Solutions: Homework Set 4

1. Problem 3.16. The paragraph before Eq. (3.52) explains why the amplitude of the
vibration, for the rotating unbalanced mass excitation case, is obtained by replacing
the constant A in Eq. (3.28) with (m/M )eω 2 /ωn2 . The amplitude of the transmitted
force found in Eq. (3.94) is for the case of harmonic force excitation, rather than
rotating unbalanced mass excitation. When the vibration amplitude is given by Eq.
(3.52) rather than Eq. (3.28), the transmitted force amplitude becomes:
hm i h i
2 2
Ftr = e (ω/ωn ) k 1 + (2ζ (ω/ωn )) |G(iω)|
M
  12
 2 
m 2
 1 + [2ζ (ω/ωn )] 
= e (ω/ωn ) k h i2
M  1 − (ω/ωn )2 + [2ζ (ω/ωn )]2 
 

Considering that (ω/ωn ) = 4, the quantity in {}’s is 1 + (8ζ)2 / 225 + (8ζ)2 , which
has a value of 1/225 when ζ = 0 and asymptotically approaches 1 as ζ increases.
Inserting the given parameters and setting Ftr equal to the limit of 250 N yields a
maximum value for ζ:
 12
1 + (8ζ)2

5
Ftr = × 0.1 × 42 × 8000 × = 250 ⇒ ζmax = 0.6026
80 (1 − 42 )2 + (8ζ)2

This corresponds to a dashpot coefficient of cmax = 2ζmax ωn M = 964.2 Nm·s . Any

value of c below cmax (even c = 0) would keep Ftr below the limit. However, some
damping may be desirable for controlling the amplitude of vibration and transmitted
force as the system accelerates through resonance to the operating speed. If vibration
during the acceleration through resonance is a significant issue, a value of c close to
cmax should be chosen. If not, a lower level of damping would reduce the amount of
heat produced in the damper at operating speed.

2. Problem 3.19. EOM: mẍ + cẋ + kx = kf (t) ⇒ ẍ + 2ζωn ẋ + ωn2 x = ωn2 f (t) where
k c
ωn2 = and 2ζωn = . From Fig. 3.36,
m m
1 1
 −A, − 2 T < t < − 4 T


f (t) = A, − 14 T < t < 14 T

−A, 14 T < t < 12 T


Since f (t) is even, the real Fourier series representation becomes

∞
a0 X
f (t) = + ap cos(pω0 t)
2 p=1

1
and the imaginary parts of Cp and Ap coefficients are zero in the alternative complex
Fourier series representations
∞ ∞
!
X A0
X
f (t) = Cp eipω0 t = + Re Ap eipω0 t
p=−∞
2 p=1

For p = 0,
" Z Z T /2 #
−T /4 Z T /4
2A
a0 = − dt + dt − dt = 0 = 2C0 = A0
T −T /2 −T /4 T /4

For p 6= 0,
" Z #
−T /4 Z T /4 Z T /2
2A
ap = − cos pω0 t dt + cos pω0 t dt − cos pω0 t dt
T −T /2 −T /4 T /4

A h pπ i 4A pπ
= 4 sin − 2 sin pπ = sin
πp 2 pπ 2
or " Z #
−T /4 Z T /4 Z T /2
A
Cp = − e−ipω0 t dt + e−ipω0 t dt − e−ipω0 t dt
T −T /2 −T /4 T /4

A h pπ i 2A pπ
= 2 sin − sin pπ = sin
πp 2 pπ 2
or Ap = 2Cp . Since ap = Cp = Ap = 0 for even p, letting p = 2r − 1 so r = (p + 1)/2
yields
∞ ∞
4A X (−1)r+1 2A X (−1)r+1 i(2r−1)ω0 t
f (t) = cos(2r − 1)ω0 t = e
π r=1 2r − 1 π r=−∞ 2r − 1
∞
" #
4A X (−1)r+1 i(2r−1)ω0 t
= Re e
π r=1 2r − 1

Response:
∞
4A X (−1)r+1
x(t) = |Gr | cos[(2r − 1)ω0 t − φr ]
π r=1 2r − 1
∞
2A X (−1)r+1
= Gr ei(2r−1)ω0 t
π r=−∞ 2r − 1
∞
( )
4A X (−1)r+1
= Re Gr ei(2r−1)ω0 t
π r=1 2r − 1
∞
( )
4A X (−1)r+1
= Re |Gr |ei[(2r−1)ω0 t−φr ]
π r=1 2r − 1

2
 where all representations are equal, the last simplifying easily to the first, for example,
and
1
Gr = 2
1 − [(2r − 1)ω0 /ωn ] + i2ζ(2r − 1)ω0 /ωn

so that
1
|Gr | = 1
{[1 − [(2r − 1)ω0 /ωn ]2 ]2 + [2ζ(2r − 1)ω0 /ωn ]2 } 2

and
2ζ(2r − 1)ω0 /ωn
φr = tan−1
1 − [(2r − 1)ω0 /ωn ]2

3. Problem 3.18. From FBD of m: mẍ = −cẋ − k1 x − k2 (x − y) which is:

k2
ẍ + 2ζωn ẋ + ωn2 x = y
m

c k1 + k2 A
where 2ζωn = , ωn2 = . Over the period 0 < t < T , y(t) = t + B so
m m T
A
the average value of y(t) is + B. This average value determines the a0 , C0 or A0
2
coefficients, and the time-dependent terms in the Fourier series represent the variation
away from the (constant) average value. Since y(t) − (A/2 + B) is an odd function of
time, the ap coefficients are zero for p > 0, or if one of the complex exponential forms
of Fourier series is used, the real parts of Cp or Ap coefficients are equal to zero for
p 6= 0. Alternative Fourier series representations are:

∞ ∞ ∞
!
a0 X X
ipω0 t A0 X
y(t) = + bp sin pω0 t = Cp e = + Re Ap eipω0 t
2 p=1 p=−∞
2 p=1

Here,

Z T Z T    
2 2 A A
a0 = 2C0 = A0 = y(t) dt = t+B dt = 2 +B
T 0 T 0 T 2

and
Z T Z T  
2 2 A A
bp = y(t) sin pω0 t dt = t + B sin pω0 t dt = −
T 0 T 0 T πp
or
Z T Z T  
1 −ipω0 t 1 A iA
Cp = y(t)e dt = t + B e−ipω0 t dt = ,
T 0 T 0 T 2πp

3
 iA
for p 6= 0, or Ap = 2Cp = for p 6= 0. EOM:
πp

∞
" #
k2 A AX1
ẍ + 2ζωn ẋ + ωn2 x = +B− sin pω0 t
m 2 π p=1 p
" ∞ #
k2 X
= Cp eipω0 t
m p=−∞
∞
" !#
k2 A X
= + B + Re Ap eipω0 t
m 2 p=1

Response:

∞
" #
k2 A AX1
x(t) = +B− |Gp | sin(pω0 t − φp )
k1 + k2 2 π p=1 p
" ∞ #
k2 X
= Cp Gp eipω0 t
k1 + k2 p=−∞
∞
" !#
k2 A A X i
= + B + Re Gp eipω0 t
k1 + k2 2 π p=1
p
∞
" !#
k2 A A X i
= + B + Re |Gp |ei(pω0 t−φp )
k1 + k2 2 π p=1
p

where you should note that all representations are equal (the last simplifies easily to
the first, for example), and

1
Gp =
1 − (pω0 /ωn )2 + i2ζpω0 /ωn

so that

1 2ζpω0 /ωn
|Gp | = 1 , φp = tan−1
{[1 − (pω0 /ωn )2 ]2 + (2ζpω0 /ωn )2 } 2 1 − (pω0 /ωn )2

4. Problem 4.2. Eq. (4.22) gives the step response of a viscously damped single-degree-
of-freedom system. Since the unit step function is the integral of the unit impulse, the
unit impulse is the time derivative of the unit step function and the impulse response
is the time derivative of the step response. Differentiating Eq. (4.22) results in the
impulse response, given in Eq. (4.11).

4
 t c
5. Problem 4.4. EOM: mẍ + cẋ + kx = t u(t). Particular solution: xp (t) = − 2.
k k
Homogeneous solution: xh (t) = e−ζωn t (C1 cos ωd t + C2 sin ωd t). Constants: C1 =
c 2ζ 2ζ 2 − 1
= , C2 = . Ramp response:
k2 kωn kωd

2ζ 2 − 1
  
1 2ζ −ζωn t 2ζ
r(t) = t− +e cos ωd t + sin ωd t u(t)
k ωn ωn ωd

2ζ
For x0 and v0 in Eq. (2.32) to agree with the homogeneous solution: x0 = kω n
=
1
−xp (0) and v0 = − k = −ẋp (0) so the ramp response is the sum of the particular
solution and a free response solution whose initial conditions cancel the initial dis-
placement and velocity associated with the particular solution.

F0
6. Problem 4.5. Since F (t) = [t u(t) − (t − T )u(t − T )], the response is
T
F0
x(t) = [r(t) − r(t − T )]
T 
2ζ 2 − 1
 
F0 2ζ −ζωn t 2ζ
= t− +e cos ωd t + sin ωd t u(t)
kT ωn ωn ωd
2ζ 2 − 1
  
2ζ −ζωn (t−T ) 2ζ
− t−T − +e cos ωd (t − T ) + sin ωd (t − T )
ωn ωn ωd

· u(t − T )

For the undamped case, this reduces to

    
F0 1 1
x(t) = t− sin ωn t u(t) − t − T − sin ωn (t − T ) u(t − T )
kT ωn ωn

which agrees with the results from the lecture.