Chapter 2 Response to Harmonic

Excitation
Introduces the important
concept of resonance

© D. J. Inman https://www.youtube.com/watch?v=SPHoumJvVtQ
1/49 University of Michigan
https://www.youtube.com/watch?v=DKql4Is8Pas
2.1 Harmonic Excitation of Undamped
Systems
• Consider the usual spring mass damper system
with applied force F(t)=F0cosωt
• ω is the driving frequency
• F0 is the magnitude of the applied force
• We take c = 0 to start with
Displacement
x
F=F0cosωt k
M

© D. J. Inman
2/49 University of Michigan
 Equations of motion
Figure 2.1
• Solution is the sum of
homogenous and
particular solution
• The particular solution
assumes form of
forcing function
(physically the input
wins):

© D. J. Inman
3/49 University of Michigan
 Substitute particular solution into
the equation of motion:

Thus the particular solution has the form:

© D. J. Inman
4/49 University of Michigan
 Add particular and homogeneous
solutions to get general solution:

(2.8)

© D. J. Inman
5/49 University of Michigan
 Apply the initial conditions to
evaluate the constants

(2.11)
© D. J. Inman
6/49 University of Michigan
 Comparison of free and forced
response
• Sum of two harmonic terms of different frequency
• Free response has amplitude and phase effected
by forcing function
• Our solution is not defined for ωn = ω because it
produces division by 0.
• If forcing frequency is close to natural frequency
the amplitude of particular solution is very large

© D. J. Inman
7/49 University of Michigan
 Response for m=100 kg, k=1000 N/m, F=100 N, ω = ωn +5
v0=0.1m/s and x0= -0.02 m.

0.05
Displacement (x)

-0.05
0 2 4 6 8 10
Time (sec)
Note the obvious presence of two harmonic signals
© D. J. Inman
University of Michigan Go to code demo
8/49
What happens when ω is near ωn?
(Eq. 2.11) after assuming x0 & v0 = 0
and using trigonometry

When the drive frequency and natural frequency

are close a beating phenomena occurs
1

0.5
Displacement (x)

0
Larger
amplitude
-0.5
Beat oscillation

-1
0 5 10 15 20 25 30
Time (sec)
© D. J. Inman Called beating
9/49 University of Michigan
 What happens when ω is ωn?
x p (t ) tX sin(ωt ) is used for paticular
= sin(ωt ) is not valid at ω ωn
solution since x p (t ) X=
substitute into eq. of motion and solve for X
x(t ) + ωn2 x(t ) =
f 0 cos(ωt )
f0
and X=
2ω
5
When the drive
frequency and natural
Displacement (x)

frequency are the

same the amplitude 0
of the vibration grows
without bounds. This
is known as a
resonance condition.
The most important -5
0 5 10 15 20 25 30
concept in Chapter 2!
Time (sec)
© D. J. Inman
10/49 University of Michigan Figure 2.5
Example 2.1.1: Compute and plot the response for
m=10 kg, k=1000 N/m, x0=0,v0=0.2 m/s, F=23 N, ω =2ωn.

© D. J. Inman
11/49 University of Michigan
Example 2.1.2 Given zero initial conditions a harmonic input of
10 Hz with 20 N magnitude and k= 2000 N/m, and measured response
amplitude of 0.1m, compute the mass of the system.
ω=2πf, ω = (k/m)1/2
v0  f0  f0
x(t )
= sin ωnt +  x0 − 2  cos ωnt + 2 cos ωt
ωn  ωn − ω 2  ωn − ω 2

(2.13)

© D. J. Inman
12/49 University of Michigan
 Example 2.1.3 Design a rectangular mount
for a security camera subjected to wind load.

0.01 x 0.01 m
in cross section
EAl = 7.1x1010 N/m2

Compute  > 0.2 m so that the mount keeps the camera from
vibrating more then 0.01 m of maximum amplitude under a wind
load of 15 N at 10 Hz. The mass of the camera is 3 kg.
© D. J. Inman
13/49 University of Michigan
Solution:Modeling the mount and camera as a
beam with a tip mass, and the wind as harmonic, the
equation of motion becomes:

From strength of materials: f 0 = F0 / m

Mass normalized force

Thus the frequency expression is:

Here we are interested computing  that will make the

amplitude less then 0.01m in equation (2.13):

© D. J. Inman
14/49 University of Michigan
Case (a) (assume aluminum for the material):

Case (b):

© D. J. Inman
15/49 Mechanical Engineering at Virginia Tech
Remembering the constraint that the length
must be at least 0.2 m, (a) and (b) yield

To check, note that

Next check the mass of the designed beam to

insure it does not change the frequency. Note
it is less then 5 % of the camera mass, so it is
reasonable to ignore the mass of the support
according to example 1.4.4.
© D. J. Inman
16/49 Mechanical Engineering at Virginia Tech
A harmonic force may also be represented by
sine or a complex exponential. How does this
change the solution?

The particular solution then becomes a sine:

Substitution of (2.19) into (2.18) yields:

Solving for the homogenous solution and evaluating the constants yields

© D. J. Inman
17/49 Mechanical Engineering at Virginia Tech
 Section 2.2 Harmonic Excitation of
Damped Systems
Extending resonance and response
calculation to damped systems

© D. J. Inman
18/49 University of Michigan
2.2 Harmonic excitation of damped
systems
(2.26)

(2.27)

Phase shift is included due to the effect of damping force. Displacement

x
k
F=F0cosωt
M
c
© D. J. Inman
19/49 University of Michigan
 Let xp have the form:

Note that we are using the rectangular form, but

we could use one of the other forms of the solution.
© D. J. Inman
20/49 University of Michigan
Substitute into the equations of motion

© D. J. Inman
21/49 University of Michigan
 Write as a matrix equation:

Solving for As and Bs:

© D. J. Inman
22/49 University of Michigan
 Substitute the values of As and Bs into xp:

Add homogeneous and particular to get total solution:

(2.37)

Note: that A and φ will not have the same values as in Ch 1,

for the free response. Also as t gets large, transient dies out.
© D. J. Inman
23/49
University of Michigan
 Things to notice about damped forced
response
• If ζ = 0, undamped equations result
• Steady state solution prevails for large t
• Often we ignore the transient term (how
large is ζ, how long is t?)
• Coefficients of transient terms (constants
of integration) are effected by the initial
conditions AND the forcing function
• For underdamped systems at resonance
the, amplitude is finite.
© D. J. Inman
24/49 University of Michigan
Example 2.2.3: ωn = 10 rad/s, ω = 5 rad/s, ζ = 0.01, F0= 1000 N, m = 100
kg, and the initial conditions x0 = 0.05 m and v0 = 0. Compare
amplitude A and phase Φ for forced and unforced case:
Using the equations on slide 23:

Differentiating yields:

The numbers in ( ) are

those obtained by
© D. J. Inman incorrectly using the free
25/49 University of Michigan
response values
 Proceeding with ignoring the transient

• Always check to make sure the

transient is not significant
• For example, transients are very
important in earthquakes
• However, in many machine
applications transients may be ignored

© D. J. Inman
26/49 University of Michigan
 Proceeding with ignoring the transient

Magnitude:
(2.39)

Frequency ratio:

Non dimensional
Form:

(2.40)
Phase:

© D. J. Inman
27/49 University of Michigan
 Magnitude plot
• Resonance is close to
r=1
• For ζ = 0, r =1 defines 40

resonance 30
• As ζ grows resonance
moves r <1, and X 20

decreases
X (dB)

10
• The exact value of r,
can be found from 0
differentiating the
-10
magnitude
-20
0 0.5 1 1.5 2
r
© D. J. Inman
Fig 2.7
28/49 University of Michigan
 Phase plot

• Resonance occurs
at Φ = π/2
• The phase changes 3.5

more rapidly when 3

ζ =0.01
ζ =0.1
the damping is ζ =0.3
2.5
small ζ =0.5
Phase (rad)

ζ =1
• From low to high 2

values of r the 1.5

phase always
1
changes by 1800 or
π radians 0.5

0
0 0.5 1 1.5 2

© D. J. Inman
r
Fig 2.7
29/49 University of Michigan
 Example 2.2.3 Compute max peak
by differentiating:

(2.41)

(2.42)

© D. J. Inman
30/49 University of Michigan
 Effect of Damping on Peak Value
30

• The top plot shows how 25

the peak value becomes 20

very large when the 15

damping level is small 10

• The lower plot shows 5

how the frequency at 0

0 0.2 0.4 0.6 0.8
which the peak value ζ
occurs reduces with 1

increased damping 0.8

• Note that the peak value 0.6

is only defined for rpeak
0.4
values ζ<0.707
0.2

0
0 0.2 0.4 0.6 0.8
© D. J. Inman ζ
31/49 University of Michigan Fig 2.9
 Section 2.3 Alternative
Representations
• A variety methods for solving differential
equations
• So far, we used the method of undetermined
coefficients
• Now we look at 3 alternatives:
a geometric approach
a frequency response approach
a transform approach
• These also give us some insight and additional
useful tools.
© D. J. Inman
32/49 University of Michigan
2.3.1 Geometric Approach
The geometric solution treats each force in the equation
of motion as vector.
F0 cos ωt
mx + cx + kx =
For the assumed solution of
=
The derivates are: x p (t ) X cos(ωt − θ )
x p (t ) X cos(ωt − θ )
=
x p (t ) =−ω X sin(ωt − θ ) =ω X cos(ωt − θ + 90)
−ω 2 X cos(ωt − θ )
x p (t ) =

Substituting the above and the equation of motion and
solving for X in terms of F0 :

© D. J. Inman
33/49
 2.3.1 Geometric Approach
and
• Position, velocity and acceleration phase shifted
each by π/2
• Therefore write each as a vector
• Compute X in terms of F0 via vector addition
Im D C

C
E
θ
ωt A B
θ
B
A Re

© D. J. Inman
34/49 University of Michigan
 Using vector addition on the diagram:

At resonance:
π It can be seen from the figure that at
(θ = ) the applied force and the damping force are
2
acting in the same direction, and the stiffness force
and inertia for are equal and in opposite direction.

© D. J. Inman
35/49 University of Michigan
2.3.2 Complex response method

(2.47)

(2.48)

Real part of this complex solution corresponds to

the physical solution

© D. J. Inman
36/49 University of Michigan
 Choose complex exponential as a solution

(2.49)

(2.50)

(2.51)

(2.52)

H ( jω ) is called the complex frequency response function

Note: These are all complex functions
© D. J. Inman
37/49 University of Michigan
 Using complex arithmetic:
Manipulating the complex numbers:

(2.53)

(2.54)

(2.55)

Has real part = to previous solution

© D. J. Inman
38/49 University of Michigan
Comments:
• Label x-axis Re(ejωt) and y-axis
Im(ejωt) results in the graphical
approach
• It is the real part of this complex
solution that is physical
• The approach is useful in more
complicated problems

© D. J. Inman
39/49 University of Michigan
Example 2.3.1: Use the frequency response
approach to compute the particular solution of an
undamped system
The equation of motion is written as

© D. J. Inman
40/49 University of Michigan
 2.3.3 Transfer Function Method
The Laplace Transform
• Changes ODE into algebraic equation
• Solve algebraic equation then compute the
inverse transform
• Rule and table based in many cases
• Is used extensively in control analysis to
examine the response
• Related to the frequency response
function
• (Read Appendix B)
© D. J. Inman
41/49 University of Michigan
The Laplace Transform approach:

• See appendix B and section 3.4 for details

• Transforms the time variable into an algebraic,
complex variable
• Transforms differential equations into an
algebraic equation
• Related to the frequency response method

© D. J. Inman
42/49 University of Michigan
The Laplace Transform
• Laplace transform of derivatives have
been calculated as the following:
L[=
x (t )] sX ( s ) − x(0)
x(t )]= s 2 X ( s ) − sx(0) − x (0)
L[ 
• Where x(0) and x (0) are the initial values
of function x(t).
• Table B.1 (Appendix B) lists the Laplace
transform of common functions.
© D. J. Inman
43/49 University of Michigan
 Take the transform of the equation of motion:
(Assume zero initial conditions)

Now solve algebraic equation in s for X(s)

To get the time response this must be “inverse

transformed”

© D. J. Inman
44/49 University of Michigan
Transfer Function Method
With zero initial conditions:
2
(ms + cs + k ) X ( s=) F ( s) ⇒
The transfer
X ( s) 1 function
= H= (s)
F ( s) ms + cs + k (2.59)
2

1
H ( jω ) = (2.60)
k − mω + cω j
2

(same as equation 2.52)

(s is a complex number. If s is restricted to lie along the imaginary, s = jω)
Example 2.3.2 Compute forced response of the
suspension system shown using the Laplace transform
Summing moments about the shaft:

Taking the Laplace transform:

Figure 2.12

Taking the inverse Laplace transform:

© D. J. Inman
46/49 University of Michigan
Notes on Phase for Homogeneous
and Particular Solutions

• Equation (2.37) gives the full solution for a

harmonically driven underdamped SDOF
oscillator to be

How do we interpret these phase angles?

Why is one added and the other subtracted?
© D. J. Inman
47/49 University of Michigan
 Non-Zero initial conditions

© D. J. Inman
48/49 University of Michigan
Zero initial displacement

© D. J. Inman
49/49 University of Michigan
Zero initial velocity

© D. J. Inman
50/49 University of Michigan
 Phase on Particular Solution

• Simple “atan” gives -π/2 < Φ < π/2

• Four-quadrant “atan2” gives 0 < Φ < π
© D. J. Inman
51/49 University of Michigan