284 4 Forced Oscillations of Systems Having a Single Degree of Freedom

K1 = 2εk1 , K2 = O(1) (4.28.1)

and let the solution of the equation be

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.28.2)

Substituting (4.28.1) and (4.28.2) into the original equation and retaining to O(ε),
we get

0 = D02 + 2εD0 D1 (x0 + εx1 ) + (x0 + εx1 )
1
− ε(D0 + εD1 )(x0 + εx1 ) + ε[(D0 + εD1 )(x0 + εx1 )]3
3
− 2εkn cos(1 T0 + θ1n ) − K2 cos(2 T0 + θ2 ) + · · ·
1
= D02 x0 + εD02 x1 + 2εD0 D1 x0 + x0 + εx1 − εD0 x0 + ε(D0 x0 )3
3
− 2εk1 cos(1 T0 + θ1 ) − K2 cos(2 T0 + θ2 ) + · · ·
= D02 x0 + x0 − K2 cos(2 T0 + θ2 )
1
+ ε[D02 x1 + x1 + 2D0 D1 x0 − D0 x0 + D0 x0 )3 − 2k1 cos(1 T0 + θ1 ) + · · ·
3
(4.28.3)

Let the coefficient of the like power of ε be zero, we get

D02 x0 + x0 = K2 cos(2 T0 + θ2 ) (4.28.4)

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − D0 x0 )3 + 2k1 cos(1 T0 + θ1 (4.28.5)
3
The solution of (4.28.4) is

x0 = AeiT0 + eiθ2 ei2 T0 + cc (4.28.6)

where A = A(T1 ) and

K2 9
= = K2 (4.28.7)
2 1 − 22 16

Therefore, the first order approximate solution of the equation is

u = A(T1 )eiT0 + cc + 2 cos(2 T0 + θ2 ) + O(ε) (4.28.8)

Substituting (4.28.6) into (4.28.5) and taking 1 ≈ 1 and 2 ≈ 1/3 into account,
we get
4.28 Exercise 4.28 (Primary and Superharmonic Resonances of Self-Excited … 285

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3 + 2k1 cos(1 T0 + θ1 )
3
1
= −(2iD1 A − iA)e − (3iA2 AeiT0 + 6i22 2 AeiT0
iT0
3
− i32 3 e3iθ2 e3i2 T0 ) + 2k1 cos(1 T0 + θ1 ) + cc + NST (4.28.9)

Let
1 3 3
= k2 , εσ1 = 1 − 1, εσ2 = 32 − 1 (4.28.10)
3 2
Equation (4.28.9) becomes

D02 x1 + x1 = −(2iD1 A − iA + iA2 Ā + 2i22 2 A

− ik2 ei(σ2 T1 +3θ2 ) − k1 ei(σ1 T1 +θ1 ) )eiT0 + cc + NST (4.28.11)

In order to eliminate secular terms from the above equation, there must be

2iA − iA + iA2 A + 2i22 2 A − ik2 ei(σ2 T1 +3θ2 ) − k1 ei(σ1 T1 +θ1 ) = 0 (4.28.12)

where the prime denotes the derivative with respect to T1 . Let

1 iβ
A= ae (4.28.13)
2

γ1 = σ1 T1 − β + θ1 , γ2 = σ2 T1 − β + 3θ2 (4.28.14)

Then from (4.28.13), we have

1 1
ia − aβ − i a + i a3 + i22 2 a − k1 eiγ1 − ik2 eiγ2 = 0 (4.28.15)
2 8
Separating the real and imaginary parts of the above equation yields

1 1
a = − 22 2 − a2 a + k1 sin γ1 + k2 cos γ2 (4.28.16)
2 8

aβ = −k1 cos γ1 + k2 sin γ2 (4.28.17)

(b) From (4.28.16) and (4.28.17), we can find that the system has a steady state
solution when and only when γ1 and γ2 are constant values, hence there are

β = σ1 = σ2 σ (4.28.18)
286 4 Forced Oscillations of Systems Having a Single Degree of Freedom

(c) Let a = 0 in (4.28.16) and (4.28.17), we can obtain

1 1
− − 22 2 − a2 a = k1 sin γ1 + k2 cos γ2 (4.28.19)
2 8

−σ a = k1 cos γ1 − k2 sin γ2 (4.28.20)

The two equations are squared and added together to obtain the frequency–
response equation as

1 1
( − 22 2 − a2 )2 a2 + σ 2 a2 = k12 + k22 + 2k1 k2 sin ψ (4.28.21)
2 8

ψ = γ1 − γ2 = θ1 − 3θ2 (4.28.22)

Solving σ from (4.28.21), we have

k12 + 2k1 k2 sin ψ + k22 1 1

σ = ±[ 2
− ( − 22 2 − a2 )2 ]1/2 (4.28.23)
a 2 8
From this, the frequency–response curve of the system can be plotted in Fig. 4.18.
In order to analyze the stability of the steady state solutions, Eqs. (4.28.16) and
(4.28.17) are rewritten as

1 1
a = − 22 2 − a2 a + k1 sin(γ + θ1 ) + k2 cos(γ + 3θ2 ) (4.28.24)
2 8

a σ − γ = −k1 cos(γ + θ1 ) + k2 sin(γ + 3θ2 ) (4.28.25)

where

Fig. 4.18 Frequency-response and frequency-phase characteristics of the steady state solution
(θ1 = 0, θ2 = π4 , k1 = 5, K2 = 1) for Exercise 4.28
4.29 Exercise 4.29 (Primary and Superharmonic Resonances of Self-Excited … 287

γ = σ1 T1 − β = σ2 T1 − β = σ T1 − β (4.28.26)

Let

a = a0 + ã, γ = γ0 + γ̃ (4.28.27)

Thus the linearized form of (4.28.24) and (4.28.25) are

1 3
ã = − 22 2 − a02 ã + k1 cos(γ0 + θ1 ) − k2 sin(γ0 + 3θ2 )]γ̃ (4.28.28)
2 8
σ 1
γ̃ = ã − k1 sin(γ0 + θ1 ) + k2 cos(γ0 + 3θ2 )]γ̃ (4.28.29)
a0 a0

Write the above two equations in matrix form as

% %
ã l11 l12 ã
= (4.28.30)
γ̃ l21 l22 γ̃

where
1 3
l11 = − 22 2 − a02 , l12 = k1 cos(γ0 + θ1 ) − k2 sin(γ0 + 3θ2 )
2 8
σ 1
l21 = , l22 = − k1 sin(γ0 + θ1 ) + k2 cos(γ0 + 3θ2 )]
a0 a0

The eigenvalues of Eq. (4.20.26) are

)
(l11 + l22 ) ± (l11 + l22 )2 − 4(l11 l22 − l12 l21 )
λ= (4.28.31)
2
The sign of the real part of the eigenvalue determines the stability of the corre-
sponding steady state solution. It follows that branches 1, 3, and 5 in Fig. 4.18 are
unstable.

4.29 Exercise 4.29 (Primary and Superharmonic

Resonances of Self-Excited Systems)

Solution: (a) Let

K1 = 2εk1 , K2 = O(1) (4.29.1)

and set the solution of the equation be

288 4 Forced Oscillations of Systems Having a Single Degree of Freedom

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.29.2)

Substituting (4.28.1) and (4.28.2) into the original equation and retaining to O(ε),
we get

1 2
0 = ü + u − ε u̇ − u̇3 − Kn cos(n t + θn )
3 n=1
2
D0 + 2εD0 D1 (x0 + εx1 ) + (x0 + εx1 )
1
− ε(D0 + εD1 )(x0 + εx1 ) + ε[(D0 + εD1 )(x0 + εx1 )]3
3
− 2εkn cos(1 T0 + θ1n ) − K2 cos(2 T0 + θ2 ) + · · ·
1
= D02 x0 + εD02 x1 + 2εD0 D1 x0 + x0 + εx1 − εD0 x0 + ε(D0 x0 )3
3
− 2εk1 cos(1 T0 + θ1 ) − K2 cos(2 T0 + θ2 ) + · · ·
= D02 x0 + x0 − K2 cos(2 T0 + θ2 )
1
+ ε[D02 x1 + x1 + 2D0 D1 x0 − D0 x0 + D0 x0 )3 − 2k1 cos(1 T0 + θ1 ) + · · ·
3
(4.29.3)

Let the coefficient of the like power of ε be zero, we get

D02 x0 + x0 = K2 cos(2 T0 + θ2 ) (4.29.4)

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − D0 x0 )3 + 2k1 cos(1 T0 + θ1 (4.29.5)
3
The solution of (4.28.4) is

x0 = AeiT0 + eiθ2 ei2 T0 + cc (4.29.6)

where A = A(T1 ) and

K2 1
= = − K2 (4.29.7)
2 1 − 22 16

Let
1 iβ
A= ae (4.29.8)
2
Therefore, the first order approximate solution of the equation is

u = a(T1 ) cos(T0 + β) + 2 cos(2 T0 + θ2 ) + O(ε) (4.29.9)

4.29 Exercise 4.29 (Primary and Superharmonic Resonances of Self-Excited … 289

Substituting (4.28.6) into (4.28.5) and taking 1 ≈ 1 and 2 ≈ 3 into account,

we get

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3 + 2k1 cos(1 T0 + θ1 )
3
1
= −(2iD1 A − iA)eiT0 − (3iA2 ĀeiT0 + 6i22 2 AeiT0
3
− 3i2 Ā eiθ2 ei(2 −2)T0 ) + cc + NST + 2k1 cos(1 T0 + θ1 )
2

(4.29.10)

Let
1
k2 = 2 , εσ1 = 1 − 1, εσ2 = 2 − 3 (4.29.11)
4
Equation (4.28.9) becomes

D02 x1 + x1 = −(2iD1 A − iA + iA2 Ā + 16ik22 A

− 4ik2 Ā ei(σ2 T1 +θ2 ) − k1 ei(σ1 T1 +θ1 ) )eiT0 + cc + NST
2
(4.29.12)

In order to eliminate secular terms from the above equation, there must be
2
2iA − iA + iA2 A + 2i22 2 A − 4ik2 A ei(σ2 T1 +θ2 ) − k1 ei(σ1 T1 +θ1 ) = 0 (4.29.13)

The prime represents the derivative with respect to T1 . Substituting (4.29.8) into
the above equation and let

γ1 = σ1 T1 − β + θ1 , γ2 = σ2 T1 − 3β + θ2 (4.29.14)

then
1 1
ia − aβ − i a + i a3 + i16k22 a − k1 eiγ1 − ik2 a2 eiγ2 = 0 (4.29.15)
2 8
Separating the real and imaginary parts of the above equation yields

1 1
a = − 16k22 − a2 a + k1 sin γ1 + k2 a2 cos γ2 (4.29.16)
2 8

aβ = −k1 cos γ1 + k2 a2 sin γ2 (4.29.17)

(b) We can find from (4.28.16) and (4.28.17) that the system has a steady state
solution when and only when γ1 and γ2 are constants, hence
290 4 Forced Oscillations of Systems Having a Single Degree of Freedom

1
β = σ1 = σ2 (4.29.18)
3

4.30 Exercise 4.30 (Simultaneous Subharmonic

and Superharmonic Resonance of Self-Excited
Systems)

Solution: (a) We assume that Kn = O(1) (n = 1, 2) and the solution of the equation
is

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.30.1)

Substituting (4.30.1) into the original equation and retaining to O(ε) yields

1 2
0 = ü + u − ε u̇ − u̇3 − Kn cos(n t + θn )
3 n=1

= D02 + 2εD0 D1 (x0 + εx1 ) + (x0 + εx1 )
1
− ε(D0 + εD1 )(x0 + εx1 ) + ε[(D0 + εD1 )(x0 + εx1 )]3
3
2
− Kn cos(n T0 + θn ) + · · ·
n=1
1
= D02 x0 + εD02 x1 + 2εD0 D1 x0 + x0 + εx1 − εD0 x0 + ε(D0 x0 )3
3

2
− Kn cos(n T0 + θn ) + · · ·
n=1

2
= D02 x0 + x0 − Kn cos(n T0 + θn )
n=1
1
+ ε[D02 x1 + x1 + 2D0 D1 x0 − D0 x0 + D0 x0 )3 + · · · (4.30.2)
3
Let the coefficient of the like power of ε be zero, we get

2
D02 x0 + x0 = Kn cos(n T0 + θn ) (4.30.3)
n=1

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3 (4.30.4)
3
4.30 Exercise 4.30 (Simultaneous Subharmonic and Superharmonic … 291

The solution of (4.30.3) is

2
x0 = AeiT0 + n eiθn ein T0 + cc (4.30.5)
n=1

where A = A(T1 ) and

K1 1 K2 9
1 = = − K2 , 2 = = K2 (4.30.6)
2 1 − 1
2 16 2 1 − 2
2 16

Let
1 iβ
A= ae (4.30.7)
2
Therefore, the first order approximate solution of the equation is

u = a(T1 ) cos(T0 + β) + 2 cos(2 T0 + θ2 ) + O(ε) (4.30.8)

Substituting (4.30.5) into (4.30.4) and taking 1 ≈ 3 and 2 ≈ 1/3 into account,
we get

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3
3
1
= −(2iD1 A − iA)e − (3iA2 AeiT0
iT0
3
+ 6i21 21 AeiT0 + 6i22 22 AeiT0
2
− i32 32 ei3θ2 ei32 T0 − 3iA 1 1 eiθ1 ei(1 −2)T0 ) + cc + NST (4.30.9)

Let

εσ1 = 1 − 3, εσ2 = 32 − 1 (4.30.10)

Equation (4.30.9) becomes

D02 x1 + x1 = −(2iD1 A − iA + iA2 Ā + 2i21 21 A + 2i22 22 A

1
− i 32 32 ei(σ2 T1 +3θ2 ) − iĀ 1 1 ei(σ1 T1 +θ1 ) )eiT0
2
3
+ cc + NST (4.30.11)

In order to eliminate secular terms from the above equation, there must be

2A − A + A2 A + 221 21 A + 222 22 A

292 4 Forced Oscillations of Systems Having a Single Degree of Freedom

1 2
− 32 32 ei(σ2 T1 +3θ2 ) − A 1 1 ei(σ1 T1 +θ1 ) = 0 (4.30.12)
3
The prime represents the derivative with respect to T1 . Substituting (4.30.7) into
the above equation, we get

1 1
a + iaβ − a + a3 + 21 21 a + 22 22 a
2 8
1 3 3 i(σ2 T1 −β+3θ2 ) 1 2
− 2 2 e − a 1 1 ei(σ1 T1 −3β+θ1 ) = 0 (4.30.13)
3 4
Let
1
k1 = 1 1 , γ1 = σ1 T1 − 3β + θ1 , γ2 = σ2 T1 − β + 3θ2 (4.30.14)
4
then
1 1 1
a − a + a3 + 16k12 a + 22 22 a + iaβ − k1 a2 eiγ1 − 32 32 eiγ2 = 0
2 8 3
(4.30.15)

Separating the real and imaginary parts of the above equation yields

1 1 1
a = − 16k12 − 22 22 − a2 a + k1 a2 cos γ1 + 32 32 cos γ2 (4.30.16)
2 8 3
1
aβ = k1 a2 sin γ1 + 32 32 sin γ2 (4.30.17)
3

(b) It can be found from (4.30.16) and (4.30.17) that the system has a steady state
solution when and only when γ1 and γ2 are constants, hence

1
β = σ1 = σ2 (4.30.18)
3

4.31 Exercise 4.31 (Self-Excited System with Both Primary

and Combined Resonances I)

Solution: (a) We assume

K1 = 2εk1 , K2 = O(1) (4.31.1)

and let the solution of the equation be

4.31 Exercise 4.31 (Self-Excited System with Both Primary and Combined … 293

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.31.2)

Substituting (4.31.1) and (4.31.2) into the original equation and retaining to O(ε),
we get

1 3
0 = ü + u − ε u̇ − u̇3 − Kn cos(n t + θn )
3 n=1

= D02 + 2εD0 D1 (x0 + εx1 ) + (x0 + εx1 )
1
− ε(D0 + εD1 )(x0 + εx1 ) + ε[(D0 + εD1 )(x0 + εx1 )]3
3
3
− 2εk1 cos(1 T0 + θ1 ) − Kn cos(n T0 + θn ) + · · ·
n=2

3
= D02 x0 + x0 − Kn cos(n T0 + θn )
n=2
1
+ ε[D02 x1 + x1 + 2D0 D1 x0 − D0 x0 + D0 x0 )3 − 2k1 cos(1 T0 + θ1 ) + · · ·
3
(4.31.3)

Let the coefficient of the like power of ε be zero, we get

3
D02 x0 + x0 = Kn cos(n T0 + θn ) (4.31.4)
n=2

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − D0 x0 )3 + 2k1 cos(1 T0 + θ1 (4.31.5)
3
The solution of (4.31.4) is

3
x0 = AeiT0 + n eiθn ein T0 + cc (4.31.6)
n=2

where A = A(T1 ) and

Kn
n = (4.31.7)
2 1 − 2n

Let
1 iβ
A= ae (4.31.8)
2
Therefore, the first order approximate solution of the equation is
294 4 Forced Oscillations of Systems Having a Single Degree of Freedom

3
u = a(T1 ) cos[T0 + β(T1 )] + 2 n cos(n T0 + θn ) + O(ε) (4.31.9)
i=2

Substituting (4.31.6) into (4.31.5) and taking 1 ≈ 1 and 22 + 3 ≈ 1 into

account, we get

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3
3
1
= −(2iD1 A − iA)eiT0 − (3iA2 AeiT0
3
+ 6iA22 22 eiT0 + 6iA23 23 eiT0
− i322 3 22 3 ei(2θ2 +θ3 ) ei(22 +3 )T0 ) + k1 ei(1 T0 +θ1 ) + cc + NST
(4.31.10)

Let

εσ1 = 1 − 1, εσ2 = 22 + 3 − 1 (4.31.11)

Equation (4.31.10) becomes

D02 x1 + x1 = −(2iD1 A − iA + iA2 Ā + 2iA22 22 + 2iA23 23

− i22 3 22 3 ei(σ2 T1 +2θ2 +θ3 ) − k1 ei(σ1 T1 +θ1 ) )eiT0
+ cc + NST (4.31.12)

In order to eliminate secular terms from the above equation, there must be

2A − A + A2 A + 222 22 A + 223 23 A

− 22 3 22 3 ei(σ2 T1 +2θ2 +θ3 ) + ik1 ei(σ1 T1 +θ1 ) = 0 (4.31.13)

The prime represents the derivative with respect to T1 . Substituting (4.30.7) into
the above equation, we get

1 1
a + iaβ − a + a3 + 22 22 a + 23 23 a
2 8
i(σ2 T1 −β+2θ2 +θ3 )
− 2 3 2 3 e
2 2
+ ik1 ei(σ1 T1 −β+θ1 ) = 0 (4.31.14)

Let

k2 = 22 3 22 3 , γ1 = σ1 T1 − β + θ1 , γ2 = σ2 T1 − β + 2θ2 + θ3 (4.31.15)

then
4.31 Exercise 4.31 (Self-Excited System with Both Primary and Combined … 295

1 1 3
a − a + a3 + 2n 2n + iaβ + ik1 eiγ1 − k2 eiγ2 = 0 (4.31.16)
2 8 n=2

Separating the real and imaginary parts of the above equation yields
,
1 2 2 1 2
3

a = − − a a + k1 sin γ1 + k2 cos γ2 (4.31.17)
2 n=2 n n 8

aβ = −k1 cos γ1 + k2 sin γ2 (4.31.18)

(b) It can be found from (4.30.16) and (4.30.17) that the system has a steady state
solution when and only when γ1 and γ2 are constants, hence

β = σ1 = σ2 σ (4.31.19)

(c) Let a = 0and β = σ in (4.31.17) and (4.31.18), we can obtain

,
1 2 2 1 2
3
− − − a a = k1 sin γ1 + k2 cos γ2 (4.31.20)
2 n=2 n n 8

−σ a = k1 cos γ1 − k2 sin γ2 (4.31.21)

The two equations are squared and added together to obtain the frequency–
response equation as
,2
1 2 2 1 2
3
− − a a2 + σ 2 a2 = k12 + k22 + 2k1 k2 sin ψ (4.31.22)
2 n=2 n n 8

ψ = γ1 − γ2 = θ1 − 2θ2 − θ3 (4.31.23)

Solving σ from (4.28.21), we get

⎡ ,2 ⎤1/2
1 2 2 1 2
3
k 2 + 2k1 k2 sin ψ + k22
σ = ±⎣ 1 − − − a ⎦ (4.31.24)
a2 2 n=2 n n 8

From this, the frequency–response curve of the system can be plotted in Fig. 4.19.
In order to analyze the stability of steady state solutions, Eqs. (4.31.17) and
(4.31.18) are rewritten as
296 4 Forced Oscillations of Systems Having a Single Degree of Freedom

Fig. 4.19 Frequency-response and frequency-phase characteristics of the steady state solution
(θ1 = 0, θ2 = θ3 = π3 , k1 = 10, K2 = K3 = 5, 2 = 0.3, 3 = 0.4) for Exercise
4.31

,
1 2 2 1 2
3

a = − − a a + k1 sin(γ + θ1 ) + k2 cos(γ + 2θ2 + θ3 )
2 n=2 n n 8
(4.31.25)

a σ − γ = −k1 cos(γ + θ1 ) + k2 sin(γ + 2θ2 + θ3 ) (4.31.26)

where

γ = σ1 T1 − β = σ2 T1 − β = σ T1 − β (4.31.27)

Around any set of steady state solution (a0 , γ0 ), we let

a = a0 + ã, γ = γ0 + γ̃ (4.31.28)

The linearized form of (4.31.25) and (4.31.26) are

,
1 2 2 3 2
3
ã = − n n − a0 ã + k1 cos(γ0 + θ1 ) − k2 sin(γ0 + 2θ2 + θ3 )]γ̃
2 n=2 8
(4.31.29)
σ 1
γ̃ = ã − k1 sin(γ0 + θ1 ) + k2 cos(γ0 + 2θ2 + θ3 )]γ̃ (4.31.30)
a0 a0

Write the above two equations in matrix form as

% %
ã l l ã
= 11 12 (4.31.31)
γ̃ l21 l22 γ̃
4.32 Exercise 4.32 (Self-Excited System with Both Primary and Combined … 297

-
3
where l11 = 1
2
− 2n 2n − 38 a02 , l12 = k1 cos(γ0 + θ1 ) − k2 sin(γ0 + 2θ2 + θ3 )
n=2

σ 1
l21 = , l22 = − k1 sin(γ0 + θ1 ) + k2 cos(γ0 + 2θ2 + θ3 )]
a0 a0

The eigenvalues of (4.31.31) are

)
(l11 + l22 ) ± (l11 + l22 )2 − 4(l11 l22 − l12 l21 )
λ= (4.31.32)
2
The stability of the corresponding steady-state solution can be determined from
the sign of the real part of the eigenvalue. It follows that branch 3 in Fig. 4.19 is
unstable.

4.32 Exercise 4.32 (Self-Excited System with Both Primary

and Combined Resonances II)

Solution: (a) We assume

K1 = 2εk1 , K2,3 = O(1) (4.32.1)

and let the solution of the equation be

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.32.2)

Substituting (4.32.1) and (4.32.2) into the original equation and retaining to O(ε),
we get

1 3
0 = ü + u − ε u̇ − u̇3 − Kn cos(n t + θn )
3 n=1

= D02 + 2εD0 D1 (x0 + εx1 ) + (x0 + εx1 )
1
− ε(D0 + εD1 )(x0 + εx1 ) + ε[(D0 + εD1 )(x0 + εx1 )]3
3
3
− 2εk1 cos(1 T0 + θ1 ) − Kn cos(n T0 + θn ) + · · ·
n=2
1
= D02 x0 + εD02 x1 + 2εD0 D1 x0 + x0 + εx1 − εD0 x0 + ε(D0 x0 )3
3
3
− 2εk1 cos(1 T0 + θ1 ) − Kn cos(n T0 + θn ) + · · ·
n=2
298 4 Forced Oscillations of Systems Having a Single Degree of Freedom

3
= D02 x0 + x0 − Kn cos(n T0 + θn )
n=2
1
+ ε[D02 x1 + x1 + 2D0 D1 x0 − D0 x0 + D0 x0 )3 − 2k1 cos(1 T0 + θ1 ) + · · ·
3
(4.32.3)

Let the coefficient of the like power of ε be zero, we get

3
D02 x0 + x0 = Kn cos(n T0 + θn ) (4.32.4)
n=2

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − D0 x0 )3 + 2k1 cos(1 T0 + θ1 (4.32.5)
3
The solution of (4.32.4) is

3
x0 = AeiT0 + n eiθn ein T0 + cc (4.32.6)
n=2

where A = A(T1 ) and

Kn
n = (4.32.7)
2 1 − 2n

Let
1 iβ
A= ae (4.32.8)
2
Therefore, the first order approximate solution of the equation is

3
u = a(T1 ) cos[T0 + β(T1 )] + 2 n cos(n T0 + θn ) + O(ε) (4.32.9)
i=2

Substituting (4.32.6) into (4.32.5) and taking 1 ≈ 1 and 2 + 3 ≈ 1 into

account, we get

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − D0 x0 )3 − 2k1 cos(1 T0 + θ1 )
3
1
= −(2iD1 A − iA)eiT0 − (3iA2 AeiT0
3
+ 6iA22 22 eiT0 + 6iA23 23 eiT0
+ 3i2 3 2 3 Aeiθ2 eiθ3 ei(2 +3 −1)T0 )
4.32 Exercise 4.32 (Self-Excited System with Both Primary and Combined … 299

+ k1 ei(1 T0 +θ1 ) + cc + NST (4.32.10)

Let

εσ1 = 1 − 1, εσ2 = 22 + 3 − 1 (4.32.11)

Equation (4.32.10) becomes

D02 x1 + x1 = −(2iD1 A − iA + iA2 A + 2iA22 22 + 2iA23 23

+ i2 3 2 3 Aei(σ2 T1 +θ2 +θ3 ) − k1 ei(σ1 T1 +θ1 ) )eiT0
+ cc + NST (4.32.12)

In order to eliminate secular terms from the above equation, there must be

2A − A + A2 A + 222 22 A + 223 23 A

+ 2 3 2 3 Aei(σ2 T1 +θ2 +θ3 ) + ik1 ei(σ1 T1 +θ1 ) = 0 (4.32.13)

The prime represents the derivative with respect to T1 . Substituting (4.32.8) into
the above equation, we get

1 1
a + iaβ − a + a3 + 22 22 a + 23 23 a
2 8
+ i2 3 2 3 aei(σ2 T1 −2β+θ2 +θ3 ) + ik1 ei(σ1 T1 −β+θ1 ) = 0 (4.32.14)

Let

k2 = 2 3 2 3 , γ1 = σ1 T1 − β + θ1 , γ2 = σ2 T1 − 2β + θ2 + θ3 (4.32.15)

then

1 1 3
a − a + a3 + 2n 2n + iaβ + ik1 eiγ1 + k2 eiγ2 = 0 (4.32.16)
2 8 n=2

Separating the real and imaginary parts of the above equation yields
,
1 2 2 1 2
3

a = − − a a + k1 sin γ1 − k2 cos γ2 (4.32.17)
2 n=2 n n 8

aβ = −k1 cos γ1 − k2 sin γ2 (4.32.18)

(b) It can be found from (4.32.17) and (4.32.18) that the system has a steady state
solution when and only when γ1 and γ2 are constants, hence
300 4 Forced Oscillations of Systems Having a Single Degree of Freedom

1
β = σ1 = σ2 (4.32.19)
2

4.33 Exercise 4.33 (Combination Resonance of Self-Excited

Systems)

Solution: (a) We assume Kn = O(1)(n = 1, 2, 3) and the solution of the equation is

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.33.1)

Substituting (4.31.1) and (4.31.2) into the original equation and retaining to O(ε),
we get

1 3
0 = ü + u − ε u̇ − u̇3 − Kn cos(n t + θn )
3 n=1

= D02 + 2εD0 D1 (x0 + εx1 ) + (x0 + εx1 ) − ε(D0 + εD1 )(x0 + εx1
1 3
+ ε[(D0 + εD1 )(x0 + εx1 )]3 − Kn cos(n t + θn )
3 n=1
1
= D02 x0 + εD02 x1 + 2εD0 D1 x0 + x0 + εx1 − εD0 x0 + ε(D0 x0 )3
3
3
− Kn cos(n T0 + θn ) + · · ·
n=1

3
= D02 x0 + x0 − Kn cos(n T0 + θn )
n=1
1
+ ε[D02 x1 + x1 + 2D0 D1 x0 − D0 x0 + D0 x0 )3 + · · · (4.33.2)
3
Let the coefficient of the like power of ε be zero, we get

3
D02 x0 + x0 = Kn cos(n T0 + θn ) (4.33.3)
n=1

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3 (4.33.4)
3
The solution of (4.33.3) is
4.33 Exercise 4.33 (Combination Resonance of Self-Excited Systems) 301

3
x0 = AeiT0 + n eiθn ein T0 + cc (4.33.5)
n=1

where A = A(T1 ) and

Kn
n = (4.33.6)
2 1 − 2n

Let
1 iβ
A= ae (4.33.7)
2
Therefore, the first order approximate solution of the equation is

3
u = a(T1 ) cos[T0 + β(T1 )] + 2 n cos(n T0 + θn ) + O(ε) (4.33.8)
i=1

Substituting (4.33.5) into (4.33.4), and taking 1 + 2 + 3 ≈ 1 into account,

we get

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3
3
1 3
= −(2iD1 A − iA)e iT0
− [3iA2 ĀeiT0 + 6iAeiT0 2n 2n )
3 n=1

− 6i1 2 3 1 2 3 ei(θ1 +θ2 +θ3 ) ei(1 +2 +3 )T0 ] + cc + NST

(4.33.9)

Let

εσ2 = 1 + 2 + 3 − 1 (4.33.10)

Equation (4.31.10) becomes

D02 x1 + x1 = −(2iD1 A − iA + iA2 Ā + 2iA21 21 + 2iA22 22 + 2iA23 23

− i1 2 3 1 2 3 ei(σ T1 +θ1 +θ2 +θ3 ) )eiT0
+ cc + NST (4.33.11)

In order to eliminate secular terms from the above equation, there must be

2A − A + A2 A + 2A21 21 + 2A22 22 + 2A23 23

− 2i1 2 3 1 2 3 ei(σ T1 +θ1 +θ2 +θ3 ) = 0 (4.33.12)
302 4 Forced Oscillations of Systems Having a Single Degree of Freedom

The prime represents the derivative with respect to T1 . Substituting (4.33.7) into
the above equation, we get

1 1
a + iaβ − a + a3 + 21 21 a + 22 22 a + 23 23 a
2 8
− 21 2 3 1 2 3 ei(σ T1 −β+θ1 +θ2 +θ3 ) = 0 (4.33.13)

Let

k = 21 2 3 1 2 3 , γ = σ T1 − β + θ1 + θ2 + θ3 (4.33.14)

then

1 1 3
a − a + a3 + 2n 2n + iaβ − keiγ = 0 (4.33.15)
2 8 n=1

Separating the real and imaginary parts of the above equation yields
,
1 2 2 1 2
3
a = − − a a + k cos γ (4.33.16)
2 n=1 n n 8

aβ = k sin γ (4.33.17)

(b) Let a = 0and β = σ in (4.33.16) and (4.33.17), we obtain

,
1 2 2 1 2
3
− − − a a = k cos γ (4.33.18)
2 n=1 n n 8

aσ = k sin γ (4.33.19)

The two equations are squared and added together to obtain the frequency–
response equation as
,2
1 2 2 1 2
3
− − a a2 + σ 2 a2 = k 2 (4.33.20)
2 n=1 n n 8

The solution of (4.33.20) is

2 1/2
k2 1 3 1
σ =± 2 − − 2n 2n − a2 (4.33.21)
a 2 n=1 8
4.33 Exercise 4.33 (Combination Resonance of Self-Excited Systems) 303

Fig. 4.20 Frequency-response and frequency-phase characteristics of the steady state solution
(K1 = 5, K2 = 5, K3 = 2, 1 = 0.2, 2 = 0.3, 3 = 0.5) for Exercise 4.33

From this, the frequency–response curve of the system can be plotted as shown
in Fig. 4.20.
In order to analyze the stability of steady state solutions, we can rewrite (4.31.17)
and (4.31.18) as
,
1 2 2 1 2
3
a = − − a a + k cos γ (4.33.22)
2 n=2 n n 8

a σ − γ = k sin γ (4.33.23)

Around any set of steady state solution (a0 , γ0 ), we let

a = a0 + ã, γ = γ0 + γ̃ (4.33.24)

The linearized form of (4.33.22) and (4.33.23) are

,
1 2 2 3 2
3
ã = − − a ã − (k sin γ0 )γ̃ (4.33.25)
2 n=1 n n 8 0

σ 1
γ̃ = ã − (k cos γ0 )γ̃ (4.33.26)
a0 a0

Write the above two equations in matrix form as

% %
ã l11 l12 ã
= (4.33.27)
γ̃ l21 l22 γ̃

where
304 4 Forced Oscillations of Systems Having a Single Degree of Freedom

1 2 2 3 2
3
l11 = − − a , l12 = −k sin γ0
2 n=1 n n 8 0

σ 1
l21 = , l22 = − k cos γ0
a0 a0

The eigenvalues of Eq. (4.31.31) are

)
(l11 + l22 ) ± (l11 + l22 )2 − 4(l11 l22 − l12 l21 )
λ= (4.33.28)
2
The stability of the corresponding steady-state solution can be determined from
the sign of the real part of the eigenvalue. It follows that branch 1 in Fig. 4.20 is
unstable.

4.34 Exercise 4.34 (Self-Excited System with Simultaneous

Primary, Superharmonic, and Subharmonic
Resonances)

Solution: (a) We assume

K1 = 2εk1 , K2,3 = O(1) (4.34.1)

and let the solution of the equation be

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.34.2)

Substituting (4.34.1) and (4.34.2) into the original equation and retaining to O(ε),
we get

1 3
0 = ü + u − ε u̇ − u̇3 − Kn cos(n t + θn )
3 n=1

= D02 + 2εD0 D1 (x0 + εx1 ) + (x0 + εx1 )
1
− ε(D0 + εD1 )(x0 + εx1 ) + ε[(D0 + εD1 )(x0 + εx1 )]3
3
3
− 2εk1 cos(1 T0 + θ1 ) − Kn cos(n T0 + θn ) + · · ·
n=2
1
= D02 x0 + εD02 x1 + 2εD0 D1 x0 + x0 + εx1 − εD0 x0 + ε(D0 x0 )3
3
4.34 Exercise 4.34 (Self-Excited System with Simultaneous Primary … 305

3
− 2εk1 cos(1 T0 + θ1 ) − Kn cos(n T0 + θn ) + · · ·
n=2

3
= D02 x0 + x0 − Kn cos(n T0 + θn )
n=2
1
+ ε[D02 x1 + x1 + 2D0 D1 x0 − D0 x0 + D0 x0 )3 − 2k1 cos(1 T0 + θ1 ) + · · ·
3
(4.34.3)

Let the coefficient of the like power of ε be zero, we get

3
D02 x0 + x0 = Kn cos(n T0 + θn ) (4.34.4)
n=2

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − D0 x0 )3 + 2k1 cos(1 T0 + θ1 (4.34.5)
3
The solution of (4.34.4) is

3
x0 = AeiT0 + n eiθn ein T0 + cc (4.34.6)
n=2

where A = A(T1 ) and

Kn
n = (4.34.7)
2 1 − 2n

Let
1 iβ
A= ae (4.34.8)
2
Therefore, the first order approximate solution of the equation is

3
u = a(T1 ) cos[T0 + β(T1 )] + 2 n cos(n T0 + θn ) + O(ε) (4.34.9)
i=2

Substituting (4.34.6) into (4.34.5) and taking 1 ≈ 1, 2 ≈ 1/3 and 3 ≈ 3 into

account, we get

1
D02 x1 + x1 = −2D0 D1 x0 + D0 x0 − (D0 x0 )3
3
1
= −(2iD1 A − iA)eiT0 − (3iA2 AeiT0 + 6iA22 22 eiT0 + 6iA23 23 eiT0
3
306 4 Forced Oscillations of Systems Having a Single Degree of Freedom

2
− 3iA 3 3 eiθ3 ei(3 −2)T0 − i32 32 ei3θ2 ei32 T0 ) + k1 ei(1 T0 +θ1 )
+ cc + NST (4.34.10)

Let

εσ1 = 1 − 1, εσ2 = 32 − 1, εσ3 = 3 − 3 (4.34.11)

Equation (4.34.10) becomes

D02 x1 + x1 = −(2iD1 A − iA + iA2 Ā + 2iA22 22 + 2iA23 23

1
− iĀ 3 3 ei(σ3 T1 +θ3 ) − i32 32 ei(σ2 T1 +3θ2 )
2
3
− k1 ei(σ1 T1 +θ1 ) )eiT0 + cc + NST (4.34.12)

In order to eliminate secular terms from the above equation, there must be
2
2A − A + A2 A + 222 22 A + 223 23 A − A 3 3 ei(σ3 T1 +θ3 )
1
− 32 32 ei(σ2 T1 +3θ2 ) + ik1 ei(σ1 T1 +θ1 ) = 0 (4.34.13)
3
The prime represents the derivative with respect to T1 . Substituting (4.34.8) into
the above equation, we get

1 1 1
a + iaβ − a + a3 + 22 22 a + 23 23 a − a2 3 3 ei(σ3 T1 −3β+θ3 )
2 8 4
1 3 3 i(σ2 T1 −β+3θ2 ) i(σ1 T1 −β+θ1 )
− 2 2 e + ik1 e =0 (4.34.14)
3
Let
1 3 3 1
k2 = , k3 = 3 3 , γ1 = σ1 T1 − β + θ1 ,
3 2 2 4
γ2 = σ2 T1 − β + 3θ2 , γ2 = σ3 T1 − 3β + θ3 (4.34.15)

then

1 1 3
a − a + a3 + 2n 2n + iaβ + ik1 eiγ1 − k2 eiγ2 − k3 a2 eiγ3 = 0 (4.34.16)
2 8 n=2

Separating the real and imaginary parts of the above equation yields
,
1 2 2 1 2
3
a = − − a a + k1 sin γ1 + k2 cos γ2 + k3 a2 cos γ3 (4.34.17)
2 n=2 n n 8
4.36 Exercise 4.36 (Simple Harmonic Response of Van Der Pol’s Oscillator) 307

aβ = −k1 cos γ1 + k2 sin γ2 + k3 a2 sin γ3 (4.34.18)

(b) It can be found from (4.34.17) and (4.34.18) that the system has a steady state
solution when and only when γ1 , γ2 and γ3 are constants, hence

1
β = σ1 = σ2 = σ3 (4.34.19)
3

4.35 Exercise 4.35 (Self-Excited System with Simultaneous

Primary, and Combinatorial Resonances)

Readers are invited to solve this problem by referring to Exercise 4.33 and Exercise
4.34.

4.36 Exercise 4.36 (Simple Harmonic Response of Van Der

Pol’s Oscillator)

Solution: (a) Let the solution to the equation be

u(t; ε) = u0 (T0 , T1 ) + εu1 (T0 , T1 ) + · · ·

(4.36.1)
z(t; ε) = z0 (T0 , T1 ) + εz1 (T0 , T1 ) + · · ·

Substituting (4.36.1) into the original equation and retaining to O(ε) yields

0 = ü + ω02 u − 2ε[(1 − z)u̇ − żu] + 2K sin t

= D02 + 2εD0 D1 (u0 + εu1 ) + ω02 (u0 + εu1 ) − 2ε(D0 + εD1 )(u0 + εu1 )
+ 2ε(z0 + εz1 )[(D0 + εD1 )(u0 + εu1 )] + · · ·
+ 2ε(u0 + εu1 )[(D0 + εD1 )(z0 + εz1 )] + 2K sin T0 + · · ·
= D02 u0 + εD02 u1 + 2εD0 D1 u0 + ω02 u0 + εω02 u1 − 2εD0 u0
+ 2εz0 D0 u0 + 2εu0 D0 z0 + 2K sin T0 + · · ·
= D02 u0 + ω02 u0 + ε(D02 u1 + ω02 u1 + 2D0 D1 u0 − 2D0 u0
+ 2z0 D0 u0 + 2u0 D0 z0 ) + 2K sin T0 + · · · (4.36.2)
308 4 Forced Oscillations of Systems Having a Single Degree of Freedom

0 = τ ż + z − u2
= τ (D0 + εD1 )(z0 + εz1 ) + (z0 + εz1 ) − (u0 + εu1 )2 + · · ·
(4.36.3)
= τ D0 z0 + τ εD1 z0 + τ εD0 z1 + z0 + εz1 − u02 − 2εu0 u1 + · · ·
= τ D0 z0 + z0 − u02 + ε(τ D0 z1 + z1 + τ D1 z0 − 2u0 u1 ) + · · ·

Considering the primary resonance, we assume.

= ω0 + εσ, K = εk (4.36.4)

Substitute this into (4.36.2) and (4.36.3) and let the coefficient of the like power
of ε be zero, we can obtain

D02 u0 + ω02 u0 = 0
(4.36.5)
τ D0 z0 + z0 = u02

D02 u1 + ω02 u1 = −2D0 D1 u0 + 2D0 u0 − 2z0 D0 u0 − 2u0 D0 z0

− 2ksin(ω0 T0 + σ T1 )
1 2
D0 z1 + z1 = −D1 z0 + u0 u1 (4.36.6)
τ τ
Solving from (4.36.5) yields

u0 = Aeiω0 T0 + cc
−1 (4.36.7)
z0 = be−T0 /τ + 2AA + (4ω02 τ 2 + 1)−1/2 A2 e−itan 2ω0 τ e2iω0 T0 + cc

where b is the real constant of integration and A = A(T1 ). Let

1 iβ
A= ae (4.36.8)
2
Therefore, the first order approximate solution of the equation is

u = acos(ω0 t + β) + O(ε) (4.36.9)

z = be−t/τ + 21 a2 + 21 cos 2ω0 t + 2β − tan−1 2ω0 τ + O(ε)

Substituting (4.36.7) into (4.36.6) yields

D02 u1 + ω02 u1 = −2D0 D1 u0 + 2D0 u0 − 2z0 D0 u0 − 2u0 D0 z0 − 2k sin(ω0 T0 + σ T1 )

= −2ω0 i(D1 A − A)eiω0 T0 − iω0 [2bAe−T0 /τ
# −1

+ 4A2 A − 2 4ω02 τ 2 1)−1/2 A2 Ae−i tan 2ω0 τ eiω0 T0
2 T0
# −1

− [− be− τ A + 4iω0 4ω02 τ 2 + 1)− 2 A2 Ae−i tan 2ω0 τ eiω0 T0
1

τ
+ iω0 keiσ T1 eiω0 T0 + cc + NST
4.36 Exercise 4.36 (Simple Harmonic Response of Van Der Pol’s Oscillator) 309

= −[2ω0 (iD1 A − iA) + 2iω0 bAe−T0 /τ + 4iω0 A2 A

−1
+ 2iω0 (4ω02 τ 2 + 1)−1/2 A2 Ae−i tan 2ω0 τ
2
− be−T0 /τ A − iω0 keiσ T1 ]eiω0 T0 + cc + NST (4.36.10)
τ
In order to eliminate secular terms from the above equation, there must be
−1
2D1 A − 2A + 4A2 A + 2(4ω02 τ 2 + 1)− 2 A2 Ae−i tan 2ω0 τ
1

T0 2 T0
+ 2bAe− τ + i be− τ A − keiσ T1 = 0 (4.36.11)
τ ω0

The prime represents the derivative with respect to T1 . Substituting (4.36.8) into
the above equation and omitting the fast decay term containing e−T0 /τ , we get

1 1 −1
a + iaβ − a + a3 + (4ω02 τ 2 + 1)−1/2 a3 e−i tan 2ω0 τ − kei(σ T1 −β) = 0
2 4
(4.36.12)

Separating the real and imaginary parts of the above equation yields
−1 2
a = {1 − 41[2 + 4ω02 τ 2 + 1)−1/2 cos
−1 tan τ
2 a −1
2ω 0 }a + k cos(σ T1 − β)
−1/2
β = − 4 [− 4ω0 τ + 1)
1 2 2
sin tan 2ω0 τ a + ka sin(σ T1 − β)
(4.36.13)

Let
−1
αr = 2 + (4ω02 τ 2 + 1)−1/2 cos
tan 2ω0 τ
(4.36.14)
αi = −(4ω02 τ 2 + 1)−1/2 sin tan−1 2ω0 τ

and change the time variable T1 to εt in (4.36.13) we can obtain

ȧ = ε 1 − 41 αr a2 a + εk cos(εσ t − β)
(4.36.15)
β̇ = − 41 εαi a2 + εka−1 sin(εσ t − β)

Let

γ = εσ t − β (4.36.16)

Equation (4.36.15) becomes

ȧ = ε 1 − 14 αr a2 a + εk cos γ
(4.36.17)
γ̇ = εσ + 41 εαi a2 − εka−1 sin γ
310 4 Forced Oscillations of Systems Having a Single Degree of Freedom

For the stead-state solutions of (4.36.17), ȧ = γ̇ = 0 so that a and γ are the solutions
of

− 1 − 41 αr a2 a = k cos γ
(4.36.18)
σ a + 41 αi a3 = k sin γ

Squaring and adding two equations in (4.36.18) together yields the frequency–
response equation as

1 1
[ 1 − αr a2 a]2 + (σ a + αi a3 )2 = k 2 (4.36.19)
4 4

Solving σ in terms of a from (4.36.19)

1 2 1 2 1
σ = − αi a ± k − [ 1 − αr a2 a]2 (4.36.20)
4 a 4

To analyze the stability of steady state (a0 , γ0 ), let

a = a0 + a1 , γ = γ0 + γ1 (4.36.21)

then the linearized form of (4.36.17) is

% %
ȧ1 l11 l12 a1
=ε (4.36.22)
γ̇1 l21 l22 γ1

where
3
l11 = 1 − αr a02 , l12 = −k sin γ0
4
σ 3 1
l21 = + αi a0 , l22 = − k cos γ0
a0 4 a0

The eigenvalues of Eq. (4.36.22) are

)
λ (l11 + l22 ) ± (l11 + l22 )2 − 4(l11 l22 − l12 l21 )
= (4.36.23)
ε 2
The stability of the corresponding steady-state solution can be determined from
the sign of the real part of the eigenvalue.
(b) For the case of hard nonresonance excitations, i.e., K = O(1), − ω0 > O(ε),
let the coefficient of the like power of ε be zero in (4.36.2) and (4.36.3), we can
obtain
4.36 Exercise 4.36 (Simple Harmonic Response of Van Der Pol’s Oscillator) 311

D02 u0 + ω02 u0 = −2K sin T0

(4.36.24)
τ D0 z0 + z0 = u02

D02 u1 + ω02 u1 = −2D0 D1 u0 + 2D0 u0 − 2z0 D0 u0 − 2u0 D0 z0

(4.36.25)
D0 z1 + τ1 z1 = −D1 z0 + τ2 u0 u1

Solving (u0 , z0 ) from (4.36.24) yields

u0 = Aeiω0 T0 + iK(ω02 − 2 )−1 eiT0 + cc (4.36.26)

T0
z0 = be− τ + 2AA + 2K 2 2 (ω02 − 2 )−2
−1
+ (4ω02 τ 2 + 1)− 2 A2 e−itan 2ω0 τ 2iω0 T0
1
e
−1
− ) (4 τ + 1)− 2 e−itan
2 −2 1
−K 2 2
(ω02 2 2 2τ i2T0
e
2 −1 − 21 −itan−1 (+ω0 )τ i(+ω0 )T0
+ 2iK(ω02 − ) [( + ω0 ) τ + 1] 2 2
Ae e
−1
+ 2iK(ω02 − 2 )−1 [( − ω0 )2 τ 2 + 1]−1/2 Ae−itan (−ω0 )τ i(−ω0 )T0
e + cc
(4.36.27)

where b is the real constant of integration and A = A(T1 ). Let

1 iβ
A= ae (4.36.28)
2
Therefore, the first order approximate solution of the equation is

u = a cos(ω0 t + β) − 2K(ω02 − 2 )−1 sin t + O(ε) (4.36.29)

Substituting (4.36.26) and (4.36.27) into (4.36.25) and keeping only the terms
containing eiω0 T0 and the terms that can produce subharmonic and superharmonic
resonance excitations at ≈ 3ω0 and 3 ≈ ω0 , and omitting the fast-decaying
term containing e−T0 /τ , we get

D02 u1 + ω02 u1 = −2D0 D1 u0 + 2D0 u0 − 2D0 (u0 z0 )

= −2iω0 (D1 A − A)eiω0 T0 − 2iω0 [2AA + 2K 2 2 ω02 − 2 )−2 Aeiω0 T0
−1
− 2iω0 (4ω02 τ 2 + 1)−1/2 A2 Ae−i tan 2ω0 τ iω0 T0
e
−1
− 4iω0 K 2 2
(ω02 − ) [( + ω0 ) τ + 1]−1/2 Ae−i tan
2 −2 2 2 (+ω0 )τ iω0 T0
e
−1
− 4iω0 K 2 2 (ω02 − 2 )−2 [( − ω0 )2 τ 2 + 1]−1/2 Aei tan (−ω0 )τ iω0 T0
e
−1
2 −3 −1/2 −i tan
− 6K 3 4
(ω02 − ) (4 τ + 1) 2 2
e 2τ i3T0
e
2 −1 −1/2
+ 4K( − 2ω0 )(ω02 − ) [( − ω0 ) τ + 1] 2 2

2 −i tan−1 (−ω0 )τ i(−2ω0 )T0

A e e
312 4 Forced Oscillations of Systems Having a Single Degree of Freedom

+ 2K( − 2ω0 )(ω02 − 2 )−1 (4ω02 τ 2 + 1)−1/2

2 −1
2ω0 τ i(−2ω0 )T0
A ei tan e + cc + NST (4.36.30)

In order to eliminate secular terms from the above equation, there must be

A − A + 2K 2 2 (ω02 − 2 )−2 A
−1
+2A2 A + (4ω02 τ 2 + 1)−1/2 A2 Ae−i tan 2ω0 τ

−1 (4.36.31)
+2K 2 2 (ω02 − 2 )−2 [( + ω0 )2 τ 2 + 1]−1/2 Ae−i tan (+ω0 )τ

−1
2K 2 2 (ω02 − 2 )−2 [( − ω0 )2 τ 2 + 1]−1/2 Aei tan (−ω0 )τ
=0

The prime represents the derivative with respect to T1 . Substituting (4.36.28) into
the above equation, we get

1
a + iaβ − a + a3 + 2K 2 2 (ω02 − 2 )−2 a
2
1 −1
+ (4ω0 τ + 1)−1/2 a3 e−i tan 2ω0 τ
2 2
4
−1
+ 2K 2 2 (ω02 − 2 )−2 [( + ω0 )2 τ 2 + 1]−1/2 ae−i tan (+ω0 )τ
−1
+ 2K 2 2 (ω02 − 2 )−2 [( − ω0 )2 τ 2 + 1]−1/2 aei tan (−ω0 )τ
=0 (4.36.32)

Separating the real and imaginary parts of the above equation yields

a = {1 − 2K 2 2 (ω02 − 2 )−2 [1+[( + ω0 )2 τ 2 + 1]−1/2 cos( + ω0 )τ

+ [( − ω0 )2 τ 2 + 1]−1/2 cos( − ω0 )τ }a
1
− [2 + 4ω02 τ 2 + 1)−1/2 cos 2ω0 τ a3 (4.36.33)
4
1
β = −[− 4ω02 τ 2 + 1)−1/2 sin 2ω0 τ a2 + 2K 2 2 (ω02 − 2 )−2
. 4 /
+ ω0 )2 τ 2 + 1]−1/2 sin( + ω0 τ − − ω0 )2 τ 2 + 1]−1/2 sin( − ω0 τ
(4.36.34)

Let

ηr = 1 − 2K 2 2 (ω02 − 2 )−2 (1 + hr )
(4.36.35)
ηi = 2K 2 2 (ω02 − 2 )−2 hi

where

hr = [( + ω0 )2 τ 2 + 1]−1/2 cos( + ω0 )τ

(4.36.36)
+[( − ω0 )2 τ 2 + 1]−1/2 cos( − ω0 )τ
4.37 Exercise 4.37 (Analysis on the Response of a Cubic Nonlinear System … 313

hi = [( + ω0 )2 τ 2 + 1]−1/2 sin( + ω0 )τ

(4.36.37)
−[( − ω0 )2 τ 2 + 1]−1/2 sin( − ω0 )τ

Considering the expressions of αr and αi in (4.36.14), (4.23.4) and (4.36.34)

become

a = ηr − 41 αr a2 a
(4.36.38)
β = ηi − 41 αi a2

Change the time variable T1 to εt in (4.36.38), we can obtain

1
ȧ = ε ηr − αr a2 a
4
(4.36.39)
1
β̇ = ε ηi − αi a2
4

Separates the variables in the first equation of (4.36.39) to give

da
= εdt (4.36.40)
ηr − 41 αr a2 a

After integration, we can obtain

ηr e2εηr t
a2 = (4.36.41)
C + 41 αr e2εηr t

where C is the integration constant.

It can be seen that when ηr > 0, the amplitude a diverges exponentially; when
ηr < 0, the amplitude a decays exponentially; and when ηr = 0, the amplitude a is
zero, which means that the external excitation does not excite the free vibration.
The result obtained above is different from the result that the Exercise asks us to
prove, so the rest two problems are left unsolved.

4.37 Exercise 4.37 (Analysis on the Response of a Cubic

Nonlinear System by the Slowly Varying Excitation)

Solution: (a) Let the solution of the equation be

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.37.1)

Substituting (4.37.1) into the original equation and retaining to O(ε) yields
314 4 Forced Oscillations of Systems Having a Single Degree of Freedom

0 = ü + ω02 u + 2εμu̇ + εαu3 − ω02 f (T1 )

= D02 + 2εD0 D1 (u0 + εu1 ) + ω02 (u0 + εu1 )
+ 2εμ(D0 + εD1 )(u0 + εu1 ) + εα(u0 + εu1 )3 − ω02 f (T1 ) + · · ·
= D02 u0 + εD02 u1 + 2εD0 D1 u0 + ω02 u0 + εω02 u1
+ 2εμD0 u0 + εαu03 − ω02 f (T1 ) + · · ·
= D02 u0 + ω02 u0 − ω02 f (T1 )

+ ε D02 u1 + ω02 u1 + 2D0 D1 u0 + 2μD0 u0 + αu03 + · · · (4.37.2)

Let the coefficient of the like power of ε be zero, we get

D02 u0 + ω02 u0 = ω02 f (T1 ) (4.37.3)

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − αu03 (4.37.4)

Since f is a function of T1 , i.e., it is slowly changing with time T0 , it can be

considered to be approximately constant with respect to T0 . The solution of (4.37.3)
is

u0 = Aeiω0 T0 + cc + f (4.37.5)

where A = A(T1 ). Therefore, the first order approximate solution of the equation is

u = A(T1 )eiω0 T0 + cc + f (T1 ) + O(ε) (4.37.6)

Substituting (4.37.5) into (4.37.4) yields

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − αu03

= −2iω0 (D1 A + μA)eiω0 T0 − 3αA2 Aeiω0 T0 − 3αf 2 Aeiω0 T0
+ cc + NST (4.37.7)

In order to eliminate secular terms from the above equation, there must be

2iω0 A + μA + 3αA2 A + 3αf 2 A = 0

i.e.,

2iω0 A + μA + 3α AA + f 2 A = 0 (4.37.8)

The prime denotes the derivative with respect to T1 .

(b) Substituting A = 21 aeiβ into the above equation, we get
4.38 Exercise 4.38 (Analysis of the Response of a Self-Excited System Under … 315

3 1
iω0 a + iaβ + μa + α a2 + f 2 a = 0 (4.37.9)
2 4

Separating the real and imaginary parts of the above equation yields

a = −μa (4.37.10)
ω0 β = 23 α 41 a2 + f 2

4.38 Exercise 4.38 (Analysis of the Response

of a Self-Excited System Under Slowly Varying
Excitation)

Solution: (a) Let the solution be

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.38.1)

Substituting (4.38.1) into the original equation and retaining to O(ε) yields

0 = ü + ω02 u − ε 1 − u2 u̇ − ω02 f (T1 )

= D02 + 2εD0 D1 (u0 + εu1 )
+ ω02 (u0 + εu1 ) − ε(D0 + εD1 )(u0 + εu1 )
+ ε(u0 + εu1 )2 (D0 + εD1 )(u0 + εu1 ) − ω02 f (T1 ) + · · ·
= D02 u0 + εD02 u1 + 2εD0 D1 u0 + ω02 u0 + εω02 u1
− εD0 u0 + εu02 D0 u0 − ω02 f (T1 ) + · · ·
= D02 u0 + ω02 u0 − ω02 f (T1 )

+ ε D02 u1 + ω02 u1 + 2D0 D1 u0 − D0 u0 + u02 D0 u0 + · · · (4.38.2)

Let the coefficient of the like power of ε be zero, we get

D02 u0 + ω02 u0 = ω02 f (T1 ) (4.38.3)

D02 u1 + ω02 u1 = −2D0 D1 u0 + D0 u0 − u02 D0 u0 (4.38.4)

Since f is a function of T1 , i.e., it is slowly changing with time T0 , it can be

considered to be approximately constant with respect to T0 . The solution of (4.38.3)
is

u0 = Aeiω0 T0 + cc + f (4.38.5)
316 4 Forced Oscillations of Systems Having a Single Degree of Freedom

where A = A(T1 ). Therefore, the first order approximate solution of the equation is

u = A(T1 )eiω0 T0 + cc + f (T1 ) + O(ε) (4.38.6)

Substituting (4.38.5) into (4.38.4) yields

D02 u1 + ω02 u1 = −2D0 D1 u0 + D0 u0 − u02 D0 u0

= −iω0 (2D1 A − A)eiω0 T0

− iω0 f 2 A + A2 A eiω0 T0 + cc + NST (4.38.7)

In order to eliminate secular terms from the above equation, there must be

2A = 1 − AA − f 2 A (4.38.8)

The prime denotes the derivative with respect to T1 .

(b) Substituting A = 21 aeiβ into the above equation, we get

1 1
a + iaβ = 1 − a2 − f 2 a (4.38.9)
2 4

Separating the real and imaginary parts of the above equation yields

a = 21 1 − f 2 − 41 a2 a
(4.38.10)
β = 0

4.39 Exercise 4.39 (Analysis of the Response

of a Self-Excited Vibrating System Under Hard
Excitation at Different Frequencies)

Solution: Let the solution be

u(t; ε) = x0 (T0 , T1 ) + εx1 (T0 , T1 ) + · · · (4.39.1)

Substituting (4.39.1) into the original equation and retaining to O(ε) yields

0 = ü + ω02 u − ε 1 − u2 u̇ − K cos t

= D02 + 2εD0 D1 (u0 + εu1 )
+ ω02 (u0 + εu1 ) − ε(D0 + εD1 )(u0 + εu1 )
+ ε(u0 + εu1 )2 (D0 + εD1 )(u0 + εu1 ) − K cos T0 + · · ·
4.39 Exercise 4.39 (Analysis of the Response of a Self-Excited Vibrating … 317

= D02 u0 + ω02 u0 − K cos T0

+ ε D02 u1 + ω02 u1 + 2D0 D1 u0 − D0 u0 + u02 D0 u0 + · · · (4.39.2)

Let the coefficient of the like power of ε be zero, we get

D02 u0 + ω02 u0 = K cos T0 (4.39.3)

D02 u1 + ω02 u1 = −2D0 D1 u0 + D0 u0 − u02 D0 u0 (4.39.4)

The solution of (4.39.3) is

u0 = A(T1 )eiω0 T0 + eiT0 + cc (4.39.5)

where A = A(T1 ), = 21 K(ω02 − 2 )−1 . Therefore, the first order approximate
solution of the equation is

u = A(T1 )eiω0 T0 + eiT0 + cc + O(ε) (4.39.6)

Substituting (4.39.5) into (4.39.4) yields

D02 u1 + ω02 u1 = −2D0 D1 u0 + D0 u0 − u02 D0 u0

= −iω0 (2D1 A − A)eiω0 T0 − iω0 A2 Aeiω0 T0 − 2iω0 2 Aeiω0 T0
2
− i3 ei3T0 − i( − 2ω0 )A ei(−2ω0 )T0
− i(ω0 − 2)2 Aei(ω0 −2)T0 − i(2 + ω0 )2 Aei(2+ω0 )T0
+ cc + NST (4.39.7)

(a) For the case of is not close to 0, 3ω0 and 13 ω0 :

In order to eliminate secular terms from the above equation, there must be

2A = A − 22 A − A2 A (4.39.8)

The prime denotes the derivative with respect to T1 .

(b) For the case of ω0 = 3 + εσ :

In order to eliminate secular terms from (4.39.7), there must be

2A − A + A2 A + 22 A + ω0−1 3 e−iσ T1 = 0

318 4 Forced Oscillations of Systems Having a Single Degree of Freedom

i.e.,

2A = A − 22 A − A2 A − ω0−1 3 e−iσ T1 (4.39.9)

(c) For the case of = 3ω0 + εσ :

In order to eliminate secular terms from (4.39.7), there must be

2
2A − A + A2 A + 22 A + ω0−1 ( − 2ω0 )A eiσ T1 = 0

i.e.,
2
2A = A − 22 A − A2 A + 2 − ω0−1 A eiσ T1 (4.39.10)

(d) For the case of = 3ω0 + εσ :

In order to eliminate secular terms from (4.39.7), there must be

2A − A + A2 A + 22 A + ω0−1 (ω0 − 2εσ )2 Ae−i2σ T1

+ω0−1 (2εσ + ω0 )2 Aei2σ T1 = 0

i.e.,

2A = A − 22 A − A2 A − 22 A cos 2σ T1 (4.39.11)

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License ([Link] which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Chapter 5
Parametrically Excited Systems

5.1 Exercise 5.1 (The Method of Strained Parameters

to Determine the Transition Curves for a Hill Equation)

Solution: Based on the Floquet theory, that the characteristic exponent is 0 or i

(i.e., the solutions have periods of 2π or 4 π ) and then determines the values of the
parameters for which the assumption is true. We seek the solutions of the original
equation having periods of 2π and 4 π and the equations for the transition curves
a = a(ε) in the form of the following perturbation expansions:

u(x; ε) = u0 (x) + εu1 (x) + ε2 u2 (x) + · · · (5.1.1)

a = a0 + εa1 + ε2 a2 + · · · (5.1.2)

In addition,

(1 − εcosx)−1 = 1 + εcosx + ε2 cos2 x

(1 − εcosx)−2 = 1 + 2εcosx + 3ε2 cos2 x
(a0 + εa1 + ε2 a2 )−1 = a0−1 (1 + εa0−1 a1 + ε2 a0−1 a2 )−1

= a0−1 [1 − εa0−1 a1 + ε2 a0−1 a2 + εa0−1 a1 + ε2 a0−1 a2 )2
= a0−1 − εa0−2 a1 − ε2 a0−2 a2 + ε2 a0−3 a12
(a0 + εa1 + ε2 a2 )−2 = a0−2 − 2εa0−3 a1 − 2ε2 a0−3 a2 + 3ε2 a0−4 a12 (5.1.3)

Substituting (5.1.1) and (5.1.2) into the original equation and retaining to O(ε2 ),
we obtain

© The Author(s) 2025 319

Z. He et al., Solved Problems in Nonlinear Oscillations,
[Link]
320 5 Parametrically Excited Systems

1
0 = u + (1 − ε cos x)−2 a−2 2(1 − ε cos x) 2 − εa2 cos x + ε2 a2 sin2 x u
4
1 1
= u + (1 − ε cos x)−1 a−2 − ε cos x u + ε2 (1 − ε cos x)−2 u sin2 x

2 4
2

= u0 + εu1 + ε u2 + 1 + ε cos x + ε cos x 2 2

1
× a0−2 − 2εa0−3 a1 − 2ε2 a0−3 a2 + 3ε2 a0−4 a12 − ε cos x u0 + εu1 + ε2 u2
2
1 2
+ ε 1 + 2ε cos x + 3ε2 cos2 x u0 + εu1 + ε2 u2 sin2 x + · · ·
4
1
+ ε2 1 + 2ε cos x + 3ε2 cos2 x u0 + εu1 + ε2 u2 sin2 x + · · ·
4
+ ε2 (u2 + a0−2 u2 − 2a0−3 a2 u0 + 3a0−4 a12 u0 − 2a0−3 a1 u0 cos x
+ ε2 (u2 + a0−2 u2 − 2a0−3 a2 u0 + 3a0−4 a12 u0 − 2a0−3 a1 u0 cos x
1
− 2a0−3 a1 u1 − u1 cos x + a0−2 u1 cos x) + · · · (5.1.4)
2
Equating coefficients of like powers of ε in the above equation yields

u0 + a0−2 u0 = 0 (5.1.5)

1
u0 + a0−2 u1 = −2a0−3 a1 u0 + u0 Cos x − a0−2 u0 Cos x (5.1.6)
2

u2 + a0−2 u2 = 2a0−3 a2 u0 − 3a0−4 a12 u0 + 2a0−3 a1 u0 cos x

1 1
+ u0 cos2 x − a0−2 u0 cos2 x − u0 sin2 x
2 4
−3 1 −2
+ 2a0 a1 u1 + u1 cos x − a0 u1 cos x (5.1.7)
2
The solution of (5.1.5) is

u0 = b cos a0−1 x + c sin a0−1 x (5.1.8)

CASE I: a0 = 2

u0 = b cos(x/2) + c sin(x/2) (5.1.9)

u0 is a periodic solution with period 2T = 4π . Substituting (5.1.9) into (5.1.6) yields

5.1 Exercise 5.1 (The Method of Strained Parameters to Determine … 321

1 1 1 1
u1 + u1 = − a1 b cos x + c sin x
4 4 2 2

1 1 1
+ b cos x + c sin x cos x
4 2 2
1 1 1 1
= − a1 b cos x − a1 c sin x
4 2 4 2
1 1 1 1
+ b cos x − c sin x + NST
8 2 8 2
1 1 1 1 1 1
= − b a1 − cos x − c a1 + sin x + NST (5.1.10)
4 2 2 4 2 2

In order to eliminate secular terms from the above equation, there must be

1 1
a1 = or a1 = − (5.1.11)
2 2
Therefore, the transition curves emanating from a = 2 is given by

1
a = 2 ± ε + O ε2 (5.1.12)
2

CASE II: a0 = 1

u0 = b cos x + c sin x (5.1.13)

u0 is a periodic solution with period T = 2π . Substituting (5.1.9) into (5.1.6) yields

1
u1 + u1 = −2a1 (bcosx + csinx) − (bcosx + csinx)cosx
2
1 1 1
= −2ba1 cosx − 2ca1 sinx − b − bcos2x − csin2x (5.1.14)
4 4 4
In order to eliminate secular terms from the above equation, there must be

a1 = 0

Then the solution of (5.1.14) is

1 1 1
u1 = − b + b cos 2x + c sin 2x (5.1.15)
4 12 12
Substituting (5.1.13) and (5.1.15) into (5.1.7) yields
322 5 Parametrically Excited Systems

1
u2 + u2 = 2a2 (b cos x + c sin x) − (b cos x + c sin x)
4
1
− (b cos x + c sin x) cos x2
4
1 1 1 1
− − b + b cos 2x + c sin 2x cos x
2 4 12 12

1 1
= 2a2 − b cos x + 2a2 − c sin x + NST (5.1.16)
3 3

In order to eliminate secular terms from the above equation, there must be

a2 = 1/6 (5.1.17)

Therefore, the transition curves emanating from a = 1 is given by

1
a = 1 + ε2 + O ε3 (5.1.18)
6

5.2 Exercise 5.2 (The Method of Strained Parameters

to Determine the Transition Curves of the Hill’s
Equation for LRC Circuits with Sinusoidally Varying
Resistance)

Solution: (a) Since

Ṙ = R0 α cos t

t
t
İ = w ddtτ exp − 21 L−1
R(ξ )d ξ − 1
2
wL−1 Rexp − 21 L−1 R(ξ )d ξ
0 0
t
= 21 w − 21 wL−1 R exp − 21 L−1 R(ξ )d ξ
0
1

Ï = [ 4 w − 2
− 1
4
w L−1 R 1
2
wL−1 Ṙ
t
− 21 w − 21 wL−1 R 21 L−1 R exp − 21 L−1 R(ξ )d ξ ]
0
= 41 2 w − 21 w L−1 R + 41 wL−2 R2 − 21 wL−1 R0 α cos t
t
× exp − 21 L−1 R(ξ )d ξ
0

Substituting the above results into the original equation

5.2 Exercise 5.2 (The Method of Strained Parameters to Determine … 323

1 1 1 1
L 2 w − L−1 Rw + L−2 R2 w − L−1 R0 αw cos t
4 2 4 2

1 1
+ R w − L−1 Rw + R0 α cos t + C −1 w = 0
2 2

i.e.,

1
w + [4−2 L−1 C −1 − −2 L−2 R20 − −2 L−2 R20 α 2
2
− 2−2 L−2 R20 α sin t 2−1 L−1 R0 α cos t
1
+ −2 L−2 R20 α 2 cos 2t] w = 0 (5.2.1)
2
Let

2αR0 4 1 R20 1 2
ε= , δ = − 1 + α (5.2.2)
L 2 LC 4L2 2

Considering t = 2τ , , (5.2.1) becomes

1 1
w + δ + ε cos 2τ − ε2 α −1 sin 2τ + ε2 cos 4τ w = 0 (5.2.3)
2 8

(b) (5.2.3) is a Hill equation with the period of the coefficients T = π . We use the
method of strained parameters to solve this problem. We seek the solutions of the
original equation having periods of π and 2 π and the equations for the transition
curves δ = δ(ε) in the form of the following perturbation expansions:

w(τ ; ε) = u0 (τ ) + εu1 (τ ) + ε2 u2 (τ ) + · · · (5.2.4)

δ = δ0 + εδ1 + ε2 δ2 + · · · (5.2.5)

Substituting the above two equations into (5.2.3) and keeping to O(ε2 ) terms, we
get
324 5 Parametrically Excited Systems

1 1
0 = w + δ + ε cos 2τ − ε2 α −1 sin 2τ + ε2 cos 4τ w
2 8
= u0 + εu1 + ε2 u2 + [δ0 + εδ1 + ε2 δ2 + ε cos 2τ
1 1
− ε2 α −1 sin 2τ + ε2 cos 4τ ] u0 + εu1 + ε2 u2 + · · ·
2 8
= u0 + δ0 u0 + ε u1 + δ0 u1 + δ1 u0 + u0 cos 2τ
1 1
+ε2 (u2 + δ0 u2 + δ2 u0 − α −1 u0 sin 2τ + u0 cos 4τ
2 8
+δ1 u1 + u1 cos 2τ ) + · · ·

Equating coefficients of like powers of ε in the above equation yields

u0 + δ0 u0 = 0 (5.2.6)

u1 + δ0 u1 = −δ1 u0 − u0 cos 2τ (5.2.7)

1 1
u2 + δ0 u2 = −δ2 u0 + α −1 u0 sin 2τ − u0 cos 4τ − δ1 u1 − u1 cos 2τ (5.2.8)
2 8
The solution of (5.2.6) is

u0 = a cos δ0 τ + b sin δ0 τ (5.2.9)

CASE I: δ 0 = 1

u0 = a cos τ + b sin τ (5.2.10)

u0 is a periodic solution with period 2T = 2π . Substituting (5.2.10) into (5.2.7)

yields

u 1 +
δ0 u1 = −δ1 (acosτ
+ bsinτ )− (acosτ + bsinτ )cos2τ (5.2.11)
= − δ1 + 21 acosτ − δ1 sinτ − 21 bsinτ + NST

In order to eliminate secular terms from the above equation, there must be

1 1
δ1 = or δ1 = − (5.2.12)
2 2
Therefore, the transition curves emanating from δ = 1 is

1
δ = 1 ± ε + O ε2 (5.2.13)
2
5.3 Exercise 5.3 (Stability Analysis of a Supported Movable Pendulum) 325

CASE II: δ 0 = 4

u0 = a cos 2τ + b sin 2τ (5.2.14)

u0 is a periodic solution with period T = π . Substituting (5.2.14) into (5.2.7) yields

u 1 + δ0 u1 = −δ1 (acos2τ + bsin2τ )

− (acos2τ + bsin2τ )cos2τ
= −δ1 acos2τ − δ1 bsin2τ + NST (5.2.10)

In order to eliminate secular terms from the above equation, there must be

δ1 = 0 (5.2.11)

Therefore, the transition curves emanating from δ = 4 is

δ = 4 + O ε3 (5.2.12)

5.3 Exercise 5.3 (Stability Analysis of a Supported Movable

Pendulum)

Solution: (a) When Y ≡ 0, the differential equation of motion of the pendulum

becomes
g
θ̈ + sin θ = 0 (5.3.1)
l
Its equilibrium position is given by sinθ = 0, i.e., θ = nπ , n is an integer.
The linearized form of (5.3.1) near the equilibrium position θ = 0 is
g
θ̈ + θ =0 (5.3.2)
l

Its eigenvalues are λ = ±i g/l, so the equilibrium position θ = 0 is the center.

The linearized form of (5.3.1) near the equilibrium position θ = π is
g
θ̈ − θ =0 (5.3.3)
l

Its eigenvalues are λ = ± g/l, so the equilibrium position θ = π is the saddle

point.
326 5 Parametrically Excited Systems

(b) When Y = εg cos t, the differential equation of motion of the pendulum
becomes

θ̈ + δ(1 − ε cos t) sin θ = 0, δ = g/l (5.3.4)

The linearized form of (5.3.4) near the equilibrium position θ = 0 is

θ̈ + δ(1 − ε cos t)θ = 0, δ = g/l (5.3.5)

This is a Hill equation with a period T = 2π/ . According to Floquet theory, the
transition curves determine the parameters which make the solution of this equation
periodic (with a period of T or 2T ). We use the method of strained parameters
to determine the transition curves for this problem. We seek the solutions of the
original equation having periods of T and 2 T and the equations for the transition
curves δ = δ(ε) in the form of the following perturbation expansions:

θ (t; ε) = u0 (t) + εu1 (t) + ε2 u2 (t) + · · · (5.3.6)

δ = δ0 + εδ1 + ε2 δ2 + · · · (5.3.7)

Substituting the above two equations into (5.3.4) and retaining to O(ε2 ), we get

0 = θ̈ + δ(1 − ε cos t)

θ = ü0 + εü1 + ε2 ü2

+ δ0 + εδ1 + ε2 δ2 (1 − ε cos t) u0 + εu1 + ε2 u2 + · · ·
= ü0 + εü1 + ε2 ü2

+ δ0 + εδ1 − εδ0 cos t + ε2 δ2 − ε2 δ1 cos t u0 + εu1 + ε2 u2 + · · ·
= ü0 + δ0 u0 + ε(ü1 + δ0 u1 + δ1 u0 − δ0 u0 cos t)
+ ε2 (ü2 + δ0 u2 + δ2 u0 − δ1 u0 cos t + δ1 u1 − δ0 u1 cos t) + · · ·

Equating coefficients of like powers of ε in the above equation yields

ü0 + δ0 u0 = 0 (5.3.8)

ü1 + δ0 u1 = −δ1 u0 + δ0 u0 cos t (5.3.9)

ü2 + δ0 u2 = −δ2 u0 + δ1 u0 cos t − δ1 u1 + δ0 u1 cos t (5.3.10)

The solution of (5.3.8) is

u0 = a cos δ0 t + b sin δ0 t (5.3.11)

5.3 Exercise 5.3 (Stability Analysis of a Supported Movable Pendulum) 327

CASE I: δ0 = 2 :

u0 = a cos t + b sin t (5.3.12)

u0 is a periodic solution with period T = 2π/ . Substituting (5.3.12) into (5.3.9)

yields

ü1 + 2 u1 = −δ1 u0 + δ0 u0 cos t

= −δ1 (a cos t + b sin t)
− 2 (a cos t + b sin t) cos t
= −aδ1 cos t − bδ1 sin t
1 1
− 2 a(1 + cos 2t) − 2 b sin 2t (5.3.13)
2 2
In order to eliminate secular terms from the above equation, there must be

δ1 = 0 (5.3.14)

Then the solution of (5.3.13) is

1 1 1
u1 = − a + a cos 2t + b sin 2t (5.3.15)
2 6 6
Substituting (5.3.15) into (5.3.10), we get

ü2 + 2 u2 = −δ2 a cos t − δ2 b sin t

1 1 1
+ δ0 − a + a cos 2t + b sin 2t cos t
2 6 6

5 2
= − δ2 + a cos t
12

1 2
− δ2 − b sin t + NST (5.3.16)
12

In order to eliminate secular terms from the above equation, there must be

5 2 1 2
δ2 = − or δ2 = (5.3.17)
12 12

Therefore, the transition curves emanating from δ = 2 is

g 5 2
δ = = 1 − ε + O ε3
2
l 12
328 5 Parametrically Excited Systems

i.e.,

g 1
= 2 1 + ε2 + O ε3 (5.3.18)
l 12

CASE II: δ0 = 41 2 :

1 1
u0 = a cos t + b sin t (5.3.19)
2 2
u0 is a periodic solution with period 2T = 4π/ . Substituting (5.3.11) into (5.3.9)
yields

1
ü1 + 2 u1 = −δ1 u0 + δ0 u0 cos t
4
1 1
= −δ1 a cos t + b sin t
2 2

1 1 1
− 2 a cos t + b sin t cos t
4 2 2

1 1
= − δ1 + 2 a cos t
8 2

1 1
− δ1 − 2 b sin t + NST (5.3.20)
8 2

In order to eliminate secular terms from the above equation, there must be

1
δ1 = ± 2 (5.3.21)
8

Therefore, the transition curves emanating from δ = 41 2 is

g 1
δ = = 1 ± ε + O ε2
2
(5.3.22)
l 8

(c) When Y = αg cos t, the differential equation of motion of the pendulum
becomes
g
θ̈ + (1 − α cos t) sin θ = 0 (5.3.23)
l
Let θ = π + u, the linearized equation near the equilibrium solution θ = π is

ü + δ(1 − α cos t)u = 0, δ = −g/l (5.3.24)

5.4 Exercise 5.4 (Analyze the Linear and Nonlinear Stability of a Rotating … 329

This is a Hill equation with a period of T = 2π/ and has exactly the same form
as (5.3.5). Therefore, the transition curves emanating from δ = 2 is

g 5 2
− = 1 − ε + O ε3
2
(5.3.25)
l 12

i.e.,

g 1
− = 2 1 + ε2 + O ε3 (5.3.26)
l 12

and the transition curves emanating from δ = 41 2 is

g 1
− = 1 ± ε + O ε2
2
(5.3.27)
l 8

5.4 Exercise 5.4 (Analyze the Linear and Nonlinear

Stability of a Rotating Pendulum Using the Method
of Strained Parameters and the Method of Multiple
Scales)

Solution: (a) The kinetic energy of the system is

1 2 2
T= m l θ̇ + l 2 2 sin2 θ (5.4.1)
2
The potential energy of the system is

V = mgl(1 − cos θ ) (5.4.2)

Substituting (5.4.2) and (5.4.1) into the Lagrange’s equation, we can obtain the
differential equation of motion of the simple pendulum attached to rotating base

g 1
θ̈ + sin θ − 2 sin 2θ = 0 (5.4.3)
l 2

(b) Let = 0 (1 + ε cos ωt), where ε << 1, , then (5.4.3) becomes

g 1
θ̈ + sin θ − 20 (1 + ε cos ωt)2 sin 2θ = 0 (5.4.4)
l 2
330 5 Parametrically Excited Systems

Linearizing the above equation near θ = 0 yields

g
θ̈ + θ − 20 (1 + ε cos ωt)2 θ = 0 (5.4.5)
l
i.e.,
g
θ̈ + − 20 − 220 ε cos ωt − 20 ε2 cos2 ωt θ = 0 (5.4.6)
l

(c) Denote δ = gl − 20 . We use the method of strained parameters to determine

the transition curves for this problem. We seek the solutions of the original equation
having periods of T and 2 T and the equations for the transition curves δ = δ(ε) in
the form of the following perturbation expansions:

θ (t; ε) = u0 (t) + εu1 (t) + ε2 u2 (t) + · · · (5.4.7)

δ = δ0 + εδ1 + ε2 δ2 + · · · (5.4.8)

Substituting the above two equations into (5.4.6) and retaining to O(ε2 ), we get

0 = θ̈ + δ − ε220 cos ωt − ε2 20 cos2 ωt θ
= ü0 + εü1 + ε2 ü2

+ δ0 + εδ1 − 2ε20 cos ωt + ε2 δ2 − ε2 20 cos2 ωt

× u0 + εu1 + ε2 u2 + · · ·

= ü0 + δ0 u0 + ε ü1 + δ0 u1 + δ1 u0 − 220 u0 cos ωt
+ ε2 (ü2 + δ0 u2 + δ2 u0 − 20 u0 cos2 ωt
+ δ1 u1 − 220 u1 cos ωt) + · · · s

Equating coefficients of like powers of ε in the above equation yields

ü0 + δ0 u0 = 0 (5.4.9)

ü1 + δ0 u1 = −δ1 u0 + 220 u0 cos ωt (5.4.10)

ü2 + δ0 u2 = −δ2 u0 + 20 u0 cos2 ωt

− δ1 u1 + 220 u1 cos ωt (5.4.11)

The solution of (5.4.9) is

u0 = a cos δ0 t + b sin δ0 t (5.4.12)

5.4 Exercise 5.4 (Analyze the Linear and Nonlinear Stability of a Rotating … 331

CASE I: δ0 = 0:

u0 = a (5.4.13)

u0 is a constant, which is the periodic solution of any period. Substituting (5.4.13)

into (5.4.10) yields

ü1 = −δ1 a + 220 a cos ωt (5.4.14)

In order to eliminate secular terms from the above equation, there must be

δ1 = 0 (5.4.15)

Then the solution of (5.4.14) is

220
u1 = − a cos ωt (5.4.16)
ω2
Substituting (5.4.16) and (5.4.13) into (5.4.11), we get

420
ü2 = −δ2 a + 20 1 − ω2
acos2 ωt
20
(5.4.17)
= −δ2 a + 220 1
4
− ω2 (1 + cos2ωt)a

In order to eliminate secular terms from the above equation, there must be

1 20
δ2 = 220 − 2 (5.4.18)
4 ω

Therefore, the transition curves emanating from δ = 0 is

g 1 20 2
δ= − 20 = 220 − 2 ε + O ε3
l 4 ω

i.e.,

g 1 20 2
= 20 1 + 2 − 2 ε + O ε3 (5.4.19)
l 4 ω

CASE II: δ0 = 41 ω2 :

1 1
u0 = a cos ωt + b sin ωt (5.4.20)
2 2
332 5 Parametrically Excited Systems

u0 is a periodic solution with period 2T = 4π/ω. Substituting (5.4.20) into (5.4.10)

yields

1 2 1 1
ü1 + ω u1 = −δ1 a cos ωt + b sin ωt
4 2 2

1 1
+ 20 a cos ωt + b sin ωt cos ωt
2
2 2
1
= − δ1 − 20 a cos ωt
2
1
− δ1 + 20 b sin ωt + NST (5.4.21)
2
In order to eliminate secular terms from the above equation, there must be

δ1 = ±20 (5.4.22)

Therefore, the transition curves emanating from δ = 41 ω2 is

g 1
δ= − 20 = ω2 ± 20 ε + O ε2
l 4
i.e.,

g 1
= 20 + ω2 ± 20 ε + O ε2 (5.4.23)
l 4

CASE III: δ0 = ω2 :

u0 = a cos ωt + b sin ωt (5.4.24)

u0 is a periodic solution with period 2T = 4π/ω. Substituting (5.4.24) into (5.4.10)

yields

ü1 + ω2 u1 = −δ1 (a cos ωt + b sin ωt)

+ 220 (a cos ωt + b sin ωt) cos ωt
= −δ1 a cos ωt − δ1 b sin ωt
+ 20 a(1 + cos 2ωt) + 20 b sin 2ω (5.4.25)

In order to eliminate secular terms from the above equation, there must be

δ1 = 0 (5.4.26)

The solution of (5.4.25) is

5.4 Exercise 5.4 (Analyze the Linear and Nonlinear Stability of a Rotating … 333

20 20 20

u1 = a − a cos 2ωt − b sin 2ωt (5.4.27)
ω2 3ω2 3ω2
Substituting (5.4.27) and (5.4.24) into (5.4.11), we get

ü2 + δ0 u2 = −δ2 (a cos ωt + b sin ωt)

+ 20 (a cos ωt + b sin ωt) cos2 ωt
2
0 20 20
+ 220 a − a cos 2ωt − b sin 2ωt cos ωt
ω2 3ω2 3ω2

3 540
= − δ2 − 20 − a cos ωt
4 3ω2

1 4
− δ2 − 20 + 02 b sin ωt + NST (5.4.28)
4 3ω

In order to eliminate secular terms from the above equation, there must be

3 520 1 2
δ2 = 20 + maybe δ2 = 20 − 02 (5.4.29)
4 3ω2 4 3ω

Therefore, the transition curves emanating from δ = ω2 is

g 3 520 2 3
δ= − 20 = ω2 + 20 + ε + O ε
l 4 3ω2

g 1 2
δ = − 20 = ω2 + 20 − 02 ε2 + O ε3
l 4 3ω

i.e.,

g 2 3 520 2
= 0 + ω + 0
2 2
+ ε + O ε3 (5.4.30)
l 4 3ω2

g 1 2
= 20 + ω2 + 20 − 02 ε2 + O ε3 (5.4.31)
l 4 3ω

(d) Expanding the Eq. (5.4.4) near θ = 0 and retaining to O(θ 3 ) yields
334 5 Parametrically Excited Systems

θ̈ + gl −1 − 20 − 220 ε cos ωt − 20 ε2 cos2 ωt θ

2 2 1 −1 4 2 2 2 2
+ − gl + 0 ε cos ωt + 0 ε cos ωt θ 3 = 0 2
(5.4.32)
3 0 6 3 3

When θ = O ε1/2 , let

θ = ε̂u, ε̂ = ε1/2 (5.4.33)

Equation (5.4.32) becomes

ü + δ − 220 ε̂2 cos ωt − 20 ε̂4 cos2 ωt u

2 2 2 1 −1 2 4 2 4 2
+ 0 ε̂ − gl ε̂ + 0 ε̂ cos ωt + 20 ε̂6 cos2 ωt u3 = 0 (5.4.34)
3 6 3 3

where
1 2
δ = gl −1 − 20 ≈ ω (5.4.35)
4
In order to determine the effect of nonlinear terms on the stability of the equation,
we use the method of multiple scales to solve (5.4.34). Setting

u(t; ε) = u0 (T0 , T1 , T2 ) + ε̂u1 (T0 , T1 , T2 ) + ε̂2 u2 (T0 , T1 , T2 ) + · · · (5.4.36)

where Tn = ε̂n t. Substituting the above equation into (5.4.34) and retaining to O(ε̂2 ),
we get

0 = D02 + 2ε̂D0 D1 + ε̂2 D12 + 2D0 D2 u0 + ε̂u1 + ε̂2 u2

+ δ − 220 ε̂2 cos ωt − 20 ε̂4 cos2 ωt u0 + ε̂u1 + ε̂2 u2

4 2 2 1 −1 2 8 2 4 4
+ 0 ε̂ − gl ε̂ + 0 ε̂ cos ωt + 20 ε̂6 cos2 ωt
3 6 3 3
(u0 + ε̂u1 + ε̂2 u2 )3

= D02 u0 + 2ε̂D0 D1 u0 + ε̂2 D12 + 2D0 D2 u0
+ ε̂D02 u1 + 2ε̂2 D0 D1 u1 + ε̂2 D02 u2 + δu0 + ε̂δu1 + ε̂2 δu2

4 1
− 220 ε̂2 u0 cos ωt + ε̂2 20 − gl −1 u03 + · · ·
3 6
2
= D0 u0 + δu0 + ε̂ D0 u1 + δu1 + 2D0 D1 u0
2

+ ε̂2 [D02 u2 + δu2 + D12 + 2D0 D2 u0 + 2D0 D1 u1
5.4 Exercise 5.4 (Analyze the Linear and Nonlinear Stability of a Rotating … 335

4 2 1 −1 3
− 220 u0 cos ωt + 0 − gl u0 ] + · · ·
3 6

Equating coefficients of like powers of ε̂ in the above equation yields

D02 u0 + ω02 u0 = 0 (5.4.37)

D02 u1 + ω02 u1 = −2D0 D1 u0 (5.4.38)

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1

1 −1 4 2 3
+ 20 u0 cosωt +
2
gl − 0 u0 (5.4.39)
6 3

where ω02 = δ. The solution of (5.4.37) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.4.40)

where A = A(T1 , T2 ). Substituting u0 into (5.4.38) yields

D02 u1 + ω02 u1 = −2iω0 D1 Aeiω0 T0 + cc (5.4.41)

In order to eliminate secular terms from the above equation, there must be

D1 A = 0 ⇒ A = A(T2 ) (5.4.42)

Thus

u1 = 0 (5.4.43)

Substituting u0 and u1 into (5.4.39) yields

D02 u2 + ω02 u2 = −2iω0 D2 Aeiω0 T0 + 20 Aei(ω−ω0 )T0

1 −1
+ gl − 420 A2 Aeiω0 T0 + cc + NST (5.4.44)
2

When δ = ω02 = gl −1 − 20 ≈ 41 ω2 , we use the detuning parameter σ to

quantitatively describe the nearness of δ to ω0 , i.e.,

ω = 2ω0 + ε̂2 σ (5.4.45)

So (5.4.44) becomes
336 5 Parametrically Excited Systems

D02 u2 + ω02 u2 = −2iω0 D2 Aeiω0 T0 + 20 Aeiσ T2 eiω0 T0

1 −1
+ gl − 40 A2 Aeiω0 T0 + cc + NST
2
(5.4.46)
2

In order to eliminate secular terms from the above equation, there must be

− 1 −
−2iω0 D2 A + 20 eiσ T2 A + gl −1 − 420 A2 A = 0 (5.4.47)
2

Let A = 21 aeiβ , we get

1 1 −1
−2iω0 a + 2ω0 aβ + 20 aei(σ T2 −2β) + gl − 40 a3 = 0
2
(5.4.48)
4 2

Separating the real and imaginary parts of the above equation yields

20
a = a sin(σ T2 − 2β)β
2ω0

1 1 −1 2
=− gl − 40 a2 − 0 cos(σ T2 − 2β)
2
8ω0 2 2ω0

Let γ = σ T2 − 2β and consider ω02 ≈ 41 ω2 , we can write the above equation as

a = 20 ω−1 a sin γ (5.4.49)

1 1
γ = σ + ω−1 gl −1 − 420 a2 + 220 ω−1 cos γ (5.4.50)
2 2

Therefore, the first order approximate solution of (5.4.34) is

1 1
u0 = a cos[ (ω − εσ )t + γ )] + O ε1/2 (5.4.51)
2 2

Let a = γ = 0 in (5.4.49) and (5.4.50), we can obtain

20 ω−1 a sin γ = 0 (5.4.52)

1 −1 1 −1
σ+ ω gl − 40 a2 + 220 ω−1 cos γ = 0
2
(5.4.53)
2 2

Therefore, the steady state solution is

1
a = 0, cos γ = − σ −2
0 ω (5.4.54)
2
5.5 Exercise 5.5 (Parametric Excitation of a Particle-String System … 337

The condition for the existence of a steady state solution of γ is

|σ | ≤ 220 ω−1 (5.4.55)

Therefore, the critical value of σ for the existence of steady state solution is

σ = ±220 ω−1 (5.4.56)

Substituting this into (5.4.45), we can obtain the critical condition for the existence
of a steady state solution

1 2
gl −1 − 20 = ω ± ε20 + O ε2 (5.4.57)
4
Combining (5.4.23), and (5.4.57), we know that the effect of the nonlinear terms
is to limit the unstable linear motion to a finite-amplitude motion, whose frequency is
one half the frequency of the excitation. In other words, a subharmonic is generated
by the system.

5.5 Exercise 5.5 (Parametric Excitation of a Particle-String

System Subjected to Axial Forces at Both Ends)

Solution: (a) The ends of the rope can move only in the horizontal direction.
According to Newton’s second law, we can obtain the equation of motion of m
along x-direction:
x
mẍ = −2T √ (5.5.1)
x2 + l2

i.e.,

mẍ + 2T0 (1 + ε sin ωt)x(l 2 + x2 )−1/2 = 0 (5.5.2)

(b) Linearizing the above equation near the equilibrium position x = 0, we get

mẍ + 2T0 (1 + ε sin ωt)l −1 x = 0 (5.5.3)

i.e.,

2T0
ẍ + ω02 (1 + ε sin ωt)x = 0, ω02 = (5.5.4)
ml
338 5 Parametrically Excited Systems

(c) (5.5.4) is a Hill equation with a time-varying parameter of period T = 2π/ω.

We use the method of strained parameters to obtain transition curves. We seek the
solutions of the original equation having periods of T and 2 T and the equations
for the transition curves ω02 = δ = δ(ε) in the form of the following perturbation
expansions:

x(t; ε) = u0 (t) + εu1 (t) + ε2 u2 (t) + · · · (5.5.5)

δ = δ0 + εδ1 + ε2 δ2 + · · · (5.5.6)

Substituting the above two equations into (5.5.4) and retaining to O(ε2 ), we get

0 = ẍ + δ(1 + ε sin ωt)x

= ü0 + εü1 + ε2 ü2

+ δ0 + εδ1 + ε2 δ2 (1 + ε sin ωt) u0 + εu1 + ε2 u2 + · · ·
= ü0 + εü1 + ε2 ü2 + δ0 u0 + εδ0 u1 + εδ0 u0 sin ωt
+ ε2 δ0 u2 + ε2 δ0 u1 sin ωt
+ εδ1 u0 + ε2 δ1 u1 + ε2 δ1 u0 sin ωt + ε2 δ2 u0 + · · ·
= ü0 + δ0 u0 + ε(ü1 + δ0 u1 + δ1 u0 + δ0 u0 sin ωt)
+ ε2 (ü2 + δ0 u2 + δ2 u0 + δ1 u0 sin ωt + δ1 u1 + δ0 u1 sin ωt) + · · ·

Equating coefficients of like powers of ε in the above equation yields

ü0 + δ0 u0 = 0 (5.5.7)

ü1 + δ0 u1 = −δ1 u0 − δ0 u0 sin ωt (5.5.8)

ü2 + δ0 u2 = −δ2 u0 − δ1 u0 sin ωt − δ1 u1 − δ0 u1 sin ωt (5.5.9)

The solution of (5.5.7) is

u0 = a cos δ0 t + b sin δ0 t (5.5.10)

CASE I: δ0 = ω2 :

u0 = a cos ωt + b sin ωt (5.5.11)

u0 is a periodic solution with period T = 2π/ω. Substituting (5.5.11) into (5.5.8)

yields
5.5 Exercise 5.5 (Parametric Excitation of a Particle-String System … 339

ü1 + ω2 u1 = − δ1 + ω2 sin ωt (a cos ωt + b sin ωt)
= −δ1 a cos ωt − δ1 b sin ωt
1 1
− ω2 a sin 2ωt − ω2 b(1 − cos 2ωt) (5.5.12)
2 2
In order to eliminate secular terms from the above equation, there must be

δ1 = 0 (5.5.13)

Then the solution of (5.5.12) is

1 1 1
u1 = − b + a sin 2ωt − b cos 2ωt (5.5.14)
2 6 6
Substituting the above results into (5.5.9), we obtain

ü2 + δ0 u2 = −δ2 u0 − δ0 u1 sin ωt

= −δ2 (a cos ωt + b sin ωt)

2 1 1 1
+ω b − a sin 2ωt + b cos 2ωt sin ωt
2 6 6

1 2
= − δ2 + ω a cos ωt
12

5 2
− δ2 − ω b sin ωt + NST (5.5.15)
12

In order to eliminate secular terms from the above equation, there must be

1 2 5 2
δ2 = − ω or ω (5.5.16)
12 12

Therefore, the transition curves emanating from δ = ω2 is

1 2 2
δ = ω2 − ω ε + O ε3
12
5 2 2
δ = ω2 + ω ε + O ε3
12
i.e.,

1 2
ω02 = ω 1 − ε + O ε3
2
(5.5.17)
12
or
340 5 Parametrically Excited Systems

5
ω02 = ω2 1 + ε2 + O ε3 (5.5.18)
12

CASE II: δ0 = 41 ω2 :

1 1
u0 = a cos ωt + b sin ωt (5.5.19)
2 2
u0 is a periodic solution with period 2T = 4π/ω. Substituting (5.5.19) into (5.5.8)
yields

1 1 1
ü1 + ω2 u1 = −δ1 a cos ωt + b sin ωt
4 2 2

1 2 1 1
− ω a cos ωt + b sin ωt sin ωt
4 2 2

1 2 1
= − δ1 a + ω b cos ωt
8 2

1 2 1
− δ1 b + ω a sin ωt + NST (5.5.20)
8 2

In order to eliminate secular terms from the above equation, there must be

δ1 a + 18 ω2 b = 0 δ1 18 ω2 a 0
⇒ = (5.5.21)
8
ω a + δ1 b = 0
1 2
8
ω δ1
1 2
b 0

The condition for having a nontrivial solution is

1
δ1 = ± ω 2 (5.5.22)
8

Therefore, the transition curves emanating from δ = 41 ω2 is

1 2 1 2
δ= ω ± ω ε + O ε2
4 8
i.e.,

1 2 1
ω02 = ω 1 ± ε + O ε2 (5.5.23)
4 2

(d) Expand the Eq. (5.5.2) near x = 0 and retain to O(x3 ), we obtain

ẍ x 1 x 2
+ ω02 (1 + ε sin ωt) [1 − ) =0 (5.5.24)
l l 2 l
5.5 Exercise 5.5 (Parametric Excitation of a Particle-String System … 341

By making θ = x/l, the above equation becomes

1
θ̈ + ω02 (1 + ε sin ωt)θ − ω02 (1 + ε sin ωt)θ 3 = 0 (5.5.25)
2
1
When x = O ε1/2 , θ = O ε 2 , let

θ = ε̂u, ε̂ = ε1/2 (5.5.26)

Equation (5.5.25) becomes

1
ü + ω02 1 + ε̂2 sin ωt u − ε̂2 ω02 1 + ε̂2 sin ωt u3 = 0 (5.5.27)
2
In order to determine the effect of the nonlinear terms on the stability of the
system, we use the method of multiple scales to solve (5.5.27). Setting

u(t; ε) = u0 (T0 , T1 , T2 ) + ε̂u1 (T0 , T1 , T2 ) + ε̂2 u2 (T0 , T1 , T2 ) + · · · (5.5.28)

where Tn = ε̂n t. Substituting the above equation into (5.5.27) and retaining to O(ε̂2 ),
we obtain
1
0 = ü + ω02 1 + ε̂2 sin ωt u − ε̂2 ω02 1 + ε̂2 sin ωt u3
2
= D02 + 2ε̂D0 D1 + ε̂2 D12 + 2D0 D2 u0 + ε̂u1 + ε̂2 u2

+ ω02 1 + ε̂2 sin ωt u0 + ε̂u1 + ε̂2 u2
1
− ε̂2 ω02 1 + ε̂2 sin ωt (u0 + ε̂u1 + ε̂2 u2 )3
2
= D02 u0 + ε̂D02 u1 + ε̂2 D02 u2 + 2ε̂D0 D1 u0 + 2ε̂2 D0 D1 u1

+ ε̂2 D12 + 2D0 D2 u0 + ω02 u0 + ε̂ω02 u1 + ε̂2 ω02 u2
1
+ ε̂2 ω02 u0 sin ωt − ε̂2 ω02 u03 + · · ·
2
= D02 u0 + ω02 u0 + ε̂ D02 u1 + ω02 u1 + 2D0 D1 u0

+ ε̂2 [D02 u2 + ω02 u2 + D12 + 2D0 D2 u0 + 2D0 D1 u1
1
+ ω02 u0 sin ωt − ω02 u03 ] + · · ·
2
Equating coefficients of like powers of ε in the above equation yields

D02 u0 + ω02 u0 = 0 (5.5.29)

D02 u1 + ω02 u1 = −2D0 D1 u0 (5.5.30)

342 5 Parametrically Excited Systems

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1
(5.5.31)
−ω02 u0 sinωt + 21 ω02 u03

The solution of (5.5.29) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.5.32)

where A = A(T1 , T2 ). Substituting u0 into (5.5.30) yields

D02 u1 + ω02 u1 = −2iω0 D1 Aeiω0 T0 + cc (5.5.33)

In order to eliminate secular terms from the above equation, there must be

D1 A = 0 ⇒ A = A(T2 ) (5.5.34)

Therefore,

u1 = 0 (5.5.35)

Substitute u0 and u1 into (5.5.31) yields

D02 u2 + ω02 u2 = −2iω0 D2 Aeiω0 T0 + 21 iω02 Aei(ω−ω0 )T0

(5.5.36)
+ 23 ω02 A2 Aeiω0 T0 + cc + NST

When ω02 ≈ 14 ω2 , we use the detuning parameter σ to quantitatively describe the

nearness of ω to 2ω0 , i.e.,

ω = 2ω0 + ε̂2 σ (5.5.37)

So (5.5.36) becomes

D02 u2 + ω02 u2 = −2iω0 D2 Aeiω0 T0 + 21 iω02 Aeiσ T2 eiω0 T0

(5.5.38)
+ 23 ω02 A2 Aeiω0 T0 + cc + NST

In order to eliminate secular terms from the above equation, there must be

1 − 3 −
−2iD2 A + iω0 A eiσ T2 + ω0 A2 A = 0 (5.5.39)
2 2

Let A = 21 aeiβ , we get

1 3
−ia + aβ + iω0 ei(σ T2 −2β) a + ω0 a3 = 0 (5.5.40)
4 16
5.5 Exercise 5.5 (Parametric Excitation of a Particle-String System … 343

Separating the real and imaginary parts of the above equation yields

1
a = ω0 a cos(σ T2 − 2β)
4
3 1
β = − ω0 a2 + ω0 sin(σ T2 − 2β)
16 4

Let γ = σ T2 − 2β and considering ω02 ≈ 41 ω2 , the above equation becomes

1
a = ω0 a cos γ (5.5.41)
4
3 1
γ = σ + ω0 a2 − ω0 sin γ (5.5.42)
8 2
Therefore, the first order approximate solution of (5.5.27) is

1 1
u0 = a cos[ (ω − εσ )t + γ )] + O ε1/2 (5.5.43)
2 2

Let a = γ = 0 in (5.5.41) and (5.5.42), we can obtain

1
ω0 a cos γ = 0 (5.5.44)
4
3 1
σ + ω0 a2 − ω0 sin γ = 0 (5.5.45)
8 2
Therefore, the steady state solution is

a = 0, sin γ = 2σ ω0−1 (5.5.46)

The condition for the existence of a steady state solution of γ is

1
|σ | ≤ ω0 (5.5.47)
2
Therefore, the critical value of σ for the existence of steady state solution is

1 1
σ = ± ω0 ≈ ± ω (5.5.48)
2 4
Substituting this into (5.5.37), we can obtain the critical condition for the existence
of a steady state solution
344 5 Parametrically Excited Systems

1 2 1
ω02 = ω 1 ± ε + O ε2 (5.5.49)
4 2

Combining (5.5.23), and (5.5.49), we know that the effect of the nonlinear terms
is to limit the unstable linear motion to a finite-amplitude motion, whose frequency is
one half of the frequency of the excitation. In other words, a subharmonic is generated
by the system.

5.6 Exercise 5.6 (Linear and Nonlinear Stability Analysis

of a Pendulum with Varying Length)

Solution: (a) The kinetic energy of the system is

1 2 1 d (r sin θ ) 2 d (r cos θ + y) 2
T= mv = m ] + ]
2 2 dt dt
1
= m[(ṙ sin θ + r θ̇ cos θ )2 + ṙ cos θ − r θ̇ sin θ + ẏ)2
2
1 2
= m ṙ + ẏ2 + r 2 θ̇ 2 + 2ṙ ẏ cos θ − 2r ẏθ̇ sin θ
2
1
= m[2ẏ2 + l − y)2 θ̇ 2 − 2ẏ2 cos θ − 2(l − y)ẏθ̇ sin θ (5.6.1)
2
The potential energy of the system is

V = mgr(1 − cos θ ) = mg(l − y)(1 − cos θ ) (5.6.2)

Substituting (5.6.1) and (5.6.2) into the Lagrange’s equation, i.e.,

d ∂T ∂T ∂V
− =− (5.6.3)
dt ∂ θ̇ ∂θ ∂θ

we can obtain the differential equation of motion of the system

(l − y)θ̈ − 2ẏθ̇ − ÿ sin θ + g sin θ = 0 (5.6.4)

(b) Linearizing the above equation near θ = 0 yields

2ẏθ̇ g − ÿ
θ̈ − + θ =0 (5.6.5)
l−y l−y

(c) When y = εl cos t, (5.6.5) becomes

5.6 Exercise 5.6 (Linear and Nonlinear Stability Analysis of a Pendulum … 345

2ε sin t ω2 + ε2 cos t g

θ̈ + θ̇ + 0 θ = 0, ω02 = (5.6.6)
1 − ε cos t 1 − ε cos t l

We use the method of strained parameters to obtain the transition curve. Let the
solution of (5.6.6) solution and parameter ω02 have the following expansion:

θ (t; ε) = u0 (t) + εu1 (t) + ε2 u2 (t) + · · · (5.6.7)

ω02 = δ0 + εδ1 + ε2 δ2 + · · · (5.6.8)

Substituting the above two equations into (5.6.5) and retaining to O(ε2 ), we get

2ε sin t
0 = ü0 + εü1 + ε2 ü2 + u̇0 + εu̇1 + ε2 u̇2
1 − ε cos t
ω02 + ε2 cos t
+ u0 + εu1 + ε2 u2
1 − ε cos t
= ü0 + εü1 + ε2 ü2

+ 2ε sin t(1 + ε cos t) u̇0 + εu̇1 + ε2 u̇2

+ ω02 + ε2 cos t 1 + ε cos t + ε2 cos2 t

u0 + εu1 + ε2 u2 + · · ·
= ü0 + εü1 + ε2 ü2 + 2εu̇0 sin t
+ 2ε2 u̇0 sin t cos t + 2ε2 u̇1 sin t
+ ω02 u0 + εω02 u0 cos t + ε2 u0 cos t
+ ε2 ω02 u0 cos2 t + ε2 2 u0 cos2 t
+ εω02 u1 + ε2 ω02 u1 cos t
+ ε2 2 u1 cos t + ε2 ω02 u2 + · · ·
= ü0 + εü1 + ε2 ü2 + 2εu̇0 sin t
+ 2ε2 u̇0 sin t cos t + 2ε2 u̇1 sin t

+ δ0 + εδ1 + ε2 δ2 u0 + ε δ0 + εδ1 + ε2 δ2 u0 cos t

+ ε2 u0 cos t + ε2 δ0 + εδ1 + ε2 δ2 u0 cos2 t
+ ε2 2 u0 cos2 t

+ ε δ0 + εδ1 + ε2 δ2 u1 + ε2 δ0 + εδ1 + ε2 δ2 u1 cos t

+ ε2 2 u1 cos t + ε2 δ0 + εδ1 + ε2 δ2 u2 + · · ·
= ü0 + εü1 + ε2 ü2 + 2εu̇0 sin t
+ 2ε2 u̇0 sin t cos t + 2ε2 u̇1 sin t
+ δ0 u0 + εδ1 u0 + ε2 δ2 u0
+ εδ0 u0 cos t + ε2 δ1 u0 cos t
346 5 Parametrically Excited Systems

+ ε2 u0 cos t + ε2 δ0 u0 cos2 t

+ ε2 2 u0 cos2 t
+ εδ0 u1 + ε2 δ1 u1 + ε2 δ0 u1 cos t
+ ε2 2 u1 cos t + ε2 δ0 u2 + · · ·
= ü0 + δ0 u0

+ ε ü1 + δ0 u1 + δ1 u0 + 2u̇0 sin t + δ0 u0 cos t + 2 u0 cos t
+ ε2 [ü2 + δ0 u2 + δ2 u0 + δ1 u0 cos t + 2u̇0 sin t cos t
+ 2u̇1 sin t + δ0 u0 cos2 t + 2 u0 cos2 t
+ δ1 u1 + δ0 u1 cos t + 2 u1 cos t] + · · ·

Equating coefficients of like powers of ε̂ in the above equation yields

ü0 + δ0 u0 = 0 (5.6.9)

ü1 + δ0 u1 = −δ1 u0 − 2u̇0 sin t

− δ0 u0 cos t − 2 u0 cos t (5.6.10)

ü2 + δ0 u2 = −δ2 u0 − δ1 u0 cos t − δ0 u0 cos2 t

− 2 u0 cos2 t − u̇0 sin 2t
− δ1 u1 − δ0 u1 cos t − 2 u1 cos t
− 2u̇1 sin t (5.6.11)

The solution of (5.6.9) is

u0 = a cos δ0 t + b sin δ0 t (5.6.12)

√
CASE I: δ0 = 21 :

1 1
u0 = a cos t + b sin t (5.6.13)
2 2
Substituting (5.6.13) into (5.6.10) yields

1 5
ü1 + 2 u1 = −δ1 u0 − 2 u0 cos t − 2u̇0 sin t
4 4
1 1
= −δ1 a cos t + b sin t
2 2

5 2 1 1
− a cos t + b sin t cos t
4 2 2
5.6 Exercise 5.6 (Linear and Nonlinear Stability Analysis of a Pendulum … 347

1 1
− 2 −a sin t + b cos t sin t
2 2
1 1
= −δ1 a cos t − δ1 b sin t
2 2
5 2 1
− a cos t cos t
4 2
5 2 1
− b sin t cos t
4 2
1 1
+ 2 a sin t sin t − 2 b sin t cos t
2 2
1 1
= − δ1 + 2 a cos t
8 2

1 2 1
− δ1 − b sin t + NST (5.6.14)
8 2

In order to eliminate secular terms from the above equation, there must be

1
δ1 = ± 2 (5.6.15)
8

Therefore, the transition curves emanating from ω02 = 41 2 is

1 2 1
ω02 = gl −1 = 1 ± ε + O ε2 (5.6.16)
4 2

√
CASE II: δ0 = :

u0 = a cos t + b sin t (5.6.17)

Substituting (5.6.17) into (5.6.10) yields

ü1 + 2 u1 = −δ1 u0 − 2u̇0 sin t

− 2 u0 cos t − 2 u0 cos t

= − δ1 + 22 cos t u0 − 2u̇0 sin t

= − δ1 + 22 cos t (a cos t + b sin t)
− 22 sin t(−a sin t + b cos t)
= −δ1 (a cos t + b sin t)
− 2 a(1 + cos 2t) − 2 b sin 2t
+ 2 a(1 − cos 2t) − 2 b sin 2t
= −δ1 (a cos t + b sin t) − 22 a cos 2t
348 5 Parametrically Excited Systems

− 22 b sin 2t (5.6.18)

In order to eliminate secular terms from the above equation, there must be

δ1 = 0 (5.6.19)

Thus, the solution of (5.6.18) is

2 2
u1 = a cos 2t + b sin 2t (5.6.20)
3 3
Substituting (5.6.17) and (5.6.20) into (5.6.11), we get

ü2 + 2 u2 = −δ2 u0 − 22 u0 cos2 t − u̇0 sin 2t

− 22 u1 cos t − 2u̇1 sin t
= −δ2 (a cos t + b sin t)
− 22 (a cos t + b sin t) cos2 t
− 2 (−a sin t + b cos t) sin 2t

2 2
− 22 a cos 2t + b sin 2t cos t
3 3

2 2
− 42 − a sin 2t + b cos 2t sin t
3 3

1
= − δ2 + 2 a cos t − δ2 + 32 b sin t + NST (5.6.21)
3

In order to eliminate secular terms from the above equation, there must be

1
δ2 = − 2 or δ2 = −32 (5.6.22)
3

Therefore, the transition curves emanating from ω02 = 2 is

1
ω02 = gl −1 = 2 1 − ε2 + O ε3 (5.6.23)
3
or

ω02 = gl −1 = 2 1 − 3ε2 + O ε3 (5.6.24)

(d) Expanding (5.6.4) near θ = 0 and retaining to O(θ 3 ) yields

2ẏθ̇ g − ÿ θ3
θ̈ − + θ− =0 (5.6.25)
l−y l−y 6
5.6 Exercise 5.6 (Linear and Nonlinear Stability Analysis of a Pendulum … 349

When y = εl cos t, , the nonlinear Eq. (5.6.25) becomes

2ε sin t ω02 + ε2 cos t θ3 g
θ̈ + θ̇ + θ− = 0, ω02 = (5.6.26)
1 − ε cos t 1 − ε cos t 6 l

When θ = O ε1/2 , let

θ = ε̂u, ε̂ = ε1/2 (5.6.27)

Equation (5.6.26) becomes

3
2ε̂2 sin t ω02 + ε̂2 2 cos t 2u
ü + u̇ + u − ε̂ =0 (5.6.28)
1 − ε̂2 cos t 1 − ε̂2 cos t 6

In order to determine the effect of nonlinear terms on the stability of the equation,
we use the method of multiple scales to solve (5.6.28). Setting

u(t; ε) = u0 (T0 , T1 , T2 ) + ε̂u1 (T0 , T1 , T2 )

+ ε̂2 u2 (T0 , T1 , T2 ) + · · · (5.6.29)

where Tn = ε̂n t. Substituting the above equation into (5.6.28) and retaining to O(ε̂2 ),
we get
3
2ε̂2 sin t ω02 + ε̂2 2 cos t 2u
0 = ü + u̇ + u − ε̂
1 − ε̂2 cos t 1 − ε̂2 cos t 6
2
= D0 + 2ε̂D0 D1 + ε̂ D1 + 2D0 D2 u0 + ε̂u1 + ε̂ u2
2 2 2

+ 2ε̂2 sin T0 (D0 + ε̂D1 ) u0 + ε̂u1 + ε̂2 u2

+ ω02 + ε̂2 2 cos T0 1 + ε̂2 cos T0

1
u0 + ε̂u1 + ε̂2 u2 − ε̂2 u03 + · · ·
6
= D02 u0 + ε̂D02 u1 + ε̂2 D02 u2 + 2ε̂D0 D1 u0

+ 2ε̂2 D0 D1 u1 + ε̂2 D12 + 2D0 D2 u0
+ 2ε̂2 sin T0 D0 u0 + ω02 u0 + ε̂ω02 u1
1
+ ε̂2 ω02 u2 − ε̂2 ω02 u03
6
+ ε̂2 ω02 u0 cos T0 + ε̂2 2 u0 cos T0 + · · ·

= D02 u0 + ω02 u0 + ε̂ D02 u1 + ω02 u1 + 2D0 D1 u0

+ ε̂2 [D02 u2 + ω02 u2 + D12 + 2D0 D2 u0
1
+ 2D0 D1 u1 − ω02 u03 + 2 sin T0 D0 u0
6
350 5 Parametrically Excited Systems

+ ω02 u0 cos T0 + 2 u0 cos T0 ] + · · ·

Equating coefficients of like powers of ε̂ in the above equation yields

D02 u0 + ω02 u0 = 0 (5.6.30)

D02 u1 + ω02 u1 = −2D0 D1 u0 (5.6.31)

1
D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1 + ω02 u03
6
− 2 sin T0 D0 u0 − ω02 u0 cos T0 − 2 u0 cos T0 (5.6.32)

The solution of (5.6.30) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.6.33)

where A = A(T1 , T2 ). Substituting u0 into (5.6.31) yields

D02 u1 + ω02 u1 = −2iω0 D1 Aeiω0 T0 + cc (5.6.34)

In order to eliminate secular terms from the above equation, there must be

D1 A = 0 ⇒ A = A(T2 ) (5.6.35)

Thus

u1 = 0 (5.6.36)

Substitute u0 , u1 into (5.6.32), we can obtain

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2 sin T0 D0 u0
1
− ω02 u0 cos T0 − 2 u0 cos T0 + ω02 u03
6
= −2iω0 D2 Aeiω0 T0 + ω0 ei(−ω0 )T0 A
1 1
− ω02 + 2 ei(−ω0 )T0 A + ω02 A2 Aeiω0 T0 + cc (5.6.37)
2 2

When ω0 = gl −1 ≈ 21 , we use the detuning parameter σ to quantitatively

describe the nearness of to 2ω0 , i.e.,

= 2ω0 + ε̂2 σ (5.6.38)

So (5.6.37) becomes
5.6 Exercise 5.6 (Linear and Nonlinear Stability Analysis of a Pendulum … 351

1 −
D02 u2 + ω02 u2 = −iD2 Aeiω0 T0 − 2 A eiσ T2 eiω0 T0
8
1 2 2 − iω0 T0
+ A Ae (5.6.39)
8
In order to eliminate secular terms from the above equation, there must be

1 − 1 −
−iD2 A − A eiσ T2 + A2 A = 0 (5.6.40)
8 8

Let A = 21 aeiβ , we get

1 1
−ia + aβ − aei(σ T2 −2β) + a3 = 0 (5.6.41)
8 32
Separating the real and imaginary parts of the above equation yields

1 1 1
a = − a sin(σ T2 − 2β)β = − a2 + cos(σ T2 − 2β)
8 32 8
Let γ = σ T2 − 2β, the above equation becomes

1
a = − a sin γ (5.6.42)
8
1 1
γ = σ + a2 − cos γ (5.6.43)
16 4
Therefore, the first order approximate solution of (5.6.28) is

1 1
u0 = a cos[ t − γ )] + O ε1/2 (5.6.44)
2 2

Let a = γ = 0 in (5.6.42) and (5.6.43), we can obtain

1
a sin γ = 0 (5.6.45)
16
1 1
σ+ a2 − cos γ = 0 (5.6.46)
16 4
Therefore, the steady state solution is

a = 0, cos γ = 4σ −1 (5.6.47)

The condition for the existence of a steady state solution of γ is

352 5 Parametrically Excited Systems

1
|σ | ≤ (5.6.48)
4
Therefore, the critical value of σ for the existence of steady state solution is

1
σ =± (5.6.49)
4
Substituting this into (5.6.38), we can obtain the critical condition for the existence
of a steady state solution

1 2 1
ω02 = 1 ± ε̂2 (5.6.50)
4 2

i.e.,

1 2 1
gl −1 = 1± ε (5.6.51)
4 2

Combining (5.6.51), we know that the effect of the nonlinear terms is to limit
the unstable linear motion to a finite-amplitude motion, whose frequency is one half
the frequency of the excitation. In other words, a subharmonic is generated by the
system.

5.7 Exercise 5.7 (Linear and Nonlinear Stability Analysis

of a Particle Sliding on a Rotating Parabola)

Solution: (a) The kinetic energy of the system is

1 2
T= m ẋ + ż 2 + 2 x2
2
1
= m 1 + 4p2 x2 ẋ2 + 2 x2 (5.7.1)
2
The kinetic energy of the system is

V = mgz = mgpx2 (5.7.2)

Substituting (5.7.1) and (5.7.2) into the Lagrange’s equation, we can obtain the
differential equation of motion of the system

m 1 + 4p2 x2 ẍ + 4mp2 xẋ2 − m2 x = −2mgpx
5.7 Exercise 5.7 (Linear and Nonlinear Stability Analysis of a Particle … 353

i.e.,

1 + 4p2 x2 ẍ + 4p2 xẋ2 + 2pg − 2 x = 0 (5.7.3)

When = 0 (1 + ε cos ωt), the

above equation
becomes.

i.e., 1 + 4p2 x2 ẍ + 4p2 xẋ2 + 2pg − 20 x − 20 2ε cos ωt + ε2 cos2 ωt x = 0.
Let ω02 = 2gp − 20 , we get

1 + 4p2 x2 ẍ + 4p2 ẋ2 x + ω02 x − 20 2ε cos ωt + ε2 cos2 ωt x = 0 (5.7.4)

(b) Linearizing the above equation near x = 0, we get

ẍ + ω02 x − 20 2ε cos ωt + ε2 cos2 ωt x = 0 (5.7.5)

We use the method of strained parameters to obtain the transition curve. Let the
solution of Eq. (5.7.5) and parameter ω02 have the following expansion:

x(t; ε) = u0 (t) + εu1 (t) + ε2 u2 (t) + · · · (5.7.6)

ω02 = δ0 + εδ1 + ε2 δ2 + · · · (5.7.7)

Substituting the above two equations into (5.7.5) and retaining to O(ε2 ), we get

0 = ẍ + ω02 x − 20 2ε cos ωt + ε2 cos2 ωt x
= ü0 + εü1 + ε2 ü2

+ δ0 + εδ1 + ε2 δ2 u0 + εu1 + ε2 u2

− 20 2ε cos ωt + ε2 cos2 ωt u0 + εu1 + ε2 u2
= ü0 + εü1 + ε2 ü2 + δ0 u0 + εδ0 u1
+ ε2 δ0 u2 + εδ1 u0 + ε2 δ1 u1
+ ε2 δ2 u0 − 2ε20 u0 cos ωt
− ε2 20 u0 cos2 ωt − 2ε2 u1 20 cos ωt

= ü0 + δ0 u0 + ε ü1 + δ0 u1 + δ1 u0 − 220 u0 cos ωt

+ ε2 ü2 + δ0 u2 + δ2 u0 + δ1 u1 − 20 u0 cos2 ωt − 220 u1 cos ωt

Equating coefficients of like powers of ε in the above equation yields

ü0 + δ0 u0 = 0 (5.7.8)

ü1 + δ0 u1 = −δ1 u0 + 220 u0 cos ωt (5.7.9)

354 5 Parametrically Excited Systems

ü2 + δ0 u2 = −δ2 u0 − δ1 u1 + 20 u0 cos2 ωt + 220 u1 cos ωt (5.7.10)

The solution of (5.7.8) is

u0 = a cos δ0 t + b sin δ0 t (5.7.11)

√
CASE I: δ0 = 0:

u0 = a (5.7.12)

Substituting (5.7.12) into (5.7.9) yields

ü1 = −δ1 a + 220 a cos ωt (5.7.13)

In order to eliminate secular terms from the above equation, there must be

δ1 = 0 (5.7.14)

Thus

u1 = −220 ω−2 a cos ωt (5.7.15)

Substituting u0 , u1 into (5.7.10) yields

ü2 = −δ2 a + 20 a cos2 ωt − 440 ω−2 a cos2 ωt

1
= −δ2 a + 20 a(1 + cos 2ωt)
2
− 240 ω−2 a(1 + cos 2ωt)

1 2 4 −2
= − δ2 − 0 + 20 ω a
2

1 2
+ − 240 ω−2 a cos 2ωt (5.7.16)
2 0

In order to eliminate secular terms from the above equation, there must be

1 1 2
δ2 − 20 + 240 ω−2 ⇒ δ2 = − 240 ω−2 (5.7.17)
2 2 0
Therefore, the transition curves emanating from ω0 = 0 is

1 2 2
ω02 = ε 0 1 − 420 ω−2 + O ε3 (5.7.18)
2
5.7 Exercise 5.7 (Linear and Nonlinear Stability Analysis of a Particle … 355
√
CASE II: δ0 = 21 ω:

1 1
u0 = a cos ωt + b sin ωt (5.7.19)
2 2
Substituting (5.7.19) into (5.7.9) yields

1 2 1 1
ü1 + ω u1 = −δ1 a cos ωt + b sin ωt
4 2 2

1 1
+ 20 a cos ωt + b sin ωt cos ωt
2
2 2
1
= − δ1 − 20 a cos
2
1
− δ1 + 20 b sin ωt + NST (5.7.20)
2
In order to eliminate secular terms from the above equation, there must be

δ1 = ±20 (5.7.21)

Therefore, the transition curves emanating from ω0 = 21 ω is

1 2
ω02 = ω ± ε20 + O ε2 (5.7.22)
4
√
CASE III: δ0 = ω:

1 1
u0 = a cos ωt + b sin ωt (5.7.23)
2 2
Substituting (5.7.23) into (5.7.9) yields

1 2 1 1
ü1 + ω u1 = −δ1 a cos ωt + b sin ωt
4 2 2

1 1
+ 20 a cos ωt + b sin ωt cos ωt
2
2 2
1 1
= − δ1 − 20 a cos − δ1 + 20 b sin ωt + NST (5.7.24)
2 2
In order to eliminate secular terms from the above equation, there must be

δ1 = ±20 (5.7.25)

Therefore, the transition curves emanating from ω0 = ω is

356 5 Parametrically Excited Systems

ω02 = ω2 ± ε20 + O ε2 (5.7.26)

(c) In the neighborhood of x = 0, the nonlinear equation of the system (5.7.4) has
the same form. When θ = O ε1/2 , let

x = ε̂u, ε̂ = ε1/2 (5.7.27)

Equation (5.7.4) becomes

1 + 4p2 ε̂2 u2 ü + 4p2 ε̂2 u̇2 u + ω02 u

− 20 2ε̂2 cos ωt + ε̂4 cos2 ωt u = 0 (5.7.28)

In order to determine the effect of the nonlinear terms on the stability of the
solution, we use the method of multiple scales to solve (5.7.28). Setting

u(t; ε) = u0 (T0 , T1 , T2 ) + ε̂u1 (T0 , T1 , T2 ) + ε̂2 u2 (T0 , T1 , T2 ) + · · · (5.7.29)

where Tn = ε̂n t. Substituting the above equation into (5.7.28) and retaining to O(ε̂2 ),
we get

0 = 1 + 4p2 ε̂2 u2 ü + 4p2 ε̂2 u̇2 u + ω02 u

− 20 2ε̂2 cos ωt + ε̂4 cos2 ωt u

= 1 + 4p2 ε̂2 u02 D02 + 2ε̂D0 D1 + ε̂2 D12 + 2D0 D2

u0 + ε̂u1 + ε̂2 u2

+ 4p2 ε̂2 (D0 u0 )2 u0 + ω02 u0 + ε̂u1 + ε̂2 u2

− 20 2ε̂2 cos ωt + ε̂4 cos2 ωt u0 + ε̂u1 + ε̂2 u2 + · · ·
= D02 u0 + ε̂D02 u1 + ε̂2 D02 u2 + 2ε̂D0 D1 u0 + 2ε̂2 D0 D1 u1

+ ε̂2 D12 + 2D0 D2 u0 + 4p2 ε̂2 u02 D02 u0
+ 4p2 ε̂2 (D0 u0 )2 u0
+ ω02 u0 + ε̂ω02 u1 + ε̂2 ω02 u2 − 2ε̂2 20 u0 cos ωt + · · ·

= D02 u0 + ω02 u0 + ε̂ D02 u1 + ω02 u1 + 2D0 D1 u0

+ ε̂2 [D02 u2 + ω02 u2 + D12 + 2D0 D2 u0 + 2D0 D1 u1

+ 4p2 u02 D02 u0 + 4p2 D0 u0 )2 u0 − 220 u0 cos ωt + · · ·

Equating coefficients of like powers of ε̂ in the above equation yields

D02 u0 + ω02 u0 = 0 (5.7.30)

D02 u1 + ω02 u1 = −2D0 D1 u0 (5.7.31)

5.7 Exercise 5.7 (Linear and Nonlinear Stability Analysis of a Particle … 357

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1
− 4p2 u02 D02 u0 − 4p2 (D0 u0 )2 u0
+ 220 u0 cosωt (5.7.32)

The solution of (5.7.30) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.7.33)

where A = A(T1 , T2 ). Substituting u0 into (5.7.31) yields

D02 u1 + ω02 u1 = −2iω0 D1 Aeiω0 T0 + cc (5.7.34)

In order to eliminate secular terms from the above equation, there must be

D1 A = 0 ⇒ A = A(T2 ) (5.7.35)

Thus

u1 = 0 (5.7.36)

Substitute u0 , u1 into (5.7.32) yields

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 4p2 u02 D02 u0
− 4p2 (D0 u0 )2 u0 + 220 u0 cos ωt
= −2iω0 D2 Aeiω0 T0 + 8p2 ω02 A2 Aeiω0 T0
+ 20 Aei(ω−ω0 )T0 + cc + NST (5.7.37)

When ω0 ≈ 21 ω, we use the detuning parameter σ to quantitatively describe the

nearness of ω to 2ω0 , i.e.,

ω = 2ω0 + ε̂2 σ (5.7.38)

So the Eq. (5.7.37) becomes

D02 u2 + ω02 u2 = −2iω0 D2 Aeiω0 T0 + 8p2 ω02 A2 Aeiω0 T0

+ 20 Aeiσ T2 eiω0 T0 + cc + NST (5.7.39)

In order to eliminate secular terms from the above equation, there must be

− −
−2iD2 A + 8p2 ω0 A2 A +20 ω0−1 A eiσ T2 = 0 (5.7.40)
358 5 Parametrically Excited Systems

Let A = 21 aeiβ , we get

1
−ia + aβ + p2 ω0 a3 + 20 ω0−1 aei(σ T2 −2β) = 0 (5.7.41)
2
Separating the real and imaginary parts of the above equation yields

1 2 −1
a = ω a sin(σ T2 − 2β)
2 0 0
1
β = p2 ω0 a2 − 20 ω0−1 cos(σ T2 − 2β)
2
Let γ = σ T2 − 2β, the above equation becomes

1 2 −1
a = ω a sin γ (5.7.42)
2 0 0

γ = σ − 2p2 ω0 a2 + 20 ω0−1 cos γ (5.7.43)

Therefore, the first order approximate solution of (5.7.28) is

1 1
u0 = a cos[ ( − εσ )t + γ )] + O ε1/2 (5.7.44)
2 2

Let a = γ = 0 in (5.7.42) and (5.7.43), we can obtain

1 2 −1
ω a sin γ = 0 (5.7.45)
2 0 0

σ − 2p2 ω0 a2 + 20 ω0−1 cos γ = 0 (5.7.46)

Therefore, the steady state solution is

a = 0, cos γ = −σ 20 ω0 (5.7.47)

The condition for the existence of a steady state solution of γ is

|σ | ≤ 20 ω0−1 (5.7.48)

Therefore, the critical value of σ for the existence of steady state solution is

σ = ±20 ω0−1 (5.7.49)

Substituting this into (5.7.38), we can obtain the critical condition for the existence
of a steady state solution
5.8 Exercise 5.8 (Nonlinear Solution to the Pure Rolling of a Cylinder … 359

1 2
ω02 = ω ± ε20 + O ε2 (5.7.50)
4
Combining (5.7.22) and (5.7.50), we know that the effect of the nonlinear terms is
to limit the unstable linear motion to a finite-amplitude motion, whose frequency is
one half the frequency of the excitation. In other words, a subharmonic is generated
by the system.

5.8 Exercise 5.8 (Nonlinear Solution to the Pure Rolling

of a Cylinder on a Circular Surface)

Solution: (a) The kinetic energy of the system is

1 2
T= M ẋ + ẏ2
2
1
+ m ẋ + (R − r)θ̇ cos θ ]2 + ẏ + (R − r)θ̇ sin θ ]2
2

1 1 2 (R − r)θ̇ 2
+ mr [ ]
2 2 r

1 1 3
= M ẋ2 + ẏ2 + m ẋ2 + ẏ2 + (R − r)2 θ̇ 2
2 2 2

+ 2(R − r)ẋθ̇ cos θ + 2(R − r)ẏθ̇ sin θ (5.8.1)

where M denotes the mass of the cylinder and the block. The kinetic energy of the
system is

V = Mgy + mg(R − r)(1 − cos θ ) (5.8.2)

Substituting (5.8.1) and (5.8.2) into the Lagrange’s equation, we can obtain the
differential equation of motion of the system

3
m[ R − r)2 θ̈ + (R − r)ẍ cos θ + (R − r)ÿ sin θ = −mg(R − r) sin θ
2
i.e.,
360 5 Parametrically Excited Systems

2 2
θ̈ + (R − r)−1 (g + ÿ) sin θ + (R − r)−1 ẍ cos θ = 0 (5.8.3)
3 3

Expanding the above equation near θ = 0 and retaining to O(θ 3 ) yields

2 1 2 1
θ̈ + (R − r)−1 (g + ÿ) θ − θ 3 + (R − r)−1 ẍ 1 − θ 2 = 0 (5.8.4)
3 6 3 2

(b) When ẍ = εK cos t, ÿ = 0„ the Eq. (5.8.4) becomes

2 1 2 1
θ̈ + (R − r)−1 g θ − θ 3 + ε(R − r)−1 K 1 − θ 2 cos t = 0 (5.8.5)
3 6 3 2

When θ = O ε1/2 , let

x = ε̂u, ε̂ = ε1/2 (5.8.6)

Equation (5.8.5) becomes

2 −1 1 2 3 2 −1 1 2 2
ü + (R − r) g u − ε̂ u + ε̂(R − r) K 1 − ε̂ u cos t = 0
3 6 3 2
(5.8.7)

Let
2
ω02 = (R − r)−1 g, K̂ = Kω02 g −1 (5.8.8)
3
Equation (5.8.7) becomes

1 1
ü + ω02 u − ε̂2 ω02 u3 − ε̂3 K̂u2 cos t = −ε̂ K̂ cos t (5.8.9)
6 2
We use the method of multiple scales to solve (5.8.9). Let

u(t; ε) = u0 (T0 , T1 , T2 ) + ε̂u1 (T0 , T1 , T2 ) + ε̂2 u2 (T0 , T1 , T2 ) + · · · (5.8.10)

where Tn = ε̂n t.. Substituting the above equation into (5.8.9) and retaining to O(ε̂2 ),
we get
5.8 Exercise 5.8 (Nonlinear Solution to the Pure Rolling of a Cylinder … 361

1 1
0 = ü + ω02 u − ε̂2 ω02 u3 − ε̂3 K̂u2 cos t + ε̂K̂ cos t
6 2
= D02 + 2ε̂D0 D1 + ε̂2 D12 + 2D0 D2 u0 + ε̂u1 + ε̂2 u2
1
+ ω02 u0 + ε̂u1 + ε̂2 u2 − ε̂2 ω02 (u0 + ε̂u1 + ε̂2 u2 )3
6
1 3
− ε̂ K̂(u0 + ε̂u1 + ε̂ u2 ) cos t + ε̂K̂ cos t
2 2
2
= D02 u0 + ε̂D02 u1 + ε̂2 D02 u2 + 2ε̂D0 D1 u0

+ 2ε̂2 D0 D1 u1 + ε̂2 D12 + 2D0 D2 u0 + ω02 u0
1
+ ε̂ω02 u1 + ε̂2 ω02 u2 − ε̂2 ω02 u03 + ε̂K̂ cos t + · · ·
6
= D02 u0 + ω02 u0 + ε̂ D02 u1 + ω02 u1 + 2D0 D1 u0 + K̂ cos T0

1
+ ε̂ D02 u2 + ω02 u2 + 2D0 D1 u1 + D12 + 2D0 D2 u0 − ω02 u03 + · · ·
2
6

Equating coefficients of like powers of ε̂ in the above equation yields

D02 u0 + ω02 u0 = 0 (5.8.11)

D02 u1 + ω02 u1 = −2D0 D1 u0 − K̂ cos T0 (5.8.12)

1
D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1 + ω02 u03 (5.8.13)
6
The solution of (5.8.11) is
−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.8.14)

where A = A(T1 , T2 ). Substituting u0 into (5.8.12) yields

1
D02 u1 + ω02 u1 = −2iω0 D0 D1 Aeiω0 T0 − K̂eiT0 + cc (5.8.15)
2

CASE I: ≈ ω0 :
We use the detuning parameter σ to quantitatively describe the nearness of to
ω0 , i.e.,

= ω0 + ε̂σ (5.8.16)

therefore,
362 5 Parametrically Excited Systems

1
D02 u1 + ω02 u1 = −2iω0 D0 D1 Aeiω0 T0 − K̂eiσ T1 eiω0 T0 + cc (5.8.17)
2
In order to eliminate secular terms from the above equation, there must be

1
−2iω0 D1 A − K̂eiσ T1 = 0 (5.8.18)
2
then
1
A= K̂ω0−1 σ −1 eiσ T1 (5.8.19)
4
Substituting this into (5.8.14), we can obtain the first order approximate solution
of the system

1
u= K̂ω0−1 σ −1 cos(ω0 T0 + σ T1 ) + O(ε̂)
2
1
= K̂ω0−1 σ −1 cos ω0 + ε1/2 σ t + O ε1/2 (5.8.20)
2

CASE II: ≈ 3ω0 :

Equation (5.8.15) becomes

1
D02 u1 + ω02 u1 = −2iω0 D0 D1 Aeiω0 T0 − K̂e3iω0 T0 + cc (5.8.21)
2
In order to eliminate secular terms from the above equation, there must be

D1 A = 0 ⇒ A = A(T2 ) (5.8.22)

Then the solution of (5.8.21) is

1 −2 3iω0 T0
u1 = ω K̂e + cc (5.8.23)
16 0
Substituting u0 , u1 into (5.8.13) yields

1
D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1 + ω02 u03
6
1 2 2 iω0 T0
= −2iω0 D2 Ae iω0 T0
+ ω0 A Ae + cc + NST (5.8.24)
2
In order to eliminate secular terms from the above equation, there must be

1 −
−2iD2 A + ω0 A2 A = 0 (5.8.25)
2
5.8 Exercise 5.8 (Nonlinear Solution to the Pure Rolling of a Cylinder … 363

Let A = 21 aeiβ , we get

1
−ia + aβ + ω0 a3 = 0 (5.8.26)
16
Separating the real and imaginary parts of the above equation yields

a = 0
1
β = − ω0 a2 (5.8.27)
16
Near u = 0, (5.8.27) has the steady state solution

a = 0, β = 0 (5.8.28)

Therefore, the first order approximate solution of the system is

u = 0 + O ε1/2 (5.8.29)

This result shows that it is not possible to generate the subharmonic resonance
response under soft excitation.
CASE III: ≈ 13 ω0 :
A similar process to the one above leads to the first order approximate solution
of the system when ≈ 13 ω0 is

u = 0 + O ε1/2 (5.8.30)

(The reader is invited to complete the procedure on their own).

This result shows that it is not possible to generate the super harmonic resonance
response under soft excitation.
(c) When ẍ = εK cos 1 t, ÿ = ε cos 2 t, (5.8.4) becomes

2 −1 1 3
θ̈ + (R − r) (g + ε cos 2 t) θ − θ
3 6

2 1
+ ε(R − r)−1 K 1 − θ 2 cos 1 t = 0 (5.8.31)
3 2

When θ = O ε1/2 , let

x = ε̂u, ε̂ = ε1/2 (5.8.32)

Equation (5.8.5) becomes

364 5 Parametrically Excited Systems

1
ü + ω02 u − ε̂2 ω02 u3 + ε̂2 αu cos 2 t = −ε̂ K̂ cos 1 t (5.8.33)
6

2 1
ü + (R − r)−1 g + ε̂2 cos 2 t u − ε̂2 u3
3 6

2 1
+ ε̂(R − r)−1 K 1 − ε̂2 u2 cos 1 t = 0 (5.8.34)
3 2

Let
2 2 2
ω02 = (R − r)−1 g, α = (R − r)−1 , K̂ = (R − r)−1 K (5.8.35)
3 3 3

Substituting the above equation into (5.8.34) and retaining to O(ε̂2 ), it becomes

1
ü + ω02 u − ε̂2 ω02 u3 + ε̂2 αu cos 2 t = −ε̂ K̂ cos 1 t (5.8.36)
6
We use the method of multiple scales to solve the Eq. (5.8.36). Set

u(t; ε) = u0 (T0 , T1 , T2 ) + ε̂u1 (T0 , T1 , T2 ) + ε̂2 u2 (T0 , T1 , T2 ) + · · · (5.8.37)

where Tn = ε̂n t. Substituting the above equation into (5.8.36) and retaining to O(ε̂2 ),
we get

1
0 = ü + ω02 u − ε̂2 ω02 u3 + ε̂2 αu cos 2 t + ε̂ K̂ cos 1 t
6
= D02 + 2ε̂D0 D1 + ε̂2 D12 + 2D0 D2 u0 + ε̂u1 + ε̂2 u2
1
+ ω02 u0 + ε̂u1 + ε̂2 u2 − ε̂2 ω02 (u0 + ε̂u1 + ε̂2 u2 )3
6
+ ε̂ α u0 + ε̂u1 + ε̂ u2 cos 2 T0 + ε̂K̂ cos 1 T0
2 2

+ ε̂2 α u0 + ε̂u1 + ε̂2 u2 cos 2 T0 + ε̂K̂ cos 1 T0
= D02 u0 + ε̂D02 u1 + ε̂2 D02 u2 + 2ε̂D0 D1 u0 + 2ε̂2 D0 D1 u1

+ ε̂2 D12 + 2D0 D2 u0 + ω02 u0 + ε̂ω02 u1 + ε̂2 ω02 u2
1
− ε̂2 ω02 u03 + ε̂2 αu0 cos 2 T0 + ε̂K̂ cos 1 T0 + · · ·
6
= D02 u0 + ω02 u0 + ε̂ D02 u1 + ω02 u1 + 2D0 D1 u0 + K̂ cos 1 T0

Equating coefficients of like powers of ε̂ in the above equation yields

D02 u0 + ω02 u0 = 0 (5.8.38)

D02 u1 + ω02 u1 = −2D0 D1 u0 − K̂ cos 1 T0 (5.8.39)

5.8 Exercise 5.8 (Nonlinear Solution to the Pure Rolling of a Cylinder … 365

1
D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1 + ω02 u03 − αu0 cos 2 T0
6
(5.8.40)

The solution of (5.8.38) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.8.41)

where A = A(T1 , T2 ). Substituting u0 into (5.8.39) yields

1
D02 u1 + ω02 u1 = −2iω0 D0 D1 Aeiω0 T0 − K̂eiT0 + cc (5.8.42)
2
In order to eliminate secular terms from the above equation, it is necessary to
consider different values of 1 , and further, if the Eq. (5.8.40) is considered, it is
also necessary to consider different values of 2 . Two cases are considered here: (i)
1 ≈ ω0 , 2 ≈ 2ω0 and (ii) 2 − 1 ≈ ω0 .
(i) 1 ≈ ω0 , 2 ≈ 2ω0 :
We use the detuning parameter σ to quantitatively describe the nearness of 1 to
ω0 , i.e.,

1 = ω0 + ε̂σ (5.8.43)

therefore,

1
D02 u1 + ω02 u1 = −2iω0 D1 Aeiω0 T0 − K̂eiσ T1 eiω0 T0 + cc (5.8.44)
2
In order to eliminate secular terms from the above equation, there must be

1
−2iω0 D1 A − K̂eiσ T1 = 0 (5.8.45)
2
then
1
A= K̂ω0−1 σ −1 eiσ T1 (5.8.46)
4
Substituting this into (5.8.14), the first order approximate solution of the system
is
1
u= K̂ω0−1 σ −1 cos(ω0 T0 + σ T1 ) + O(ε̂)
2
1
= K̂ω0−1 σ −1 cos ω0 + ε1/2 σ t + O ε1/2 (5.8.47)
2
366 5 Parametrically Excited Systems

Therefore, when ≈ ω0 is adopted, 2 ≈ 2ω0 is not required.

(ii) 2 − 1 ≈ ω0 :
Equation (5.8.39) becomes

1
D02 u1 + ω02 u1 = −2iω0 D1 Aeiω0 T0 − K̂ei1 T0 + cc (5.8.48)
2
In order to eliminate secular terms from the above equation, there must be

D1 A = 0 ⇒ A = A(T2 ) (5.8.49)

Thus the solution of (5.8.48) is

1
u1 = 2 K̂ei1 T0 + cc (5.8.50)
2 1 − ω02

Substituting u0 , u1 into (5.8.40) yields

1
D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1 + ω02 u03
6
− αu0 cos 2 T0
1
= −2iω0 D2 Aeiω0 T0 + ω02 A2 Aeiω0 T0 + cc + NST (5.8.51)
2
1
D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1 + ω02 u03 − αu0 cos 2 T0
6
In order to eliminate secular terms from the above equation, there must be

1 −
−2iD2 A + ω0 A2 A = 0 (5.8.52)
2

Let A = 21 aeiβ , we get

1
−ia + aβ + ω0 a3 = 0 (5.8.53)
16
Separating the real and imaginary parts of the above equation yields

a = 0
1
β = − ω0 a2 (5.8.54)
16
(5.8.36) has the steady state solution near u = 0, i.e.,
5.9 Exercise 5.9 (The Method of Harmonic Balance to Solve Duffing’s … 367

a = 0, β = 0 (5.8.55)

Therefore, the first order approximate solution of the system is

u = 0 + O ε1/2 (5.8.56)

This result shows that it is not possible to generate the combined resonant response
of 2 − 1 ≈ ω0 under soft excitation.

5.9 Exercise 5.9 (The Method of Harmonic Balance

to Solve Duffing’s Equation and Stability Analysis)

Solution: (a) Use the method of harmonic balance to find the first order approximate
solution of the Duffing equation. Let the solution of the equation be

u ≈ u0 = a cos ωt (5.9.1)

Substituting it into the original equation, we get

0 = ü + ω02 u + αu3 − K cos ωt

= −aω2 cos ωt + ω02 a cos ωt
+ αa3 cos3 ωt − K cos ωt
= −aω2 cos ωt + ω02 a cos ωt
3 1
+ αa3 cos ωt + αa3 cos 3ωt − K cos ωt
4
4
3 3
= −aω + ω0 a + αa − K cos ωt + · · ·
2 2
(5.9.2)
4

Let the coefficient of cos ωt be zero, we can obtain the frequency–response

equation

2 3
ω0 − ω2 a + αa3 = K (5.9.3)
4

(b) To examine the stability of the steady state solution, u0 , let

u = u0 + x (5.9.4)

where x is a small term. Substituting it into the original Duffing equation and keeping
linear terms in x yields
368 5 Parametrically Excited Systems

ẍ + ω02 x + 3αu02 x + ü0 + ω02 u0 + αu03 = K cos ωt

Considering ü0 + ω02 u0 + αu03 ≈ K cos ωt, we get

ẍ + ω02 + 3αu02 x = 0 (5.9.5)

(c) Substitute u0 = a cos ωt into (5.9.5), we obtain

ẍ + ω02 + 3αa2 cos2 ωt x = 0

i.e.,

3 2 3 2
ẍ + ω0 + αa + αa cos 2ωt x = 0
2
(5.9.6)
2 2

Thus, the stability of the periodic solution is transformed into determining the
stability of the solution of the Mathieu Eq. (5.9.6).
(d) We use the method of strained parameters to determine the transition curves of
equation. Since x << u0 , let the solution and parameters of the equation have the
following expansion:

x(t; ε) = εx1 (t) + ε2 x2 (t) + ε3 x3 (t) + · · · (5.9.7)

3
ω02 + αa2 = δ = δ0 + εδ1 + ε2 δ2 + · · · (5.9.8)
2
In order to let the parametric excitation term appearing in the control equation as
the higher order term, let

α = εα̂ (5.9.9)

Substituting the above three equations into (5.9.6) and retaining to O(ε3 ), we get
5.9 Exercise 5.9 (The Method of Harmonic Balance to Solve Duffing’s … 369

3
0 = ẍ + δ + εα̂a2 cos 2ωt x
2

= εẍ1 + ε ẍ2 + ε3 ẍ3 + δ0 + εδ1 + ε2 δ2 εx1 + ε2 x2 + ε3 x3
2

3
+ εα̂a2 εx1 + ε2 x2 + ε3 x3 cos 2ωt + · · ·
2
= εẍ1 + ε2 ẍ2 + ε3 ẍ3 + εδ0 x1 + ε2 δ0 x2 + ε3 δ0 x3 + ε2 δ1 x1 + ε3 δ1 x2 + ε3 δ2 x1
3 3
+ ε2 α̂a2 x1 cos 2ωt + ε3 α̂a2 x2 cos 2ωt + · · ·
2 2
3 2
= ε(ẍ1 + δ0 x1 ) + ε ẍ2 + δ0 x2 + δ1 x1 + α̂a x1 cos 2ωt
2
2

3
+ε ẍ3 + δ0 x3 + δ1 x2 + δ2 x1 + α̂a x2 cos 2ωt + · · ·
3 2
2

Equating coefficients of like powers of ε in the above equation yields

ẍ1 + δ0 x1 = 0 (5.9.10)

3
ẍ2 + δ0 x2 = −δ1 x1 − α̂a2 x1 cos 2ωt (5.9.11)
2
3
ẍ3 + δ0 x3 = −δ1 x2 − δ2 x1 − α̂a2 x2 cos 2ωt (5.9.12)
2
The solution of (5.9.10) is

x1 = b cos δ0 t + c sin δ0 t (5.9.13)

δ0 = ω 2 :

x1 = b cos ωt + c sin ωt (5.9.14)

x1 is a periodic solution with the period of T = 2π/ω. Substituting (5.9.14) into

(5.9.11) yields

ẍ2 + ω2 x2 = −δ1 (bcosωt + csinωt)

3
− α̂a2 (bcosωt + csinωt)cos2ωt
2

3
= − δ1 + α̂a2 (bcosωt + csinωt) + NST (5.9.15)
4

In order to eliminate secular terms from the above equation, there must be

3
δ1 = − α̂a2 (5.9.16)
4
370 5 Parametrically Excited Systems

Therefore, the transition curves emanating from δ = ω2 is

3 3
ω02 + αa2 = ω2 − εα̂a2 + O ε2
2 4
i.e.,

9
ω02 = ω2 − εα̂a2 + O ε2 (5.9.17)
4
Let

ω = ω0 + εσ

then the Eq. (5.9.17) can be written as

9 9
(ω − ω0 )(ω + ω0 ) − αa2 ≈ 0 ⇒ 2εσ ω0 (ω − ω0 ) − αa2 ≈ 0
4 4
i.e.,

9α̂a2
σ− ≈0 (5.9.18)
8ω0

5.10 Exercise 5.10 (The Method of Harmonic Balance

for Solving Pure Cubic Nonlinear Equations)

Solution: (a) Solve the given equation by the method of harmonic balance. Let the
approximate solution of the equation be

u ≈ u0 = a cos ωt + b cos 3ωt (5.10.1)

Substituting above equation into the original equation, we get.

0 = ü + u3 − K cos ωt
= −ω2 a cos ωt − 9ω2 b cos 3ωt + a3 cos3 ωt
+ 3a2 b cos2 ωt cos 3ωt
+ 3ab2 cos ωt cos2 3ωt + b3 cos3 3ωt − K cos ωt
3
= −ω2 a cos ωt − K cos ωt − 9ω2 b cos 3ωt + a3 cos ωt
4
5.10 Exercise 5.10 (The Method of Harmonic Balance for Solving Pure Cubic … 371

1 3 3
+ a3 cos 3ωt + a2 b cos 3ωt + a2 b cos ωt
4 2 4
3 2 3 2 3 2
+ a b cos 5ωt + ab cos ωt + ab cos 5ωt
4 2 4
3 2 3 3 1 3
+ ab cos 7ωt + b cos ωt + b cos 3ωt
4 4 4
3 3 3 3 3
= −ω a − K + a + a b + ab + b cos ωt
2 3 2 2
4 4 2 4

1 3 1
+ −9ω2 b + a3 + a2 b + b3 cos 3ωt + · · · (5.10.2)
4 2 4

Let the coefficients of cos ωt and cos 3ωt be zero, we get

−ω2 a − K + 34 a3 + 34 a2 b + 23 ab2 + 43 b3 = 0
(5.10.3)
−9ω2 b + 41 a3 + 23 a2 b + 41 b3 = 0

(b) To examine the stability of the steady state solution u0 , let

u = u0 + x (5.10.4)

where x is a small term. Substituting into the original equation and keeping linear
terms in x yields

ẍ + 3u02 x + ü0 + u03 = K cos ωt

Considering ü0 + u03 ≈ K cos ωt, we get

ẍ + 3u02 x = 0 (5.10.5)

Substituting u0 into (5.10.5) yields

ẍ + 3(a cos ωt + b cos 3ωt)2 x = 0

i.e.,

3 2
ẍ + a + b2 + a2 + 2ab cos 2ωt + 2ab cos 4ωt + b2 cos 6ωt x = 0 (5.10.6)
2
Thus, the stability of the steady state solution u0 is transformed into determining
the stability of the solution of (5.10.6).
(c) We use the method of strained parameters to determine the transition curves of
Eq. (5.10.6). Since x << u0 , it can be assumed that x is of order O(ε). In order to
make the time-varying parameter terms appear in the higher-order control equations
of x, the 3 periodic terms, cos 2ωt, cos 4ωt and cos 6ωt are of order O(ε). So, let the
372 5 Parametrically Excited Systems

solution and parameters of the equation have the following expansion:

x(t; ε) = εx1 (t) + ε2 x2 (t) + ε3 x3 (t) + · · · (5.10.7)

3 2
a + b2 = δ = δ0 + εδ1 + ε2 δ2 + · · · (5.10.8)
2
3 2 3 2
a + 2ab = εα1 , 3ab = εα2 , b = εα3 (5.10.9)
2 2

Substituting the above three equations into (5.10.6) and retaining to O(ε3 ), we get

0 = ẍ + (δ + εα1 cos 2ωt + εα2 cos 4ωt + εα3 cos 6ωt)x

= εẍ1 + ε2 ẍ2 + ε3 ẍ3

+ δ0 + εδ1 + ε2 δ2 εx1 + ε2 x2 + ε3 x3

+ εα1 εx1 + ε2 x2 + ε3 x3 cos 2ωt

+ εα2 εx1 + ε2 x2 + ε3 x3 cos 4ωt

+ εα3 εx1 + ε2 x2 + ε3 x3 cos 6ωt
= ε(ẍ1 + δ0 x1 )
+ ε2 (ẍ2 + δ0 x2 + δ1 x1 + α1 x1 cos 2ωt + α2 x1 cos 4ωt + α3 x1 cos 6ωt)
+ ε3 (ẍ3 + δ0 x3 + δ1 x2 + α1 x2 cos 2ωt + α2 x2 cos 4ωt + α3 x2 cos 6ωt)

Equating coefficients of like powers of ε in the above equation yields

ẍ1 + δ0 x1 = 0 (5.10.10)

ẍ2 + δ0 x2 = −δ1 x1 − α1 x1 cos 2ωt − α2 x1 cos 4ωt − α3 x1 cos 6ωt (5.10.11)

ẍ3 + δ0 x3 = −δ1 x2 − α1 x2 cos 2ωt − α2 x2 cos 4ωt − α3 x2 cos 6ωt (5.10.12)

The solution of (5.10.10) is

x1 = c cos δ0 t + d sin δ0 t (5.10.13)

CASE I:δ0 = ω2 :

x1 = c cos ωt + d sin ωt (5.10.14)

x1 is a periodic solution with period 2T = 2π/ω. Substituting (5.10.14) into (5.10.11)

yields
5.10 Exercise 5.10 (The Method of Harmonic Balance for Solving Pure Cubic … 373

ẍ2 + 4ω2 x2 = −δ1 (c cos ωt + d sin ωt)

− α1 (c cos ωt + d sin ωt) cos 2ωt
− α2 (c cos ωt + d sin ωt) cos 4ωt
− α3 (c cos ωt + d sin ωt) cos 6ωt

1
= − δ1 + α1 c cos ωt
2

1
− δ1 − α1 d sin ωt + NST (5.10.15)
2

In order to eliminate secular terms from the above equation, there must be

1
δ1 = ± α1 (5.10.16)
2

Therefore, the transition curves emanating from δ = ω2 is

3 2 1
a + b2 = δ = ω2 ± εα1 + O ε2 + · · ·
2 2
i.e.,

3 2 3
ω2 = a + b2 ± a2 + 2ab + O ε2 (5.10.17)
2 4

CASE II:δ0 = 4ω2 :

x1 = c cos 2ωt + d sin 2ωt (5.10.18)

x1 is a periodic solution with period T = π/ω. Substituting (5.10.18) into (5.10.11)

yields

ẍ2 + 4ω2 x2 = −δ1 (c cos 2ωt + d sin 2ωt)

− α1 (c cos 2ωt + d sin 2ωt) cos 2ωt
− α2 (c cos 2ωt + d sin 2ωt) cos 4ωt
− α3 (c cos 2ωt + d sin 2ωt) cos 6ωt

1
= − δ1 + α2 c cos 2ωt
2

1
− δ1 − α2 d sin 2ωt + NST (5.10.19)
2

In order to eliminate secular terms from the above equation, there must be
374 5 Parametrically Excited Systems

1
δ1 = ± α2 (5.10.20)
2

Therefore, the transition curves emanating from δ = 4ω2 is

3 2 1
a + b2 = δ = 4ω2 ± εα2 + O ε2 + · · ·
2 2
i.e.,

3 2
ω2 = a + b2 ± ab + O ε2 (5.10.21)
8

5.11 Exercise 5.11 (The Quadratic Parametric Excitation

of a Linear Viscous Damping System)

Solution: (a) We use the method of multiple scales to solve the first order approximate
periodic solution of the system. Let the solution of the equation be

u(t; ε) = u0 (T0 , T1 , T2 ) + εu1 (T0 , T1 , T2 ) + ε2 u2 (T0 , T1 , T2 ) + · · · (5.11.1)

Substituting the above equation into the original equation and retaining to O(ε2 ),
we get

0 = ü + ω02 u + 2εμu̇ + 2εu2 cos 2t

= D02 + 2εD0 D1 + ε2 D12 + 2D0 D2 u0 + εu1 + ε2 u2

+ ω02 u0 + εu1 + ε2 u2 + 2εμ(D0 + εD1 ) u0 + εu1 + ε2 u2
+ 2ε(u0 + εu1 + ε2 u2 )2 cos 2t
= D02 u0 + εD02 u1 + ε2 D02 u2 + 2εD0 D1 u0

+ 2ε2 D0 D1 u1 + ε2 D12 + 2D0 D2 u0
+ ω02 u0 + εω02 u1 + ε2 ω02 u2 + 2εμD0 u0
+ 2ε2 μD0 u1 + 2ε2 μD1 u0
+ 2εu02 cos 2t + 4ε2 u0 u1 cos 2t + · · ·
= D02 u0 + ω02 u0

+ ε D02 u1 + ω02 u1 + 2D0 D1 u0 + 2μD0 u0 + 2u02 cos 2T0

+ ε2 [D02 u2 + ω02 u2 + D12 + 2D0 D2 u0 + 2D0 D1 u1
+ 2μD0 u1 + 2μD1 u0 + 4u0 u1 cos 2T0 ] + · · ·
5.11 Exercise 5.11 (The Quadratic Parametric Excitation of a Linear Viscous … 375

Equating coefficients of like powers of ε in the above equation yields

D02 u0 + ω02 u0 = 0 (5.11.2)

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − 2u02 cos 2t (5.11.3)

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1
(5.11.4)
−2μD0 u1 − 2μD1 u0 − 4u0 u1 cos2t

The solution of (5.11.2) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.11.5)

where A = A(T1 , T2 ). Let

1 iβ
A= ae (5.11.6)
2
Substituting this into (5.11.5), we can obtain the first order approximate solution
of the given equation

u ≈ u0 = a cos(ω0 t + β) + O(ε) (5.11.7)

Substituting u0 into (5.11.3) yields

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − 2u02 cos 2T0

= −2iω0 D1 Aeiω0 T0 − 2iω0 μAeiω0 T0 − 2AAe2iT0
2
− A2 ei(2ω0 −2)T0 − A ei(2−2ω0 )T0 + cc + NST (5.11.8)

CASE I: When ω0 is far from 2 or 23 :

In order to eliminate secular terms from (5.11.8), we need

−A − μA = 0 (5.11.9)

The prime denotes the derivative with respect to T1 . Substituting A = 21 aeiβ into
the above equation gives

a + iaβ + μa = 0 (5.11.10)

Separating the real and imaginary parts of the above equation yields

a = −μa, β = 0 (5.11.11)
376 5 Parametrically Excited Systems

CASE II: When ω0 = 2 + εσ :

In order to eliminate secular terms from (5.11.8), we need
−
−2iω0 A − 2iω0 μA − 2A A e−iσ T1 − A2 eiσ T1 = 0 (5.11.12)

The prime denotes the derivative with respect to T1 . Substituting A = 21 aeiβ into
the above equation gives

1 1
iω0 a − ω0 aβ + iω0 μa + a2 e−i(σ T1 +β) + a2 ei(σ T1 +β) = 0 (5.11.13)
2 4
Separating the real and imaginary parts of the above equation yields

1
a = −μa + ω0−1 a2 sin(σ T1 + β) a
4
3 −1 2
β = ω a cos(σ T1 + β)
4 0
Considering ω0 ≈ 2, the above equation becomes

1 3
a = −μa + a2 sin γ , aβ = a2 cos γ (5.11.14)
8 8
where

γ = σ T1 + β (5.11.15)

CASE III: When 3ω0 = 2 + εσ :

In order to eliminate secular terms from (5.11.8), we need

−2
−2iω0 A − 2iω0 μA − A e−iσ T1 = 0 (5.11.16)

The prime denotes the derivative with respect to T1 . Substituting A = 21 aeiβ into
the above equation gives

1
iω0 a − ω0 aβ + iω0 μa + a2 e−i(σ T1 +3β) = 0 (5.11.17)
4
Separating the real and imaginary parts of the above equation yields

1 1
a = −μa + ω0−1 a2 sin(σ T1 + 3β)a β = ω0−1 a2 cos(σ T1 + 3β)
4 4

Considering ω0 ≈ 23 , the above equation becomes

5.11 Exercise 5.11 (The Quadratic Parametric Excitation of a Linear Viscous … 377

3 3
a = −μa + a2 sin γ , aβ = a2 cos γ (5.11.18)
8 8
where

γ = σ T1 + 3β (5.11.19)

(b) We only determine the stability of the steady-state solution in the second case
above, readers are invited to analyze the rest of the cases on their own. Writing the
Eq. (5.11.14) as

3 9
a = −μa + a2 sin γ , aγ = aσ + a2 cos γ (5.11.20)
8 8

Let a = γ = 0 in (5.11.20), we obtain

3 9
−μa + a2 sin γ = 0, aσ + a2 cos γ = 0 (5.11.21)
8 8
The steady state solutions can be obtained:
(1) a = 0, γ are arbitrary constants;
(2) a = 0, γ are certain constants.
(1) a = 0 and γ = γ 0 are arbitrary constants.
Let

a = 0 + a1 , γ = γ0 + γ1 (5.11.22)

Substituting (5.11.22) into (5.11.20), we can obtain the corresponding linearized

equation

a1 = −μa1 (5.11.23)

So a1 = e−μT1 = e−εμt → 0, therefore this steady state solution is stable.

(2) a = a0 = 0, γ = γ 0
For this case, the frequency–response equation is

1 9 4
(μa)2 + (aσ )2 = a (5.11.24)
9 64
Let

a = a0 + a1 , γ = γ0 + γ1 (5.11.25)
378 5 Parametrically Excited Systems

Substituting (5.11.25) into (5.11.20), we can obtain the corresponding linearized

equation

a1 −μ + 43 a0 sin γ0 38 a02 cos γ0 a1
= (5.11.26)
γ1 σ a0−1 + 94 cos γ0 − 98 a0 sin γ0 γ1

For any set of steady state solutions a = a0 = 0, γ = γ 0 , one can calculate

the eigenvalues of Eq. (5.11.26), and the stability of the corresponding steady state
solutions can be determined from the sign of the real part of the two eigenvalues.

5.12 Exercise 5.12 (The Cubic Parametric Excitation

of a Linear Viscous Damping System)

Solution: (a) We use the method of multiple scales to solve the first order approximate
periodic solution of the system. Let the solution of the equation be

u(t; ε) = u0 (T0 , T1 , T2 ) + εu1 (T0 , T1 , T2 ) + ε2 u2 (T0 , T1 , T2 ) + · · · (5.12.1)

Substituting the above equation into the original equation and retaining to O(ε2 ),
we get

0 = ü + ω02 u + 2εμu̇ + 2εu3 cos 2t

= D02 + 2εD0 D1 + ε2 D12 + 2D0 D2

u0 + εu1 + ε2 u2

+ ω02 u0 + εu1 + ε2 u2

+ 2εμ(D0 + εD1 ) u0 + εu1 + ε2 u2
+ 2ε(u0 + εu1 + ε2 u2 )3 cos 2t
= D02 u0 + εD02 u1 + ε2 D02 u2 + 2εD0 D1 u0

+ 2ε2 D0 D1 u1 + ε2 D12 + 2D0 D2 u0
+ ω02 u0 + εω02 u1 + ε2 ω02 u2 + 2εμD0 u0
+ 2ε2 μD0 u1 + 2ε2 μD1 u0
+ 2εu03 cos 2t + 6ε2 u0 u1 cos 2t + · · ·
= D02 u0 + ω02 u0

+ ε D02 u1 + ω02 u1 + 2D0 D1 u0 + 2μD0 u0 + 2u03 cos 2T0

+ ε2 [D02 u2 + ω02 u2 + D12 + 2D0 D2 u0 + 2D0 D1 u1
+ 2μD0 u1 + 2μD1 u0 + 6u0 u1 cos 2T0 ] + · · ·
5.12 Exercise 5.12 (The Cubic Parametric Excitation of a Linear Viscous … 379

Equating coefficients of like powers of ε in the above equation yields

D02 u0 + ω02 u0 = 0 (5.12.2)

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − 2u03 cos 2T0 (5.12.3)

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1
− 2μD0 u1 − 2μD1 u0 − 6u0 u1 cos2t (5.12.4)

The solution of (5.12.2) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.12.5)

where A = A(T1 , T2 ). Let

1 iβ
A= ae (5.12.6)
2
and substitute this into (5.12.5), we can obtain the first order approximate solution
of the given equation

u ≈ u0 = a cos(ω0 t + β) + O(ε) (5.12.7)

Substituting u0 into (5.12.3) yields

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − 2u03 cos2T0

= −2iω0 D1 Aeiω0 T0 − 2iω0 μAeiω0 T0
2
− 3AA ei(2−ω0 )T0 + cc
3
− A ei(2−3ω0 )T0 − A3 ei(3ω0 −2)T0 + NST (5.12.8)

CASE I: When ω0 is far from 1 or 21 :

In order to eliminate secular terms from (5.12.8), we need

−A − μA = 0 (5.12.9)

The prime denotes the derivative with respect to T1 . Substituting A = 21 aeiβ into
the above equation gives

a + iaβ + μa = 0 (5.12.10)

Separating the real and imaginary parts of the above equation yields
380 5 Parametrically Excited Systems

a = −μa, β = 0 (5.12.11)

CASE II: When ω0 = 1 + εσ :

In order to eliminate secular terms from (5.12.8), we need

−2
−2iω0 A − 2iω0 μA − 3A A e−i2σ T1 − A3 ei2σ T1 = 0 (5.12.12)

The prime denotes the derivative with respect to T1 . Substituting A = 21 aeiβ into
the above equation gives

3
iω0 a − ω0 aβ + iω0 μa + a3 e−i(2σ T1 +2β)
8
1 3 i(2σ T1 +2β)
+ a e =0 (5.12.13)
8
Separating the real and imaginary parts of the above equation and taking into
account ω0 ≈ 1, we get

1 1
a = −μa + a3 sinγ , aβ = a3 cosγ (5.12.14)
4 2
where

γ = 2σ T1 + 2β (5.12.15)

CASE III: When 2ω0 = 1 + εσ :

In order to eliminate secular terms from (5.12.8), we need

−3
−2iω0 D1 Aeiω0 T0 − 2iω0 μAeiω0 T0 − A e−2iσ T1 eiω0 T0 = 0 (5.12.16)

Substituting A = 21 aeiβ into the above equation gives

1
iω0 a − ω0 aβ + iω0 μa + a3 e−i(2σ T1 +4β) = 0 (5.12.17)
8
The prime denotes the derivative with respect to T1 . Separating the real and
imaginary parts of the above equation and taking into account ω0 ≈ 21 yields

1 1
a = −μa + a3 sinγ , aβ = a3 cosγ (5.12.18)
4 4
where

γ = 2σ T1 + 4β (5.12.19)
5.12 Exercise 5.12 (The Cubic Parametric Excitation of a Linear Viscous … 381

(b) We only determine the stability of the steady-state solution in the second case
above, readers are invited to analyze the rest of the cases on their own. Writing the
Eq. (5.12.14) as

1
a = −μa + a3 sinγ , aγ = 2σ a + a3 cosγ (5.12.20)
4

Let a = γ = 0 in (5.12.20), we obtain

1
−μa + a3 sinγ = 0, 2σ a + a3 cosγ = 0 (5.12.21)
4
The steady state solutions can be obtained:
(1) a = 0, γ are arbitrary constants;
(2) a = 0, γ are certain constants.

(1) a = 0 and γ = γ 0 are arbitrary constants.

Let

a = 0 + a1 , γ = γ0 + γ1 (5.12.22)

Substituting (5.12.22) into (5.12.20), we can obtain the corresponding linearized

equation

a1 = −μa1 (5.12.23)

So a1 = e−μT1 = e−εμt → 0, therefore this steady state solution is stable.

(2) a = a0 = 0, γ = γ 0
For this case, the frequency–response equation is

16(μa)2 + (σ a)2 = a6 (5.12.24)

Let

a = a0 + a1 , γ = γ0 + γ1 (5.12.25)

Substituting (5.12.25) into (5.12.20), we can obtain the corresponding linearized

equation

a1 −μ + 43 a02 sinγ0 41 a03 cosγ0 a1
= (5.12.26)
γ1 2σ + 3a02 cosγ0 −a03 sinγ0 γ1
382 5 Parametrically Excited Systems

For any set of steady state solutions a = a0 = 0,γ = γ 0 , one can calculate the
eigenvalues of Eq. (5.12.26), and the stability of the corresponding steady state
solutions can be determined from the sign of the real part of the two eigenvalues.

5.13 Exercise 5.13 (The High Order Nonlinear Parametric

Excitation of a Linear Viscous Damping System)

Solution: (a) We use the method of multiple scales to solve the first order approximate
periodic solution of the system. Let the solution of the equation be

u(t; ε) = u0 (T0 , T1 , T2 ) + εu1 (T0 , T1 , T2 ) + ε2 u2 (T0 , T1 , T2 ) + · · · (5.13.1)

Substituting the above equation into the original equation and retaining to O(ε2 ),
we get

0 = ü + ω02 u + 2εμu̇ + 2εun cos 2t

= D02 + 2εD0 D1 + ε2 D12 + 2D0 D2

u0 + εu1 + ε2 u2

+ ω02 u0 + εu1 + ε2 u2

+ 2εμ(D0 + εD1 ) u0 + εu1 + ε2 u2
+ 2ε(u0 + εu1 + ε2 u2 )n cos 2t
= D02 u0 + εD02 u1 + ε2 D02 u2 + 2εD0 D1 u0

+ 2ε2 D0 D1 u1 + ε2 D12 + 2D0 D2 u0
+ ω02 u0 + εω02 u1 + ε2 ω02 u2 + 2εμD0 u0
+ 2ε2 μD0 u1 + 2ε2 μD1 u0
+ 2εu0n cos 2t + 2nε2 u0n−1 u1 cos 2t + · · ·
= D02 u0 + ω02 u0

+ ε D02 u1 + ω02 u1 + 2D0 D1 u0 + 2μD0 u0 + 2u0n cos 2t

+ ε2 [D02 u2 + ω02 u2 + D12 + 2D0 D2 u0 + 2D0 D1 u1
+ 2μD0 u1 + 2μD1 u0 + 2nu0n−1 u1 cos 2t] + · · ·

Equating coefficients of like powers of ε in the above equation yields

D02 u0 + ω02 u0 = 0 (5.13.2)

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − 2u0n cos 2T0 (5.13.3)

5.13 Exercise 5.13 (The High Order Nonlinear Parametric Excitation … 383

D02 u2 + ω02 u2 = − D12 + 2D0 D2 u0 − 2D0 D1 u1
− 2μD0 u1 − 2μD1 u0 − 2nu0n−1 u1 cos2t (5.13.4)

The solution of (5.13.2) is

−
u0 = Aeiω0 T0 + A e−iω0 T0 (5.13.5)

where A = A(T1 , T2 ). Let

1 iβ
A= ae (5.13.6)
2
Substituting this into (5.13.5), we can obtain the first order approximate solution
of the given equation

u ≈ u0 = a cos(ω0 t + β) + O(ε) (5.13.7)

Substituting u0 into (5.12.3) yields

D02 u1 + ω02 u1 = −2D0 D1 u0 − 2μD0 u0 − 2u0n cos 2T0

= −2iω0 D1 Aeiω0 T0 − 2iω0 μAeiω0 T0
− (Aeiω0 T0 + Ae−iω0 T0 )n ei2T0 + cc
= −2iω0 D1 Aeiω0 T0 − 2iω0 μAeiω0 T0
n
k
− Cnk An−k A ei[(n−2k)ω0 +2]T0 + cc (5.13.8)
k=0

From the above equation, when the following equation is satisfied

(n − 2k)ω0 + 2 ≈ ω0 (5.13.9)

i.e.,

2
ω0 ≈ , k = 1, 2, . . . , n (5.13.10)
1 + 2k − n

parametric resonance occurs and the first-order solution of Eq. (5.13.8) can be
obtained. Since ω 0 ≥ 0 is required,1
+ 2k − n ≥ 1, i.e., k is the integer greater
than or equal to 2n , where 2n denotes the largest integer less than or equal to 2n .
Therefore, for the case of n is an odd number, parameter resonance occurs when.

2 2 2 2
ω0 ≈ 1, , , , ..., for odd n (5.13.11)
4 6 8 n+1