscillations of magnetic levitation. J. Sound Vib.319, 515–530 (2009)

8. Marinkovic, B., Koser, H.: Smart sand-A wide bandwidth vibration energy harvesting platform. Appl. Phys. Lett.94, 103505 (2009)

9. Ramlan, R., Brennan, M.J., Mace, B.R., Kovacic, I.: Poten- tial benefits of a non-linear stiffness in an energy harvesting device. Nonlinear Dyn.59, 545–558 (2010)

10. Stanton, S.C., McGehee, C.C., Mann, B.P.: Reversible hys- teresis for broadband magnetopiezoelastic energy harvest- ing. Appl. Phys. Lett.95, 174103 (2009)

11. Daqaq, M.: Transduction of a bistable inductive generator driven by white and exponentially correlated Gaussian noise.

J. Sound Vib.330, 2554–2564 (2011)

12. Erturk, A., Hoffmann, J., Inman, D.J.: A piezomagnetoe- lastic structure for broadband vibration energy harvesting.

Appl. Phys. Lett.94, 254102 (2009)

13. Erturk, A., Inman, D.J.: Broadband piezoelectric power gen- eration on high-energy orbits of the bistable Duffing oscil- lator with electromechanical coupling. J. Sound Vib.330, 2339–2353 (2011)

14. Erturk, A., Inman, D.J.: Piezoelectric Energy Harvesting.

Wiley, New York (2011)

15. Van Blarigan, L., Danzl, P., Moehlis, J.: A broadband vibra- tional energy harvester. Appl. Phys. Lett. 100, 253904 (2012)

16. Quinn, D.D., Triplett, A.L., Vakakis, A.F., Bergman, L.A.:

Energy harvesting from impulsive loads using intestinal essential nonlinearities. J. Vib. Acoust.133, 011004 (2011) 17. Gu, L., Livermore, C.: Impact-driven, frequency up- converting coupled vibration energy harvesting device for low frequency operation. Smart Mater. Struct.20, 045004 (2011)

18. Hagedorn, P., DasGupta, A.: Vibrations and Waves in Con- tinuous Mechanical Systems. Wiley, England (2007) 19. Priya, S.: Modeling of electric energy harvesting using

piezoelectric windmill. Appl. Phys. Lett.87, 184101 (2005) 20. Abdelkefi, A., Nayfeh, A., Hajj, M.: Modeling and analysis of piezoaeroelastic energy harvesters. Nonlinear Dyn.67, 925–939 (2011)

21. Abdelkefi, A., Nayfeh, A.H., Hajj, M.R.: Design of piezoaeroelastic energy harvesters. Nonlinear Dyn.68, 519–

530 (2012)

22. Bibo, A., Daqaq, M.F.: Energy harvesting under combined aerodynamic and base excitations. J. Sound Vib.332, 5086–

5102 (2013)

23. Hamdi, M., Belhaq, M.: Quasi-periodic vibrations in a delayed van der Pol oscillator with time-periodic delay amplitude. J. Vib. Control (2015). doi:10.1177/

1077546315597821

24. Nayfeh, A.H., Mook, D.T.: Nonlinear Oscillations. Wiley, New York (1979)

25. Wirkus, S., Rand, R.H.: The dynamics of two coupled van der Pol oscillators with delay coupling. Nonlinear Dyn.30, 205–221 (2002)

26. Shampine, L.F., Thompson, S.: Solving delay differential equations with dde23.http://www.radford.edu/~thompson/

webddes/tutorial(2000)

27. Belhaq, M., Houssni, M.: Quasi-periodic oscillations, chaos and suppression of chaos in a nonlinear oscillator driven by parametric and external excitations. Nonlinear Dyn.18, 1–24 (1999)

28. Rand, R.H., Guennoun, K., Belhaq, M.: 2:2:1 resonance in the quasi-periodic Mathieu equation. Nonlinear Dyn.31, 187–193 (2003)

29. Hamdi, M., Belhaq, M.: Quasi-periodic oscillation envelopes and frequency locking in excited nonlinear sys- tems with time delay. Nonlinear Dyn.73, 1–15 (2013)

Références

Documents relatifs

total number of surgeries total number of operating rooms total number of surgeons duration of surgery i set of surgeries for surgeon s required quantity of kits of type k for surgery

a été menée dans le bute de rechercher les métabolites secondaires de type lactones sesquiterpéniques et flavonoïdes. Ces deux classes de substances naturelles sont

62 D’après la figure qui représente la variation de coefficient de friction moyen le long de la paroi inférieure à différente vitesse (nombre de Reynolds) pour les deux canaux

Comparing with the power spectra measured on the bridge (Figure 7(left)) reveals a resonant behaviour of the pipe at frequency 14.5 Hz..

For the bridge example considered, the predicted performances of a validated model of piezoelectric harvesting device are subsequently discussed in relation with the

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

This chapter presents a three-dimensional monolithic mixed-hybrid finite element formulation with six independent unknown fields—velocities, stresses, potential rate,

The research aims to utilize flow induced structural vibrations to harvest useful electrical energy with the help of piezoelectric material and develop a numerical model to solve